Proxy reporting in education surveys: factors influencing accurate reporting in the 2012 Qatar Education Study

Abstract

Proxy reporting is a common practice during survey data collection to increase response rates while reducing fieldwork costs, and agreement between proxies and self-reports is critical to make reliable and valid inferences. This study is the first to unpack what influences proxy accuracy in a non-Western setting using data from the 2012 Qatar Education Study. We find that agreement is a function of a student’s grade in school, grades, a parent’s level of education, and the interaction between immigration status and parent education. These findings suggest in multicultural contexts, agreement may vary based on factors beyond what is typically accounted when examining the components of reporting error as a result of using proxies over self-reports.

Keywords:

• Proxy reporting
• measurement error
• Middle East and North Africa
• immigration
• cross-cultural survey design

Acknowledgements

The authors would like to thank Darwish Alemadi, Ahmed Alemadi, and Abdoulaye Diop for making this work possible.

Notes

1. Although a direct test of this is not possible in the current study due to data limitations in the QES 2012.

2. In early 2016 the Supreme Council of Education was reorganized into the Ministry of Education and Higher Education.

3. The questionnaires were designed in English and translated into Arabic by professional translators, and then checked by bilingual staff. The questionnaires were pre-tested in four randomly selected schools and revised based on the results of the pre-tests.

4. Special attention was paid to designing the questionnaire in ways to minimize item non-response, particularly during question writing and paper layout. The questionnaire was also piloted and revised to further reduce the incidence of item non-response. A total of 1848 students and 1409 parents completed the QES 2012 questionnaire. An additional 385 cases were dropped using listwise deletion because of missing values on one of the variables used in the analysis.

5. A recent meta-analysis of self-reported grades found that self-reported grades generally predict outcome variables in the same manner as actual grades (Kuncel et al., 2005), thus we use self-reported grades as a surrogate for academic performance in this analysis.

6. The proportion of Qatari students whose parents have some college experience is .17 in comparison to .42 for non-Qatari students (p < .001).