Indexes and Scales Why use a “composite” measure of a concept? ▫ There is often no clear “single” indicator ▫ Increase the range of variation ▫ Make data analysis more “efficient”
Indexes vs. Sclaes Index ▫ Simply “add up” single indicators to form a composite measure Most common in social science Scale ▫ Assign score to pattern of responses (not all items are equal) Commonality: rank order composite measures
Index Construction Item selection Face validity Unidimensionality General/specific Variance considerations Scoring Variance versus “adequate number of cases” in categories Equal weight or not for the items in a composite index? Empirical examinations Bivariate versus multivariate Dealing with missing data Index validation
Scales/Scaling Empirically driven Responses can be scaled, not survey items/questions Reflect “intensity” of particular items Is there a pattern in how the items tap a concept? “easy and hard” questions
Scale Construction Key = not all indicators are equal ▫ “intensity structures” among indicators Better assurance of “ordinality” Examples ▫ Bogardus Social Distance Scale ▫ Thurstone Scale ▫ Likert Scale ▫ Semantic Differential ▫ Guttman Scale
Typologies Allows a researcher to summarize the overlap/intersection of two or more variables ▫ Typically driven by some combination of theory and data life-course criminality probation officer example Difficult to use a typology as the dependent variable ▫ Why do particular people fall into a particular typology (predicting typologies).