Validating the knowlegde intensive working environment survey target 2.0 latent variable measurement interpretation
MetadataShow full item record
- Institutt for psykologi 
This thesis primarily concerns validation of the Knowledge Intensive Working Environment Survey Target (KIWEST) 2.0 measurement theory as a valid account of the observations generated by administering KIWEST 2.0 on a sample from its target population (N = 12170). A working model of validation is developed by combining validity- and latent variable theory. A distinction is drawn between latent variable measurement interpretations (weak claims) and identity interpretations (strong claims). KIWEST consists of 119 items, and its measurement theory specifies between 27 to 33 latent variables to account for observed (co)variation, depending on whether its multifaceted constructs are represented by single or multiple factors. Following data integrity treatment, 7643 cases and 118 items were retained. The method employed was maximum likelihood confirmatory factor analysis, employing the alternative-models strategy of Jöreskog (1993), comparing the fit of- and selecting among 16 nested models accounting for item (co)variation. The least parsimonious model was retained and subjected to evaluation of parameter estimates as evidence of validity. The results indicate that the model comprehensively account of the observations, but suffers from lack of parsimony. The discussion develops a number of suggestions for altering the interpretation to fit the observations (i.e., changes to the KIWEST theory; proximal remedies), and for altering the questionnaire to produce observations that fit the interpretation (i.e., changes to the KIWEST questionnaire; distal fixes). The conclusion of the thesis is that changes ought to be made to either the questionnaire or interpretation before proceeding with validation of the KIWEST latent variable identity interpretations.