ABSTRACT
Consumer psychologists create knowledge by developing and testing theories about relationships among constructs. A factor analysis will allow determining for each individual scale item whether there is more variance shared with the latent construct than variance that is unique/error. The average variance extracted represents the amount of variance that is due to the latent construct. The observed responses are an effect of the latent construct, and these measures are referred to as effect indicators or “reflective” measures. Classical measurement theory is based on the premise that there are latent and that these measures include both some “true score” variance and some “error” variance. Domain-sampling model provides the basis for important concepts in developing measures and assessing their validity. Once the researcher has a set of scale items that appear to evidence the properties of content validity and unidimensionality, it is time to assess reliability and discriminant validity.
