0

I have two hierarchical models with continuous variables. In the first block, one of the variables is significant. However, in the second block, when I add three more variables the first variable loses its significance. I've checked for multicollinearity using VIF and they're all less than 5.

What else could be contributing to this result and how can I address it?

Statistics_3280
  • 1
  • 1
    What is a "block" ? What is the purpose of your model (prediction or inference ?), and why do you need to "address it"? – Robert Long Dec 07 '23 at 15:55
  • 1
    This is almost a duplicate of https://stats.stackexchange.com/questions/52067/does-adding-more-variables-into-a-multivariable-regression-change-coefficients-o but I'm leaving it open as the part about significance is new. – Peter Flom Dec 07 '23 at 16:28
  • The block refers to each model (with the control and predictor variables). The first block only contains one predictor. I need to address it in terms of explaining the change in significance once the new variables are added. – Statistics_3280 Dec 07 '23 at 16:34

1 Answers1

0

First, don't concentrate on significance and non-significance. See e.g. Andrew Gelman's paper "The difference between significant' and not significant' is not, itself, statistically significant". Instead look at effect sizes and how they change. There is nothing really to address.

In addition, see this thread which answers other aspects of your question.

Peter Flom
  • 119,535
  • 36
  • 175
  • 383