Mental health, reporting bias and economic transitions[Semiparametric estimation with mismeasured dependent variables: an application to duration models for unemployment spells]

Measures of mental health are heavily relied upon to identify at-risk individuals. However, self-reported mental health metrics might be unduly affected by mis-reporting (perhaps stemming from stigma effects). In this article, we consider this phenomenon by focusing upon the mis-reporting of mental health using UK panel data from 1991 to 2018. In separate analyses of males and females, we examine how inaccurate reporting of the General Health Questionnaire (GHQ-12) measure, specifically its sub-components, can adversely affect the distribution of the index. The analysis suggests that individuals typically over report their mental health (especially so for males). The results are then used to adjust the GHQ-12 score to take mis-reporting into account. We then compare the effects of the adjusted/unadjusted GHQ-12 index when modelling a number of important economic transitions. Using the original index typically leads to an underestimate of the effect of poor mental health on transitions into improved economic states, for example, unemployment to employment.