Exercise

# How Factorable is our Dataset?

As mentioned in the video, before reducing dimensions with EFA, we first need to make sure that our dataset is *factorable*. In other words, the first step in performing `EFA`

is to check whether it is even worth doing it. This dilemma is captured by the following question: Is there sufficient correlation among the observed variables of our dataset to allow for dimensionality reduction in the first place?

`hsq_polychoric`

, calculated with the `hetcor()`

function of the `polycor`

package, is the correlation matrix of the **Humor Styles Questionnaire** [`HSQ`

] dataset that you will be working throughout this and part of the next chapter.

In this exercise, your mission is to decide whether `HSQ`

is factorable enough to allow an `EFA`

.

Instructions

**100 XP**

- Apply the
`Bartlett sphericity test`

on`hsq_polychoric`

. For an`EFA`

to be considered suitable, the`Bartlett sphericity test`

result must be less than 0.05 to be deemed statistically significant. - The second test we will use is the
`Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy`

. Recall that the closer the value to 1 is the more effectively and reliably the reduction will be. Notice the indices for each variable in the output.