A Comparison of Moderated Regression Techniques Considering Strength of Effect

Journal of Management - Tập 8 Số 2 - Trang 35-47 - 1982
Arthur L. Darrow1, Douglas R. Kahl2
1Bowling Green State University
2The University of South Dakota

Tóm tắt

Moderated regression is a commonly used technique within the behavioral sciences. The power of such analyses, however, is dependent on the strength of the moderator relationship and the distribution of the moderator variable. This study compares the traditional moderated technique with a technique designed to increase the probability of the indication of a moderator variable. The results indicate that, often, the detection of moderator variables is not so much dependent on their existence but, rather, is dependent on their strength. The results also indicate that the ability to detect moderators also depends on the distribution of the moderator variable. Finally, the results indicate that a higher probability of detecting a moderator exists if the interaction is entered into the regression first.

Từ khóa


Tài liệu tham khảo

10.1016/0030-5073(82)90254-9

Blood, M.R., 1977, Where have all the moderators gone? The perils of Type II error

Cohen, J.B. Sociocultural change and behavior patterns in disease etiology: An epidemiology study of coronary disease among Japanese Americans. Doctoral dissertation, University of California at Berkeley, 1974.

Cohen, J.B., 1975, Applied multiple regression/correlation analysis for the behavioral sciences

McGrath, J.E., 1979, Research in Organizations

Neter, J., 1974, Applied linear statistical models

10.1111/j.1540-5915.1979.tb00009.x

10.1177/001316445601600205

10.2307/3150970

10.1037/h0031543

Thông tin
Thông tin xuất bản

Journal of Management

Tập 8 Số 2

35-47

Thông tin tác giả