When The Hypothesis Is Invalidated, Does It Matter? A New Approach To Hierarchical Multiple Regression Analysis

摘要

This study original utilized hierarchical multiple regression analysis to isolate the main effects of organizational knowledge structure on the knowledge-creating process and to independently assess how classical strategy and procession strategy moderated the relationship between organizational knowledge structure and the knowledge-creating process. However the hypothesis of "a classical strategy will moderate the relationship between the symbolic-analyst-dependent organization and externalization within the new product developing (NPD) team's knowledge-creating process" was invalidated. Therefore in order to examine the reasons why the data showed the no support for this hypothesis, the taxonomy of independent t-test was recycled and sequentially the hierarchical regression multiple analysis was used to confirm the relationships between all variables on each single attribute. Perhaps the moderating effects of classical strategy or procession strategy were hidden or diluted in the data. Finally this study found classical strategy did positive moderate the relationship between symbolic-analyst-dependent organization and externalization in the female group and did support this hypothesis. Even the sample size was small and the representativeness might be insufficient, the method of recycled the taxonomy of independent t-test coupled with the hierarchical regression multiple analysis did assist to disinter the hidden evidences to support the invalidated hypotheses in original study.