Re: Logistic Regression - variable become significant when including controls

In article <962ad690-4405-4530-9571-1694a5d695e9@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Stefan.Duke@xxxxxxxxx <Stefan.Duke@xxxxxxxxx> wrote:
Hello,
I conduct a logistic regression and if I just use one independent
variable (continious) this one is not significant (p-value .105). But
when I include two controls (one binary and one continous) it suddenly
becomes signficiant (p-value .03) and the coefficients becomes
stronger. (The model makes theoretical sense)
This kind of confuses me, because I knew when including controls,
variables that are bivariate significant can become insignificant, but
not the other way around. I do have a small sample (n=78).

Thanks for any advice!
Best,
Stefan

This is not at all unusual. Even directions of effects
can change when using controls.

Also, statistical significance means far less than you
seem to think it does.


--
This address is for information only. I do not claim that these views
are those of the Statistics Department or of Purdue University.
Herman Rubin, Department of Statistics, Purdue University
hrubin@xxxxxxxxxxxxxxx Phone: (765)494-6054 FAX: (765)494-0558
.

Relevant Pages