New Mexico PreK Impacts: Regression-Discontinuity Results Overall and by Child Race/Ethnicity and Cohort

• In Event: 3-044 - Evaluating Publicly-Funded Preschool Programs: Understanding Impacts, Impact Variation, and Lasting Effects

Integrative Statement

Expansion of U.S. state-funded prekindergarten programs, variation between states in prekindergarten models, and population differences create a context where policymakers need more information about the effectiveness of individual state approaches. Research provides evidence of positive child outcomes attributable to a year of participation in statewide (e.g., Gormley et al., 2008; Wong et al., 2008) and citywide (Weiland & Yoshikawa, 2013) prekindergarten settings during the year before kindergarten. Yet further research is needed about the overall effectiveness of different prekindergarten models, and few studies have moved beyond examining impacts for overall prekindergarten populations to examine potential benefits for subgroups of children who may experience different types of risk. Our study uses data from five successive cohorts of participants in the New Mexico PreK initiative to examine overall child impacts for that state’s prekindergarten population, and also subgroup impacts based on race and ethnicity.

Children were randomly selected and assessments were conducted by trained researchers at the start of each school year, representing the first 5 years of New Mexico PreK. Reflecting the overall state population, children in our sample (total N = 5,219) were primarily Hispanic (60%), White (19%), or Native American (16%). English was the home language for 78% of these children. Child outcomes were measured for language (PPVT-III; Dunn & Dunn, 1997), literacy (Test of Preschool Early Literacy; Lonigan et al., 2007), and math (Woodcock-Johnson Applied Problems subtest; Woodcock et al., 2001) skills, with Spanish versions of each instrument used as appropriate. Similar to studies in several other states (e.g., Gormley et al., 2008; Lipsey et al., in press), we used regression-discontinuity (RD) models to compare two groups of children – a treatment group of entering kindergartners who had just completed PreK and a comparison group of children just entering PreK. These comparisons allow us to estimate PreK impacts because the state uses a birth date cutoff to determine eligibility for PreK and kindergarten.

Data across all five years were pooled and preliminary RD analyses were conducted. Table 1 shows that New Mexico PreK participants demonstrated statistically significant (p < .001) language, literacy, and math gains. This was the case using instrumental variables models employing samples with 12-, 6-, and 3-month windows around the eligibility cutoff date. Preliminary analyses show that language and literacy impacts vary significantly based on children’s ethnic and racial group membership. However, no differences by race/ethnicity were found on our math measure. Analyses, including further sensitivity analyses, will be finalized prior to this presentation.

These results suggest that New Mexico PreK participants enter kindergarten more academically ready to succeed. Due to the high level of diversity in both New Mexico and its PreK program, we had sufficiently large subgroups to examine outcomes separately for Hispanic, White, and Native American PreK participants. Consistent with prior findings (Gormley et al., 2005; Weiland & Yoshikawa, 2013), positive impacts were identified for children in each group. Policy implications for prekindergarten programs in New Mexico and other states with highly diverse populations will be explored as part of the presentation.

Authors

• Jason T Hustedt, University of Delaware
  Presenting Author

• Kwanghee Jung, NIEER/Rutgers University
  Non-Presenting Author

• Allison Friedman-Krauss, NIEER/Rutgers University
  Non-Presenting Author

• W. Steven Barnett, NIEER/Rutgers University
  Non-Presenting Author

• Gerilyn Slicker, University of Delaware
  Non-Presenting Author