The Psychometric Properties of the Early Development Instrument: A Rasch Analysis Based on Swedish Pilot Data

The Early Development Instrument (EDI) is a population measure and an indicator of children’s developmental health before entering the school system. EDI-Sweden was translated and adapted from EDI-Canada. In 2011 a pilot study was conducted, as a first step of the preparations for nationwide implementation of EDI in Sweden. The purpose of the study is to analyse the psychometric properties of EDI-Sweden. Data about 116 5-year-old children were collected at ten preschools in two municipalities. EDI consists of 104 core items in five domains: physical health and well-being, social competence, emotional maturity, language and cognitive development, communication skills and general knowledge. Preschool teachers completed a web based questionnaire for each child. The data were analysed using the unidimensional Rasch model. With exception for the domain of physical health and well-being the Rasch analysis showed satisfying psychometric properties of EDI after removal of some misfitting items. In these four domains no items showed disordered thresholds and the reliability was good, indicated by person separation index values of 0.73 or higher. Tentative analyses of Differential Item Functioning (DIF) showed that some items didn’t work invariantly across genders, suggesting that the DIF-items should be split into gender specific items. Due to the relatively small sample size the results can’t provide definite answers but tentative indications of the psychometric properties of the EDI-Sweden. As a whole the Rasch analysis provides ground for cautious optimism for large scale assessment of EDI-Sweden enabling more thorough and finer level analysis of the instrument.     

Factors relevant to a return to work: a multivariate approach

This paper contributes to the understanding of the factors that are significant for returning to work, and identifies factors that might be used early on in a period of sick leave to discern whether people are likely to work again. In the design, the multivariate Partial Least Squares (PLS) of Latent Structures method was used to analyse information from a questionnaire containing socio-demographic items, and information on symptoms, consequences on daily life, expectations and psychosocial factors. Data about the incidence of sickness over a four-year period was included. 121 adults aged 18-64 years on sick leave participated, irrespective of their diagnoses. A reliable prediction of a return to work required the combination of many factors: individual psychosocial instruments are not useful when considered in isolation. The strongest predictive factors for a return to work concern the individuals' expectations, the number of days of sick leave taken in the past, somatic disorders, and a high level of life satisfaction and sense of coherence. Many factors influence the outcome for people on sick leave: PLS analysis demonstrated that a multivariate approach using this method could predict the long-term outcome early on in a period of sick leave.