Internet Learning Volume 1, Number 1, Fall 2012 | Page 11

10

Internet Learning

Archival data were acquired from the APUS Office of Institutional Assessment through a
request for data. Information requested was provided to the researchers through an Excel file for
all end-of-course survey responses. Data sets were examined for integrity, resulting in the
removal of 10,028 entries, leaving 103,166 data sets in the final analysis.
Data were analyzed in three separate linear regressions, using the forward method of
entry. In the regressions, mean scores for aggregated Teaching, Social, and Cognitive Presence
items served as the criterion variable. Predictor variables consisted of a binary variable
representing gender, a binary variable representing traditional versus non-traditional status, and
four dummy binary variables representing ethnicity (Caucasian, Hispanic, Black, and Asian). As
the use