0

Kindly need your advice on this important aspect.

I am running the logistic regression and testing the effect of a moderator variable on another indp.variable. As highlighted in 'red' the moderator is causing a very big change in p-value as well as 'B' of the indp variable 'prev_business_exp" so I am little bit concerned about it whether its plausible or not. The negative value of moderator variable is also shown. enter image description here

Muhammad Shahzad Hanif
  • 33
  • 6
  • This is answered for multiple regression at https://stats.stackexchange.com/questions/28474/how-can-adding-a-2nd-iv-make-the-1st-iv-significant/28493#28493. Everything said there applies to logistic regression. Related discussion and analyses can be found at https://stats.stackexchange.com/questions/31841/coefficients-change-signs/32237#32237. – whuber Dec 12 '17 at 14:36

1 Answers1

0

Whether it is plausible or not is really a substantive issue, not a statistical one. Interactions can certainly cause huge changes in the main effects so there is no statistical reason to be suspicious here.

Peter Flom
  • 94,055
  • 35
  • 143
  • 276