Students Who Choose and the Schools They Leave: Examining Participation in Intradistrict Transfers

Abstract

Data from one urban school district is analyzed to examine equity-based arguments about school choice as they pertain to intradistrict transfer policies. We specifically examine which factors influence the propensity for parents to participate in choice, and how choice schools differ from the schools that students are zoned to attend. We find that advantaged and disadvantaged parents make similar choices in that they are both likely to choose more affluent schools with better academic records than the schools they are zoned to attend. However, these choices operate in different spheres, as advantaged parents choose the most affluent schools with the best academic records, and disadvantaged parents choose away from the least affluent schools with the worst academic records to schools that are slightly better.

NOTES

Notes

¹ In this study, the term “disadvantaged” is used to identify low-income, economically disadvantaged students.

² A student who is unwilling to adhere to certain codes of conduct will sometimes be denied the opportunity to choose another school; however, this is rare. Otherwise, the district has never encountered the need to deny students attendance in their first-choice school.

³ Because of the nature of our study, we omitted students with disabilities who received special services from the district as well as students who qualified for and were attending specialized programs. Both sets of students generally attended schools other than their zoned schools: However, their choices were not made through the intradistrict transfer choice process. Because these students were likely to choose schools for reasons very different than intradistrict transfer choosers, they were excluded from all analyses.

⁴ The National School Lunch Program is a federally assisted meal program. To be eligible, a student must be from a household with an income at or below 130 percent of the poverty threshold for free lunch, or between 130 and 185 percent of the poverty threshold for reduced-price lunch (Aud et al. 2011). Even though using participation in FRL as a measure of student poverty is a relatively simple measure of poverty, education researchers use it often. In fact, some researchers claim it is the most commonly used measure of poverty in education research (Kurki, Boyle, and Aladjem 2005), and it is often used in federally funded studies, reports, and is described as an indicator of economic advantage in the NCLB legislation.

⁵ Reference groups are applied only to the regression analyses conducted in this study. Other descriptive and inferential statistics do not require reference groups. Therefore, the groups identified in the text as “reference groups” are included in these analyses.

⁶ Ideally, we would like to compare the distance and travel times between home and zoned schools as well as home and choice schools, but the limitations of our data do not allow us to access student addresses for the purpose of creating such measures. However, the distance and travel time between zoned and choice schools act as reasonable proxies for measures of the distance and additional time considerations associated with choosing a school versus remaining in a zoned school.

⁷ In our regression analyses, the percentage of students who are proficient in language arts is highly correlated with the percentage of students who are proficient in mathematics. Because of subsequent problems with multicollinearity, we only include proficiency levels in mathematics in our analyses. Mathematics was chosen over language arts because research has demonstrated that “mathematics is the academic area most influenced by schooling and least affected by home factors” (Lee and Bryk 1989:179).

⁸ We also assume that the relationship between row and column predictors is fixed. During our preliminary analysis, we explored the possibility that the relationship between row and column predictors was random, meaning that zoned school effects were allowed to vary across students, and student effects are allowed to vary across time points. However, these relationships were not significant and hindered the calculation of our models. Therefore, we presume these relationships to be fixed, and in doing so, we do not underestimate the standard error, which is a possibility when these associations are incorrectly assumed to be fixed (Raudenbush and Bryk 2002).

⁹ Because the outcome of whether or not a student participated in open enrollment is a dichotomous measure with a Bernoulli distribution, a log transformation was made to estimate the models. Therefore, in reporting the results of the models, all coefficients are exponentiated and interpreted as event rate ratios.

¹⁰ In addition to the zoned school characteristics described here, other characteristics were initially included in the models—adequate yearly progress (AYP) status and Title I status. However, they did not yield significant results and were excluded from the models to maximize the degrees of freedom in the column–level of the HCM models.

¹¹ All differences between advantaged and disadvantaged choosers presented on Figure 1 and discussed in this section of the article are statistically significant at the p < 0.001 level.