Selecting the best model
Now use your knowledge of finding and interpreting absolute and relative model fit statistics to select the best model for your data. When I introduced this dataset I said that the items were theorized to load onto five factors, but you may have noticed that your scree plot indicated six factors. You might be wondering which you should trust. Not to worry - you can use fit statistics to make am empirical decision about how many factors to use.
First, you'll use the bfi_EFA dataset to run EFAs with each of the hypothesized number of factors. Then, you can look at the BIC, which is a relative fit statistic, to compare models. Remember, the lowest BIC is preferred!
Diese Übung ist Teil des Kurses
Factor Analysis in R
Anleitung zur Übung
- Run both EFAs on the
bfi_EFAdataset - one with five factors according to the theory, and one with six factors according to the eigenvalues. - Take a look at the BIC value for each of the models. The BIC is stored in the
BIClist element of the results object.
Interaktive Übung
Versuche dich an dieser Übung, indem du diesen Beispielcode vervollständigst.
# Run each theorized EFA on your dataset
bfi_theory <- ___(___, nfactors = ___)
bfi_eigen <- ___(___, nfactors = ___)
# Compare the BIC values
___
___