1. Incorporating prior theory in covariance structure analysis: A bayesian approach
2. This sample size is based on the policy mood time series reported by Stimson (1991), which runs through 1989. The time series has been updated, but for my purposes the original data are a better choice because the smaller number of observations in these data more clearly reveals the advantages of similarity coefficients. Each item is based on a comparison of the percentage of liberal and conservative responses to different survey items (for details see Stimson 1991).
3. Hunter's similarity coefficient may be interpreted as a correlation between vectors of correlations only at the extremes. That is, when |φij | = 1 this indicates perfect correlation between the correlation vectors ri and rj. However, when |φij | < 1 a correlational interpretation is no longer appropriate.