1
$\begingroup$

I ran a glm looking at population changes over years, and am unsure which results I need to include for a poster presentation. I understand the p value is important for significance but other than that what numbers are necessary and how would I report them?

img

I do not completely understand what each of the values mean. I am trying to show how populations have changed over the years. I am trying to say something in a university poster along the lines of "I found there to be an overall increase in the population from 2015-2021 (p=0.00155)" but I'm unsure what else would go in the brackets.

Improve this question
$\endgroup$
4
  • 1
    $\begingroup$ What is the reason you are doing the regression? What do you want to learn from your model? Those will drive what you want to include on the poster. $\endgroup$
    – Dave
    Commented Oct 7, 2021 at 22:11
  • $\begingroup$ Do you understand what each of those numbers mean? $\endgroup$
    – Firebug
    Commented Oct 7, 2021 at 22:43
  • 1
    $\begingroup$ Answer depends on a lot of things, for instance the audience for the presentation! Please edit to tell us! $\endgroup$
    – kjetil b halvorsen ♦
    Commented Oct 8, 2021 at 17:25
  • $\begingroup$ I do not fully understand what the number mean. I am trying to show changes in a population over years given patrol data. I am presenting at a university conference. $\endgroup$
    – Alex
    Commented Oct 8, 2021 at 23:07

2 Answers 2

Reset to default
2
$\begingroup$

The information you include on a poster depends on what you want to demonstrate and the contextual knowledge of your audience. Since your goal is to show the change in the population of lemurs over the years, there are a few basic things I would suggest you include:

  • A time-series plot showing the lemur population each year.

  • Your point estimate for the regression coefficient years and an appropriate confidence interval around this point. (A confidence interval is typically more useful than giving a p-value for the hypothesis test for a non-zero coefficient; see Wasserstein and Lazar 2016.)

You should ensure that you are happy with your model and make some overall commentary on the fit of the model and whether you think it represents the data well. If you are presenting to a statistically-skilled audience, and you wish to go into more detail on your model and its fit properties, you could further augment this with some information about the overall goodness-of-fit of your model (e.g., McFadden's coefficient of determination) and possibly some plots of deviance residuals. This would usually be excessive for a poster presentation but it might be part of an underlying paper or supplementary material for the presentation.

Improve this answer
$\endgroup$
1
  • 1
    $\begingroup$ +1 for mentioning the paper. $\endgroup$
    – User1865345
    Commented Oct 22, 2023 at 3:31
0
$\begingroup$

I would add the estimate and standard error, for sure, not just the p-value. The estimate of the increase per year and its precision are important, too.

Improve this answer
$\endgroup$
2
  • $\begingroup$ To be clear, that data would be from the "year" row, not the intercept right? $\endgroup$
    – Alex
    Commented Oct 12, 2021 at 15:38
  • $\begingroup$ Yes, though if you want people to be able to use the model, you need to give both, e.g. by presenting Lemurs=Intercept+Slope*Year. This is uncommon for posters, though. So you could also go with: "I found there to be an overall increase in the population from 2015-2021 (Slope(SE)=0.06(0.02), p=0.00155).". $\endgroup$
    – Aolon
    Commented Oct 13, 2021 at 5:00

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