3
$\begingroup$

I have to present my study as a poster and, in order to make my results more visual, I thought it was a good idea to report them as a forest plot. However, it is ok to use a forest plot to report the crude odds ratio (obtained using uni-variable binary logistic regression) or can this type of graphs only be used in meta-analysis?

$\endgroup$
1
  • 3
    $\begingroup$ Some people call them effect plots outside meta-analysis. $\endgroup$
    – mdewey
    Apr 20 at 12:12

1 Answer 1

5
$\begingroup$

You can. I have done it in peer-reviewed work such as Figure 1 at https://doi.org/10.1177/19485506231162401.

If you're an R user, there's a number of packages that call these "coefficient plots" or coefplot() for short. The popular package {arm} has a function for this. There's even a package called {coefplot}. How I make mine is by using emmeans::emmeans() to get point estimates and 95% confidence intervals, coerce that to a data.frame, and do {ggplot2} from there.

I also use these plots for best-worst scaling (i.e., MaxDiff) results. I recommend them in a vignette for my package {bwsTools} (for an applied example, see page 8 here).

$\endgroup$
2
  • 3
    $\begingroup$ Mark, could you check your doi id? $\endgroup$ Apr 20 at 14:53
  • 2
    $\begingroup$ @MarkWhite I took the liberty of correcting the DOI, as the link was pointing to an error page. $\endgroup$
    – J-J-J
    Apr 20 at 19:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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