Why Do Immigrants Overestimate Their Language Proficiency?

The use of self-reported language proficiency as an indicator for immigrants’ actual language proficiency is a commonly used measure in the social sciences. However, the validity of self-reported language proficiency is highly disputed. Based on a Rational Choice approach of respondent behavior this contribution analyses the determinants of self-ratings and the conditions under which self-reports of language proficiency are accurate or not. For the analysis we use data of the German National Educational Panel Study (NEPS) of starting cohort 4 (grade 9) which contains tested as well as self-reported first and second language proficiency of students of either Russian (N=628) or Turkish (N=809) origin. Our results confirm that self-reports seriously deviate from test results, with a predominant tendency of students overestimating their language proficiency. The bias of self-reports systematically vary by gender, ethnic origin and duration of stay in destination country. Further analyses indicate that the accuracy of self-reports depends on learning experience, language use, the available frames of references of the students.