0

in my thesis i had a significant interaction effect between two variables and their main effect is insignificant. A variable had a p-value of 0.636 B variable had a p-value of 0.219 A*B had a p-value of 0.022 A is employee training practices, B is Autonomy, and the response is explicit knowledge creation. my thesis is about evaluation of the impact of quality management practices on knowledge creation in a public corporation in Jordan.

how can i explain this result ???

Ahmad
  • 51
  • 4
  • This is fairly common, and I have heard as a rule of thumb that you can/should avoid interpreting main effects if interaction effects are significant. Thinking intuitively about your problem, it does seem like it might make sense that in isolation, training practices and autonomy might not contribute *statistically* significantly to knowledge creation but together they help a lot. Sort of like sleeping poorly and eating poorly on their own contribute a little bit to bad health, and together can make you sick quickly (trivial example). – HFBrowning Mar 14 '14 at 19:06
  • For a more statistical treatment, maybe check out this author's discussion on disordinal interactions: http://pages.uoregon.edu/stevensj/interaction.pdf – HFBrowning Mar 14 '14 at 19:07
  • @HFBrowning: Less a rule of thumb; more an observation that when the effect of a unit change in the predictor $x_1$ depends on the value of another predictor, $x_2$, the effect of a unit change in $x_1$ in the particular case when $x_2=0$ is not usually of especial interest. – Scortchi - Reinstate Monica Mar 14 '14 at 20:50

0 Answers0