Model Policy Manual

Tier B: one strong causal study or consistent quasi-experiments

Finding. Causal studies find that participation pays: Houston's randomized arts-access initiative (42 schools, more than 10,000 students) reduced disciplinary infractions, raised writing achievement and increased empathy and, for elementary students, school engagement and college aspirations; Title IX's forced expansion of girls' sports raised women's college attendance and employment (Stevenson); German sports-club participation improved grades and health. Longitudinal work (Fredricks and Eccles) links breadth of participation to higher GPA, college enrollment and lower risk behavior. Fees are the main access barrier: average required fees in 2018 to 2019 were $161 for sports, $86 for arts and $46 for clubs, lower-income students are twice as likely not to participate, and only 7% of parents have ever requested a waiver. (arts RCT: discipline referrals down, writing up, empathy up; Title IX sports expansion: +1 pp college attendance per 10 pp participation; lower-income nonparticipation about 2x; waiver uptake 7%) ^{1,2,3,4,5,6,7}

Where the evidence is mixed. Most of the participation literature is correlational and selection runs strongly (motivated, better-resourced students join), so effect sizes from longitudinal studies overstate causal returns. The Houston RCT found no effect on attendance, math, reading or science scores, and only elementary students showed engagement gains. Zdroik and Veliz found Michigan schools with pay-to-play fees had similar participation rates to schools without fees, but the schools able to charge fees were larger, whiter and lower-poverty, which is the selection problem itself rather than evidence that fees are harmless. Fee elimination has a real budget cost and the research does not show which activities yield the most per dollar. Sports carry injury and concussion costs that the achievement studies do not count.

Standard. The district shall guarantee that no student is excluded from any school-sponsored activity for inability to pay, shall grant automatic fee waivers to students eligible for free or reduced-price meals without a separate application, shall provide weekly arts instruction in every elementary grade, and shall report participation by subgroup to the board annually.

Indicator. Participation rate in at least one school-sponsored activity by free-and-reduced-lunch status, race and grade (student information system activity rosters); number and dollar value of fee waivers granted versus eligible students; share of elementary students receiving weekly arts instruction; list of activities cut or added each year with enrollment.

Red flags in real manuals. participation fees are required and non-refundable; fee waivers are available upon application with documentation of hardship; students with outstanding fees or fines may not participate; activities may be cancelled when enrollment or funding is insufficient, with no notice standard; arts instruction is offered as resources permit; no participation reporting by subgroup; transportation to activities is the responsibility of the family

Law. Title IX, 20 U.S.C. 1681 and 34 CFR 106.41, requires equal athletic opportunity by sex. California prohibits fees for any educational activity including extracurriculars (Hartzell v. Connell, 35 Cal.3d 899 (1984); Education Code 49010 to 49013, with a required complaint process), while most states permit fees subject to waiver rules; several, including Massachusetts and Ohio, require waivers or caps. The McKinney-Vento Act (42 U.S.C. 11431 et seq.) requires that homeless students not be excluded from activities by fees or documentation. State athletic association eligibility rules govern academic eligibility and may conflict with no-pay-no-play provisions.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. When districts automatically place every student who meets an objective readiness threshold into the next advanced course, with parent opt-out instead of opt-in, advanced course enrollment rises and gaps by race and income narrow without lowering grades or graduation; Washington's Academic Acceleration policies raised the chance of taking any advanced course by about 5 percent district-wide and more among proficient students who had been passed over. Effects on later college enrollment and persistence from the high school version of the policy are null so far. (WA Academic Acceleration: advanced course enrollment up, gaps narrowed, GPA and graduation unchanged; college enrollment and persistence unchanged. Wake County objective math placement: income and race no longer predicted placement; on-track-for-algebra share up. Earning AP credit raises degree completion (regression discontinuity).) ^{8,9,10,11,12,13,14,15}

Where the evidence is mixed. The best-studied high school automatic enrollment policy (Washington, grades 11 and 12) did not change college enrollment, first-year coursetaking, or persistence to a second year (Austin et al. 2026). Wake County's objective placement improved access but accelerated students earned lower course grades in middle school and the acceleration did not raise test scores or college readiness measures (Dougherty et al. 2017). Expanding AP and dual enrollment seats without supports is associated with wider, not narrower, racial participation gaps across districts (Xu, Solanki and Fink 2021). Access is necessary but not sufficient; the courses must be taught well and students supported to pass them.

Standard. The policy shall name an objective readiness threshold, require automatic enrollment of every student who meets it in the next most rigorous course offered, allow only written parent opt-out, and require annual public reporting of participation and success by student group.

Indicator. Share of students meeting the readiness threshold who are enrolled in the corresponding advanced course the next term, by race, ethnicity, income, English learner and disability status; AP, IB and dual-credit participation and pass rates by group (state report card, College Board and IB data); opt-out counts by school.

Red flags in real manuals. teacher recommendation required for honors, AP, or IB placement; application, essay, or interview required for advanced courses; students must opt in to advanced courses; prerequisite grade thresholds above B or above the state standard; AP exam fees or dual-credit tuition charged to students without a waiver; no reporting of advanced course participation by student group

Law. Washington RCW 28A.320.195 required every district board to adopt an academic acceleration policy by 2021-22 that automatically enrolls students who meet the state standard on the grade 8 or high school ELA or math assessment in the next most rigorous course, with parent opt-out. Illinois Accelerated Placement Act (105 ILCS 5/14A-32), amended by Public Act 103-0743 in 2024, requires automatic enrollment of high school students who exceed state standards in the next most rigorous course and an option to enroll for students who meet them, with district policies amended by 2027-28. North Carolina G.S. 115C-81.36 requires placement of students who score at the highest level on the math end-of-grade or end-of-course test in the advanced course for their next math course. Texas SB 2124 (2023) requires automatic enrollment in advanced sixth grade math of students in the top 40 percent on the fifth grade STAAR math test, with parent opt-out.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Structured CTE pathways, meaning multi-year sequences with employer partners, work-based learning, and a credential, raise graduation and earnings in causal studies: the MDRC career academies RCT found earnings 11 percent higher eight years after high school (17 percent for young men), Massachusetts and Connecticut regression-discontinuity studies find 7 to 10 point gains in on-time graduation and, in Connecticut, 32 percent higher quarterly earnings for men, and New York City's lottery-based P-TECH evaluation found associate degree completion up 5 percentage points with dual enrollment more than doubled. Effects come from depth in one field, not scattered introductory courses, and are concentrated among young men. (Career academies: earnings +11 percent per year over 8 years (+17 percent men), attainment unchanged. Massachusetts regional vocational schools: on-time graduation +7 to +10 points. Connecticut technical high schools: graduation +10 points and quarterly earnings +32 percent for men, no effect for women. P-TECH: associate degree +5 points, internships +38 points. Upper-level CTE courses: about +2 percent wages per additional year; introductory courses 0.) ^{16,17,18,19,20,21,22,23}

Where the evidence is mixed. Gains are uneven. Career academies did not raise educational attainment, and the Connecticut technical schools had no measurable effect on women. P-TECH's degree effects were concentrated among young men, and its interim credits were mostly in CTE and non-academic subjects. Earnings returns vary widely by field, with health care and IT far above others and some fields near zero (Ecton and Dougherty 2023), and returns accrue only to upper-level coursework in a single field (Kreisman and Stange 2020). The strongest effects come from whole-school or regional CTE models with oversubscribed admissions; a single elective or a program without employer partners should not expect these results. Kemple warned that partial implementation of the academy model may not reproduce the impacts.

Standard. Every high school shall offer at least one multi-year CTE pathway aligned to a regional high-wage field with an employer partner, work-based learning, dual credit, and an industry credential; admission to oversubscribed pathways shall be by lottery or objective criteria, and completion and post-graduation outcomes shall be reported by student group.

Indicator. Share of graduates who completed a CTE program of study (concentrators: two or more courses in one pathway) and who earned an industry-recognized credential or 12 or more aligned college credits, by race, ethnicity, income, English learner, disability and sex (Perkins V accountability data); share of concentrators with a paid work-based learning placement; concentrator graduation rate and post-graduation employment or enrollment from state wage and Clearinghouse matches.

Red flags in real manuals. CTE offered only as single-semester electives with no sequence; no employer partner or work-based learning requirement; CTE used as a lower track or as an alternative to college-preparatory courses; placement in CTE by counselor or staff recommendation without objective criteria; credential exam fees charged to students; no reporting of concentrator outcomes by student group or sex

Law. The Strengthening Career and Technical Education for the 21st Century Act (Perkins V, 2018) requires each recipient to conduct a comprehensive local needs assessment every two years, align programs of study to labor market demand, and report concentrator outcomes (graduation, academic proficiency, postsecondary placement, credentials, and participation in nontraditional fields) by subgroup. State graduation requirements increasingly allow CTE sequences or credentials to count toward a diploma endorsement or as a college-and-career-readiness indicator under ESSA.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Missing 10 percent or more of school days lowers achievement and engagement from kindergarten on, and the share of students chronically absent nearly doubled after the pandemic. Controlled studies show that early, supportive, family-facing responses (mailed and texted attendance information, caseworkers who remove material barriers, structured adult mentoring) cut chronic absence by roughly 8 to 23 percent at modest cost, while no rigorous US study shows that fines, prosecution or court referral improve attendance. (Family mailings and texts: chronic absence -10% to -15% (2 to 7 points); school-based material-needs caseworkers: -8%; Check & Connect mentoring: -4.2 days (-23%) in grades 5-7, nothing in grades 1-4; pooled attendance interventions small (g = 0.09 to 0.25)) ^{24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39}

Where the evidence is mixed. Average effects are small and uneven. Eklund et al. (2022) pooled g = 0.25 across 22 controlled studies and the family-school partnership category's confidence interval included zero; Li et al. (2026) pooled g = 0.09 across 13 post-2020 studies, with weaker effects in majority students-of-color samples. Check & Connect had no detectable effect in grades 1 to 4 and none on test scores or GPA; texting cut chronic absence but not achievement (Heppen et al. 2020). On punishment: Maynard et al. (2012) found court-based truancy programs did not differ from school- or community-based ones on short-term attendance, and an Australian RCT of a police-school partnership raised attendance, so 'punitive responses never work' overstates the record. What the evidence does support is narrower: no rigorous US study shows fines or prosecution improve attendance; Texas truancy filings fell about 90 percent after 2015 decriminalization with no drop in statewide attendance (descriptive only); and in a 131,312-student RCT the standard legalistic, punitively worded truancy notice was outperformed by a plain, supportive one.

Standard. Policy requires monthly identification of every student at or trending toward 10 percent absence, a documented supportive family contact and barrier assessment before any penalty, a written plan at Tier 2 and case management at Tier 3, and quarterly public reporting by building and subgroup; court or truancy referral is permitted only after the supportive steps are documented.

Indicator. Percent of enrolled students missing 10 percent or more of enrolled days (the state chronic absence measure in the EDFacts or state report card file), reported by school, grade and subgroup; the share of chronically absent students who received a documented Tier 2 contact within 10 school days of crossing the threshold; and the number of truancy referrals filed, each tracked monthly.

Red flags in real manuals. truancy referral, citation or fine triggered automatically at a fixed number of unexcused absences with no supportive contact first; loss of credit, grade reduction or denial of course credit for absences; out-of-school suspension or exclusion from activities as a consequence for absence; attendance tracked only as average daily attendance or only unexcused absences, with no chronic absence (10 percent, all absences) measure; no named role responsible for outreach and no reporting frequency; attendance letters written in legal or threatening language with no offer of help

Law. ESSA requires states to report chronic absenteeism on report cards and most states use it as a school-quality indicator. Every state has a compulsory attendance statute, and many still define truancy as a status offense with court jurisdiction over students or parents; Texas HB 2398 (2015) decriminalized failure to attend school and requires districts to apply truancy prevention measures before court referral. Several states require supportive interventions before court referral (for example Connecticut General Statutes 10-198a); California Education Code 60901 defines a chronic absentee as a pupil absent 10 percent or more of enrolled days.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Every rigorous evaluation finds public pre-K raises school readiness at kindergarten entry (Boston 0.4 to 0.6 SD in language and math, Tulsa large gains, Tennessee gains at end of pre-K), but what happens next depends on the program and the schools that follow: Boston's lottery cohorts showed no later test-score gains yet higher high school graduation, SAT taking and college enrollment and less juvenile incarceration, Tulsa's cohorts kept middle school math and course-taking advantages, and Tennessee's statewide program, a randomized trial, left participants worse off than controls on achievement, discipline and special education by sixth grade. A meta-analysis of 22 causal studies finds average long-run reductions in special education placement (8 points) and grade retention (8 points) and higher graduation (11 points). (Kindergarten entry: +0.4 to +0.6 SD (Boston); long run: special education -8.1 points, grade retention -8.3 points, high school graduation +11.4 points (meta-analysis); Boston lottery: college enrollment up, juvenile incarceration down, test scores null; Tennessee RCT: sixth-grade achievement, discipline and special education all worse than controls) ^{40,41,42,43,44,45,46,47,48,49,50}

Where the evidence is mixed. The only large randomized trial of a scaled state program (Tennessee Voluntary Pre-K, n 2,990) found that children offered pre-K had lower state test scores in grades 3 through 6, more disciplinary infractions, worse attendance and more special education designations than wait-listed controls. Test-score fadeout by grade 3 is the norm even in positive studies (Bailey et al. 2017), and the Boston long-run gains came from a program with coached teachers, a shared curriculum and small classes that most districts do not replicate. The Tulsa evidence relies on propensity-score weighting, not randomization. Pre-K is among the most expensive grades to add, and poor-quality programs can be worse than the child-care arrangements they replace.

Standard. District pre-K uses an adopted evidence-based curriculum, certified lead teachers with ongoing coaching, class sizes of 20 or fewer with two adults, and a full instructional day, and the district tracks each pre-K cohort through third grade and reports readiness and later outcomes by subgroup.

Indicator. Share of age-eligible four-year-olds enrolled in district pre-K, by income and attendance area (enrollment file); kindergarten entry assessment results for pre-K attendees versus non-attendees by subgroup; pre-K class size, teacher certification and curriculum adoption; third-grade reading proficiency, grade retention and special education referral rates for the pre-K cohort tracked longitudinally.

Red flags in real manuals. pre-K classes above 20 children or with a single adult; lead teachers without early childhood certification; no adopted curriculum or coaching provision; half-day pre-K only; enrollment by first-come first-served with no priority for low-income or English learner children; no longitudinal reporting on pre-K cohorts

Law. Federal Head Start (42 USC 9831 et seq.) serves children below the poverty line with its own performance standards (class size 20 or fewer, two staff). State pre-K programs set eligibility, class size and teacher credential rules; 44 states and the District of Columbia fund some pre-K, and universal programs exist in Oklahoma, Florida, Georgia, Vermont, West Virginia, Wisconsin and the District of Columbia. IDEA Part B section 619 requires preschool special education for eligible three- to five-year-olds in every district.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Two independent lottery-based randomized studies of early college high schools (North Carolina, followed 17 years, and the AIR national study, followed six years past graduation) find higher high school graduation, college enrollment, and degree completion, with very large effects on associate degrees and bachelor's gains for low-income and first-generation students. A Tennessee RCT of a dual-credit math course shifted students from remedial to precalculus and AP math and from two-year to four-year colleges, though it did not raise overall college enrollment. The What Works Clearinghouse rates dual enrollment as having positive effects on degree attainment, credit accumulation, and completing school. (Early college: college enrollment +7 percentage points (84 vs 77 percent at year 10 in NC), bachelor's +5 points (30 vs 25 percent), associate degree effects very large; AIR study: enrollment and degree completion up within six years. Dual-credit math RCT: no change in overall enrollment, shift toward four-year colleges for mid-achievers. Observational dual enrollment: degree attainment gains largest for low-income students.) ^{51,52,53,54,55,56,57,58,59}

Where the evidence is mixed. Early college effects come from a whole-school model with college-going culture, supports, and a community college partner on site; a la carte dual enrollment courses are not the same intervention. The Tennessee RCT of a single dual-credit course found no effect on overall college enrollment and only limited evidence of better college math performance. Dual enrollment and AP participation gaps by race exist in most districts, and gaps widen where seats expand without equity supports (Xu, Solanki and Fink 2021). Credits earned may not transfer to the student's eventual college, and some AIR early college impacts were stronger for students with higher prior achievement. Early college GPA at completion and double-majoring did not differ from controls.

Standard. Every high school shall offer dual enrollment with a public college partner at no cost to students, with automatic eligibility for students meeting an objective threshold, transportation and textbooks provided, and participation and credit outcomes reported annually by student group; districts of sufficient size shall operate or partner with an early college pathway.

Indicator. Share of graduates who earned at least one transferable college credit in high school and share who earned 12 or more, by race, ethnicity, income, English learner and disability status (state report card and National Student Clearinghouse); dual enrollment course pass rates; share of credits that transferred; postsecondary enrollment and completion of dual enrollment participants versus non-participants.

Red flags in real manuals. students pay tuition, fees, or textbook costs for dual enrollment; dual enrollment limited to students with a minimum GPA above the college's own placement standard; counselor or principal approval required with no objective criteria; dual enrollment offered only off campus with no transportation; no articulation agreement or no record of credit transfer; no reporting of participation by student group

Law. Nearly every state has a dual enrollment statute; eligibility rules, funding, and who pays tuition vary by state (for example Washington's Running Start under RCW 28A.600.300 and Texas Education Code 28.009). Title IV of ESSA permits Title I and Title IV-A funds for dual enrollment and early college, and Perkins V allows CTE funds for dual credit in CTE programs of study. Several states require that dual-credit general education courses transfer to public universities.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Systematic, explicit phonics instruction in K-1 raises word reading, spelling and comprehension by a moderate amount (about 0.4 SD) and is the only reading-disability treatment whose efficacy is confirmed in randomized trials; curriculum-based screening in K-2 predicts later reading well enough to find at-risk readers early. State 'science of reading' laws raise high-stakes elementary reading scores by a small amount so far, mainly where the policy is comprehensive, and the gains on low-stakes NAEP are not yet statistically clear. (Phonics d = 0.41 overall, 0.55 when begun in K-1; intensive K-3 interventions about 0.3 to 0.4 SD; state literacy laws +0.025 to +0.10 SD on state tests, NAEP effect not significant) ^{60,61,62,63,64,65,66,67,68,69}

Where the evidence is mixed. Suggate (2016) found phonics-only gains fade more at follow-up than phonemic awareness and comprehension gains, so phonics is necessary but not sufficient. Bowers (2020) reviewed 12 meta-analyses and argues the advantage of systematic phonics over alternative methods is smaller and less consistent than usually claimed; Torgerson et al. (2018) rate the effect moderate and the evidence base thinner than policy rhetoric implies. Westall and Cummings (2023) find that state early literacy laws move high-stakes scores by only 0.025 to 0.05 SD on average, that effects fade by middle school unless the policy is comprehensive, and that NAEP gains are not statistically significant; Mississippi's rise from 49th to the national average on fourth-grade NAEP reading (2013 to 2019) coincided with a package of screening, LETRS teacher training, coaching and third-grade retention, so no single component can claim the credit. Screening accuracy varies by tool and by how it is administered.

Standard. The policy shall require that core K-3 reading instruction be explicit and systematic in phonemic awareness, phonics, fluency, vocabulary and comprehension, shall prohibit three-cueing as the primary word-reading strategy, and shall require a valid universal screener three times a year in K-3 with intervention for every student flagged at risk.

Indicator. Share of K-3 students screened within the first 30 school days of each year and share flagged at risk (district screener data, e.g. DIBELS 8 or Acadience); share of flagged students receiving a documented intervention within 15 school days; third-grade state ELA proficiency rate and the subgroup gaps; percent of K-3 teachers with completed structured-literacy training.

Red flags in real manuals. balanced literacy or guided reading named as the core K-3 program; three-cueing, MSV, or 'use the picture' strategies in the reading policy or adopted materials; screening only for students referred by a teacher rather than universal; no timeline between screening and the start of intervention; parent notification of reading risk only at report card time; phonics limited to kindergarten or treated as an intervention rather than core instruction

Law. By 2021, 41 states and DC had early literacy laws (Westall and Cummings). Mississippi's Literacy-Based Promotion Act (2013) requires K-3 screening, intervention, LETRS-trained teachers and third-grade retention. Ohio HB 33 (2023) requires science-of-reading core curricula and bans three-cueing. Texas HB 3 (2019) requires K-3 teachers to complete reading academies. Most states now mandate K-3 dyslexia or reading-risk screening; check the state's screener approval list before scoring a district's choice of tool.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Lottery studies show dual-language immersion raises English reading achievement (Portland: +13 percentile points by grade 5 and +22 by grade 8, with no loss in math or science; Charlotte: +0.04 to 0.05 SD per year in math and reading for English learners and non-learners alike) and raises the share of English learners reclassified by grade 6. Regression-discontinuity studies of reclassification show that the rule matters more than the label: reclassifying students on a single lenient test score harmed later outcomes, while multi-measure criteria that keep supports in place removed the harm, and timely reclassification before the grade 5 and grade 8 transitions improved on-time graduation. (Dual-language immersion: English reading +13 to +22 percentile points by grade 5 to 8 (Portland lottery); +0.04 to 0.05 SD per year (Charlotte lottery). Reclassification effects: +0.25 SD English test scores for marginal grade 2 to 4 reclassifiers in Los Angeles; harmful under single-test lenient criteria, null under multi-measure criteria.) ^{70,71,72,73,74,75,76,77,78,79,80}

Where the evidence is mixed. Dual-language immersion effects are concentrated in English reading; the Portland lottery found no effect on math or science and the Charlotte study found no effect on EL status. Reclassification effects vary by district: Cimpian, Thompson and Makowski (2017) found that where reclassification cuts off services abruptly, marginal reclassified students lost ground, and Pope (2016) found no benefit for students reclassified in later grades. Johnson (2019) found no effect of initial EL classification on graduation or college entry, so the classification itself is not the lever; program quality and exit rules are. Dual immersion students in San Francisco scored lower in English language arts in early grades before catching up and passing peers by middle school (Valentino and Reardon 2015), which can look like failure on short-horizon accountability measures. Program seats are limited by the supply of bilingual teachers.

Standard. The district shall offer a dual-language or bilingual program where enrollment supports it, use multiple measures (state proficiency test plus academic performance and teacher input) for reclassification, continue academic monitoring and supports for at least four years after exit, and place English learners in grade-level core content with language support rather than in separate lower-track courses.

Indicator. Annual EL reclassification rate by years in program and by program model; share of long-term English learners (6 or more years); state English language proficiency assessment growth; ELA and math achievement of current and former ELs tracked for four years after reclassification; EL enrollment in grade-level core courses and advanced courses (student information system course rosters); dual-language seats offered versus applicants.

Red flags in real manuals. reclassification based solely on the state English proficiency test; no monitoring period for reclassified students; English learners scheduled into ELD in place of a core subject; program described only as 'English immersion' with no bilingual option; no named administrator responsible for English learner services; parent waiver required to receive bilingual instruction; no annual report of EL outcomes to the board

Law. Lau v. Nichols (1974) and Castaneda v. Pickard (5th Cir. 1981) require programs based on sound theory, adequately resourced and evaluated; the Equal Educational Opportunities Act of 1974 codifies the duty. ESSA Title III requires standardized statewide entrance and exit procedures and four years of monitoring of former ELs for accountability. The 2015 joint Department of Justice and Department of Education Dear Colleague letter sets expectations for identification, program access, reclassification and monitoring. California Proposition 58 (2016) repealed English-only mandates; Arizona Proposition 203 and its four-hour ELD block were loosened by SB 1014 (2019); Texas requires bilingual education where 20 or more ELs share a language at a grade level (Texas Education Code 29.053).

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Full-day kindergarten raises achievement at the end of the kindergarten year by about 0.25 to 0.35 SD compared with half-day programs (meta-analysis of 40 studies; Community Guide systematic review; an Indiana lottery experiment found 0.31 SD in literacy, larger for Hispanic children and low-entering-skill children), but in most US studies the test-score advantage fades by third grade. (End of kindergarten: +0.25 SD (meta-analysis) to +0.35 SD (Community Guide; verbal +0.46, math +0.24); lottery experiment: literacy +0.31 SD, Hispanic children +0.70 SD; advantage largely gone by grade 3 in ECLS-K studies) ^{81,82,83,84,85,86,87}

Where the evidence is mixed. Fadeout is the consistent finding: Cannon, Jacknowitz and Painter (2006) and Cooper et al. (2010) both find the achievement difference disappears by third grade, and the 2006 study found no extra benefit for children below the poverty line. Cooper's meta-analysis also found suggestive evidence of less positive attitudes toward school and more behavior problems in full-day programs. Ontario's play-based full-day program showed lasting gains through grade 2 and grade 3 reading, so persistence may depend on program design and the quality of the grades that follow. Full-day kindergarten roughly doubles the staffing and space cost of the grade.

Standard. The district offers tuition-free full-day kindergarten, with the same daily instructional minutes as grade 1, to every eligible child in every attendance area, and reports enrollment and end-of-year readiness results by subgroup.

Indicator. Share of kindergarten students enrolled in a free full-day program (enrollment file); daily kindergarten instructional minutes compared with grade 1; kindergarten entry assessment and end-of-year literacy and numeracy screening results by subgroup; kindergarten attendance and chronic absence rate.

Red flags in real manuals. full-day kindergarten offered only as a tuition or fee program; full-day seats limited to certain buildings or by lottery with no plan to expand; kindergarten day shorter than grade 1 by more than 30 minutes; no kindergarten readiness or end-of-year reporting; kindergarten listed as optional with no full-day provision

Law. State law sets whether districts must offer kindergarten and whether full-day is funded. Education Commission of the States counted 15 states plus the District of Columbia requiring districts to offer full-day kindergarten as of 2018; Washington and Rhode Island have since added requirements. Most states fund full-day kindergarten at the full per-pupil rate but do not require it, and six states do not require districts to offer any kindergarten. Check the state code for the required daily minutes and attendance age.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Replacing teacher or parent referral with universal screening of every student, with no change in the eligibility cutoff, raised gifted identification in Broward County from 3.3 to 5.5 percent of second graders, with the largest gains for low-income, Black, Hispanic and English learner students who had been systematically under-referred. Black students with the same test scores are referred less often than White students, especially by non-Black teachers, which is why referral-based systems miss them. (Broward universal screening: identification rate +67 percent overall; Hispanic identification roughly +130 percent, Black +80 percent, with eligibility cutoffs unchanged. Local (school) norms would further raise Black and Latinx representation. Gifted services themselves: achievement effects for marginal students near zero in RD studies, though high-achiever classrooms raised Black and Hispanic scores by 0.5 SD in one district.) ^{88,89,90,91,92,93,94}

Where the evidence is mixed. Universal screening fixes who is found, not what they get: regression-discontinuity and lottery studies find little or no short-run achievement gain for students at the margin of gifted eligibility (Bui, Craig and Imberman 2014), and nationally representative data show Black and low-income students do not see the small gains their peers get from gifted services (Redding and Grissom 2021). Broward suspended universal screening during budget cuts and identification fell back, showing the policy needs a funded test every year. Screening with a single nonverbal test can still under-identify English learners unless local norms or multiple measures are used.

Standard. Every student shall be screened for advanced learning services with a validated instrument at least once in the primary grades and once in middle school, without a prior referral, using multiple measures and local norms, and identification rates by student group shall be published annually.

Indicator. Share of each grade cohort screened (target 100 percent at the screening grade); identification rate by race, ethnicity, income, and English learner status compared with each group's share of enrollment (representation index); share of identified students receiving services; state gifted data collection and the federal Civil Rights Data Collection gifted enrollment counts.

Red flags in real manuals. identification requires a teacher or parent referral or nomination; testing only for students recommended by staff; a single national-norm cutoff (for example 130 IQ or 97th percentile) as the only criterion; screening or testing fees charged to families; no reporting of identification by student group; services limited to pull-out enrichment with no advanced curriculum

Law. No federal mandate to identify or serve gifted students; the Jacob K. Javits Gifted and Talented Students Education Act funds research and demonstration only. State mandates vary widely: some states require identification and services and set identification rules (for example Texas Education Code chapter 29 subchapter D and Florida Rule 6A-6.03019 Plan B for underrepresented groups), others require neither. Gifted enrollment by race, sex, disability and English learner status is reported to the federal Civil Rights Data Collection.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Regression discontinuity studies of test-based third-grade retention paired with intensive reading intervention (Florida, Indiana, Mississippi) show large short-run reading gains, with Indiana and Mississippi gains persisting into middle school. Long-run evidence is split: Florida retainees reached high school with higher GPAs and less remediation and no graduation penalty, while a 2026 Texas study finds retained third graders were 9 points less likely to graduate and earned 19 percent less at 26, and retention in later grades (Chicago grade 8, Louisiana grade 8) raises dropout and juvenile crime. (Short-run reading gains of roughly 0.2 to 0.6 SD that fade or persist by state; Texas: graduation -9 pp, earnings at 26 -19%; older grades: dropout up; disciplinary incidents up short-run (Florida)) ^{95,96,97,98,99,100,101,102,103,104,105}

Where the evidence is mixed. Zhong (2026, Texas) is the longest follow-up of a modern test-based policy and finds net harm: higher absenteeism, violent behavior and juvenile crime, lower graduation and earnings despite temporary score gains. Ozek (2015) finds Florida retention raised disciplinary incidents and suspensions in the short run, concentrated among poor and male students. Jacob and Lefgren (2009) find grade 8 retention in Chicago raises dropout. Eren, Depew and Barnes (2017) find grade 8 retention in Louisiana raises juvenile crime. The pre-2000 literature summarized by Jimerson (2001) associated retention with worse achievement and adjustment, though Allen et al. (2009) show the negative estimates shrink toward zero in better-designed studies. The positive results all come from retention bundled with summer reading camps, high-quality teachers and intervention plans; nothing in the evidence supports retention alone. Retention costs taxpayers about one additional year of schooling per retained student, less than previously estimated (Winters 2022) but still substantial.

Standard. The policy shall prohibit retention as a stand-alone response, shall require that any retention in grades K-3 be preceded by documented intervention and followed by a written individual reading plan with a different and more intensive program, shall provide good-cause exemptions, and shall prohibit retention in grades 6 through 8 as a consequence of test scores alone.

Indicator. Number and share of students retained by grade, school, race, disability and EL status each year (student information system); share of retained students who had a documented intervention plan before the retention decision; reading proficiency of retained students one and three years later compared with barely promoted peers; attendance and discipline of retained students.

Red flags in real manuals. retention at the discretion of the principal or teacher with no documented intervention first; promotion requirements based on a single cut score with no good-cause exemptions; retention permitted in middle school grades as a consequence of test scores or behavior; no individual plan or change of program for the repeated year; parent notification only after the retention decision; no limit on the number of times a student may be retained

Law. Florida (2002), Mississippi (2013), Indiana (2024 SEA 1), Alabama (effective 2023-24), Tennessee (effective 2023) and roughly a dozen other states require or strongly presume third-grade retention for students below a reading cut score, nearly all with good-cause exemptions. Michigan repealed its retention mandate effective 2024 and Ohio made retention optional in 2023 (HB 33). District policy must at least match state exemption and notification rules; where state law is silent, the district standard above governs.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. Grades measure a mix of academic mastery and behavior and vary widely from teacher to teacher. The causal evidence favors two things: higher grading standards raise learning (Algebra I students of tougher graders learn more and keep the gain in later courses), and a standards-based system with formative feedback and reassessment for full credit raised ninth-grade math test scores by 0.33 SD in a cluster RCT. The specific rules districts fight over, no zeros, 50-percent floors and late-work penalties, have almost no outcome research; the one study of a minimum-grade policy is from a single school. (Higher teacher grading standards: positive effect on test scores and later math courses; PARLO standards-based grading RCT: +0.33 SD; minimum-grade and no-zero policies: no causal outcome evidence) ^{106,107,108,109,110,111,112,113,114,115,116,117}

Where the evidence is mixed. Carey and Carifio (2012) found no grade inflation or social promotion under a 50-point floor at one urban high school over seven years, but it is a single site with no comparison group. Reeves (2004) argues against zeros on arithmetic grounds (a zero on a 100-point scale is a six-point penalty on a 4-point scale), which is a measurement argument, not outcome evidence. Welsh, D'Agostino and Kaniskan (2013) find standards-based report-card grades only moderately converge with state test scores and vary more within schools than between them. Hjorth-Trolle, Rosenqvist and Hed (2021) show that changing grading practice (Sweden's move from relative to absolute grades) widened the gap between students of low- and high-educated parents, so grading reforms can move equity in either direction. Gershenson, Holt and Tyner (2024), Betts and Grogger (2003) and Figlio and Lucas (2004) all point the other way from lenient grading: students learn more under higher standards. Equity-grading programs such as Feldman's have practitioner and pilot data but no peer-reviewed causal evaluation.

Standard. The policy shall require every course to publish, before the first grading period, how grades are computed, shall separate academic achievement from behavior and compliance in the reported grade, shall allow at least one reassessment of a summative standard for credit, and shall not mandate or forbid numeric floors without the district's own grade-to-test data.

Indicator. Correlation between course grades and state or end-of-course test scores by school and course (gradebook export joined to assessment file); spread of failure rates across teachers of the same course; share of students with a D or F who passed the end-of-course exam and vice versa; count of grading appeals.

Red flags in real manuals. grading left entirely to individual teacher discretion with no published criteria; attendance, participation or behavior deductions built into the academic grade; no reassessment permitted or reassessment capped at a failing score; a zero required for any missing work with no window to submit; a mandated numeric floor adopted without any district grade-to-test review; different weighting across sections of the same course

Law. Few states regulate K-12 grading beyond GPA calculation for scholarships and transcripts. South Carolina enacted a 2026 law prohibiting districts from requiring a minimum grade above a student's actual performance (reported by Education Week, August 2026; verify the statute before citing). A 2024 RAND survey found about 27 percent of teachers reported a 'no zeros' policy at their school or district.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Across 96 randomized trials, tutoring raises achievement by about 0.29 SD, with the largest effects when teachers or paraprofessionals tutor, sessions run at least three days a week, and tutoring happens during the school day. Post-pandemic programs run at district scale have produced much smaller gains (0.04 to 0.09 SD in reading and no math effect in Nashville), and a 263-trial reanalysis puts the realistic effect of large, test-targeted programs at 0.16 to 0.22 SD. (+0.29 SD pooled in RCTs; +0.18 to +0.40 SD math in Chicago Saga RCTs; +0.16 to +0.22 SD expected at scale; Nashville district program +0.04 to +0.09 SD reading, 0 in math) ^{118,119,120,121,122,123,124}

Where the evidence is mixed. Effects shrink sharply with program scale: Kraft, Schueler and Falken (2026) find pooled effects fall from 0.40 SD in the full sample to 0.16 to 0.22 SD in studies that resemble large, standardized-test-targeted programs. Metro Nashville's district-run program, delivered partly before and after school by a mix of teachers and volunteers, produced only small reading gains and no math or grade effects (Kraft, Edwards and Cannata 2024). Opt-in and on-demand online tutoring is rarely used (under 1 percent of eligible students in New Mexico) and reaches higher-performing students. Pre-pandemic RCT effects came from programs costing several thousand dollars per student; thinner programs should expect thinner results.

Standard. Every student below benchmark on the district assessment shall be scheduled for tutoring in groups of four or fewer, at least three sessions a week of 30 minutes or more, during the school day, with a consistent trained tutor, and the district shall track delivered dosage and report it by school and student group.

Indicator. Share of students below benchmark on the district's interim assessment who received tutoring at least three sessions a week for at least ten weeks during the school day, by school and student group; sessions delivered versus scheduled; tutor consistency (share of sessions with the student's assigned tutor); growth of tutored versus matched non-tutored students on the interim assessment.

Red flags in real manuals. tutoring offered only after school, before school, or online on demand; tutoring available on request or by referral only; no dosage, frequency, or group-size requirement; tutoring assigned to unpaid volunteers without training; no reporting of who received tutoring or how much

Law. No federal mandate; ESSER funds that paid for most pandemic tutoring had to be obligated by September 30, 2024, so programs now depend on local, Title I, and state funds. Texas HB 4545 (2021), as amended by HB 1416 (2023), requires accelerated instruction, including supplemental instruction in groups of no more than four, for students who do not pass STAAR. Several states (for example Tennessee's TN ALL Corps) fund tutoring with dosage conditions.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Which textbook a district adopts can move achievement as much as many costlier reforms: in California the best-performing elementary math series outscored its rivals by 0.05 to 0.08 SD at near-zero marginal cost, a randomized IES trial found differences of 4 to 7 percentile points between first and second grade math curricula, and a field experiment giving teachers high-quality off-the-shelf lessons raised math achievement by 0.06 SD at about 6 dollars per student. Effects depend on the specific materials and their use with aligned professional development; districts' formal adoption processes rarely examine evidence of effectiveness. (Textbook choice: 0.05 to 0.08 SD (California), roughly 0.1 to 0.17 SD between best and worst in the IES RCT. Online lessons plus supports: +0.06 SD math (RCT). Curriculum plus curriculum-based PD: positive effects on science achievement (RCT). Six-state Common Core era study: no detectable differences among textbooks.) ^{125,126,127,128,129,130,131,132,133,134}

Where the evidence is mixed. Blazar et al. (2020) examined elementary math textbooks across six states after the Common Core transition and found no meaningful differences in achievement among the most widely used books, a null that contrasts with the earlier California and Indiana findings; one explanation is that textbook use had become fragmented and teachers supplemented heavily. Effects in the IES trial and in California varied by grade and did not always persist. Standards alignment on the cover is unreliable: Polikoff (2015) found textbooks labeled Common Core aligned were poorly aligned. Materials without aligned professional development and without time to implement show weak effects, and Jackson and Makarin's gains were concentrated among teachers who actually used the lessons. Evidence for English language arts and secondary materials is thinner than for elementary math.

Standard. The district shall adopt core instructional materials for each subject and grade through a published process that requires an independent quality review, consideration of effectiveness evidence, a classroom pilot, and a curriculum-specific professional development plan, and shall maintain a public inventory of materials in use with achievement tracked after each adoption.

Indicator. Inventory of core materials in use by subject and grade with the adoption year (curriculum office records); independent quality rating of each adopted core program (EdReports or state review) at the time of adoption; share of classrooms using the adopted core program as the primary resource (teacher survey or walkthrough); hours of curriculum-specific professional development per teacher in the adoption year; student achievement and growth before and after each adoption, tracked by the district for at least three years.

Red flags in real manuals. adoption process based on teacher preference or publisher presentation alone; no independent quality review required; no pilot or evidence summary before adoption; materials adopted without a professional development plan; no inventory of materials in use; adoption cycle longer than seven years or unspecified; policy addresses only objectionable content review, not quality

Law. About 20 states run state textbook adoption lists that districts must choose from or justify departing from (for example Texas Education Code Chapter 31, California Education Code 60200 et seq. for K-8, Florida Statute 1006.28 to 1006.42). Several states now require or fund high-quality instructional materials: Louisiana's tiered review, Tennessee's 2019 to 2021 ELA adoption and the Tennessee Literacy Success Act, Nebraska and Rhode Island HQIM laws, and Texas House Bill 1605 (2023) funding state-approved materials. Many state laws since 2021 add parent review and objection procedures for materials; these address content, not quality.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. Homework has a small to moderate positive association with achievement in secondary grades and little or none in elementary grades; the relation runs through how often and how well homework is assigned, not how many hours students spend on it. Beyond roughly ten minutes per grade level per night the returns flatten, stress rises, and the benefit accrues more to higher-income students. (Cooper 2006: d about 0.6 in the few experimental comparisons, secondary correlations r about 0.25, elementary near zero; Fan 2017: small positive effect in math and science; high-performing high schools average 3 hours per night with higher stress) ^{135,136,137,138,139,140,141,142,143,144}

Where the evidence is mixed. The meta-analyses pool mostly correlational studies, and time on homework may reflect struggling as much as learning (Trautwein 2007 finds frequency, not time, predicts achievement). Daw (2012) and Ronning (2011) find that homework widens socioeconomic gaps because higher-income students gain more from each hour, and Ronning finds low-SES students may be harmed. Galloway, Conner and Pope (2013) report that high-performing high school students averaging more than three hours a night show more academic stress, physical complaints and less balance. Pressman et al. (2015) find primary school children receive about three times the ten-minute guideline and that family stress rises with load. Eren and Henderson (2008) find positive effects that are concentrated among high and low achievers, with a share of students responding negatively to more homework. No randomized trial of district homework policy exists.

Standard. The policy shall state maximum nightly homework time by grade band (none required in K-2, no more than ten minutes per grade level through grade 8, no more than two hours total in high school), shall cap homework at a minor share of the academic grade, and shall require that assigned homework be practice of taught material that a student can complete without adult help or paid resources.

Indicator. Expected and reported nightly minutes by grade band (annual student and family survey); share of final grade weight assigned to homework by course (gradebook categories); share of students reporting more than two hours per night in high school; homework completion rate by free-lunch status.

Red flags in real manuals. no stated time limit by grade band; homework required in kindergarten or grade 1 beyond reading; homework weighted at 20 percent or more of the grade; zeros for missing homework with no make-up window; homework assigned over breaks or holidays; assignments that require printing, purchased supplies or parent participation; homework used as a disciplinary consequence

Law. No federal law. California AB 2999 (Healthy Homework Act, Chapter 751, Statutes of 2024, signed September 27, 2024) directs the state to post homework guidelines and encourages, but does not require, districts to adopt a homework policy by the 2027-28 school year that weighs student health, time and equity. Elsewhere homework is a district matter; the National PTA and NEA ten-minutes-per-grade guideline is the common reference.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. The evidence favors objective, transparent placement into accelerated math with instructional support, and cuts against both extremes. Placing moderately prepared students into Algebra I in eighth grade without extra support lowered their pass rates and later math course completion in North Carolina and reduced tenth grade math scores in California's push for algebra for all; placing students by an objective prior-achievement rule in Wake County removed race and income from placement decisions, and a randomized Dee and Huffaker study found that putting below-grade-level ninth graders into Algebra I with teacher training raised grade 11 math achievement by up to 0.2 SD. Delaying algebra for everyone, as San Francisco did, cut AP math participation 15 percent at first and left ethnoracial gaps unchanged. (Universal early algebra without support: negative on test scores and later course passing. Objective placement rule: equity of placement improved, test scores unchanged, grades lower. Detracking plus teacher training (Algebra I for below-grade-level ninth graders): +0.2 SD grade 11 math, attendance up. SFUSD delay of Algebra I: AP math participation initially down 6 percentage points, gaps unchanged. Double-dose algebra for low scorers: college enrollment up.) ^{145,146,147,148,149,150,151,152,153}

Where the evidence is mixed. Results depend on who is accelerated and what support comes with it. Clotfelter, Ladd and Vigdor (2015) found that Charlotte-Mecklenburg students accelerated into eighth grade algebra scored lower on the Algebra I exam and were less likely to pass geometry and Algebra II later. Domina et al. (2015) found that raising eighth grade algebra enrollment lowered tenth grade math achievement in large California districts. Chicago's curriculum-for-all policy raised algebra credits but also failure rates with no test score gains (Allensworth et al. 2009). Wake County's objective rule improved access but accelerated students did not gain on test scores or college readiness and earned lower grades (Dougherty et al. 2017). On the other side, removing eighth grade algebra for all students in San Francisco did not close gaps and initially reduced AP calculus participation (Huffaker, Novicoff and Dee 2025). High-achiever classrooms raised Black and Hispanic scores by 0.5 SD in one district (Card and Giuliano 2016), so grouping by readiness can help the students it is meant to help when seats are allocated by objective criteria.

Standard. Math placement shall use published objective criteria from multiple measures, place every student who meets them in the accelerated course with parent opt-out, provide a supported path (double period, tutoring, or summer bridge) for students near the cutoff, and never bar a readiness-qualified student from Algebra I in eighth grade.

Indicator. Share of eighth graders completing Algebra I or higher with a passing end-of-course score, by race, ethnicity, income, English learner and disability status; share of students meeting the placement criterion who are placed; Algebra I repeat rate; share of graduates completing a course beyond Algebra II and calculus or statistics; placement appeals and overrides by school.

Red flags in real manuals. Algebra I not offered before grade 9; placement based on teacher recommendation alone; placement criteria not published or not applied uniformly across schools; tracks or ability groups assigned in elementary grades with no annual reassessment; acceleration of all students into Algebra I with no support course; no reporting of math course completion by student group

Law. California Math Placement Act of 2015 (SB 359) requires districts to adopt a fair, objective and transparent math placement policy for students entering grade 9 with multiple objective measures and annual reporting. Texas SB 2124 (2023) requires automatic enrollment in advanced sixth grade math of students in the top 40 percent on fifth grade STAAR math, with parent opt-out, to prepare them for Algebra I in grade 8. North Carolina G.S. 115C-81.36 requires that students scoring at the highest level on the math end-of-grade test be placed in the advanced course, and that such seventh graders be placed in a high school math course in grade 8; removal requires written parent consent. Illinois Accelerated Placement Act requires accelerated placement policies and automatic enrollment of high school students who exceed state standards.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. The components of a tiered system work when they are done well: K-3 supplemental reading interventions average about 0.4 SD across dozens of rigorous studies, early K-2 screeners predict later reading, and a randomized trial shows that starting intervention immediately after screening beats waiting a semester (d = 0.36). The one large evaluation of RtI as schools actually ran it (Balu et al. 2015, 146 schools in 13 states) found that first graders just below the cutoff who were pulled into Tier 2 or 3 scored lower in spring than peers just above it, with no effect in grades 2 and 3. (Tier 2 and intensive K-3 reading interventions ES about 0.3 to 0.5; immediate versus delayed intervention d = 0.36; national RtI evaluation: grade 1 negative (about -0.1 SD), grades 2-3 null) ^{154,155,156,157,158,159,160,161,162,163,164}

Where the evidence is mixed. Balu et al. (2015) is the only large-scale evaluation of RtI practice and its result was null to negative; Gersten, Jayanthi and Dimino (2017) and Fuchs and Fuchs (2017) argue the design captured students near the cutoff who were often pulled from core instruction into undifferentiated groups, so it tests weak implementation rather than the model, but the finding stands as a warning that labels and tiers do not help by themselves. Burns et al. (2005) reported very large effects for RtI models, but from a small set of mostly non-experimental field studies. Most district MTSS policies describe structures (teams, tiers, forms) and not instructional content, intensity or dosage, which is where the measurable effects live. Universal screening generates false positives at the K-1 level and its accuracy depends on the measure and administration quality.

Standard. The policy shall name the screener, the screening windows, the timeline from flag to intervention, the minimum dosage of Tier 2 and Tier 3 instruction delivered in addition to core instruction, the progress-monitoring interval and the decision rule for moving students between tiers.

Indicator. Share of K-3 students screened each window and share identified at risk (screener vendor reports); share of at-risk students with intervention logged within 15 school days; progress-monitoring frequency per Tier 2 and Tier 3 student (every two weeks or better); share of Tier 2 students who exit to Tier 1 within a semester; special education referral and identification rates by race and school.

Red flags in real manuals. MTSS described only as a team or meeting structure with no dosage, frequency or timeline; intervention delivered during core reading or mathematics instruction; progress monitoring 'as needed' or quarterly; a required number of weeks in RtI before a special education referral may be made; screening only for students referred by teachers; no decision rule or exit criterion for tiers

Law. IDEA 2004 and 34 CFR 300.307 permit states to allow a response-to-intervention process for identifying specific learning disabilities and forbid using it to delay or deny an evaluation (OSEP memo 11-07, 2011). ESSA references MTSS as an allowable use of Title funds. Most states require K-3 reading screening by statute; several (e.g. Texas, Mississippi, Ohio) specify screening windows and parent notification timelines that a district policy must meet.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Receiving special education services causally raises achievement and attainment for students with disabilities (Texas: removing services lowered high school completion and college enrollment; NYC and Texas: entering special education raised test scores by roughly 0.1 SD), and large correlational and meta-analytic literatures find that students with disabilities who spend most of the day in general education classrooms do better academically with no detectable cost to their peers (d = 0.35 for students with general learning difficulties; peers d = 0.12 to null). Co-teaching shows a medium positive effect in a 2021 meta-analysis. The causal evidence on placement itself, as opposed to services, is thin and contested, so the best-supported policy is a documented, data-driven least restrictive environment decision for each student rather than a placement quota in either direction. (Special education services: about +0.1 SD achievement; service removal lowered high school completion and college entry. Inclusive placement: d = 0.35 cognitive outcomes for students with learning difficulties (meta-analysis of mostly non-randomized studies); peers d = 0.12 (small positive) to null. Co-teaching: medium positive effect on achievement for students with disabilities.) ^{165,166,167,168,169,170,171,172,173,174,175,176,177,178}

Where the evidence is mixed. Gilmour (2018) and Gilmour, Fuchs and Wehby (2018) show that the inclusion-achievement correlation is confounded by selection: students with stronger skills and fewer behavior challenges are placed in general education, and the reading gap for students with disabilities is about 1.2 SD regardless of setting. Hanushek, Kain and Rivkin (2002) found special education raised math for students with learning and emotional disabilities but found no difference between general and special classroom settings. Krämer, Möller and Zimmermann (2021) report no psychosocial benefit of inclusion (d = 0.00) and note selection effects in the underlying studies. Correlational work finds higher teacher turnover and more management time in classes with larger shares of students with emotional and behavioral disabilities. Co-teaching effects depend on implementation; much co-teaching in practice is one-teach, one-assist with little specialized instruction. Inclusion done by moving students without moving services or training is not supported by any of this evidence.

Standard. Every placement decision shall start from the general education classroom with supplementary aids and services, document in the IEP the specific reasons a more restrictive setting is required, and be reviewed against the student's academic and behavioral data at least annually; the district shall report placement and outcome data for students with disabilities by school each year.

Indicator. Share of students with IEPs in general education 80 percent or more of the day, 40 to 79 percent, and under 40 percent (IDEA Section 618 educational environment data), disaggregated by disability category, race and school; achievement and growth of students with disabilities on state tests versus peers; graduation rate and dropout rate of students with disabilities; suspension rates of students with disabilities; share of IEPs documenting consideration of supplementary aids and services before a more restrictive placement; co-taught sections staffed by a certified special educator for the full period.

Red flags in real manuals. placement determined by program or disability label rather than individual IEP; separate 'resource' or 'self-contained' track described as the default for a category; no mention of supplementary aids and services or least restrictive environment; co-teaching defined as paraprofessional support; inclusion stated as a target percentage without a data review process; placement contingent on 'space available' or staffing; no annual report of special education outcomes to the board

Law. IDEA, 20 U.S.C. 1412(a)(5), requires education in the least restrictive environment to the maximum extent appropriate, with removal from general education only when education there cannot be achieved satisfactorily with supplementary aids and services; 34 CFR 300.114 to 300.120 govern placement and a continuum of alternative placements. Section 504 of the Rehabilitation Act and the ADA bar exclusion on the basis of disability. States report educational environment data under IDEA Section 618 and set targets in State Performance Plan Indicator 5. Texas's former 8.5 percent special education enrollment benchmark was found to violate IDEA by the U.S. Department of Education in 2018.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Voluntary summer programs with at least five weeks and three hours of daily academic instruction produce modest gains for students who attend regularly: a meta-analysis of 37 studies finds +0.10 SD in math, a meta-analysis of 41 summer reading interventions finds significant reading gains concentrated among low-income children, and the RAND five-district randomized study found a near-term math benefit after one summer with persistent math and language arts advantages for high attenders after two. Extending the regular school day for literacy in low-performing Florida schools raised reading 0.05 SD in a regression-discontinuity design. (Summer math programs +0.10 SD (meta-analysis); summer reading interventions significant, larger for low-income children; RAND RCT: math gain after one summer that faded by fall, high attenders ahead in math and language arts through spring of the following year; Florida extra literacy hour +0.05 SD reading) ^{179,180,181,182,183,184,185,186,187,188}

Where the evidence is mixed. Intent-to-treat effects are small and fade: in the RAND study the causal benefit of being offered the program dissipated by the next fall, offering a second summer showed no significant causal effect, and three years later high-attender advantages were no longer statistically significant. Attendance is the binding constraint, and the students who attend 20 or more days differ from those who do not, so the high-attender estimates are correlational. Von Hippel and Hamrock (2019) show that much of the canonical summer-learning-loss literature was distorted by test scaling artifacts, and gaps form mainly before school entry, which weakens the rationale for summer programs as a gap-closing tool. Extended-day and extended-year studies are mostly weak in design (Patall et al. 2010), and added time costs staff pay and transportation.

Standard. The district offers a free summer program of at least five weeks with at least three hours of daily academic instruction by certified teachers, prioritized for students below grade level, and reports attendance and achievement growth for attenders each year.

Indicator. Summer program offered (yes or no), its weeks and daily academic hours (program plan); enrollment and share of enrollees attending 20 or more days, by subgroup (attendance file); fall-to-spring and spring-to-fall growth on interim assessments for attenders versus matched non-attenders; for extended-day programs, the added minutes and how they are used, with the same growth comparison.

Red flags in real manuals. summer school limited to credit recovery or retention remediation with fees; summer program under four weeks or under two hours of daily academics; no attendance target or outreach provision; extended day used for homework or supervision with no instructional plan; no reporting of attendance or outcomes

Law. No federal mandate. Title I, Part A and the 21st Century Community Learning Centers program (ESEA Title IV-B) fund summer and extended learning. The ESSER III requirement that districts spend at least 20 percent of funds on learning recovery, including summer programs, expired with the September 30, 2024 obligation deadline. Some states require summer reading intervention for students not proficient by grade 3 under third-grade reading laws (for example Florida, Michigan, Mississippi and Tennessee); check the state code.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Difference-in-differences studies across Oregon, a multi-state NWEA sample and a RAND study of Idaho, New Mexico and Oklahoma agree that moving to a four-day week lowers achievement growth modestly (spring reading about -0.07 SD and fall-to-spring growth about -0.06 SD in math and reading), with the loss concentrated in non-rural districts and in schedules that cut total instructional time; the saving is small, about 1 to 3 percent of operating spending, and there is no evidence of better attendance. (Oregon: -0.042 SD math after switching; multi-state: spring reading -0.07 SD, fall-to-spring growth -0.06 SD; RAND: slower achievement growth than matched five-day districts; expenditures -1 to -3 percent; juvenile crime +20 percent in Colorado counties; no attendance gain) ^{189,190,191,192,193,194,195,196,197}

Where the evidence is mixed. Anderson and Walker (2015) found positive effects on Colorado elementary math and reading, and Morton (2022) found no detectable effect on Oklahoma high school ACT scores or attendance and 31 to 39 percent fewer fighting and bullying incidents (partly mechanical, since students are in school less). Thompson and Ward (2022) show the harm is near zero in four-day districts that keep total hours at five-day levels, so the schedule itself is less the problem than lost time. Families, students and teachers report high satisfaction, and RAND found qualitative support for teacher recruitment in rural districts. Districts under severe revenue loss may use the small savings to protect instructional spending.

Standard. A district adopts or retains a four-day week only after showing that total instructional hours will equal or exceed its prior five-day calendar, with a published pre-adoption analysis, fifth-day meal and supervision provisions, and a dated outcome review that triggers reconsideration if achievement growth falls.

Indicator. Total annual instructional hours per grade span compared with the district's prior five-day calendar and the state minimum (board-adopted calendar); student achievement growth on state tests and interim assessments in the three years after adoption compared with matched five-day districts; chronic absence rate; documented savings as a share of operating expenditure; fifth-day childcare and meal access for low-income students.

Red flags in real manuals. four-day calendar adopted for cost savings with no instructional-hours comparison; total annual hours set at the state minimum rather than the prior calendar; no provision for meals or supervision on the fifth day; no scheduled outcome review after adoption; calendar policy silent on instructional hours

Law. State calendar law governs. Missouri SB 727 (2024) requires districts in charter counties or cities over 30,000 to win majority voter approval to start or keep a four-day week, with approval lasting ten years, effective 2026-27. Oklahoma SB 441 (2019) limits four-day weeks to districts meeting state performance benchmarks and documenting cost savings, with 1,080 instructional hours required; the State Board granted waivers through 2022. Most four-day states set minimum annual hours rather than days (for example, Texas 75,600 minutes), which is what makes shortened calendars legal.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Remote instruction delivers less learning than in-person instruction and widens gaps. In 2020 to 2021, districts that stayed in person lost 13 percentage points less in math pass rates than remote districts (Jack et al.), remote instruction was the primary driver of widened gaps by race and poverty (Goldhaber et al.), and the national decline reached 0.20 to 0.27 SD in math (Kuhfeld et al.). Randomized trials of online delivery at West Point (-0.215 SD in grades) and in Chicago algebra credit recovery (lower pass rates) confirm the direction outside the pandemic. A few weather closures do not measurably hurt achievement, so a remote day mainly needs to beat a day off, which it fails to do when attendance collapses. (remote vs in-person: about -0.2 SD (RCTs and pandemic natural experiments), largest for low-income and Black students; a handful of closure days: about 0) ^{198,199,200,201,202,203,204}

Where the evidence is mixed. No study isolates occasional virtual snow days specifically; the pandemic evidence covers months of remote schooling under crisis conditions and likely overstates the cost of a single day. Goodman (2014) found that a few snow days did not reduce Massachusetts test scores because schools made the time up, while Marcotte and Hemelt (2008) found that about five unscheduled closures before Maryland's spring tests lowered third-grade proficiency rates by nearly 3 percent, so the cost of lost days is real when they are not recovered before assessment. Blended programs that keep in-person contact perform about as well as in-person classes (Escueta et al.). Remote days do protect calendar end dates and staff pay, and for a multi-day emergency closure a well-run remote day beats nothing.

Standard. The district shall use remote instruction days only for weather or emergency closures, shall cap them at five per year counted toward instructional time, shall require live synchronous instruction and attendance taking on every remote day, shall confirm device and internet access for every student in advance, and shall make up any closure beyond the cap with in-person days.

Indicator. Number of remote instruction days used per year and the reason for each; student attendance on remote days (share logging into synchronous instruction) by school and by free-and-reduced-lunch status, compared with the in-person attendance rate; share of households confirmed to have a device and connectivity; minutes of live instruction delivered per remote day.

Red flags in real manuals. remote learning days may be used at the superintendent's discretion for any purpose; no cap on the number of remote days; asynchronous assignments posted online constitute a remote instruction day; attendance is not taken or is counted as present if the student submits work within a week; no device or connectivity confirmation; no make-up provision when the cap is exceeded; special education services are suspended on remote days

Law. State law controls whether remote days count toward minimum instructional time: as of 2025 to 2026, Arkansas, Massachusetts, Connecticut, New Jersey and DC do not count them (with narrow emergency exceptions), about 23 states cap them (typically 3 to 5 days, up to 10 in Virginia and Kentucky), and about 23 states leave it to district discretion. Indiana limits asynchronous days to three. Some states, such as New York, set device, connectivity and special-education service requirements for remote days. Confirm the current state rule before adopting a cap.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Adolescent circadian timing shifts later, and middle and high schools that start at or after 8:30 a.m. get measurably more sleep from students; a Seattle district-wide delay of nearly an hour produced 34 more minutes of sleep per night, a 4.5 percent rise in median grades and better attendance, and a Florida time-zone natural experiment found each hour of later start relative to sunrise raised adolescent math scores by 0.08 SD. Meta-analyses and a Cochrane review agree on the sleep and attendance effects; achievement effects are positive but smaller and less consistent. (+34 min sleep per school night (Seattle); grades +4.5 percent; math +0.08 SD and reading +0.06 SD per hour later start (Florida); fewer absences and tardies; achievement gains small or mixed in some district studies) ^{205,206,207,208,209,210,211,212,213,214,215,216}

Where the evidence is mixed. Achievement gains are modest and uneven: Bastian and Fuller (2018) found mixed positive and negative test-score associations across North Carolina high schools even where suspensions fell and grades rose, and the Cochrane review (2017) rated the underlying studies as low quality. Costs are real: later secondary starts usually force bus-tier changes, and RAND (2017) modeled roughly 150 dollars per student per year in added transportation cost plus a one-time facilities cost, with net benefits that turn positive only after several years in high-cost scenarios. Moving secondary schools later can push elementary schools earlier, and after-school activities and student work hours are compressed. Florida enacted a 2023 mandate and then, in 2025, replaced it with a report-and-consider requirement after districts objected to bus and staffing costs.

Standard. Instruction in every school serving any of grades 6 through 12 begins no earlier than 8:30 a.m., and the policy names who reports compliance and by when.

Indicator. First-bell instructional start time for every building serving grades 6 through 12 (master schedule); share of secondary students starting before 8:30 a.m.; median school-night sleep from the district climate survey or YRBS item; chronic absence and first-period tardy rates by building.

Red flags in real manuals. start times set solely by transportation efficiency or bus tiers; zero period or mandatory before-school instruction for secondary students; high school start time earlier than middle or elementary start time; no reporting of building start times to the board; policy delegates bell schedule to the Superintendent with no floor

Law. California SB 328 (2019) requires high schools to begin no earlier than 8:30 a.m. and middle schools no earlier than 8:00 a.m., effective July 1, 2022, with rural districts exempt. Florida enacted the same floors in 2023 (effective 2026) but in 2025 (SB 296) replaced the mandate with a requirement that districts report by June 1, 2026 that they considered later start times; districts that file the report are deemed compliant. No federal law applies.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Whole-school anti-bullying programs reduce bullying perpetration by about 19 to 20 percent and victimization by about 15 to 16 percent on average across roughly 100 evaluations, with RCT-only estimates smaller (d about 0.15) but still showing mental-health gains. Written policies matter too: students in states whose anti-bullying laws contain the core US Department of Education components had 24 percent lower odds of being bullied and 20 percent lower odds of cyberbullying, and enumerated policies that name sexual orientation and gender identity are associated with lower suicide-attempt risk among LGB youth. (Perpetration -19 to -20%; victimization -15 to -16% (program meta-analyses); RCT pooled d = -0.15 bullying, -0.21 mental health; compliant state laws: 24% lower odds of bullying, 20% lower odds of cyberbullying.) ^{217,218,219,220,221,222,223,224,225,226,227,228,229,230}

Where the evidence is mixed. Program effects shrink with age: a three-level meta-analysis found effects near zero for grade 8 and above, so a program that works in elementary school may do nothing in high school. RCT-only pooled effects are small (d about 0.15). The policy-content findings (Hatzenbuehler 2013, 2015; Meyer 2019; Kull 2016) are cross-sectional associations between state or district policy and student reports, not causal estimates. Most trials are European; KiVa's large cluster-randomized effects in Finland have not been replicated at the same scale in US districts. Self-report prevalence varies several-fold with the definition and recall window used, so year-to-year comparisons require a fixed instrument.

Standard. The policy shall name a specific evidence-based whole-school program with a written policy and classroom rules, enumerate protected characteristics including sexual orientation and gender identity, require a published reporting channel with an investigation deadline, and require an annual prevalence survey and report to the Board.

Indicator. Annual student-reported bullying victimization prevalence (past 30 days or past year, fixed instrument such as the YRBS item or the Olweus questionnaire) by school, grade and demographic group; number of reports received, number investigated within the policy timeline, number substantiated; share of students who say they would report; all compared to the prior three years.

Red flags in real manuals. bullying prohibited only in general terms with no enumerated characteristics; no investigation timeline or no named investigator; reporting channel described only as 'tell a trusted adult' with no form, email or anonymous option; program named as 'character education' or an assembly with no whole-school component; no requirement to measure prevalence; only counts of reports; retaliation not addressed; cyberbullying excluded when it occurs off campus even where it disrupts school

Law. All 50 states have anti-bullying statutes (Montana was last, 2015); most require a district policy and many require cyberbullying coverage and a reporting procedure. The US Department of Education's 2011 analysis identified 16 policy components; the three associated with lower bullying in Hatzenbuehler 2015 are a statement of scope, a description of prohibited behaviors and a requirement that districts adopt local policies. Roughly 20 states enumerate sexual orientation and gender identity. Title IX and Title VI obligations attach when bullying is based on sex, race or national origin, and Section 504 and the IDEA when it affects a student with a disability (OCR Dear Colleague Letters of 2010 and 2014).

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. The meta-analysis specific to school corporal punishment (Visser et al. 2022) finds it associated with more externalizing behavior (r = 0.27, 21 studies, 67,400 children), more internalizing problems (r = 0.16) and lower school performance (r = -0.11, 977,367 children); a propensity-matched sample of young adults from the 19 states where it was legal shows lower high school GPA and more depressive symptoms; the 75-study meta-analysis of physical punishment finds 13 of 17 outcomes significantly worse and none better. No study has found that corporal punishment improves behavior or learning, and no study has found that ending it increases misbehavior. (School CP: externalizing r = 0.27, internalizing r = 0.16, school performance r = -0.11; physical punishment generally: 13 of 17 outcomes worse (mental health problems d = 0.53, aggression d = 0.37), immediate compliance not even reliably improved (d = 0.14, n.s.).) ^{231,232,233,234,235,236,237,238,239,240,241,242,243,244}

Where the evidence is mixed. The evidence is meta-analytic but correlational: no randomized or natural-experiment study of school corporal punishment exists, and none is likely, so the direction of causation is argued from consistency, dose and the absence of any benefit rather than from an experiment. Font and Cage (2018) found only physical abuse, not mild corporal punishment, predicted declines in cognitive test performance, though every form predicted lower school engagement; do not claim that mild corporal punishment measurably lowers test scores on that study. There is no US difference-in-differences study of suspensions or achievement before and after a state or district ban, so a district that ends paddling without a replacement behavior system could shift to suspension; the cross-national evidence that bans are followed by less adolescent fighting (Elgar et al. 2018) is ecological. The commonly repeated claim that states permitting corporal punishment have lower ACT scores has no locatable primary source and should not be used.

Standard. Corporal punishment is prohibited in policy for every student in every setting, with physical force limited to the restraint standard, and the prohibition is communicated to staff and families annually.

Indicator. Instances of corporal punishment reported to the Civil Rights Data Collection (should be zero), with a district count reconciled to building discipline logs; presence of a written prohibition; where state law still permits it, the written opt-in consent rate, witness records and the race, sex and disability breakdown of students struck.

Red flags in real manuals. corporal punishment listed among permitted consequences; 'corporal punishment may be administered by the principal or designee'; parental consent, opt-in or opt-out forms for paddling; 'reasonable force' defined to include paddling or spanking; a witness, instrument size or number-of-licks rule, which signals the practice is retained; 'in accordance with state law' with no district prohibition where state law permits the practice

Law. Ingraham v. Wright, 430 U.S. 651 (1977) held that the Eighth Amendment does not apply to school corporal punishment and that no hearing is required before it, leaving substantive due process claims for excessive force open in the circuits. As of September 2026 corporal punishment in public schools is legal by statute in 17 states (Alabama, Arizona, Arkansas, Florida, Georgia, Indiana, Kansas, Kentucky, Louisiana, Mississippi, Missouri, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, Wyoming); in Kentucky and North Carolina every district has prohibited it by policy. Five states bar it for students with disabilities while permitting it otherwise (Louisiana 2017, Tennessee 2018, Arkansas 2019, Mississippi 2019, Oklahoma 2025). Where state law permits it, district policy is the operative prohibition. The Protecting Our Students in Schools Act (H.R. 3265, 2025) would bar it in federally funded schools and has not had a hearing. The AAP (2023) and the Secretary of Education (2023) have called for state bans.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. The best causal evidence on uniforms (school fixed effects in a large urban district) finds a modest attendance gain in secondary grades and no effect on achievement or discipline, and national longitudinal data show no behavior or attendance advantage; the evidence that dress rules improve learning is weak. The evidence on enforcement is clearer: GAO's national review found dress codes restrict items typically worn by girls more than boys, strict enforcement in more than four in five predominantly Black schools against about one third of predominantly white schools, informal class removals in 44 percent of codes, and body-measuring rules in 60 percent; Black girls receive disproportionate office referrals for dress code and other subjective infractions. (Uniforms: small secondary attendance gain, no achievement effect (Gentile and Imberman); no behavior or attendance link in national elementary data (Ansari et al.); physical-activity-enabling uniforms: no significant MVPA gain in a 42-school cluster RCT. Enforcement: strict codes in more than 80% of predominantly Black schools vs about 33% of predominantly white schools (GAO).) ^{245,246,247,248,249,250,251,252,253,254,255}

Where the evidence is mixed. Gentile and Imberman (2012) do find an attendance improvement in middle and high school and large increases in elementary teacher retention after uniform adoption, and Ansari et al. (2022) find low-income students in uniform schools attended better, so uniforms are not harmful on the evidence; they are simply not a lever for achievement. Cross-national PISA correlations (Baumann and Krskova 2016) link uniforms with better classroom discipline, but they cannot separate uniforms from the schools that choose them. Brunsma and Rockquemore (1998) found a small negative achievement association that later commentators disputed. The GAO findings describe enforcement patterns, not outcomes, and the harm case for strict dress codes rests on the exclusionary-discipline evidence (see exclusionary-discipline) rather than on dress-code studies themselves. The one RCT in this area (Nathan et al. 2021) tested sports uniforms for physical activity and found no significant primary effect.

Standard. Dress rules are limited to health, safety and the prohibition of hate speech and obscenity, apply identically to all students regardless of sex, do not require measuring bodies or clothing, protect religious and cultural dress and natural hair, and are enforced without removal from instruction or exclusion from school, with enforcement data reported by student group.

Indicator. Dress code referrals, class removals and suspensions per 100 students by building, race and sex (student information system; informal removals must be logged to be counted); instructional minutes lost to dress code enforcement; share of the code's rules that are gender-specific or require measuring a student's body; parent and student complaints about dress enforcement.

Red flags in real manuals. 'fingertip length', 'two inches above the knee', 'three-finger straps' or other measurement rules; rules naming girls' items only (skirts, leggings, tank tops, spaghetti straps) or 'distracting' clothing; bans on head wraps, durags, braids, locs, hairstyles or hair length; dress code violation listed as grounds for in-school or out-of-school suspension; 'students will be sent home to change' with no requirement to log the removal; 'administration has final authority to determine appropriate dress' with no stated criteria; uniform requirement with no free-uniform provision for low-income students

Law. Tinker v. Des Moines, 393 U.S. 503 (1969) protects student expression absent substantial disruption, which limits content-based dress rules. Title IX and Title VI expose sex-specific and racially disparate enforcement to federal complaint, and GAO (2022) asked the Department of Education to issue equity guidance; the recommendations remained open as of August 2025. More than twenty states have enacted CROWN Acts barring discrimination based on hair texture and protective styles in schools. Religious dress is protected under the First Amendment and state religious-freedom statutes. The federal government sets no dress code standard; state statutes commonly authorize but do not require district dress codes and uniforms, and several (for example California Education Code 35183) require an opt-out or free uniforms when a uniform is adopted.

Dispute this verdict or submit evidence

Tier D: professional standard or expert consensus

Finding. No outcome study shows that restraint or seclusion improves behavior or learning; the GAO found hundreds of allegations of abuse and death linked to their use in schools between 1990 and 2009, with untrained staff and non-aggressive children with disabilities as the recurring pattern. In 2021-22 about 105,700 students were restrained or secluded, and students served under IDEA were 14 percent of enrollment but 76 percent of those physically restrained and 68 percent of those secluded. A systematic review of 15 studies finds that school-wide prevention approaches reduce the frequency of restraint, but none is an RCT. (IDEA students: 14% of enrollment, 76% of physical restraints, 68% of seclusions (2021-22); earlier CRDC analysis: students with disabilities 7 times as likely to be restrained and 4 times as likely to be secluded; reduction programs: most of 15 studies report significant decreases, magnitudes uncontrolled.) ^{256,257,258,259,260,261,262,263,264,265,266}

Where the evidence is mixed. The evidence base is thin and the tier reflects that. Harm evidence is case-based (GAO's closed cases, media investigations) and from medical literature on restraint deaths rather than from controlled school studies. The reduction literature (Hodgkiss and Harding 2023; Trader et al. 2017) is mostly single-site pre-post designs without control groups. Federal counts are unreliable: GAO found about 70 percent of districts reported zero incidents in 2015-16 and that large districts had reported zeros in error, so trend claims from the CRDC are weak. Practitioners argue that restraint remains necessary in genuine emergencies to prevent injury, and no study claims otherwise; the standard here limits rather than bans physical restraint while banning seclusion, mechanical and chemical restraint and restraints that restrict breathing.

Standard. Seclusion, mechanical restraint, chemical restraint and any restraint that restricts breathing are prohibited; physical restraint is limited to imminent danger of serious physical harm, performed by trained staff, with same-day parent notice, a written incident report, a debrief, a behavior-plan review after repeated use, and quarterly public reporting.

Indicator. Incidents of physical restraint, mechanical restraint and seclusion per 100 students by building and by disability status, race and sex, reconciled between the district's incident log and the Civil Rights Data Collection submission (a zero submission from a district with special education programs is itself a red flag); injuries to students and staff during incidents; share of incidents with same-day parent notification; share of incidents involving a student with no behavior intervention plan.

Red flags in real manuals. seclusion, 'time-out room', 'calm room' or 'isolation room' permitted with the door closed or blocked; restraint or seclusion listed as a disciplinary consequence; prone, supine or mechanical restraint not expressly prohibited; restraint permitted to prevent property damage or for 'disruption' or 'noncompliance'; parent notification 'within a reasonable time' or not required; no training requirement, or training 'as resources allow'; no incident report, debrief or reporting to the Board; 'reasonable force' clause with no imminent-harm limit

Law. There is no federal statute on restraint or seclusion in schools; the Keeping All Students Safe Act (H.R. 6617 and S. 3448, introduced December 11, 2025) would ban seclusion and mechanical, chemical and breathing-restricting restraint in federally funded schools and has not advanced. The Department of Education's 2012 Resource Document sets fifteen principles that most state laws track. Every district must report restraint and seclusion incidents to the Civil Rights Data Collection, and GAO (2019, 2020) found that reporting unreliable. State law governs: Kern et al. (2024) find most states have a statute or guidance, with wide variation; Georgia prohibits seclusion outright; many states permit seclusion with conditions. IDEA requires that a student's IEP team consider positive behavioral interventions when behavior impedes learning (34 CFR 300.324(a)(2)(i)), which is the legal hook for a behavior plan review after repeated restraint.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. School climate is a medium-sized correlate of achievement (meta-analysis of 37 K-12 studies), buffers the achievement penalty of low socioeconomic status, and predicts adolescent mental health, engagement and graduation; within-school improvements in safety, leadership and academic expectations measured on New York City's annual survey track lower teacher turnover and higher achievement in a design the authors call consistent with causation. Climate is changeable: a cluster-randomized trial of schoolwide PBIS reduced office referrals by 33 percent. What the evidence does not show is that surveying alone changes anything; the survey is the instrument that makes the rest of the manual's climate commitments testable. (School climate to achievement: medium pooled correlation (37 studies); safety and expectations gains linked to achievement gains and lower teacher turnover within schools over time; climate intervention RCT: office referrals -33%; no direct effect estimated for the act of surveying.) ^{267,268,269,270,271,272,273,274,275,276,277,278}

Where the evidence is mixed. Nearly all climate-outcome evidence is correlational and most reviews say so; schools with high achievement may produce good climate reports rather than the reverse. Definitions and instruments vary widely, which inflates heterogeneity across studies. Self-report scales show reference bias and normative developmental declines after grade 6 that can be misread as a worsening school, so year-over-year comparisons need a fixed instrument and cohort-aware interpretation. Using climate scores for high-stakes accountability may distort responses; the CORE districts use them for improvement, not sanctions. Response rates below about 70 percent make school-level scores unreliable, and no study evaluates whether mandating a survey, by itself, improves outcomes.

Standard. The policy shall require a validated climate survey of students, staff and families every year, with a fixed instrument, a minimum response rate, results published by school and demographic group, and a requirement that each school improvement plan respond to its lowest climate domain.

Indicator. Annual climate survey response rate by school and respondent group (target at least 70% of students and staff); domain scores (safety, relationships, engagement, environment) by school, grade band and demographic group with three-year trend; correlation of school climate scores with chronic absenteeism, suspension rate and achievement; the number of school improvement plans that cite a climate finding.

Red flags in real manuals. climate survey 'may' be administered or is left to principal discretion; no named or validated instrument; no response-rate requirement; results reported only districtwide, not by school or demographic group; no link from survey results to the school improvement plan; survey used for staff evaluation or school sanctions; instrument changed every year so trends cannot be read

Law. ESSA (20 USC 6311(c)(4)(B)(v)) requires each state accountability system to include at least one indicator of school quality or student success, and several states use climate surveys for it. California's Local Control and Accountability Plan requires districts to report on school climate (Priority 6), typically with the California Healthy Kids Survey. Illinois (105 ILCS 5/2-3.153) requires a biennial statewide learning-conditions survey (5Essentials) of students and teachers. Maryland, Georgia, Nevada and others administer statewide climate surveys and publish school-level results. The federal EDSCLS instrument is free. Student surveys that ask about protected topics may trigger the Protection of Pupil Rights Amendment (20 USC 1232h) notice and opt-out requirements.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Universal, sequenced social-emotional learning programs improve social-emotional skills, behavior and achievement: the 2011 meta-analysis of 213 programs found skills gains of 0.57 SD and an 11 percentile-point achievement gain in the 35 studies that measured it, follow-up effects persist at 0.13 to 0.33 SD six months to years later, and the 2023 meta-analysis of 424 studies with 575,361 students confirms positive effects across skills, behavior, climate and achievement. Effects are smaller in independent, at-scale trials than in developer-led efficacy trials and depend on implementation quality; some universal mental-health curricula have produced null or harmful results. (Skills +0.57 SD, achievement +0.27 SD (11 percentile points), conduct problems -0.22 SD, emotional distress -0.24 SD (Durlak 2011); follow-up academic +0.33 SD in the few studies measuring it (Taylor 2017); reading and math about +0.25 SD (Corcoran 2018); effects roughly halve in independent at-scale trials (Wigelsworth 2016).) ^{279,280,281,282,283,284,285,286,287,288,289}

Where the evidence is mixed. The 11 percentile-point figure comes from 35 of 213 studies, mostly developer-led, and Wigelsworth's 2016 meta-analysis found effects shrink substantially when trials are run independently and at scale. The largest independent UK trial of PATHS (45 schools, 4,516 children) found only small teacher-rated gains in social competence and nothing on academic outcomes, and effects of the 4Rs program in New York were modest after two years. Universal mental-health-flavored programs have failed outright: the MYRIAD mindfulness trial was null in 85 schools and a DBT-skills program worsened depression and anxiety. Program adoption without fidelity monitoring is the common failure mode; Durlak's SAFE criteria (sequenced, active, focused, explicit) and implementation quality moderate every meta-analytic estimate. SEL has also become politically contested in several states, and districts that cannot explain what is taught lose parent trust.

Standard. A sequenced, explicitly taught SEL program with an independent evidence rating is delivered in every grade K-12 at its designed dose, with annual implementation-fidelity checks, published scope and sequence for families, and annual measurement on the climate survey.

Indicator. Adoption of a named, sequenced SEL program with a CASEL SELect or What Works Clearinghouse evidence rating in every grade; share of classrooms delivering the program at the scheduled dose, from fidelity walk-throughs; climate survey items on belonging, trusted adults and peer support (state climate survey or Panorama); office discipline referrals and suspensions by school.

Red flags in real manuals. social-emotional learning is prohibited or restricted in the curriculum; SEL may be integrated at the discretion of the teacher (no program, no dose); character education assemblies substitute for a sequenced curriculum; no measurement of climate or implementation; program selected without an evidence rating; no family-facing scope and sequence or opt-out process

Law. Illinois was the first state to mandate K-12 SEL standards (105 ILCS 5/2-3.153, Children's Mental Health Act of 2003) and most states now publish SEL standards or competencies, usually as guidance rather than mandate. Since 2022 several states have restricted or renamed SEL content in curriculum law or standards reviews (including Oklahoma and Florida), so program materials must be transparent to families. ESSA Title IV-A funds may support SEL programs.

Dispute this verdict or submit evidence

Tier D: professional standard or expert consensus

Finding. This domain is governed by constitutional standards rather than outcome research: Goss v. Lopez requires notice and an opportunity to respond before any suspension of ten days or less, New Jersey v. T.L.O. permits school searches on reasonable suspicion when justified at inception and reasonable in scope, and Safford v. Redding holds that a search of a student's underwear requires suspicion that the item is dangerous or hidden there. The one federal randomized trial of random drug testing found lower self-reported use among tested students (16 versus 22 percent) with no spillover to untested students, while national correlational studies find no association between testing and student drug use. (Legal standards, not effect sizes. Drug testing: tested students 16% vs 22% past-30-day use in the federal RCT; no effect on untested students or on intentions; national surveys show no association.) ^{290,291,292,293,294,295,296,297,298}

Where the evidence is mixed. Outcome evidence on searches and hearings is almost nonexistent; nobody has randomized due process. The drug-testing literature is the only causal evidence and it is mixed: the federal RCT (James-Burdumy et al. 2010) found less reported use among students actually subject to testing in seven volunteer districts, with no generalizable national estimate, while Yamaguchi et al. (2003) found no association between school testing and student drug use in national Monitoring the Future data, the SATURN prospective trial of athlete testing found no reduction in past-month use, and Sznitman and Romer (2014) found that a positive school climate, not testing, predicted less use. Courts have upheld suspicionless drug testing of students in extracurricular activities (Vernonia 1995; Earls 2002), so the limit on it here is a policy choice grounded in weak benefit evidence, not a legal requirement. Procedural-justice research in schools is correlational: students who see discipline as fair report less misconduct and more engagement, but the designs cannot show that hearings cause the difference.

Standard. Policy states the Goss procedures for every short suspension and a formal hearing with written notice, representation, evidence and a written decision for any exclusion beyond ten days; searches require individualized reasonable suspicion documented in writing, strip searches are prohibited, and searches and hearings are logged and reported by student group.

Indicator. Share of suspensions for which the student record documents notice of charges, the evidence explained and the student's response before removal (discipline record audit); share of expulsion hearings with written notice, a hearing officer independent of the referring school, and a written decision; number of student searches logged with the basis for suspicion, by building, race and disability; strip searches (should be zero); complaints and appeals filed and their outcomes.

Red flags in real manuals. 'the principal may suspend a student without a conference' or no mention of notice and opportunity to respond; expulsion decided by the same principal who referred the student, with no independent hearing; 'students have no expectation of privacy' applied to their person as well as lockers; search authority stated without a reasonable-suspicion standard or with 'any reason'; strip search or 'search of undergarments' permitted by any employee; suspicionless drug testing of all students or as a condition of parking or activities without a stated basis; no appeal procedure, or an appeal available only 'at the discretion of the superintendent'; no record-keeping requirement for searches

Law. Goss v. Lopez, 419 U.S. 565 (1975) sets the minimum process for suspensions of ten days or less and signals that longer exclusions need more. New Jersey v. T.L.O., 469 U.S. 325 (1985) sets the reasonable-suspicion standard for searches by school officials; Safford v. Redding, 557 U.S. 364 (2009) bars underwear searches absent suspicion of danger or concealment there; Vernonia (1995) and Earls (2002) permit suspicionless drug testing of students in athletics and competitive extracurriculars. IDEA (34 CFR 300.530 to 300.536) requires a manifestation determination within ten school days of a decision to change the placement of a student with a disability for discipline, and Section 504 imposes a parallel duty. FERPA governs access to discipline records and the student's right to review them for a hearing. Many state statutes prescribe expulsion hearing procedures and timelines that are more specific than Goss; the clause above is a floor that state law may raise. Searches by school resource officers or police are judged under ordinary Fourth Amendment standards in some circuits, so policy should state which standard applies when an officer conducts the search.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Out-of-school suspension lowers the suspended student's own math and reading achievement (Philadelphia student panel with instrumental variables), and attending a school with a one standard deviation higher suspension rate raises the chance of adult arrest and incarceration by 15 to 20 percent (North Carolina boundary-change natural experiment). Reforms that ended suspension for minor nonviolent conduct raised achievement and attendance in New York City and Chicago, while reforms that cut suspensions without a replacement behavior system lowered achievement and raised absences for most students in Los Angeles and Philadelphia. (Suspended student: lower math and reading scores; strict school: +15 to 20% adult arrest and incarceration; NYC reform: +0.05 SD math over three years; LAUSD reform: lower scores and GPA, more absences for most students; Chicago reform: about +1 day attendance per year.) ^{299,300,301,302,303,304,305,306,307,308,309,310,311,312,313,314,315}

Where the evidence is mixed. Peer effects cut both ways. Pope and Zuo (2023) find Los Angeles' near-elimination of suspensions lowered test scores and GPAs and raised absences and teacher turnover for most students, with gains only for the highest-risk students. Lacoe and Steinberg (2018) find Philadelphia's 2012 reform did not cut total suspensions, and serious misconduct, truancy and achievement all moved the wrong way district-wide; the companion paper shows peers were unharmed only in schools that fully implemented the reform. Hwang and Domina (2021) find classmates' math scores rise when disruptive peers are suspended in a low-suspension district, and Carrell, Hoekstra and Kuka (2018) show one disruptive elementary classmate lowers peers' adult earnings 3 percent. Sorensen et al. (2022) find harsher principals do reduce reported minor misconduct, at the cost of more juvenile complaints and lower graduation for everyone in the building. The 2025 Campbell meta-analysis finds school-based programs reduce exclusion only modestly, and out-of-school suspension is the least responsive outcome. The lesson across studies is that removing suspension without an alternative system harms peers; pairing it with a tiered behavior system (see pbis-restorative) does not.

Standard. Out-of-school suspension is limited to conduct that endangers persons or seriously disrupts instruction after documented alternatives, is barred for defiance, dress code, tardiness and attendance offenses, is capped in length, continues instruction during exclusion, and is reported annually with disparity ratios by building.

Indicator. Out-of-school suspensions per 100 students and instructional days lost, by building and by race, disability, sex and English-learner status (district student information system, reconciled to the Civil Rights Data Collection); share of suspensions for subjective or nonviolent categories such as defiance, disrespect and dress code; risk ratios for Black students and students with disabilities against white and non-disabled peers; attendance and achievement of non-suspended peers in buildings that reduce suspensions.

Red flags in real manuals. mandatory or automatic suspension for a first offense other than weapons or serious violence; suspension listed as a consequence for defiance, disrespect, insubordination, profanity or dress code; 'zero tolerance' language applied beyond the federal firearm requirement; no cap on the length of a suspension or on cumulative days per year; no requirement to provide assignments or instruction during suspension; no requirement to report suspensions by building or by student group; 'the principal may suspend for any conduct the principal deems disruptive'

Law. Goss v. Lopez, 419 U.S. 565 (1975) requires notice and an opportunity to respond before any suspension of ten days or less, with immediate removal allowed only for danger or ongoing disruption and the hearing to follow as soon as practicable; longer exclusions require more formal process. The Gun-Free Schools Act (20 U.S.C. 7961) requires a one-year expulsion for bringing a firearm to school, modifiable case by case by the superintendent. IDEA (34 CFR 300.530) requires a manifestation determination before removals beyond ten school days for students with disabilities. Several states now restrict suspension for minor conduct by statute, for example California Education Code 48900(k) bars 'willful defiance' suspensions in grades K-8 and, through 2029, grades 9-12, and Illinois Public Act 99-0456 (2015) requires exhaustion of alternatives before exclusion.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. School-wide PBIS has the strongest evidence of any discipline framework: a meta-analysis of 29 group studies including seven RCTs finds small-to-medium reductions in discipline referrals and gains in achievement, and the five-year Maryland RCT found children in PBIS schools 33 percent less likely to receive an office referral with fewer suspensions. Restorative practices are promising but mixed: the Pittsburgh RCT cut days lost to suspension 36 percent against 18 percent in controls and narrowed racial gaps, but did not improve achievement and worsened it in grades 6 to 8, while the Maine middle school RCT found no significant intent-to-treat effects. (PBIS: 33% lower odds of an office referral (Maryland RCT); meta-analytic effects small to medium on discipline and achievement. Restorative: suspension days -36% vs -18% control (Pittsburgh RCT), academic harm in grades 6 to 8, null in Maine.) ^{316,317,318,319,320,321,322,323,324,325,326,327}

Where the evidence is mixed. PBIS effects depend on implementation fidelity and are modest; most RCT evidence is from elementary schools, and a 2026 meta-analysis limited to children through grade 5 found no statistically significant pooled effect, with hints of larger effects in pre-K and kindergarten. PBIS is a framework rather than a curriculum, so naming it in policy without the tiered systems, data review and coaching does not reproduce the trial results. For restorative practices, the two RCTs disagree: Pittsburgh showed suspension reductions in elementary grades but no improvement in middle grades, for male students, for students with IEPs or for violent incidents, no reduction in arrests and lower middle-grade achievement; Maine showed no intent-to-treat effect, and reviews (Darling-Hammond et al. 2020; Lodi et al. 2021) rest mostly on non-experimental studies. Trauma-informed whole-school approaches had no eligible rigorous evidence as of a 2019 Campbell review. The honest position is PBIS as the spine, restorative practices as a complement for repair after incidents, and two or more years of coaching for either.

Standard. Every building operates a school-wide tiered behavior support system named in policy, with a leadership team, taught expectations, data review at least monthly, Tier 2 and Tier 3 interventions, annual fidelity assessment with a recognized instrument, and public reporting of referrals, exclusions and fidelity by building; restorative practices are adopted as a complement with the same reporting.

Indicator. Office discipline referrals and out-of-school suspensions per 100 students by building and student group (student information system); annual fidelity score on a recognized instrument (Tiered Fidelity Inventory, with 70 percent or higher at Tier 1 as the benchmark); share of buildings with functioning Tier 2 and Tier 3 teams; student and teacher climate survey results; referral risk ratios for Black students and students with disabilities.

Red flags in real manuals. no tiered behavior framework named anywhere in the discipline policy; a 'discipline matrix' of consequences with no mention of prevention, teaching expectations or Tier 2 and 3 supports; 'PBIS' named without a fidelity measure, a team, data review or coaching; restorative practices named as the sole discipline approach with no tiered system behind them; 'schools are encouraged to' language with no requirement or reporting; no reporting of referrals or exclusions by building or student group

Law. IDEA regulations require that when a child's behavior impedes learning the IEP team consider, and where necessary include, positive behavioral interventions and supports (34 CFR 300.324(a)(2)(i)); the Department's August 1, 2016 Dear Colleague Letter explains that duty. ESSA (20 U.S.C. 7118) allows Title IV-A funds for school-wide positive behavioral interventions and supports. Some states require tiered supports or restorative alternatives before exclusion, for example Illinois Public Act 99-0456 (2015), and California Education Code 48900.5 requires that other means of correction be documented before suspension for most offenses.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Schools that use the Comprehensive School Threat Assessment Guidelines (CSTAG, formerly the Virginia guidelines) resolve student threats with far less exclusion: in a randomized trial, threatened students in CSTAG schools were about one third as likely to receive a long-term suspension (OR 0.35) and four times as likely to receive counseling, and statewide Virginia comparisons show lower odds of suspension (OR 0.59) and law-enforcement action (OR 0.47) than under generic threat procedures, with almost no threats carried out and no racial disparity in outcomes among assessed students. (Long-term suspension OR 0.35 and alternative placement OR 0.13 (RCT); suspension OR 0.59, police action OR 0.47 versus generic approaches (statewide); about 1 in 100 threats attempted; no Black-White gap in suspension among assessed students.) ^{328,329,330,331,332,333,334,335,336}

Where the evidence is mixed. Nearly all outcome research comes from the model's developer (Cornell and colleagues at the University of Virginia), who has a financial interest in CSTAG training. The single RCT is small (201 students in 40 schools) and measures disciplinary responses, not whether violence was prevented; no study can show that threat assessment reduces attacks because attacks are rare. In Florida's statewide data, Black students are over-referred for threat assessment relative to enrollment even though outcomes after referral show smaller disparities than general discipline, so parity after referral does not mean parity in who gets referred. Students receiving special education were about a third of Virginia cases, well above their enrollment share. Teams that are poorly trained reclassify transient threats as substantive and the exclusion benefit disappears; fidelity predicted lower suspension in the RCT.

Standard. The policy shall establish a multidisciplinary threat assessment team at every school using a named, evidence-based model (CSTAG or equivalent), require initial and refresher training for every team member, require that the team's default response is support and supervision rather than exclusion, and require annual reporting of case counts, outcomes and demographic referral rates to the Board.

Indicator. Number of threat assessments per 1,000 students by school and demographic group; share classified transient versus substantive; share of assessed students who received out-of-school suspension, expulsion, alternative placement, law-enforcement referral, or counseling and a support plan; referral rate by race, disability status and gender compared with enrollment; team training completion; number of assessed threats later carried out.

Red flags in real manuals. threat assessment described as an investigation whose outcome is suspension or expulsion; any threat treated as automatic exclusion ('zero tolerance for threats'); team consists only of administrators or only of law enforcement; no training requirement or no named model; no distinction between transient and substantive threats; police notification required for every threat regardless of classification; no reporting of referral rates by demographic group

Law. Virginia Code 22.1-79.4 (2013) was the first statewide mandate for threat assessment teams in every public school. Florida Statutes 1006.07(7) (Marjory Stoneman Douglas High School Public Safety Act, 2018) requires a threat assessment team in every school using a state-standardized instrument. Texas Education Code 37.115 (SB 11, 2019) requires safe and supportive school teams for threat assessment. Roughly a dozen other states have since mandated or funded teams. FERPA permits sharing education records with the team under the health-or-safety emergency and school-official exceptions; the IDEA manifestation determination still applies to any disciplinary change of placement for a student with a disability.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Hazing is common in high school (48 percent of students in groups report hazing activities, 17 percent of grade 6 to 12 athletes in one survey, and only 14 percent of hazed students call it hazing), but no controlled study shows that any hazing policy or program reduces hazing; the evidence supports clear definitions, mandatory reporting and consent-does-not-matter language as a standard. For concussion, state remove-from-play and return-to-play laws produced a rise in reported concussions (better identification) followed by falling recurrent-concussion rates about 2.6 years after passage, and mandated preseason heat-acclimatization rules cut exertional heat illness in football by roughly half. (Hazing: prevalence 17% (athletes) to 48% (all group members); no measured program effect. Concussion laws: reported concussions up (reporting 47% to 71%), recurrent concussion rate declining 2.6 years post-law; incident rate trend reversed from +1.85% to -1.08% per month. Heat rules: exertional heat illness IRR about 0.45 (95% CI 0.23 to 0.87).) ^{337,338,339,340,341,342,343,344,345,346,347,348,349,350}

Where the evidence is mixed. The hazing half of this domain is tier D: prevalence surveys are old (Alfred 2000, Hazing in View 2008, Gershel 2003) and self-reported, and the only published high school prevention evaluation is a two-school pilot that measured knowledge, not behavior. For concussion laws, all evaluations are ecological interrupted time series; the initial rise in reported concussions is almost certainly detection, and the later decline in new concussions could partly reflect reporting fatigue rather than prevention. Laws do not include prevention components (contact limits), and recurrent-concussion declines were larger but not statistically significant in states that named the clearing provider. Emergency action plans, athletic trainer coverage and sudden cardiac arrest protocols rest on professional-body standards (NATA, Korey Stringer Institute) with policy rankings rather than outcome trials.

Standard. The policy shall define hazing by listed behaviors regardless of consent, require every coach, sponsor and student participant to receive annual training, require coaches to report suspected hazing to the principal within 24 hours, and require immediate removal from play after a suspected concussion with written clearance by a licensed health care provider, a preseason heat-acclimatization schedule, and a venue-specific emergency action plan rehearsed annually.

Indicator. Hazing: number of hazing reports and substantiated cases per season by team or group; climate-survey item on initiation experiences; share of coaches and activity sponsors with current hazing training. Athletics safety: concussions recorded in the athletic trainer injury log per 1,000 athlete-exposures, share with documented written clearance before return, days to return; exertional heat illness and cardiac events per season; share of venues with a rehearsed emergency action plan and an AED within three minutes; athletic trainer coverage by sport.

Red flags in real manuals. hazing addressed only as 'initiation activities are discouraged' or only within a code-of-conduct list; consent or voluntary participation treated as a defense; no training requirement for coaches or sponsors; return to play after concussion permitted on coach or parent judgment; no mention of return-to-learn or academic accommodations after concussion; no heat-acclimatization schedule or no emergency action plan requirement; no athletic trainer or no AED requirement for competitions

Law. All 50 states and the District of Columbia enacted youth sports concussion laws between 2009 (Washington's Lystedt Law) and 2014, nearly all with the three core components: education of athletes, parents and coaches; immediate removal from play; and written clearance by a licensed provider before return. Forty-four states have criminal anti-hazing statutes (Alaska, Hawaii, Montana, New Mexico, South Dakota and Wyoming do not); many apply to secondary schools and some require district policies. The federal Stop Campus Hazing Act (December 2024) applies only to higher education. Several state athletic associations mandate heat-acclimatization periods and emergency action plans; the Korey Stringer Institute ranks state high school policies at https://koreystringer.institute.uconn.edu/state-high-school-sports-safety-policies/.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Announced, procedure-only lockdown drills paired with training raise students' and staff's sense of preparedness and correct execution (classroom procedural accuracy up 27 percentage points after training) without raising measured anxiety, and students in one district reported less fear and lower perceived risk after drills. High-intensity simulations with actors, gunfire or deception are a different intervention: a 114-school interrupted time series of 54 million social media posts found anxiety and stress up 42 percent and depression up 39 percent after active shooter drills, and the American Academy of Pediatrics calls for ending children's participation in high-intensity drills and prohibiting deception. (Procedure-only drills with training: preparedness up, lockdown accuracy +27 points, state anxiety not higher (one study lower) post-drill; simulation-style drills: anxiety and stress +42%, depression +39% in community social media after drills; 60% of youth in a national text poll say drills make them feel scared and hopeless.) ^{351,352,353,354,355,356,357,358,359,360,361,362,363}

Where the evidence is mixed. No randomized or controlled comparison of drill types exists; the preparedness and anxiety findings come from pre-post surveys without control groups, mostly from one research team (Schildkraut and Nickerson) in one large urban district and one rural high school. The Everytown and Georgia Tech social media study cannot separate the effect of the drill from news coverage or the type of drill, and was funded by an advocacy organization. Avoidance behaviors increased after drills even as fear fell. No study shows that drills reduce deaths; the real-event analysis of lockdowns is descriptive. Younger children, students with disabilities and students with trauma histories are under-studied, and the effects of monthly drill frequencies mandated in some states are unknown.

Standard. The policy shall require lockdown drills to be announced in advance to staff, students and families, to teach the procedure without simulated gunfire, actors, props or deception, to be developmentally adjusted and accommodate students with disabilities and trauma histories, to include a debrief, and to be observed and reported annually; student participation in any high-intensity simulation shall be prohibited.

Indicator. Number and type of drills held per school per year (procedure-only versus simulation, announced versus unannounced); share of classrooms executing the lockdown procedure correctly in observed drills; student and staff perceived-preparedness and anxiety items on the annual climate survey before and after drills; counseling referrals in the week after each drill; parent notification compliance.

Red flags in real manuals. unannounced lockdown drills or drills 'conducted as if real'; use of blanks, airsoft, simunition, actors, masks or fake blood with students present; students instructed to attack or throw objects at an intruder (ALICE-style counter training) in elementary grades; no parent notification requirement; no accommodation for students with disabilities or trauma histories; no debrief or follow-up; no record of drills reported to the Board

Law. Most states mandate some number of lockdown or security drills per year (for example New York Education Law 807 requires four lockdown drills and eight evacuation drills; New Jersey requires one security drill per month). New York's 2024 regulation amendments require advance notice to parents and prohibit drills that mimic an actual shooting with actors or props. Florida Statutes 1006.07(4) requires active assailant drills at the same frequency as fire drills and, since 2023, requires age-appropriate, trauma-informed conduct and parent notification. Several states (for example Illinois, Oregon, Texas) have adopted trauma-informed drill requirements since 2021; districts should check their state's current drill statute before adopting a frequency.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. The strongest causal studies agree that placing police in schools increases suspensions, expulsions, police referrals and arrests, with the largest increases for Black students, male students and students with disabilities; a Texas quasi-experiment ties federal school-police grants to a 6 percent rise in middle school discipline, a 2.5 percent drop in high school graduation and a 4 percent drop in college enrollment. Safety effects are mixed: a national regression discontinuity finds SROs reduce some violence but not gun incidents, while a matched COPS-grant study finds no safety gain and more recorded drug and weapon offenses. (Discipline +6% (Texas); graduation -2.5%, college enrollment -4%; exclusionary discipline rate ratio 1.21 (meta-analysis); suspensions, referrals and arrests up with largest effects for Black students and students with disabilities; some non-gun violence down (RD), gun incidents unchanged.) ^{364,365,366,367,368,369,370,371,372,373}

Where the evidence is mixed. Sorensen et al. 2023 find real reductions in some forms of in-school violence, and Owens 2017 finds SROs improve measured safety and help police make drug arrests, so the claim that SROs do nothing for safety is not supported. Much of the recorded-offense increase may be detection and reporting rather than more misbehavior. Several studies identify effects off federal COPS grants, which may not generalize to districts that fund officers locally. The 2023 Campbell review found no true experiments and high heterogeneity, with the direction of effects depending on how officers are used; surveys show many students and staff feel safer with an officer present. Officer role (law enforcement only versus mixed mentoring role) changes recording and referral behavior, which is why the MOU, not the headcount, is the policy lever.

Standard. If the district places law-enforcement officers in schools, the policy shall require a written memorandum of understanding that bars officer involvement in routine discipline, defines the narrow conditions for arrest or referral, requires youth-specific training, and requires annual public reporting of arrests, referrals and use of force by school and demographic group; the Board shall review the data before renewing any officer assignment.

Indicator. Annual counts and rates per 1,000 students of school-based arrests, citations, law-enforcement referrals and use of force, by school and by race, disability status and gender (the Civil Rights Data Collection reports referrals and arrests); share of referrals for non-criminal code-of-conduct matters; out-of-school suspension rate by school; student and staff perceived-safety items on the annual climate survey; complaints against officers.

Red flags in real manuals. officers authorized to enforce the code of conduct or 'assist with discipline'; no memorandum of understanding, or an MOU not reviewed by the Board; no restriction on questioning or searching students without an administrator present; no reporting of arrests or referrals by demographic group; officer stationed in the discipline office; mandatory police notification for low-level offenses such as disorderly conduct or 'disruption'; no training requirement beyond state peace-officer certification

Law. No federal mandate. Florida Statutes 1006.12 (2018) requires a safe-school officer at every public school. Texas SB 11 (2019) funds officers and mandates safety committees. Several states and large districts (for example Illinois Public Act 101-0282 and California AB 2316 proposals, and Minneapolis, Denver, Oakland board actions in 2020) regulate or removed officer assignments. The Civil Rights Data Collection requires every district to report school-related arrests and law-enforcement referrals, which is the baseline data source. The Fourth Amendment reasonable-suspicion standard for school searches (New Jersey v. T.L.O.) does not automatically extend to searches directed by police officers, and Miranda warnings apply to custodial questioning by officers (J.D.B. v. North Carolina, 2011).

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Heat cuts learning: across 10 million PSAT retakers, each 1 degree F hotter school year reduced that year's learning by about 1 percent in schools without air conditioning, and air conditioning offset nearly all of it; hot school days fall hardest on Black and Hispanic students and explain roughly 5 percent of the racial achievement gap. Poor ventilation costs attendance and scores: in 162 California classrooms each additional 1 L/s per person of outdoor air cut illness absence 1.6 percent, a 140-classroom study found math scores rose 0.5 percent per L/s-person up to the 7.1 L/s-person standard and a further 0.5 percent per degree C cooler within 20 to 25 C, and a double-blind crossover experiment found 3 to 7 percent more correct answers when outdoor air rose from 1.7 to 6.6 L/s-person. (Learning -1% per +1 F school-year temperature without AC, mostly offset by AC; illness absence -1.6% per +1 L/s-person ventilation; math +0.5% per L/s-person and +0.5% per 1 C cooler (20 to 25 C); test speed +12% and errors -2% when classroom CO2 falls from 2100 to 900 ppm (pooled model); most US classrooms ventilate below the ASHRAE 62.1 rate.) ^{374,375,376,377,378,379,380,381,382,383,384,385,386,387,388,389}

Where the evidence is mixed. The heat evidence is tier A (several large natural experiments agree); the ventilation evidence is weaker: Mendell and Haverinen-Shaughnessy are prospective observational studies, Petersen is a small crossover with four classrooms, and Wargocki's 12 percent figure is a pooled model prediction rather than a measured effect. Graff Zivin, Hsiang and Neidell found heat lowers math performance on the day of the test but found no long-run effect of early-life temperature exposure. Stafford's Texas study found renovations that fixed indoor air problems raised test scores, but renovations change many things at once. Air conditioning and ventilation upgrades are capital-intensive (an elementary school HVAC replacement runs into the millions) and raise energy costs, and ASHRAE 241 is a voluntary standard. CO2 is an indicator of ventilation, not a pollutant at classroom levels.

Standard. Every occupied classroom is kept between 68 and 78 F during instruction, every classroom is ventilated at or above the ASHRAE 62.1 outdoor air rate with MERV-13 filtration and continuous CO2 monitoring held below 1,000 ppm, and the district has a written heat response plan with a trigger temperature, cooling priorities and a capital plan to air-condition every classroom.

Indicator. Share of classrooms with mechanical cooling and share of instructional hours above 78 F, from building management system or classroom data loggers; classroom CO2 (share of occupied hours above 1,000 ppm) from continuous monitors; measured outdoor air rate per person at commissioning; illness absence by building; number of heat-related early dismissals.

Red flags in real manuals. thermostats shall be set no lower than 78 degrees in cooling season (energy policy overriding learning); windows shall remain closed at all times (security policy with no ventilation substitute); portable air cleaners and fans are prohibited in classrooms; no heat response plan or trigger temperature; ventilation addressed only as a maintenance item without a rate or CO2 standard; energy savings policy with no classroom temperature range

Law. No federal classroom temperature or ventilation standard; ASHRAE 62.1 is adopted through state building and mechanical codes for new construction only, and ASHRAE 241 (2023) is voluntary. California Title 24 requires CO2 sensors in classrooms using demand-controlled ventilation and sets a 7.1 L/s-person school ventilation rate. New York's 2024 school heat law (effective September 2025) requires relief measures when a classroom reaches 82 F and bars occupancy at 88 F or above. OSHA has no indoor temperature standard for schools. ESSER funds (through 2024) paid for many HVAC upgrades; the EPA Indoor Air Quality Tools for Schools program is voluntary guidance.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Lead harms learning at every measured level: in Rhode Island, each 1 ug/dL lower average preschool blood lead cut the chance of scoring substantially below proficient in third-grade reading by about 1 percentage point on a 12 percent base, Swedish and North Carolina natural experiments find the same for school grades, crime and suspensions, and Flint's water crisis lowered math achievement and raised special education classification district-wide. Lead in school water is common: in seven state testing programs 13 to 81 percent of tested schools had at least one outlet above 5 ppb, and 37 percent of US districts that tested found elevated lead, but no study links school-water lead itself to achievement, so testing policy rests on the blood-lead evidence plus prevalence data. (Below-proficient reading -0.96 pct points per 1 ug/dL lower blood lead (base 12%); Flint: math achievement down and special education classification up across the district; 37% of testing districts found elevated lead (GAO); 4 of 7 states found more than 20% of schools with an outlet above 5 ppb.) ^{390,391,392,393,394,395,396,397,398,399,400,401,402,403}

Where the evidence is mixed. The causal chain is indirect: the strong studies measure blood lead from all sources (paint, soil, water) and no study isolates school drinking water's contribution to blood lead or achievement, and Flint's effects appeared even in homes with copper service lines, which points to stress and disruption as well as lead. Water is a smaller lead source than paint and dust for most children, and a child drinks only part of the day's water at school. Sampling results depend heavily on protocol (first draw versus flushed, stagnation time, which outlets), so a single clean test does not prove safety and a single high test may reflect one fixture. Remediation costs are real (fixture replacement, filtered bottle fillers, re-testing every few years) and there is no agreed health-based action level for schools: EPA uses 15 ppb for water systems and recommends schools act at 5 ppb or lower, AAP recommends 1 ppb.

Standard. Every outlet used for drinking or cooking in every building is tested for lead with first-draw samples at least every 3 years and after any plumbing work, any outlet above 5 ppb is taken out of service within 24 hours until fixed or replaced, results are posted publicly within 30 days, and every building has filtered drinking stations.

Indicator. Share of drinking and cooking outlets tested within the last 3 years, and share above 5 ppb, from the district's published results; days between a result above the action level and outlet shutoff; number of filtered bottle-filling stations per building; share of families notified within 30 days; blood lead screening rates from the county health department where available.

Red flags in real manuals. testing shall be conducted as required by the local water utility only; action level set at 15 or 20 parts per billion; flushing shall be used as the sole remediation; results provided to parents upon request only; testing limited to a sample of outlets per building; no retest interval or no requirement to test after plumbing work; no provision for lead in water at all

Law. Schools on municipal water are not regulated under the Safe Drinking Water Act; the 2021 Lead and Copper Rule Revisions (40 CFR 141.92) require community water systems to offer sampling of five outlets in each elementary school and child care facility over a five-year cycle and on request for secondary schools, and the October 2024 Lead and Copper Rule Improvements continue the schools program with lists of eligible facilities due to states by November 2027. The WIIN Act of 2016 funds a voluntary EPA grant program for school testing and remediation. State laws vary: New York requires testing of every outlet every 3 years with a 5 ppb action level (10 NYCRR Subpart 67-4, amended 2022), New Jersey requires testing every 3 years, Illinois (415 ILCS 5/17.12) requires testing in schools built before 2000, California AB 746 (2017) required water systems to test schools built before 2010, and Michigan's 2023 Filter First law requires filtered stations in every school.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Meta-analyses of experimental studies find that physical activity delivered through physical education improves classroom behavior and academic outcomes, with the gains concentrated in programs that raise the quality and intensity of PE lessons (cognition g 0.38, academic performance g 0.15) rather than in simply adding minutes (academic g 0.09, not significant). Instrumental-variable estimates using state PE mandates show more PE time increases physical activity and lowers obesity among fifth-grade boys without crowding out test scores. SHAPE America's standard is 150 minutes per week in elementary and 225 in secondary school. (Quality-focused PE interventions: cognition +0.38 SD, academic performance +0.15 SD; PA interventions: math +0.21 SD, on-task behavior +0.77 SD; more PE time: no crowd-out of test scores, lower BMI z-score for boys; added minutes alone: academic +0.09 SD (null)) ^{404,405,406,407,408,409,410,411,412,413}

Where the evidence is mixed. Adding PE minutes without changing lesson quality has little measurable academic effect. Packham and Street (2019) used a regression discontinuity on Texas Fitness Now, a 37 million dollar daily-PE mandate for low-income middle schools, and found no overall fitness improvement, no test-score or attendance gain, and some evidence of lower attendance and more disciplinary incidents. Cawley, Meyerhoefer and Newhouse (2007) found binding state PE requirements raised activity but not weight outcomes for high school students. Singh et al. (2019) rated only 11 of 58 PA intervention studies as high quality and called the evidence on overall academic performance inconclusive, while confirming math benefits. Observed PE lessons average well under the 50 percent moderate-to-vigorous activity target, so minutes on a schedule overstate activity delivered.

Standard. Every student receives at least 150 minutes per week of PE in elementary and 225 minutes per week in secondary grades, taught by a certified physical educator, with no waivers for athletics or other courses and no PE time withheld as punishment.

Indicator. Scheduled PE minutes per week by grade and building (master schedule); share of PE sections taught by a certified physical educator; observed share of PE lesson time in moderate-to-vigorous activity from a sampled SOFIT or equivalent observation each year; waiver and substitution counts; fitness assessment results where the state requires them.

Red flags in real manuals. PE waived for athletics, band, JROTC or academic courses; PE minutes defined as a range or 'as scheduling permits'; recess or classroom activity breaks counted as PE; PE taught by classroom teachers without certification; physical activity used as punishment or withheld as punishment; no weekly minute floor stated for any grade span

Law. No federal minimum; the federal Local School Wellness Policy rule (7 CFR 210.31) requires districts in the school meal programs to set physical activity goals. State minute mandates vary: California Education Code 51210 requires 200 minutes per 10 school days in grades 1-6 and 51222 requires 400 minutes per 10 days in grades 7-12; Texas Education Code 28.002(l) requires 30 minutes daily or 135 minutes weekly of structured activity in K-5. Most states require PE in unspecified amounts and allow substitutions; see SHAPE America's Shape of the Nation state profiles.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Daily recess is associated with better classroom behavior and attention: in a national sample of 11,000 third graders, children with at least one daily recess of 15 or more minutes had better teacher-rated class behavior, a within-subject experiment found fourth graders more on task and less fidgety on recess days, and a cluster-randomized trial of a structured-recess program (Playworks) in 27 to 29 schools reduced teacher-reported bullying and exclusion and cut transition time back to learning. The AAP and CDC state that recess is a necessary break and should not be withheld as punishment or to complete work. (Daily recess of 15 or more minutes: better teacher-rated class behavior (n about 11,000); recess days: more on task, less fidgeting, all ADHD children benefited; Playworks RCT: less bullying and exclusion, fewer minutes lost in transition; LiiNK four 15-minute recesses: off-task behavior down, attentional focus up) ^{414,415,416,417,418,419,420,421,422,423}

Where the evidence is mixed. No randomized trial varies the amount of recess itself; the Playworks RCT tests a coached recess program, and its accelerometer and student-survey measures of physical activity showed null or marginal effects even where teacher reports were positive. The LiiNK results come from two private schools with no random assignment. Brez and Sheets (2017) found no clear classroom benefit of recess timing in their small experiment. The large national association (Barros et al. 2009) is cross-sectional and children without recess were disproportionately poor, Black and urban, so the comparison is confounded. Recess competes with instructional minutes and requires supervision staffing.

Standard. Every elementary student receives at least 20 minutes of daily supervised recess, scheduled separately from physical education and lunch, which no staff member may withhold as a disciplinary consequence or to complete academic work.

Indicator. Scheduled daily recess minutes for every elementary building (master schedule); count of recess-withholding incidents logged in the student information system as a disciplinary action or make-up-work consequence; share of K-5 students with at least 20 minutes of recess every day.

Red flags in real manuals. recess may be withheld as a consequence; loss of recess listed among disciplinary consequences; recess may be used to complete missing work; recess counted toward physical education minutes; recess at teacher discretion or as time permits; no minimum daily recess minutes stated

Law. No federal law. Florida Statutes 1003.455 requires 100 minutes per week of unstructured free-play recess for K-5 with at least 20 consecutive minutes daily (traditional public schools only). New Jersey (2018) requires a daily 20 minute recess for K-5. Washington RCW 28A.230.295 (2024-25) requires 30 minutes daily for K-5. Pennsylvania HB 1505 requires 30 minutes daily for K-5 from 2027-28. Several other states, including Arizona, Connecticut, Missouri, Rhode Island and Virginia, set daily recess or activity minimums or bar withholding recess as punishment; check the state code.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. A full-time registered nurse in every school pays for itself: the Massachusetts cost-benefit study found $2.20 in avoided medical costs, teacher time and parent lost wages for every $1 spent. Schools that add a full-time nurse show lower illness absence, with the clearest gains for students with asthma (23 percent fewer days missed with a full-time versus part-time nurse in Toledo; fewer illness absences and emergency visits after San Jose added full-time nurses), but the attendance evidence is quasi-experimental, integrative reviews rate it weak, and no study shows a nurse effect on test scores. (Benefit-cost ratio 2.2 (Massachusetts); students with asthma -23% days absent with full-time vs part-time nurse; illness absence fell in schools adding full-time RNs; teachers in schools without a full-time nurse spent about an hour a day on student health; achievement effects unproven.) ^{424,425,426,427,428,429,430,431,432,433,434}

Where the evidence is mixed. No randomized or natural-experiment study of nurse staffing exists; the two best quasi-experiments (Telljohann 2004, Rodriguez 2013) are small and matched rather than randomized, and Yoder's 2019 review found nurse presence associated with less absence and missed class time but not with achievement. A matched 46-school study of licensed practical nurses (not RNs) found at best negligible attendance gains at $68,228 per school per year (Leach 2022), which suggests the benefit depends on scope of practice and on what the nurse is asked to do. The Massachusetts benefit-cost figure rests on modeled productivity savings, not measured outcomes. NASN itself abandoned the fixed 1:750 ratio in 2020 in favor of a workload and acuity analysis, so there is no single staffing number the research supports.

Standard. Every school building has a licensed registered nurse on site for the full instructional day, every day school is in session, with additional nursing staff assigned from an annual workload and acuity analysis; LPNs and unlicensed health aides work only under RN delegation and supervision.

Indicator. Share of school buildings with a registered nurse on site every full school day, from the district's staffing roster; nurse-to-student ratio and acuity-adjusted workload by building; illness-related absence and chronic absence for students with asthma, diabetes, seizure disorders and food allergies, from the student information system; emergency 911 calls from schools.

Red flags in real manuals. a school nurse may be shared among buildings as funds permit; the health office may be staffed by clerical staff or a health aide in the absence of a nurse; unlicensed personnel may administer medication without delegation by a registered nurse; school nurse services shall be provided to the extent funding is available; nurse services provided by contract on an as-needed basis only; no requirement for a nurse on site during the school day

Law. No federal mandate for school nurses. IDEA and Section 504 require health services needed for an eligible student to access education. State requirements vary widely: Pennsylvania's Public School Code (24 P.S. 14-1402) sets a maximum of 1,500 students per certified school nurse; many states require only access to nursing services. Since the 2014 CMS reversal of the free-care rule, districts may bill Medicaid for school health services to enrolled students, and most states have expanded school-based Medicaid billing since 2023.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Mental health services delivered in school by school staff reduce student mental health problems (Hedges g = 0.39 across 43 controlled trials with 49,941 elementary students; g = 0.76 for targeted services), and placing licensed clinicians in schools raised treatment use 8 percent and cut self-reported suicide attempts about 15 percent across 263 Minnesota schools. Counselors have causal effects: an additional counselor reduces disciplinary incidents and raises achievement (Carrell and Hoekstra), individual counselors move graduation and college attendance as much as teachers do (Mulhern), and one more high school counselor raises four-year college enrollment by up to 10 points in a regression-discontinuity design. Universal suicide prevention programs that teach help-seeking (YAM in the SEYLE trial, SOS) roughly halve new suicide attempts, but the largest Sources of Strength trial found no overall reduction and harm where fidelity was poor. (Services: g = 0.39 overall, 0.76 targeted; in-school clinicians: treatment use +8%, suicide attempts -15%; counselors: fewer disciplinary incidents and higher attainment, effects comparable to teacher effects; YAM: new suicide attempts and severe ideation about -50% at 12 months; Sources of Strength: no overall effect on attempts across 40 schools.) ^{435,436,437,438,439,440,441,442,443,444,445,446,447,448,449,450,451,452}

Where the evidence is mixed. The ASCA 250:1 ratio is a professional benchmark, not a research finding: the 2021 meta-analysis found only a small negative effect of higher ratios with serious design bias in the underlying studies, and ratio evidence does not say which staff or services matter. Universal programs can backfire: the MYRIAD trial (85 UK schools) found no benefit from school mindfulness training, a 1,071-student DBT-skills program (WISE Teens) worsened depression (d = -0.22) and anxiety (d = -0.28), and in the 40-school Sources of Strength trial poorly implemented schools saw more new suicide attempts among 9th graders than controls. The Minnesota suicide-attempt result is self-reported and the suspension result there is weaker. Counselor effects vary widely between individuals (Mulhern), so headcount alone does not guarantee results. Universal screening finds more students with depression but creates a referral load the district must be able to meet.

Standard. Every school meets or has a funded plan to reach 250 students per counselor, 500 per school psychologist and 250 per school social worker; every school has a licensed mental health clinician on site at least weekly through district staff or a community partner; secondary students are screened annually for depression with parental consent and a same-week referral pathway; every secondary school runs an evidence-based suicide prevention program with annual fidelity monitoring and a written postvention protocol.

Indicator. Students per school counselor, school psychologist and school social worker by building (district staffing roster against October 1 enrollment); number and share of students receiving school-based mental health services from a licensed clinician on site; share of students screened and share referred who begin services within 30 days; student survey items on depression, suicidal ideation and attempts (YRBS or state climate survey); suspension rate; suicide prevention program fidelity checks.

Red flags in real manuals. counselors may be assigned test coordination, scheduling, or administrative duties; mental health services shall be provided to the extent that funds are available; students shall be referred to outside providers (with no on-site provision); no suicide prevention policy or one limited to crisis response only; universal screening prohibited without a referral pathway stated; student-to-counselor ratios set above 400:1 or left unstated; school psychologists limited to special education evaluations

Law. No federal ratio mandate. The Bipartisan Safer Communities Act (2022) funded school-based mental health professional pipelines through the Department of Education, and 2023 CMS guidance expanded school Medicaid billing for mental health services. More than 20 states require annual suicide prevention training for school staff (Jason Flatt Act statutes), and several (including California AB 2246 and Texas) require every secondary school to adopt a suicide prevention policy. The Protection of Pupil Rights Amendment and most state laws require parental consent for universal mental health screening. ESSA Title IV-A funds may be used for school-based mental health services.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Children with fewer than 20 minutes seated to eat consume less of their lunch, especially fruit and vegetables: in a randomized crossover trial, 10 minutes of seated time versus 20 cut fruit consumption by 11 percentage points and vegetables by 14 points, and observational cafeteria studies in Massachusetts and California show the same pattern at scale. CDC guidance is at least 20 minutes seated after being served. (10 vs 20 seated minutes: fruit consumption -11 percentage points, vegetables -14 points (RCT); under 20 minutes: entree, milk and vegetable consumption all lower (observational); longer lunch period raises odds of eating fruit 1.4x and vegetables 1.5x) ^{453,454,455,456,457,458}

Where the evidence is mixed. The only randomized trial (Burg et al. 2021) enrolled 38 children at a university summer camp, not a school cafeteria, and found no difference in entree or beverage consumption between 10 and 20 minutes. The school-based studies are cross-sectional and cannot rule out that schools with longer lunches differ in other ways. No study links seated lunch minutes to achievement or attendance; the outcome evidence is dietary intake and plate waste. Longer lunches cost instructional minutes or require staggered periods and more cafeteria supervision.

Standard. Every building schedule provides each student at least 20 minutes seated to eat lunch after being served, and the district measures and reports seated time, not period length.

Indicator. Seated minutes per lunch period by building, measured from the time the last student in line is served to the end of the period (master schedule plus a cafeteria line-time observation each semester); share of buildings meeting 20 seated minutes; plate-waste or production-record data from food service.

Red flags in real manuals. lunch period of 20 minutes or less with no reference to seated time; lunch period length defined only as bell-to-bell; lunch may be shortened for testing, assemblies or discipline; students may be held from lunch to complete work; no building-level reporting of lunch time

Law. No federal minimum. USDA school meal rules set nutrition standards but not seated time. Connecticut General Statutes 10-221o requires a lunch period of at least 20 minutes for full-day students. Louisiana law requires no less than 20 minutes of seated lunch time. California has state guidance recommending 20 minutes seated but no statute. Most states leave lunch length to districts; CDC reports about half of districts neither require nor recommend 20 minutes.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Making meals free for every student (the Community Eligibility Provision or a district-wide universal free lunch) raises school-meal participation and produces small but real gains in test scores and fewer suspensions: the New York City difference-in-differences study found math and ELA gains of 0.03 to 0.08 SD, state CEP rollouts found elementary math gains of about 0.02 to 0.06 SD, and CEP cut suspensions by roughly 17 percent for white elementary boys with negative but imprecise estimates elsewhere. Moving breakfast after the bell reliably raises breakfast participation (from about 43 to 74 percent of students in one 446-school district), but the best causal studies of breakfast-in-classroom find little or no test-score gain from the service change itself. (Universal free lunch: math +0.03 to +0.06 SD, ELA +0.03 to +0.08 SD; participation +5 to +11 pct points; suspensions about -17% for one elementary subgroup; no BMI increase. Breakfast after the bell: participation roughly +30 pct points; test-score effect near zero in NYC and in a 446-school propensity study.) ^{459,460,461,462,463,464,465,466,467,468,469,470}

Where the evidence is mixed. Breakfast-in-classroom raised participation in New York City with no detectable gain in achievement or attendance (Corcoran, Elbel and Schwartz 2016), and the 446-school Anzman-Frasca study found better participation and slightly better attendance but no achievement change; only Imberman and Kugler (2014) found test-score gains from in-class breakfast, and they could not separate nutrition from the timing of the test. The 2024 JAMA Network Open systematic review of CEP rated the attendance evidence low certainty (no or slight improvement in 2 studies) and the suspension and obesity evidence very low certainty. Gordon and Ruffini could not rule out large positive or negative discipline effects for middle and high schools, and the math gains in CEP studies are small (0.02 to 0.06 SD) and concentrated among elementary students and students newly gaining access. Districts with an identified student percentage near the 25 percent floor receive less than full federal reimbursement under CEP and must cover the gap from general funds.

Standard. Every school with an identified student percentage at or above the federal CEP threshold operates under CEP or a state universal-meals program so that no student is charged for breakfast or lunch, and every school serves breakfast after the start of the instructional day (in-classroom, grab-and-go, or second-chance) so that breakfast participation reaches at least 70 percent of lunch participation.

Indicator. Average daily participation in breakfast and lunch as a share of enrollment, by school, from the district's monthly USDA reimbursement claims; share of eligible schools operating under CEP; chronic absence and out-of-school suspension rates from the state report card, compared across years of adoption.

Red flags in real manuals. students with unpaid meal balances shall be served an alternate meal; breakfast shall be served before the start of the school day only; the district may participate in the Community Eligibility Provision at the discretion of the Superintendent; students may be denied a meal until the balance is paid; meal debt may be referred to a collection agency; no language on breakfast service or CEP at all in a district with high-poverty schools

Law. CEP is federal (Healthy, Hunger-Free Kids Act of 2010, 42 U.S.C. 1759(a)); USDA lowered the identified student percentage threshold from 40 to 25 percent in 2023 (7 CFR 245.9). Several states fund universal free meals statewide (including California, Maine, Colorado, Minnesota, New Mexico, Vermont, Michigan and Massachusetts). Several states require breakfast after the bell in high-poverty schools (including Texas, Nevada, Colorado, New Mexico, West Virginia and Illinois). USDA requires districts to have a written unpaid meal charge policy (SP 46-2016).

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Screening every student and then delivering eye exams and free glasses on site improves reading: the Baltimore cluster randomized trial of 2,304 students in grades 3 to 7 found a 0.09 SD reading gain after one year, with gains of 0.28 SD for students in the lowest baseline quartile and 0.25 SD for students in special education; rural China trials found math gains of 0.11 to 0.16 SD from free glasses. The gains fade by the second year unless glasses are replaced and wear is followed up, and screening alone, without exams and glasses, does nothing for achievement. (Reading +0.09 SD at 1 year overall, +0.28 SD for lowest-quartile readers (Baltimore RCT); math +0.11 to +0.16 SD (China cluster RCTs); FCAT pass rate +2 pct points averaged over all students (Florida RCT); no sustained effect at 2 years without follow-up.) ^{471,472,473,474,475,476,477}

Where the evidence is mixed. Effects fade: Baltimore found no sustained impact at two years and no effect on the state PARCC test, and the Florida trial also reports fade-out. Screening without a guaranteed exam and glasses did not raise achievement in Florida, and the 2018 Cochrane review found only low-certainty evidence that school screening programs by themselves improve outcomes. Compliance is the binding constraint: many students stop wearing or lose their glasses within months, and the Baltimore and Los Angeles programs needed replacement pairs and classroom follow-up. Math effects in the US trials were small (0.03 SD in Baltimore elementary grades) and the Los Angeles study was observational.

Standard. Every student is screened for distance and near vision at school entry and at least every two years thereafter (and whenever a teacher refers), every student who fails is guaranteed a comprehensive eye exam and glasses at no cost to the family, delivered on site or through a partner within 90 days, with a replacement pair and a documented wear check each semester.

Indicator. Share of students screened each year by grade; share of students who fail screening who receive a comprehensive eye exam within 60 days and glasses within 90 days (from the school nurse's referral log); share still wearing glasses at a 3-month classroom check; reading growth of students who received glasses, from the district's benchmark assessment.

Red flags in real manuals. vision screening may be conducted as resources allow; parents shall be notified of screening failures (with no provision for exams or glasses); screening is limited to distance acuity only; the district assumes no responsibility for follow-up care; screening frequency longer than every two years; no named role responsible for referral follow-up

Law. Most states require school vision screening at specified grades, with varying frequency and methods; Illinois requires a comprehensive eye examination by an optometrist or ophthalmologist for kindergarten entry (105 ILCS 5/27-8.1). Medicaid EPSDT covers eye exams and glasses for enrolled children, so a district program can bill or refer for most low-income students. The USPSTF recommends screening children ages 3 to 5 for amblyopia and finds insufficient evidence for or against screening older children, which is about health screening, not the achievement effect of correction.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Large randomized trials show that plain, personalized letters and text messages telling families how many days their child has missed, and why it matters, reduce chronic absence by about 10 to 15 percent (2 to 7 percentage points) at a cost of roughly $6 to $9 per student or $5 to $10 per added day of attendance. Supportive, simply worded messages outperform legalistic ones, and automated alerts about missed classes and assignments also cut course failure in secondary school. (Mailings: chronic absence -10% (K-12) to -15% (K-5); texts: -2 to -7 points in grades 3-5, -20% in Head Start; supportive truancy notice: 40% more effective than standard notice; cost $5 to $10 per added day) ^{478,479,480,481,482,483,484,485,486,487,488,489}

Where the evidence is mixed. Effects on attendance are real but modest: a 10 percent reduction in chronic absence leaves nine in ten chronically absent students still absent, and the adaptive-text trial found no achievement gain after one year. Message content matters less than expected (benefit-framed and consequence-framed texts performed alike), while delivery details matter a lot: an opt-in default for an alert platform produced near-zero take-up compared with opt-out (Bergman, Lasky-Fink and Rogers 2020). Nudges that worked in district trials have failed when scaled by states or national organizations in another domain (FAFSA completion), so district-level ownership and data linkage appear to be preconditions. Check & Connect mentoring produced larger per-student gains than letters but at roughly 50 times the cost per added day, so messaging is a complement to, not a substitute for, case management for the most absent students.

Standard. Policy requires the district to send every family a plain-language, personalized absence update (days missed, why it matters, whom to call) at least when a student reaches 5 and 10 percent absence and again at each term, by mail and text in the family's preferred language, with supportive rather than legalistic wording, and to maintain current family contact information; legal truancy notices required by state law must use the same plain-language format.

Indicator. Share of enrolled students whose family received at least one personalized absence message (days missed to date, in the family's preferred language) within 10 school days of reaching 5 percent absence, from the student information system messaging log; chronic absence rate by school and subgroup; and family contact information completeness (percent of students with a working phone number or address on file).

Red flags in real manuals. absence letters sent only after the statutory truancy threshold, with no earlier personalized notice; notices that cite statute, fines or court action as the opening message; family communication only in English, or translation 'where feasible' with no named responsibility; parent portal or alert system that requires families to opt in with no default enrollment; no requirement to verify family phone numbers and addresses each year; messages that give a school-wide attendance exhortation but not the child's own days missed

Law. Most state compulsory attendance laws require written notice to parents after a set number of unexcused absences (for example California Education Code 48260.5, Texas Education Code 25.095); these notices are the legally required floor and the evidence favors rewriting them in plain, supportive language rather than adding a separate system. Title VI of the Civil Rights Act and the Equal Educational Opportunities Act require that notices reach limited-English families in a language they understand. The TCPA permits informational school texts to numbers families provide.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Quasi-experimental evaluations of the two largest community school systems find consistent gains in attendance and behavior and smaller, slower gains in achievement: New York City's 2014 initiative cut chronic absence by 7 to 8 percentage points, raised graduation 4.9 points over three years and credits earned by 1.3 per year, reduced disciplinary incidents in elementary and middle schools, and raised grade 3-8 math by 0.13 SD only in year three with no ELA effect; California's 2022 grantee cohort cut chronic absence about 30 percent and suspensions about 15 percent relative to matched schools with math +0.06 SD and ELA +0.05 SD in the first year. Longitudinal studies of City Connects, the best-studied integrated-support model, show higher middle-school test scores and roughly half the odds of high school dropout for students served in elementary school. (NYC: chronic absence -7.3 points (K-8) and -8.3 points (HS), graduation +4.9 points, credits +1.3/year, math +0.13 SD (year 3 only), ELA null; California: chronic absence -30%, suspensions -15%, math +0.06 SD, ELA +0.05 SD; City Connects: dropout 9.2% vs 16.6%; Michigan caseworkers: chronic absence -8%) ^{490,491,492,493,494,495,496,497,498,499}

Where the evidence is mixed. No randomized trial of the whole-school community schools model exists; the NYC and California estimates rest on matched-comparison designs that cannot rule out selection into the initiative, and the California window is one year. The one RCT of a core component, Communities In Schools case management (24 secondary schools, two years), found no effect on attendance, achievement, progress or behavior despite higher service receipt, improving only attitudes and relationships; MDRC's quasi-experimental study of the whole-school CIS model found attendance gains in elementary but not secondary schools. Heers et al. (2016), reviewing the literature, concluded that causal evidence for community schools as such is lacking and that extracurricular activities do not appear to raise achievement. Achievement effects where found are small (0.05 to 0.13 SD) and arrive after two to three years; ELA effects are repeatedly null. Costs run from a few hundred dollars per student for a coordinator to several thousand with wraparound services, and the Michigan material-needs program shows effects concentrated in schools with higher dosage and, uncomfortably, in wealthier and less diverse schools.

Standard. Policy defines the community school strategy by its four pillars (integrated student supports, expanded learning time, family and community engagement, collaborative leadership), requires each designated school to have a funded full-time coordinator, an annual needs assessment, a universal student support review, and a lead partner agreement, and requires annual reporting of attendance, discipline and achievement against matched comparison schools.

Indicator. For each designated community school: presence of a full-time coordinator and a signed lead-partner agreement; share of students with an identified need who received a service within 30 days (needs assessment and service log); chronic absence, suspension and course-passing rates compared with matched district schools, reported annually; per-pupil cost of the strategy.

Red flags in real manuals. community school designation with no funded coordinator position; wraparound services listed without a needs assessment or a service log; partnerships 'as resources permit' with no written agreement; no outcome reporting against comparison schools; the strategy limited to after-school programs or a single health clinic without integrated student review

Law. ESSA Title IV Part F authorizes the federal Full-Service Community Schools grant program (20 U.S.C. 7275) and Title I Section 1115 allows schoolwide programs to fund integrated supports. California's Community Schools Partnership Program (Education Code 8900 et seq.) has committed over $4 billion since 2021; New York's Foundation Aid includes a community schools set-aside; Maryland's Blueprint funds Concentration of Poverty grants requiring a coordinator and health services. Several states (for example New Mexico and Vermont) have community school statutes defining the four pillars.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. Randomized trials show that specific, frequent, two-way teacher-to-family communication about a child's own effort, attendance and missed work raises engagement and achievement: daily teacher messages increased homework completion 40 percent, weekly one-sentence notes cut course failure 41 percent, automated missed-assignment alerts cut course failures 27 percent, and parent-school meetings in France reduced truancy and discipline sanctions by about 0.15 SD. Meta-analyses of family-school partnership programs find small to moderate positive associations with achievement (r about 0.12 for involvement overall; about 0.3 SD for structured programs), strongest for communication about schoolwork and parental expectations rather than attendance at events. (Teacher-family communication RCTs: homework completion +40%, course failure -27% to -41%, truancy and sanctions -0.15 SD; involvement meta-analyses: r = 0.12 overall, programs about +0.3 SD; achievement effects of information nudges concentrated in grades, not state tests) ^{500,501,502,503,504,505,506,507,508,509,510,511,512,513,514}

Where the evidence is mixed. Most of the 'family engagement' literature is correlational and the strong associations in meta-analyses (Jeynes 2012, Castro 2015) partly reflect which families are already engaged; Hill and Tyson (2009) found school-based involvement (volunteering, attending events) and homework help have the weakest or even negative links to achievement in middle school, so policies that count event attendance measure the wrong thing. The causal trials show effects on grades, homework and course passing but generally not on state test scores (Bergman and Chan 2021, Avvisati et al. 2014), and they involve information about the specific child, delivered by the teacher or an automated system, not generic newsletters or parent nights. Eklund et al. (2022) found the family-school partnership category of attendance interventions had a confidence interval including zero. Take-up collapses under opt-in enrollment, and immigrant and limited-English families report the most barriers, so an unfunded two-way expectation can widen gaps rather than close them.

Standard. Policy sets a measurable district-wide expectation (not only Title I schools) of at least two individualized two-way contacts per family per year about the child's own progress, attendance and missed work, in the family's preferred language, logged in the student information system; requires default enrollment of all families in progress alerts; and requires annual public reporting of contact rates, conference participation and family survey results by school and subgroup.

Indicator. Percent of students whose family received at least two documented, individualized two-way contacts (phone, text exchange, conference or home visit, logged in the student information system with date, mode, language and topic) in the year, by school and subgroup; share of families enrolled by default in the alert system and reachable by verified phone; parent-teacher conference participation rate; annual family survey response rate and results on 'the school tells me how my child is doing'.

Red flags in real manuals. family engagement defined only as volunteering, PTA, fundraising or attendance at school events; a parent and family engagement policy that applies only to Title I schools; communication expectations with no logging, no frequency and no named role; 'parents are encouraged to' language with no obligation on the school; parent portal or alert enrollment that is opt-in only; conferences offered only during the school day; home language communication 'when feasible' or 'as resources allow'

Law. ESEA Section 1116 (20 U.S.C. 6318) requires every Title I district and school to adopt a written parent and family engagement policy developed with parents, a school-parent compact, an annual meeting, and information 'in a format and, to the extent practicable, in a language the parents can understand'; districts receiving more than $500,000 must reserve at least 1 percent for engagement, with 90 percent going to schools. IDEA (34 CFR 300.322 and 300.503) requires parent participation in IEP meetings and notices in the native language. Title VI and the Equal Educational Opportunities Act require meaningful communication with limited-English parents.

Dispute this verdict or submit evidence

Tier D: professional standard or expert consensus

Finding. Federal civil rights law (Title VI, the Equal Educational Opportunities Act and Lau v. Nichols) requires districts to communicate with limited-English parents in a language they understand, and IDEA requires native-language notices and interpreters for special education. Nationally representative data show that immigrant parents report the most barriers to school involvement and that English ability predicts how involved they are; the largest communication RCT that improved attendance (131,312 students) delivered its notices in five languages, and a vendor-funded matched study of a translated two-way messaging platform reports attendance gains concentrated among English learner and Latino families. (Compliance obligation; involvement gap by English proficiency documented in ECLS-K and PISA; translated two-way messaging platform (vendor study): absence -24% in Tulsa, larger gains for ELs; no causal estimate of translation itself) ^{515,516,517,518,519,520,521,522,523,524,525,526,527}

Where the evidence is mixed. No randomized or quasi-experimental study isolates the effect of translating or interpreting school communication on student outcomes; the trials that used home-language messages did not report results by language, and the strongest platform evidence (TalkingPoints) is produced by the vendor from matched comparisons. Machine translation can carry errors in high-stakes documents (discipline notices, IEPs, consent forms), and Cheatham (2011) documents interpretation in special education that fails IDEA's participation and consent standards even when an interpreter is present. Federal enforcement has weakened: Executive Order 14224 (March 2025) revoked EO 13166, the Department of Justice rescinded its LEP guidance in July 2025, and the Department of Education rescinded the January 2015 Dear Colleague letter in August 2025; the statutes and Lau remain binding but complaint-driven enforcement is less predictable, so the obligation increasingly rests on district policy and state law.

Standard. Policy identifies each family's preferred language at enrollment, translates all vital documents into every language spoken by 5 percent or 1,000 families (whichever is less) and provides free qualified interpretation on request for any language, prohibits reliance on children or untrained staff as interpreters, names a district language access coordinator, and reports delivery rates by language annually.

Indicator. Share of families with a recorded home language other than English who received each required notice category (enrollment, discipline, special education, attendance, health, emergency) in their language, from the student information system; interpreter requests fulfilled within the policy timeline; conference and IEP meeting participation rates for limited-English families compared with English-speaking families; family survey results by language.

Red flags in real manuals. translation or interpretation provided 'when feasible', 'as resources allow' or 'upon availability'; students or family members used as interpreters for discipline, special education or health matters; written translation limited to a single language regardless of enrollment; no named role responsible for language access and no complaint route; notices deemed delivered when sent in English only; fees charged for translation or interpretation

Law. Title VI (42 U.S.C. 2000d) and the EEOA (20 U.S.C. 1703(f)) remain in force and privately enforceable; Lau v. Nichols (1974) is the controlling precedent. EO 14224 (March 1, 2025) designated English the official language and revoked EO 13166; DOJ rescinded its 2002 LEP guidance (July 14, 2025) and the Department of Education rescinded the January 2015 EL and LEP-parent Dear Colleague letter (August 2025); neither action changes the statutes. ESSA Section 1116(f) requires Title I information in a language parents understand to the extent practicable; IDEA regulations (34 CFR 300.503(c), 300.322(e), 300.9) require native-language notices, interpreters at IEP meetings and consent in the native language. Many states go further: California Education Code 48985 requires translation when 15 percent of students speak a language; Illinois 105 ILCS 5/14-8.02 addresses special education interpretation; Washington's language access chapter (RCW 28A.183) sets interpreter and translation standards for schools. Districts with 5 percent or more families speaking one language are the usual enforcement trigger under former OCR practice.

Dispute this verdict or submit evidence

Tier C: consistent correlational evidence

Finding. Matched-comparison studies of relationship-building teacher home visits (the Parent Teacher Home Visits model and the DC Family Engagement Partnership) consistently find that visited students are less likely to be chronically absent (21 to 24 percent lower odds or fewer absences), somewhat more likely to score proficient in reading, and have fewer disciplinary incidents, with small positive effects on grades (d about 0.12) and math (about 0.11 SD). Schools that visit at least 10 percent of families show school-wide attendance gains that reach non-visited students. (Chronic absence odds -21% (visited students) and -22% (schools visiting 10%+ of families); ELA proficiency odds +35% at school level; attendance +0.35 points and disciplinary incidents 9.3% vs 12.2% in DC; math +0.11 SD; grades d = 0.12 to 0.13) ^{528,529,530,531,532,533,534,535,536}

Where the evidence is mixed. No randomized trial of teacher home visits in US K-12 has been published; every outcome study uses matched or regression-adjusted comparison groups, and families who accept a visit (or teachers who volunteer to make them) likely differ from those who do not, so the estimates are upper bounds. Effects on achievement are small and inconsistent: Sheldon and Jung (2018) found ELA gains at the school level only in some districts and math gains in two of four; the DC study found math but not ELA gains; Wright et al. (2018) found higher positive-behavior points and grades but no difference in discipline points. Most evaluations were commissioned by the program or its funders. Cost is real: visits take 30 to 60 minutes of paid teacher time outside the school day per family plus training, and the model's own fidelity rules (voluntary for both sides, no targeting of 'problem' students, no agenda beyond hopes and dreams) limit who gets visited. The well-known RCT evidence on home visiting (nurse and parent-educator visits to families of infants) is a different intervention and should not be cited for teacher visits.

Standard. Policy authorizes and funds voluntary, paid, relationship-building teacher home visits following a recognized model (visits in pairs, trained staff, family's language, no targeting of 'problem' students), sets a reach target of at least 10 percent of families per school with priority on transition grades and chronically absent students who volunteer, and requires annual reporting of reach and attendance outcomes.

Indicator. Percent of enrolled families receiving at least one relationship-building home visit per year, by school and grade band (program log in the student information system); chronic absence rate of visited versus non-visited students in the same schools, adjusted for prior-year attendance; share of visits conducted in the family's home language with a bilingual visitor or interpreter; teacher participation rate.

Red flags in real manuals. home visits assigned as a consequence for absence or misconduct, or used to serve truancy notices; visits made by a single staff member, or by attendance officers rather than teachers; no compensation for staff time outside the contract day; visits restricted to 'at-risk' lists; home visit used to check for residency or truancy evidence rather than relationship-building; no language provision for visits to limited-English families

Law. No federal or state law requires teacher home visits for general education students; some states fund them through family engagement or attendance grants, and California's Community Schools Partnership Program lists them as an allowable strategy. Residency-verification home visits are a different practice governed by state enrollment law and should not be conflated with this domain. Home visiting for infants and toddlers under MIECHV (42 U.S.C. 711) is a separate health program.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. The Tennessee STAR experiment shows that classes of 13 to 17 students in K-3 raise test scores by about 4 percentile points in the first year and roughly 1 point a year after, with larger gains for Black students and students on free lunch, and the same students were 2.7 percentage points more likely to attend college. Large-scale rollouts deliver less than the experiment because they dilute teacher quality and cost a great deal per student. (STAR: +0.2 SD in K-3, college attendance +2.7 pp (+7.3 pp in the poorest third of schools); Campbell review pooled effect on reading small, on math null; California CSR gains partly offset by inexperienced teachers) ^{537,538,539,540,541,542,543,544,545,546}

Where the evidence is mixed. Hoxby (2000) finds no effect of class size in Connecticut and rules out even modest effects. Leuven and Lokken (2020) find no average long-run effect of class size in Norway. The Campbell systematic review (Filges et al. 2018) finds at best a small effect on reading and a non-significant negative effect on math across ten studies. Jepsen and Rivkin (2009) show that California's statewide reduction to 20 students raised achievement but the rush to hire 25,000 new teachers brought in uncertified and inexperienced staff, which dampened the gains most in high-poverty, high-minority schools. Chetty et al. (2011) find small classes raised college attendance but the effect on earnings at 27 is not statistically significant. Class size reduction is among the most expensive interventions per unit of effect; most reviews conclude money is better spent on teacher quality once classes are below the mid-20s.

Standard. The policy shall set a numeric K-3 class-size cap (not above 20 students, or 22 with a full-time aide), shall apply the cap first in the highest-poverty schools, and shall pair any class-size reduction with a certification requirement for the teachers hired to meet it.

Indicator. Average and maximum section size in K-3 by school each October (student information system); share of K-3 sections above the cap; share of new K-3 teachers without full certification; K-3 reading and math benchmark growth by school.

Red flags in real manuals. class size caps only as averages across the district rather than per section; caps expressed as 'goals' or 'targets' with no reporting; caps that may be waived by the principal without Board notice; no cap for kindergarten or grade 1 while upper grades have one; class size reduction funded by hiring uncertified or emergency-permit teachers; combination classes used to circumvent the cap

Law. Florida's 2002 constitutional amendment (Art. IX s.1) caps core classes at 18 in K-3, 22 in grades 4-8 and 25 in 9-12. California's 1996 K-3 Class Size Reduction program funded classes of 20 or fewer; most states have no statutory cap but set caps or averages through collective bargaining or accreditation standards. Many state caps are averages or include waivers, so a district policy that is stricter than state law is meaningful.

Dispute this verdict or submit evidence

Tier A: meta-analysis or several causal studies agree

Finding. A meta-analysis of 60 causal studies finds that one-on-one instructional coaching improves teaching practice by 0.49 SD and student achievement by 0.18 SD, far more than the one-off workshops that make up most district professional development, which show no achievement effect in large randomized trials. Effects shrink as programs scale (programs over 100 teachers: 0.34 SD practice, 0.10 SD achievement) and depend on coaches observing real lessons and giving specific feedback in cycles over a year. (Coaching: +0.49 SD on instructional practice, +0.18 SD on student achievement (meta-analysis of 60 causal studies); at scale +0.34 and +0.10 SD. Content-focused workshops without coaching: null on achievement in IES RCTs. STEM PD meta-analysis: +0.21 SD when PD is paired with curriculum materials.) ^{547,548,549,550,551,552,553,554,555,556}

Where the evidence is mixed. Not all coaching works: Kraft and Blazar (2017) found large practice gains but no achievement gain in one year; Blazar and Kraft (2015) found the same program succeeded in one cohort and failed in the next when coaching dosage and school conditions changed; Kraft and Hill (2020) found web-based coaching improved observed instruction and student perceptions without a detectable achievement effect. Campbell and Malkus (2011) found math coaches had no effect in their first year and positive effects only in years two and three. Two IES randomized trials of intensive content PD (Garet et al. 2016; Jacob, Hill and Corey 2017) raised teacher knowledge but did not raise student achievement. Coaching is expensive (one coach per 10 to 20 teachers) and scaling dilutes coach quality. Studies are mostly elementary literacy and math with short follow-up.

Standard. District professional development shall be organized around sustained, content-specific coaching cycles for teachers in their first three years and for any teacher rated below effective, with coach caseloads of no more than 20 teachers, and the district shall report coaching dosage and teacher growth annually.

Indicator. Share of professional development hours delivered as job-embedded coaching cycles (observation, feedback, modeling) versus one-day workshops (PD calendar and coach logs); coach caseload (teachers per full-time coach); number of coaching cycles completed per teacher per year; observed instructional practice scores before and after coaching; student growth of coached teachers versus comparable uncoached teachers; teacher survey ratings of PD usefulness.

Red flags in real manuals. professional development defined as a required number of workshop or in-service hours; coaches assigned administrative or evaluative duties; coach caseloads above 30 teachers or coaches split across many schools; no coaching for new teachers; PD plan listing vendor sessions with no classroom follow-up; no reporting on PD outcomes; coaching offered only to teachers on improvement plans

Law. ESSA Section 8101(42) defines professional development as sustained, intensive, collaborative, job-embedded, data-driven and classroom-focused, and excludes one-day or short-term workshops; Title II-A funds must meet that definition. Most states set required in-service hours for license renewal (for example 150 clock hours per five years in Texas, 120 points per five years in Florida) but few specify format. Some states fund coaching directly, such as Mississippi's literacy coaches under the Literacy-Based Promotion Act.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. The one large randomized trial (Glazerman et al. 2010, 1,009 teachers in 17 districts) found that two years of comprehensive induction raised student achievement in the teachers' third year (+4 percentile points reading, +8 math; about 0.11 and 0.20 SD) but one year did not, and neither version changed retention. Nationally representative surveys and a field experiment on structured peer pairing find that first-year mentoring, common planning time and collaboration predict lower attrition and that pairing low-performing teachers with high-performing colleagues raised their students' achievement by about 0.12 SD. (Two-year comprehensive induction: +0.11 SD reading, +0.20 SD math in year 3 (RCT); retention: null in RCT, lower turnover in survey studies (comprehensive package associated with roughly half the turnover of no support). Structured peer mentoring: +0.12 SD student achievement for low-performing teachers (field experiment).) ^{557,558,559,560,561,562,563,564}

Where the evidence is mixed. The IES randomized trial found no effect on retention after one, two or three years and no effect on observed classroom practice; its achievement effect appeared only for the two-year treatment and only in year three, and a reanalysis has questioned it. One-year induction produced nothing. Rockoff (2008) found that New York City's mentoring program had limited effects on retention except when the mentor had worked in the same school, though more mentoring hours were associated with higher achievement. The retention findings of Smith and Ingersoll (2004) and Ronfeldt and McQueen (2017) are correlational; teachers who receive more support may be in better-run schools. Induction that consists of an assigned mentor with no release time, no training and no common planning is the common form and has no evidence behind it.

Standard. Every teacher in the first two years of service shall receive a trained mentor in the same subject or grade band with scheduled release time, weekly contact in year one, common planning time, and structured observation with feedback; the district shall track retention and mentor contact and report both annually.

Indicator. One-year, three-year and five-year retention of new hires in the district and in the school (HR records), compared by induction participation and by school poverty; share of first-year and second-year teachers with an assigned trained mentor in the same subject or grade; mentor contact hours logged per new teacher per month; new teacher survey on induction quality; student growth of teachers in years 1 to 3.

Red flags in real manuals. mentor assigned 'when available' or 'as resources permit'; induction limited to a one-day orientation; no release time or stipend for mentors; mentor not required to share subject or grade; induction lasts one year only for teachers new to the profession; no retention reporting; mentor doubles as evaluator

Law. More than half of states require some form of induction or mentoring for new teachers, often tied to the move from a probationary or initial license to a professional license (for example California's two-year Teacher Induction Program for a clear credential, Connecticut's TEAM program, New York's required mentored experience for professional certification, Texas Education Code 21.458 mentoring program with state funding). ESSA Title II-A funds may be used for induction and mentoring. Requirements rarely specify mentor training, release time or same-subject matching.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Well-run evaluation with trained observers and substantive feedback can improve teaching: Cincinnati teachers' students gained about 0.1 SD in math in the years after a rigorous peer evaluation, and Washington DC's IMPACT raised the performance of low-rated teachers by 0.27 SD and pushed the lowest performers out, with their replacements raising achievement. But the 2010s wave of state evaluation reforms, taken as a whole, produced precisely estimated null effects on achievement and attainment, rated fewer than 1 percent of teachers below proficient, and reduced the supply of new teachers. (Cincinnati peer evaluation: +0.11 SD math in years after evaluation. DC IMPACT: +0.27 SD teacher performance for low-rated teachers facing dismissal threat; turnover of low performers raised student achievement by 0.08 to 0.14 SD. Statewide reforms 2009 to 2017: null, ruling out effects as small as 0.015 SD on achievement and 1 percentage point on graduation or college enrollment.) ^{565,566,567,568,569,570,571,572,573,574,575}

Where the evidence is mixed. Bleiberg et al. (2025) find the 44-state evaluation reform wave had no average effect on achievement, graduation or college enrollment regardless of design rigor. Kraft and Gilmour (2017) found that under the new systems fewer than 1 percent of teachers were rated below proficient in most states, little different from the systems they replaced. The 575 million dollar Gates Intensive Partnerships produced no gains in achievement, graduation or access to effective teachers (Stecher et al. 2018). Kraft et al. (2020) show the reforms reduced the number of new teacher candidates without improving their quality. The IES feedback trial (Garet et al. 2017) improved one observation measure but not reading, and math gains were not sustained in year two. Steinberg and Sartain (2015) found Chicago's Danielson pilot raised reading only in higher-achieving, lower-poverty schools. Phipps and Wiseman (2021) show teachers raise their practice during scheduled observation windows, so observation scores partly measure performance for the observer. Positive effects came from a few high-capacity districts with trained observers, real consequences and compensating pay, not from rubrics alone.

Standard. Evaluation shall rest on multiple trained and calibrated observers, at least two observations per year with at least one unannounced, written feedback within five school days, and a growth plan with coaching for any teacher rated below effective; the district shall publish rating distributions and evaluator calibration results annually and shall not use a single test-based measure as more than one third of a rating.

Indicator. Distribution of summative ratings by school and by evaluator (share below proficient, share at each level); inter-rater agreement on calibration exercises; number of observations per teacher per year and share unannounced; days from observation to written feedback; share of teachers with a written growth plan tied to the evaluation; retention of teachers by rating level; student growth by rating level (does the rating predict growth); teacher survey on feedback usefulness.

Red flags in real manuals. evaluation consisting of a single announced observation per year; no observer training or calibration requirement; no timeline for written feedback; tenured teachers evaluated only every three or more years with no interim feedback; student test scores set at 50 percent or more of the rating; no growth plan or support attached to a low rating; no reporting of rating distributions

Law. Race to the Top (2009) and ESEA flexibility waivers (2011 to 2015) drove 44 states and DC to require evaluations that include student growth; ESSA (2015) removed the federal requirement and many states have since loosened test-score weights (for example New York's 2019 repeal of mandatory state test use, Tennessee and Florida reductions). Most states still set minimum observation counts, rating categories and the link between ratings and tenure or dismissal (for example Texas T-TESS, Tennessee TEAM, Florida Statute 1012.34, Illinois PERA). Collective bargaining agreements often govern observation notice and feedback timelines.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Design determines the effect. In a field RCT with nearly 1,000 high-school math students, unrestricted GPT-4 access raised practice scores (+48%) but cut unaided exam scores by 17% relative to no access; a version with guardrails that withheld answers removed the harm. A college physics RCT found a purpose-built AI tutor produced about double the learning gain of active-learning classes in less time, and an essay-writing RCT found gains persisted a week later only when students used the model as a tutor rather than to write for them. The K-12 evidence base is still one school, one subject and a few weeks. (unrestricted chatbot: -17% on unaided exam; guardrailed tutor: harm removed; purpose-built tutor (college): about 2x learning gain; effects persist only for tutor-style use) ^{576,577,578,579,580,581}

Where the evidence is mixed. Only one K-12 RCT exists (Bastani et al., one Turkish high school, four sessions covering 15% of the curriculum), and the positive tutor results come from Harvard undergraduates and Middlebury undergraduates, not children. The most-cited 2025 meta-analysis reporting large positive effects of ChatGPT on learning was retracted in April 2026, so there is no reliable pooled estimate. Lehmann, Cornelius and Sting (working paper) find harm concentrated in answer-generation use and no gain from explanation use, a pattern consistent with the RCTs but not peer reviewed. Detection tools are unreliable and produce false accusations; student-facing products carry data-privacy and age-limit (13+) constraints; teacher-facing uses (lesson planning, feedback) have essentially no outcome research. Policies that ban AI outright push use off campus where no guardrails exist, while policies that mandate adoption buy unproven products.

Standard. The district shall maintain a published list of approved AI tools, each with a stated instructional purpose, a guardrail configuration that withholds direct answers during practice, a data-privacy review and an age gate, shall require every course to state in writing when and how students may use AI, shall prohibit discipline based solely on detection software, and shall report usage and unaided-assessment results to the board annually.

Indicator. Inventory of approved AI tools with the guardrail configuration of each (answer-withholding on or off); share of courses with a written AI-use expectation on the syllabus; unaided assessment results in courses using AI tools versus matched courses; count of academic-integrity cases involving AI and how many were overturned.

Red flags in real manuals. all use of artificial intelligence by students is prohibited; AI detection scores are sufficient evidence of plagiarism; teachers may adopt any AI tool at their discretion; no minimum age or vendor-terms check for student accounts; student work or data may be used to train or improve vendor models; no requirement that tools withhold answers during practice; no annual review date despite the pace of change

Law. Executive Order 14277 (April 23, 2025) on AI education and the July 22, 2025 Department of Education Dear Colleague letter govern use of federal funds; by the end of 2025 roughly 34 state education agencies had issued K-12 AI guidance (counts vary by source), most of it non-binding. COPPA bars most consumer AI accounts for children under 13 without parental consent, and vendor terms often set 13 or 18 as the minimum age. FERPA applies to any student record entered into an AI tool. Several states (for example Tennessee SB 1711, 2024) require districts to adopt an AI use policy.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Giving every student a device does not by itself raise achievement: the large randomized One Laptop per Child trial and the Swedish 1:1 rollout found no effect on test scores, and the Swedish study found losses for children of less-educated parents. Gains appear only where the device carries targeted software (computer-assisted learning, especially in math) and teachers get sustained support; the one positive US district study (Mooresville) found no short-run effect and +0.13 SD in math after several years alongside changed teaching practice. (access alone: about 0; targeted computer-assisted learning in math: positive; well-supported district program: +0.13 SD math after 4 to 5 years) ^{582,583,584,585,586}

Where the evidence is mixed. The 2016 meta-analysis (Zheng et al.) reports positive average effects in English, writing, math and science, but it pools only 10 mostly non-randomized studies and the authors attribute gains to programs that supported teachers as well as students. Causal studies disagree on direction: Cristia et al. (Peru RCT, 318 schools) found nothing on math or language; Hall, Lundin and Sibbmark (Sweden) found no average effect and worse math outcomes and less college-preparatory enrollment for students with less-educated parents; Hull and Duch found gains, but neighboring districts without laptops improved at similar rates and the study cannot isolate which component mattered. Escueta et al. find that fully online courses underperform in-person ones. Device programs also carry recurring cost, screen-time and data-privacy trade-offs that the achievement literature does not price.

Standard. The district shall not purchase or renew a 1:1 device program or instructional software above a set dollar threshold without a written evidence review, a named instructional purpose, a teacher training plan and a usage-and-outcome report to the board each year, and shall retire products whose usage or results do not meet the stated purpose.

Indicator. For each device program and each licensed software product: an evidence summary on file before purchase; per-product usage data (share of licensed students active weekly, minutes per student) reported each semester; achievement trend by subgroup for the cohorts exposed versus the pre-adoption baseline on state assessments.

Red flags in real manuals. every student will be issued a device, with no stated instructional purpose or evaluation; software adoptions are approved by the technology department without board or academic review; no usage reporting or renewal criteria; device program funded as a one-time purchase with no refresh or support budget; screen time during the instructional day is unlimited or unmentioned; no data-privacy review tied to each product

Law. ESSA Section 8101(21) defines evidence tiers and federal Title funds used for technology require an evidence basis; E-rate and state device grants carry CIPA filtering obligations; state student-privacy laws (see student-data-privacy) govern each software contract.

Dispute this verdict or submit evidence

Tier B: one strong causal study or consistent quasi-experiments

Finding. Banning phone possession during the school day raises test scores modestly, with the gain concentrated among lower-achieving students (England: +0.07 SD overall, about +0.14 SD for the lowest quintile). Norway's middle-school bans cut bullying for both sexes and improved girls' grades, academic-track placement and mental-health care use, with no average effect on boys; the pooled effect across five European studies is small (d = 0.16) and larger for social wellbeing than for achievement. (+0.07 SD test scores (England), +0.14 SD for low achievers; Norway girls' GPA up and bullying down; pooled d = 0.16; mental-health effects unproven) ^{587,588,589,590,591,592,593,594}

Where the evidence is mixed. Sweden found no effect of use bans on achievement (Kessel et al. 2020), and the difference between banning possession and banning use may explain the split. The largest UK study (SMART Schools, 30 schools, 1,227 pupils) found restrictive school policies reduced in-school phone time but were not associated with better wellbeing, sleep, or attainment, because total daily phone use was unchanged; it is cross-sectional and cannot establish causation either way. Early US evidence (NSCH 2016 to 2024, three ban states, synthetic DiD) shows no clear change in screen time or wellbeing but is underpowered. The Dutch national evaluation is school-leader self-report, not outcome data. Bans impose real costs for students with medical or accessibility needs and for families who rely on reachability, so exceptions and a communication path matter.

Standard. Students shall keep personal phones and smartwatches powered off and stored away from their person from the first bell to the last bell, with written exceptions limited to documented medical, IEP or Section 504 needs and teacher-directed instructional use, and the district shall report compliance and referral data to the board annually.

Indicator. Share of classrooms observed with phones stored (walkthrough sample each semester); device-violation referrals per 100 students; student survey item on in-class distraction; achievement trend for the lowest-achieving quartile versus the district average (state assessment).

Red flags in real manuals. phones may be used during passing time, lunch or free periods; use is at the discretion of each teacher; a student may keep the phone on silent in a pocket or backpack; confiscated devices will be returned only to a parent, with no stated timeline; suspension for a first device violation; no exception language for IEP, 504 or medical needs; no way stated for a parent to reach a student during the day

Law. As of August 2025 about 35 US states had laws or state board rules restricting student phone use, and roughly 18 states plus DC required bell-to-bell restrictions (several, such as Georgia and Florida, only through grade 8); counts move each session, so check the state statute. Federal law (IDEA, Section 504, ADA) requires exceptions for devices that are part of a documented accommodation, for example glucose monitors paired to a phone.

Dispute this verdict or submit evidence

Tier D: professional standard or expert consensus

Finding. Audits find that most school-assigned apps send student data to third parties (Internet Safety Labs: 96% of apps used in a sample of 663 US schools; Human Rights Watch: 146 of 164 government-endorsed products worldwide, 89%, engaged in practices that risked children's rights), and breaches are frequent and large (99 K-12 breaches reported to the GAO sample 2016 to 2020; the December 2024 PowerSchool breach reached tens of millions of student records). There is no outcome research showing that any particular policy reduces harm; the standard rests on federal and state law and professional consensus. (not measurable as a student outcome; harm prevention standard (96% of school apps share data with third parties absent controls)) ^{595,596,597,598,599}

Where the evidence is mixed. No study links district privacy policy strength to measured student harm or benefit, and the audit figures count data flows, not demonstrated injury. Strict contracting raises procurement cost and can delay useful tools; blanket bans on consumer tools push use to personal accounts that the district cannot govern. Over-collection often starts with the district itself (surveillance and monitoring software), not the vendor, and that trade-off between safety monitoring and privacy has no empirical resolution.

Standard. No product that collects student data shall be used until a signed agreement is on file that limits collection to the stated educational purpose, bars sale, advertising and model training, requires deletion on request or at contract end, and sets breach notice within 72 hours, and the district shall publish the product inventory and report compliance to the board annually.

Indicator. Share of software products in use with a signed data privacy agreement on file (target 100%); public inventory of products and data elements collected; number of breach notifications and days to notify; annual count of vendor contracts reviewed and terminated; parent data-access requests fulfilled within the statutory window.

Red flags in real manuals. teachers may use free online tools at their discretion; no vendor agreement requirement or an agreement limited to FERPA boilerplate; student data may be shared with partners for product improvement; no breach notification timeline; no named privacy officer or responsible role; no inventory of products or data elements; directory information defined broadly with opt-out buried or absent; monitoring software collects off-campus activity with no retention limit

Law. FERPA, 20 U.S.C. 1232g and 34 CFR Part 99 (school official exception governs vendor access; annual notice and directory-information opt-out required). COPPA, 15 U.S.C. 6501-6506 and 16 CFR Part 312 (FTC amended rule published 2025; schools may consent on parents' behalf only for educational purposes, not commercial use). PPRA, 20 U.S.C. 1232h (surveys). State laws: 42 states had passed student privacy statutes by 2023, for example California SOPIPA (Bus. and Prof. Code 22584), New York Education Law 2-d and 8 NYCRR Part 121 (vendor data security and privacy plan required), Illinois SOPPA (105 ILCS 85, contract and public posting requirements), Texas SB 820 (cybersecurity framework). Check the state's contract-term requirements; several mandate specific clauses and public posting of agreements.

Dispute this verdict or submit evidence