We assessed non‐response bias using wave analysis as suggested by Rogelberg and Stanton (2007). Participants who provided responses after the initial two weeks of data collection (when reminders were sent) were identified as late responders. Survey responses from these late responders were then compared with those given by those who responded in the first two weeks of data collection. T‐tests were conducted on a random selection of six survey items. We found no significant differences in responses, thus suggesting that non‐response bias was not of concern in this study.
We analyzed our data for missing values using PASW 18. We conducted Little's missing completely at random test. Results of this test suggest that missing data are not dependent upon either the observed data or other missing data. Missing values represented only 0.8 percent of the total survey item responses and the estimation, maximization algorithm was used to estimate and impute missing values.