Tuesday, May 12, 2009

Regression t-value

For my question one examination, I ran a simple linear regression as the first step of a mediation model. I included the overall model's significance (R square, F-value, p-value), as well as a sentence that included the unstandardized beta, standard error of the unstandardized beta, standardized beta, t-value, and p-value for the t-value. A comment on my examination said that I could exclude the t-value. I have two questions related to this comment.

First, does this mean I remove the p-value from this sentence, too, or do I need it there for the betas and standard error? I include the F's p-value earlier.

Second, when I run the multiple regression for the third step of my mediation model, will I need to report the t-values then, as there are multiple predictors and, thus, t-values?

Thanks!

2 comments:

Mari said...

The t's are typically not reported in a regression. The p's for the betas must be reported, however.

Justin said...

Thanks!