Measuring socioeconomic status among middle school students: does atypicality of responses matter for predicting academic achievement?

Abstract

Measurement of SES among school students often relies on the indicators of parental education, occupation, and income. However, this approach often results in large proportions of non-response and a lack of consistency in student answers. An alternative way is to use composite measures with multiple items indicating family wealth, cultural possessions, and home educational resources, which solves the problem of non-response. However, little is known about the quality of students’ responses to such measures. The aim of our study was (1) to explore the dimensionality of middle school students’ answers on a composite measure of SES, and (2) to assess the predictive utility of extracted dimensions. Using 2012-2016 data from the National Survey of Student Achievement (samples of 8th and 6th grade students, N1=2987; N2=1701; N3=2252) we conducted multiple correspondence analysis for exploring the dimensionality of responses to SES items (k=9). The results showed that at least two dimensions are required to acceptably explain the variance of SES items. Inspection of the dimensions revealed that the first reflected low vs. high SES, while the second reflected typical vs. atypical (incongruent) students’ answers to SES items. The typicality dimension added substantial explanatory power to the prediction of reading and math scores.