Item response drift in the Family Affluence Scale: A study on three consecutive surveys of the Health Behaviour in School-aged Children (HBSC) survey

Comparable data on socio-economic position (SEP) is essential to international studies on health inequalities. The Health Behaviour in School-aged Children (HBSC) has used the Family Affluence Scale (FAS) on material assets. The present study used data collected from adolescents in eight countries in 2002, 2006 and 2010, and examined the construct validity of the FAS, by focusing on changes in item responses over time. The analyses reported the changes in means in item responses, and fitted models which estimated differential item functioning (DIF), and local dependency (LD) between items. DIF and LD were analysed by Graphical Log-Linear Rasch Models (GLLRMs), and changes in the measurement properties of the FAS over time and between countries were assessed. The results showed that the FAS items have changed their measurement properties between 2002 and 2010, and caution is warranted in studies comparing the FAS between different time points.