2
$\begingroup$

Say I have the following results from a glm (gaussian distribution) in R:

Coefficients:
             Estimate Std. Error t value Pr(>|t|)    
(Intercept) 210.63710   60.32309   4.220 0.000124 ***
sexm         -0.47692    0.31811  -1.460 0.167893    
year         -0.19032    0.03461  -4.194 0.000356 ***

I am testing the difference of my dependent variable between sex and year. In this case, there is no significant difference in my dependent variable between sexes, but there is a significant difference in my dependent variable between years.

I'm wondering how to report these results in a scientific paper?

Normally, in a scientific paper, you write something along the lines of: There was no significant difference in dependent variable between sex (Name of test; F = , p = ).

So for my glm, I would probably write: (GLM (maybe include the distribution and links here?); XXX, p = )

But what would go in the place of XXX in this case? From this output, we don't have an F-value, so I'm unsure of how to report this.

Any help would be appreciated!

Improve this question
$\endgroup$
3
  • 1
    $\begingroup$ I think that you can call it a t-test (or whatever) and report both the test stat and the p-value, since your paper should discuss the details of how you did your test. Ultimately, however, do it how others in that journal do it. $\endgroup$
    – Dave
    Commented Jan 31, 2020 at 1:17
  • $\begingroup$ to add to that^, you can run general F-test's comparing a reduced model to full model. For instance, you could test the null that only sex is important in modeling the dependent variable vs. the alternative that a model with sex and year does a better job. This would give you your desired f-test, which you can compute in R using anova("reduced", "full"). $\endgroup$
    – creutzml
    Commented Jan 31, 2020 at 3:25
  • $\begingroup$ How you report it would depend on the journal's conventions; these vary from area to area (and sometimes, journal to journal). Some journals say very little (presumably allowing a degree of leeway), others are extremely specific. $\endgroup$
    – Glen_b
    Commented Feb 2, 2020 at 3:42

0

Reset to default