Self-assessments or tests? Comparing cross-national differences in patterns and outcomes of graduates' skills based on international large-scale surveys

In this paper an analysis is carried out whether objective tests and subjective self-assessments in international large-scale studies yield similar results when looking at cross-national differences in the effects of skills on earnings, and skills patterns across countries, fields of study and gender. The findings indicate that subjective skills measures do not correlate well with objective measures of similar constructs when looking at cross-national differences. Countrywise associations between subjective skills measures and earnings do not correlate well with those found using objective skills measures. Moreover, cross-national differences in the level of subjective skills measures do not correlate well with cross-national differences in skill levels based on objective tests. Nor do gender differences found using subjective skills measures correlate with those found using objective skills measures. This does not mean that self-assessments cannot be used, but they need to be restricted to analysing within-country differences. Within countries, self-assessments do a good job in predicting skills differences across fields of study and also in predicting the effect of skills on earnings. When comparing gender differences in skills levels within countries, however, one needs to be aware that females tend to overestimate their skills levels in typical ‘female’ domains like literacy.