| Parcels are composed of two or more indicators whose scores are mathematically combined, usually by averaging the values to create a single score. This thesis begins by analyzing the variance of simple parcels composed of averaged indicators. Then, we consider more complex parcels that contain correlated residuals or cross-loadings. Finally, we analyze the covariance between parcels and other model variables. While parceling is a somewhat controversial topic among researchers, this discussion intends to add to the understanding of the effects of parceling on the variances of the original indicators and to assist in sound decision-making for use of parceling in SEM analysis. |
