Session Ready
Exercise

CFA, EFA & replication

When validating a scale, it's a good idea to split the survey results into two samples, using one for EFA and one for CFA. This works as a sort of cross-validation such that the overall fit of the model is less likely due to chance of any one sample's makeup. Let's practice this using the brand_rep survey. psych and lavaan have been loaded into your environment.

Instructions 1/4
undefined XP
  • 1
  • 2
  • 3
  • 4
  • Split the data into two halves using base R vector indexing. The odd-numbered rows will be used for EFA and the even-numbered rows for CFA.