Opt-Out and Socioeconomic Status Across U.S. School Districts

• In Event: Socioeconomic Status and Standardized Tests: Achievement Gaps, Alternate Assessments, and Opting Out

Abstract

Unprecedented numbers of students opted out of accountability testing in the 2014-15 school year. Thirteen states failed to meet the 95% participation requirement according to the Department of Education (Bennett, 2016). Students’ and families’ motivation for opting out of state testing is not well understood; however, there is some evidence that students who opt out of tests tend to be more white and live in more affluent school districts (Bennet, 2016; Chingos, 2016; Pizmony-Levy & Green Saraisky, 2016). Our study provides population-based analyses of the development of opt out from 2012-13 to 2014-15, investigates the association between district opt out rates district characteristics, and examines how opt out varies across the gender, racial and socioeconomic subgroups served within districts. Evidence from this analysis can inform policy by quantifying how opt out biases comparisons of district or state performance from federal accountability testing.

We estimate opt out rates for nearly every school district in the U.S. using state accountability participation and assessment data from the EDFacts database, a Department of Education data collection initiative. The EdFacts data include counts of students who took the assessments, as well as counts of students who were currently enrolled in each school district by grade, subject, and year. In total, we estimate participation rates for approximately 11,000 districts from the 2012-13 through 2014-15 school years, in the third through eighth grades in mathematics and ELA.

There are significant differences in the state-level participation rates on accountability assessments. In the 2014-15 school year, states range from having overall participation rates as low as 75% (e.g., New York), well below the federal guideline of 95%, to nearly 100% (e.g., West Virginia). We further find substantial variation among districts within states in the percentage of students opting out, and that the variation dramatically increased from 2012-13 to 2014-15 for some states. For example, Figure 1 shows the distribution of district participation rates on mathematics assessments in New York over grades and years. In New York, there were almost no districts with lower than 95% participation rates in 2012-13, but by 2014-15 there are a substantial number of districts with rates ranging between 10% and 95%.

We use state-fixed effects models to estimate the relationship between opt out rates and district characteristics within states. We focus on the socioeconomic composition of districts using data from the American Community Survey and Common Core of Data, including the percentage of students receiving free lunch, segregation of poor and non-poor students, and the socioeconomic status of families with children in public schools. On average, we find districts with more free lunch students have lower opt out rates, while more affluent districts and districts with more economic school segregation have higher opt out rates. However, we find significant variability across states in these direction and magnitude of these correlations, suggesting that the mechanisms driving opt out and the students who are opting out are different across states.

Authors

• Erin Michelle Fahle, Stanford University

• Sean F. Reardon, Stanford University

• Demetra Kalogrides, Stanford University

• Sam Trejo, Stanford University