Two surveys were mailed to 2,754 Professional employees of international and regional accounting firms in seven states. Responses were received from 504 professional employees (18.3%). Because of the time interval between the two surveys, the possibility of non-response bias may exist. To examine the possible existence of non-response bias, analysis of variance (ANOVA) was used to compare the means between the respondents from the first and second surveys. Except for perceived environmental uncertainty, there were no significant differences between the means of the first or second surveys on the independent variables (centralization, complexity, and formalization) and on the moderator variables (task complexity, and environmental uncertainty). The minimum sample size required to reject a equal .05 at the 90%
Probability interval, with r2 set at 10%, for 17 independent variables is 245 (Cohen & Cohen 1983). A sample size of 504 is more than adequate for power analysis at the 90% interval level.