Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program
Open accessThe DOI matches its Crossref record; 58 of the 58 references it cites were found in a registry or an index; 3 of the 29 sources its claims cite were checked for support.
Abstract
Social experiments are powerful sources of information about the effectiveness of interventions. In practice, initial randomization plans are almost always compromised. Multiple hypotheses are frequently tested. "Significant" effects are often reported with p-values that do not account for preliminary screening from a large candidate pool of possible effects. This paper develops tools for analyzing data from experiments as they are actually implemented.We apply these tools to analyze the influential HighScope Perry Preschool Program. The Perry program was a social experiment that provided preschool education and home visits to disadvantaged children during their preschool years. It was evaluated by the method of random assignment. Both treatments and controls have been followed from age 3 through age 40.Previous analyses of the Perry data assume that the planned randomization protocol was implemented. In fact, as in many social experiments, the intended randomization protocol was compromised. Accounting for compromised randomization, multiple-hypothesis testing, and small sample sizes, we find statistically significant and economically important program effects for both males and females. We also examine the representativeness of the Perry study.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 6 Not valid · 24 Not checked · 120 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“The validity of this procedure follows from the analysis of Romano and Wolf (2005), who provide general results on the use of stepdown multiple-hypothesis testing procedures.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > 4.6 Multiple-hypothesis testing: The stepdown algorithm
What the sources say
- Supportedcites Romano and Wolf (2005) matched to “Exact and Approximate Stepdown Methods for Multiple Hypothesis Testing” (2005) source
The Introduction (Section 1) says the paper develops valid stepdown methods and reduces their construction to that of valid single tests. The Abstract likewise describes general results for controlling the familywise error rate, subject to conditions such as monotonicity of critical values. The cited paper therefore supports the attribution of general stepdown-validity results, though the quoted sentence does not describe the citing paper’s particular procedure. (LLM, conf=0.87)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Similar to traditional multiple-hypothesis testing procedures, such as the Bonferroni or Holm procedures (see, e.g., Lehmann and Romano (2005), for a discussion of these procedures), the stepdown algorithm of Romano and Wolf (2005) exhibits strong FWER control, in contrast with the classical tests like the F or χ2.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > Benefits of the stepdown procedure
What the sources say
- Supportedcites Romano and Wolf (2005) matched to “Exact and Approximate Stepdown Methods for Multiple Hypothesis Testing” (2005) source
The Nonasymptotic Results section, Theorem 2.1, states that the stepdown algorithm controls the family-wise error rate for any configuration of true hypotheses when its critical values satisfy the specified condition. The paper also establishes asymptotic control for procedures using estimated critical values. Page numbers are unavailable in our copy. (LLM, conf=0.92)
- Not matchedcites Lehmann and Romano (2005)
An author–year citation the paper’s reference list doesn’t settle. We don’t search these by name.
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“Romano and Wolf (2005) show that their monotonicity assumption is weaker than the subset pivotality assumption.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 7. Relationship of this study to previous research
What the sources say
- Supportedcites Romano and Wolf (2005) matched to “Exact and Approximate Stepdown Methods for Multiple Hypothesis Testing” (2005) source
Section 2 says that assumptions involving subset pivotality are stronger than the monotonicity condition (5). Example 2.3 demonstrates that condition (5) can hold even when subset pivotality fails. Page numbers are not supplied in our copy. (LLM, conf=0.94)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption not shown here.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We also reject the null hypotheses of no treatment effect on age-40 employment outcomes individually.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 5. Empirical results > Employment and earnings
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This pattern is largely found in the p-values for males.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 5. Empirical results > Main results
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
85 more claims
Create a free account to see all 153 claims and every source verdict.
- It may state two separate claims as one. 44
- Probably not a substantial claim. 25
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Despite this, children who received the intervention went on to enjoy more successful lives in adulthood including greater labour market success, reduced crime involvement and better health30,33,34.”
From A systematic review and meta-analysis of effects of early life non-cognitive skills on academic, psychosocial, cognitive and health outcomes (2018) · cites this paper as [34]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > 4.2 Setup and notation
“The set 𝒟 is the support of the vector of random assignments, namely 𝒟 = [0, 1] × ⋯ × [0, 1], 123 times, so 𝒟 = [0, 1]123.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > 4.1 Randomized experiments
“Di = 1 if treatment occurs, Di = 0 otherwise.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > 4.4 Exchangeability and the permutation-based tests
“Thus, under the null, a critical value can be computed by taking the α quantile of the set {T(Y, gD) : g ∈ 𝒢X}.”
§ Analyzing social experiments as implemented: A reexamination of the evidence from the HighScope Perry Preschool Program > 4. Methods > 4.4 Exchangeability and the permutation-based tests
“The goal of our test is to control for a Type-I error at significance level α, that is, Pr(reject Hypothesis H-1|Hypothesis H-1 is true)= Pr(T(Y, D) > c|Hypothesis H-1 is true) ≤ α.”
Citations
Cites
Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing
Benjamini, Y., & Hochberg, Y. (1995) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1995.tb02031.x · Cited by 109,482 · cited 3×
“(40)For further discussion of stepdown and its alternatives, see Westfall and Young (1993), Benjamini and Hochberg (1995), Romano and Shaikh (2004, 2006), Romano and Wolf (2005), and Benjamini, Krieger, and Yekutieli (2006). 31”
BackgroundThe central role of the propensity score in observational studies for causal effects
ROSENBAUM, P. R., & RUBIN, D. B. (1983) · Biometrika · doi:10.1093/biomet/70.1.41 · Cited by 31,241 · cited 2×
“To overcome this problem, Rosenbaum and Rubin (1983) propose propensity score matching, in which matches are made based on a propensity score, that is, the probability of being treated conditional on observed covariates.”
The Bell Curve: Intelligence and Class Structure in American Life.
Preston, S. H., Herrnstein, R. J., & Murray, C. (1995) · Population and Development Review · doi:10.2307/2137757 · Cited by 3,998
The Central Role of the Propensity Score in Observational Studies for Causal Effects
Rosenbaum, P. R., & Rubin, D. B. (1983) · Biometrika · doi:10.2307/2335942 · Cited by 3,261
Adaptive linear step-up procedures that control the false discovery rate
Benjamini, Y., Krieger, A. M., & Yekutieli, D. (2006) · Biometrika · doi:10.1093/biomet/93.3.491 · Cited by 3,228 · cited 1×
“43 Tables 3 and 4 report results for females, 40 For further discussion of stepdown and its alternatives, see Westfall and Young (1993), Benjamini and Hochberg (1995), Romano and Shaikh (2004, 2006), Romano and Wolf (2005), and Benjamini, Krieger, and Yekutieli (2006).”
The Economics and Econometrics of Active Labor Market Programs
Heckman, J. J., Lalonde, R. J., & Smith, J. A. (1999) · Handbook of Labor Economics · doi:10.1016/s1573-4463(99)03012-6 · Cited by 3,059 · cited 1×
“18 Heckman and Smith (1995), Heckman, LaLonde, and Smith (1999), and Heckman and Vytlacil (2007) discussed randomization bias and substitution bias.”
Self-citationAnderson, M. L. (2008) · Journal of the American Statistical Association · doi:10.1198/016214508000000841 · Cited by 2,410 · cited 11×
“Others express the concern that previous analyses selectively report statistically signifi-cant estimates, biasing the inference about the program (Anderson (2008)).”
InfluentialCharacterizing Selection Bias Using Experimental Data
Heckman, J., Ichimura, H., Smith, J., & Todd, P. (1998) · Econometrica · doi:10.2307/2999630 · Cited by 2,286
Self-citationResampling-Based Multiple Testing: Examples and Methods for p-Value Adjustment.
Martin, M. A., Westfall, P. H., & Young, S. S. (1994) · Biometrics · doi:10.2307/2533464 · Cited by 1,699 · cited 2×
“43 Tables 3 and 4 report results for females, 40 For further discussion of stepdown and its alternatives, see Westfall and Young (1993), Benjamini and Hochberg (1995), Romano and Shaikh (2004, 2006), Romano and Wolf (2005), and Benjamini, Krieger, and Yekutieli (2006).”
Heckman, J., Pinto, R., & Savelyev, P. (2013) · American Economic Review · doi:10.1257/aer.103.6.2052 · Cited by 1,628
Self-citation
Research works printed but not linked: 2
- Hotz VJ Manski C Garfinkel I Designing an evaluation of the job training partnership act Evaluating Welfare and Training Programs 1992 Cambridge, Massachusetts Harvard University Press 76 114
- Schweinhart LJ Montie J Xiang Z Barnett WS Belfield CR Nores M Lifetime Effects: The High/Scope Perry Preschool Study Through Age 40 2005 Ypsilanti, Michigan High/Scope Press
Cited by
Chetty, R., Hendren, N., & Katz, L. F. (2016) · American Economic Review · doi:10.1257/aer.20150572 · Cited by 2,315
Heckman, J. J., & Kautz, T. (2012) · Labour Economics · doi:10.1016/j.labeco.2012.05.014 · Cited by 2,007
“Nevertheless, the program improved outcomes for both boys and girls, resulting in a statistically significant rate of return of around 6-10% per annum for both boys and girls (see Heckman et al., 2010a,b).”
BackgroundSelf-citationMeasuring the Impacts of Teachers II: Teacher Value-Added and Student Outcomes in Adulthood
Chetty, R., Friedman, J. N., & Rockoff, J. E. (2014) · American Economic Review · doi:10.1257/aer.104.9.2633 · Cited by 1,717
Heckman, J., Pinto, R., & Savelyev, P. (2013) · American Economic Review · doi:10.1257/aer.103.6.2052 · Cited by 1,628
Self-citationPersonality Psychology and Economics
Almlund, M., Duckworth, A. L., Heckman, J., & Kautz, T. (2011) · Handbook of the Economics of Education · doi:10.1016/b978-0-444-53444-6.00001-8 · Cited by 1,204
Self-citation
Metadata
- Authors
- Heckman, J. J., Moon, S. H., Pinto, R. R., Savelyev, P. A., & Yavitz, A.
- Affiliations
- University College Dublin (IE)
- Venue
- Quantitative Economics
- Publisher
- The Econometric Society
- Published
- 2010
- Volume
- 1
- Issue
- 1
- Pages
- 1-46
- ISSN
- 1759-7323 (ISSN-L)
- DOI
- 10.3982/qe8
- PMC
- PMC3524308
- OpenAlex
- W3123199285
- Cited by
- 304
- References
- 56
- Citations by year
- 2026: 2 · 2025: 16 · 2024: 17 · 2023: 24 · 2022: 26 · 2021: 24 · 2020: 13 · 2019: 22 · 2018: 22 · 2017: 18 · 2016: 31 · 2015: 27 · 2014: 15 · 2013: 23 · 2012: 14
- FWCI
- 12.82
- Type
- journal article
- Language
- en
- Open access
- diamond
- doi.org
- Licence
- cc-by-nc
- doi.wiley.com/10.1002/tdm_license_1.1 (tdm, 2069 d)
- Locations
- Quantitative Economics (publishedVersion, cc-by-nc)
- PubMed (publishedVersion)
- Europe PMC (PubMed Central) (acceptedVersion)
- PubMed Central (acceptedVersion)
- Econstor (Econstor) (submittedVersion, cc-by-nc)
- W&M Publish (College of William & Mary) (submittedVersion, cc-by-nc)
- Funders
- Pew Charitable Trusts
- American Bar Foundation
- J.B. and M.K. Pritzker Family Foundation
- Susan Thompson Buffett Foundation
- Economic and Social Research Council (RES-544-28-5001)
- Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD043411, R37HD065072, R01 HD054702)
- Topics
- Physical Sciences › Mathematics › Statistics and Probability › Advanced Causal Inference Techniques
- Social Sciences › Public Administration › Evaluation and Performance Assessment
- Social Sciences › Education › Early Childhood Education and Development
- Keywords
- social experiment, multiple comparisons, p-values, small sample size, program evaluation, preschool education, home visiting, disadvantaged children, representativeness, gender differences
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-29, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-29 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.