2

Suppose you have a linear model where the model is significant but some coefficients are not. How does one interpret the model when some coefficients are not significant?

phil12
  • 1,091
  • 2
  • 11
  • 21
  • You might have to explain a bit more about your model. Also, do you mean that some coefficients _are_ significant whilst others not? – Sootica Apr 08 '13 at 09:45
  • 3
    This is normal situation. [Similar question](http://stats.stackexchange.com/q/51583/3277) (you may ignore that the talk there is about stepwise selection) – ttnphns Apr 08 '13 at 12:58
  • Its a basic linear model with some coefficients significant at alpha = 0.05 and other coefficients not. – phil12 Apr 08 '13 at 15:32
  • You may find the following related thread helpful as well: [how can a regression be significant yet all predictors be non-significant?](http://stats.stackexchange.com/questions/14500/). – gung - Reinstate Monica Apr 09 '13 at 21:30

2 Answers2

4

If collinearity is not a major problem, you interpret "non-significant" coefficients exactly the same as you interpret "significant" ones, with confidence intervals.

Frank Harrell
  • 74,029
  • 5
  • 148
  • 322
1

This may be a sign of high collinearity among your predictors/covariates---if the overall or omnibus test is statistically significant but none of the individual covariates are significant. Check the condition indexes and their associated variance decomposition proportions.

Scott Millis
  • 61
  • 5