Measuring subjective health among adolescents in Sweden: A Rasch-analysis of the HBSC-Instrument

The cross-national WHO-study Health Behaviourin School-Aged Children (HBSC) is acomprehensive adolescent survey ongoing inEurope based on a public health perspective.The present study, examining theHBSC-instrument on subjective health, uses theunidimensional Rasch model. Items are analysedwith respect to their operating characteristicsacross the whole range of the subjective healthscale and the empirical operation of theresponse categories intended to be ordered forall items. The study is based oncross-sectional data collected in Sweden duringthe 1980s and 1990s among students in yearsfive, seven and nine.The analyses reveal that the symptom checklistin the HBSC-instrument does not workconsistently with the Rasch model when alleight items are analysed simultaneously. Inparticular, the response categories do not workas intended. Hence, the original set of eightitems should not be used to construct a latentmeasure of subjective health. In order tobring the instrument to meet the requirementsof the Rasch model, three items were removed. The reduced set of five items did workconsistently with the model with respect to theresponse categories, and did show relativeinvariance across the latent trait. Since a fewof the remaining items showed lack ofinvariance across genders and grades thatproblem should be solved, if the reduced itemset is to be used for post-hoc analyses.Furthermore, the analysis of the reduced set ofitems suggests that both ``somatic'''' and``psychological'''' complaints might be consideredas parts of one higher order dimension ofsubjective health.In order to improve the questionnaire, furtherattention should be paid to the response formatof the items.