Construction and Validation of a New Proxy Socioeconomic Status Measure for NAEP Grade 12

• In Event: 41.066 - The Development and Applications of Alternative Student Socioeconomic Status Measures

Abstract

Currently, no commonly used composite measure of socioeconomic status (SES) exists for the National Assessment of Educational Progress (NAEP). This is unfortunate as SES is an important contextual factor that should be accounted for as best as possible in research involving NAEP. In its absence, NAEP researchers often use eligibility for Free or Reduced Price Lunch as an SES proxy. Having only one crude proxy for family income, which in itself represents only one of three components of SES (the other two of the “big three” are parental education and occupational prestige), as an SES control involving NAEP is likely to underestimate the association of SES with NAEP scores. Likewise, this can lead to an overestimation of other factors’ relationships with NAEP scores that are being studied (net of an SES control in a multiple regression analyses, for example).
Previous research done by the authors using the overlap sample of the High School Longitudinal Study of 2009 (HSLS:09) which contains parent reports of the “big three” SES components and the 2013 Grade 12 (NAEP) Mathematics Assessment has demonstrated that many NAEP SES proxy variables have significant associations with one or more of the “big three” SES components; even with occupational prestige which is not directly assessed by NAEP.
The research goal of this paper is to develop a simple, intuitive SES index constructed from available NAEP proxy SES variables that would aim to be of comparable quality than an index based on the “big three” SES components measured through the HSLS:09. This comparison is made possible by the NAEP-HSLS overlap sample which contains both the “big three” SES variables as part of the HSLS parent questionnaire in addition to the NAEP proxy SES measures for students.
The NAEP proxy variables used in this paper are: student reported parental education, books in the home, household possessions, and NSLP status. The criteria to judge the different indices’ performance are: (1) the amount of variance explained for grade 12 NAEP mathematics by the respective SES index controlling for gender and race/ethnicity; and (2) the difference between the observed achievement gap and the remaining “non-explained achievement gap”, after controlling for SES, with the different SES indices.
Findings from this paper suggest that the NAEP proxy SES index not only performed as well as the HSLS SES index, it actually performed better. The NAEP proxy index had a higher R-squared value and the achievement gap net of the respective SES control (i.e. the remaining “non-explained achievement gap”) for Black and Hispanic students was lower when using the NAEP proxy SES index. We also found that a slightly expanded NAEP proxy SES index, that includes the percentage of students eligible for NSLP at the school level, worked very well in the full NAEP mathematics grade 12 sample. It also has a fairly strong correlation with grade 12 NAEP mathematics performance, has an acceptable internal consistency, and functions quite similarly for major student subgroups.

Authors

• Markus Broer, American Institutes for Research

• Qingshu Xie, MacroSys LLC

• George W. Bohrnstedt, American Institutes for Research