SOURCES: MDRC calculations are from baseline and follow-up results on the Stanford Achievement Test Series, 10th ed. (SAT 10) abbreviated battery. National norming sample calculations are from the SAT 10 (2002 norming sample): Stanford Achievement Test Series: Tenth Edition: Technical Data Report (Harcourt Assessment, 2004, pp. 312-338).
NOTES: The growth line for the enhanced program group is based on the observed mean baseline and follow-up test scores of students assigned to the enhanced after-school program for two consecutive years (baseline is Fall 2005; follow-ups are Spring 2006 and Spring 2007). The growth line for the regular program group represents the test scores that students in the enhanced program group would have obtained had they not been assigned to the enhanced program (calculated as the mean test score for the enhanced program group minus the estimated impact at a given time point). The growth line for the national norming sample is based on the average SAT 10 total reading scores for a nationally representative sample of students with the same grade composition in each period as the two-year sample. Specifically, at each point in time (the fall baseline, the first spring, and the second spring), the SAT 10 national norm scores for second-, third-, and fourth-graders are averaged weighting each grade average score according to their proportion in the two-year study sample at baseline. This creates an expected two-year improvement of nationally representative students at the same grade levels as this study’s sample. The baseline for the national norming sample is set relative to the average baseline score of the enhanced program group. Estimated impacts on follow-up results are regression-adjusted using ordinary least squares, controlling for indicators of random assignment, baseline reading total scaled score, race/ethnicity, gender, free-lunch status, age, overage for grade, single-adult household, and mother's education. Among those who did not reapply to the study in the second year, nonresponse weights are used to account for those students for whom follow-up data was not collected. Statistical significance is indicated by (*) when the p-value is less than or equal to 5 percent.