Exercise

Adjusting for non-normality

Multivariate normality is not an assumption of EFA, but it is an assumption of CFA. Change in model fit due to non-normality is more likely on smaller datasets, so let's work with 50 responses from our c_sat survey. This is saved in your environment as c_sat_50, as are the psych and lavaan packages.

Instructions

100 XP

Conduct Mardia's test for multivariate normality using the relevant psych function. The output will include a plot. Looking at the p-values, is this data normally distributed?
Using cfa(), fit the model, as stored in c_sat_model, to the data, using robust standard errors.
Print a summary of c_sat_cfa_mlr. Include standardized estimates and fit measures.