If this
is not the first book or article by the first author that you have
come across, you might know that he has been a constant (perhaps tiresome)
advocate of the argument that data is not ready to analyze until it
is clean and missing data has been dealt with. In fact, he wrote
an entire book about all the different—legitimate—things
a researcher can do to improve the quality of their data and the results
that come from analysis of those data.
Exploratory factor analysis
is no exception, but there are different issues, and different priorities
when dealing with EFA. Residuals, a common tool used to identify
outliers in inferential statistics, contradict the exploratory nature
of EFA and thus they cannot be used for EFA. Instead,
a number of other tools and techniques are used to identify overly
influential cases. Missing data is also increasingly problematic to
EFA. The factor analysis model cannot handle missing data and deletes
such cases by default. This process can introduce systematic error
and bias into the results. Therefore, missing data should be reviewed
and potentially dealt with, via various methods, prior to EFA.
In this chapter we aim
to provide a brief look at the issue of data cleaning for EFA. Specifically,
we seek to discuss two particular threats to EFA—outliers and
missing data—and the methods for dealing with them. A thorough
review of this topic and the associated SAS procedures is beyond the
scope of this book. Interested readers are encouraged to seek out
additional information on their own.