From cream skimming to push-out: The effect of eliminating the practice of selective admissions in Chile on student mobility

• In Event: Compararative approaches to higher education strategies of teaching and learning

Proposal

While the literature on cream skimming in public school choice in the US is mixed, there is more consistent international evidence that private schools use various procedures in school admissions to filter out students who do not “fit” the desired student profile (West, Ingram, & Hind, 2006; Perry-Hazan, 2019; West, Barham & Hind, 2011). In Chile, there is evidence that this filtration has also taken place in publicly funded private schools and elite public schools (Contreras, Sepúlveda, & Bustos, 2010; Zancajo, 2019). While this is not the only cause of school segregation, there is evidence that it has contributed to segregation (Carrasco, Gutierrez, & Flores, 2017; Contreras, Sepúlveda, & Bustos, 2010).
The School Inclusion law changed the way students are assigned to schools, switching from a decentralized system where each school followed their own individual admission procedures to a national standardized admission system (NSAS) based on the Gale and Shapely mechanism whereby schools are no longer able to exclude interested students.
There is evidence that this has caused modest changes in the demographics of students attending previously selective schools (Carrasco & Honey, 2019; Correa et al., 2019; Grau, 2019; Sillard, Garay, & Troncoso, 2018). While schools are not permitted to arbitrarily expel students, they may not go out of their way to accommodate students who are not happy in the school. School climate and comparison with peers are known to be driving forces in student push-out and student mobility (Dauter & Fuller, 2011; Sanclemente, 2008).
School mobility is a frequent phenomenon in Chile, previous studies calculate that, depending on how one measures it, the proportion of students transferred are about 11% (Perticara & Sanclemente, 2008; Sanclemente, 2008; Román & Perticará, 2012). There is some mixed evidence regarding the consequences of student mobility, but prior research suggests that mobility is on average harmful to students and schools (Hanushek et al., 2004).
This paper investigates the effects of the NSAS by focusing on the amount of school mobility across schools within and between years. Does the NSAS effect mobility patterns for students and if so, are there differential effects for low-income students or schools with higher representation of low-income students? Answering these questions is critical in assessing the capacity of centralized deferred assignment systems to meet the educational needs of a diverse set of students.
Research Questions:
RQ1: To what extent do schools experience changes in student mobility when school admission systems are standardized and selectivity is eliminated?
RQ2: To what extent do schools experience changes in mobility of low-income students the elimination of school selectivity in a standardized admission system?
RQ3: To what extent is there a differential effect of a standardized admission system without selectivity based on the proportion of low-income students in the school?
Data & Methods:
Administrative records provided by the Ministry of Education from the 2016-2020 school including nationwide information about each student school and grade at the beginning and the end of each academic year, alongside information about the school (public vs private, religious orientation, copayment) and an indicator for student SES.
This paper uses multiple models to build from a simple analysis of associated trends towards a quasi-causal analysis controlling for potential confounding factors and time trends. We begin with a descriptive analysis of the demographic makeup and level of school mobility the year prior to implementation of the NSAS and the first year of implementation. For the main analyses we use fixed effects models, where the analysis focuses on school-grades that switch from the old admission system to the new system. By focusing on grades that switch admission systems we are able to compare the same school grade with itself and overcome the endogeneity constraints from constant un-measured variables that could influence student mobility patterns. We benefit from the arguably exogenous assignment of the new admission system (treatment) for individual grades within schools, as the system has a phased implementation over time to different grades, as well as the exogenous assignment of the same grades in different regions in different years due to the phased implementation across regions. This method allows us to isolate the association between the NSAS and the demographics of a grade within a school. We do so by taking advantage of multiple natural discontinuities in assignment of “treatment” (the NSAS) across grades, regions, and over time. The other benefit of the phased implementation is that we are able to include year fixed effects in order to control for exogenous changes due to the passing of time.
We run separate models for elementary school grades (1st-6th grade), middle school grades (7th & 8th grade), or high school grades (9th-12th grade) in order to detect potential differences in trends in mobility at different educational levels. According to the admission system the entry level grades in Chile can be considered Pre-K, kindergarten, 1st, 7th, and 9th grades. However, there are some schools that offer comprehensive education from Pre-K to 12th grade. Thus, in our models examining mobility from one year to the next we only include grades within schools where the school offers the subsequent grade. In this way we attempt to limit conflating structural switches from non-structural switches.
Findings:
We find the NSAS is associated with a modest statistically significant increase in student mobility within years, between years, with the increase specifically detected at a slightly higher magnitude for low-income students, and with the amount of increase being positively related to the level of representation of low-income students at the beginning of the year with the effects being most pronounced in the elementary school grades.
These findings can help inform policymakers as they consider both the intended and unintended consequences of policies related to how students are assigned to schools and the additional considerations that may need to be addressed once students are in schools so they feel welcome, accommodated, and do not feel driven to find alternatives. And it fills a gap in the literature regarding the effect of centralized admissions on student mobility in systems of school choice.

Author

• Ngaire Noelle Honey, Pontificia Universidad Católica de Chile