1
$\begingroup$

I conducted a regression analysis using R's lm() function. One of the independent variables shows no significance (p = 0.89), which contradicts the hypothesis that is should have a significantly positive effect on the dependent variable.

How do you interpret that? Can you say that it has no positive effect on the dependent variable, just because it is not significant - even though it is not significantly negative?

$\endgroup$
2
$\begingroup$

You can't say that. You can only say you've not been able to confirm with a reasonable degree of confidence that it has a positive effect. Looking at a confidence interval of the regression coefficient will be more informative - you may be able to say you're confident it has a negligible effect.

$\endgroup$
  • $\begingroup$ Scorthi, thanks for your reply. I understand your point. I have a question regarding the word count devoted to insignificant regressors: Say I have two regressors: one that is significant and goes against the hypothesis and another that is non significant. How much would you write for interpreting each regressor: 50/50? I was going to do 80/20, writing mostly about the significant regressor and just a little about the non significant one... What do you think? $\endgroup$ – Tomi Seus Apr 12 '13 at 16:43
  • 1
    $\begingroup$ What an odd question! It depends entirely on your purpose in carrying out the regression, as common sense should tell you. But in general I tend to write/talk more about why expected effects weren't found than why expected ones were. $\endgroup$ – Scortchi - Reinstate Monica Apr 12 '13 at 18:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.