1: Estimation of
residuals assumes an accurate model, and we cannot make that assumption
in EFA.
[return]
2: In this context, we will use
the term “outlier” to loosely describe a thing that
does not belong. There are more technical definitions for different
contexts, but you will have to indulge us here. There are also some
in the scholarly community who insist that removing outliers (and
indeed, data cleaning in general) is detrimental to the scientific
process. The first author has published many articles, given numerous
talks and, ultimately, wrote an entire book to empirically demonstrate
why this position is wrong-headed.
[return]
3: Please note, parts
of this section are adapted from Osborne, J.W., & Blanchard, M.
R. (2011).
[return]
4: One
option, used in this particular data set included having 20 questions
with four choices: A-D, with other questions scattered throughout
the test, and particularly near the end, with items that contain only
three (A-C) or two (A-B) legitimate answers. Students or respondents
choosing illegitimate answers one or more times can be assumed to
be randomly responding, as our results show.
[return]