Measuring participation of social-support clients: : validity and reliability of IPA-MO

This study evaluates the reliability and validity of the Impact on Autonomy and Participation instrument (IPA) for heterogeneous populations of social support clients. Decentralisation of social support and accompanying budget cuts spurred interest in outcome-related payment systems to foster efficiency of social support. This prompted the need to have insight in outcomes of social support, defined as ‘self reliance and participation’. Eight municipalities in different parts of the Netherlands used an adapted version of IPA (IPA-MO) to collect self-reported outcome measures among cohorts of inhabitants receiving social or income support. Participants included people with mild physical, severe physical, cognitive or mental impairments and people depending on income support. Survey data were combined in a single database (N=4.120). Multivariate analysis was used to analyse reliability and validity of IPA-MO. The original IPA model distinguished five scales (‘participation domains’): Autonomy indoors, Family Role, Autonomy outdoors, Social life and Work & education, each scale loading on a separate factor. Due to high non-response on Work & education, our analysis mostly focused on the remaining four scales. These were confirmed, with minor changes, for IPA-MO. Financial autonomy was found as a new participation domain, composed by two new items added to the original single one. Five items of the original IPA were eliminated for duplicity and high correlation with other items. Confirmatory factor analysis confirmed construct validity of the five-scale IPA-MO model (CFI .936, TLI .925, SRMR .051). Internal reliability was confirmed for all scales (Cronbach alpha >.80, item-test correlation >.50 for all items). Exploratory factor analysis revealed a four-factor structure, with two scales (Family role and Autonomy outdoors) located on one factor. Yet, model fit is better when treated as separate scales. Two approaches to create more homogeneous groups were tested: impairment-based and age-based groups. The IPA-MO model as found for total research population, proved valid for both types of groups. The Work & education scale was tested for a small number of participants (N=234). One item was eliminated for duplicity. Exploratory factor analysis showed six scales loading onto six factors. Model fit was acceptable (CFI .915, TLI .903, SRMR .067) We conclude that the IPA-MO model is a valid and reliable instrument for local governments to assess participation of heterogeneous social-support populations. Further research is needed to test if Financial autonomy sufficiently covers clients’ perspectives