1

I have run a model for a data set that looks like the following:

site      date species         id success_rfid length success_seed seed_num
1  YB2 2/16/2019    BCCH 0110174217            0      2            1        1
2  YB2 2/16/2019    BCCH 0110174857            1      3            1        1
3  YB2 2/16/2019    BCCH 0110174217            0      1            1        1
4  YB2 2/16/2019    BCCH 0110174857            0      2            1        1
5  YB2 2/16/2019    WBNU 0110171AA5            0      2            1        1
6  YB2 2/22/2019    WBNU 0110171AA5            1      4            1        1

mod3 <- glmer(success_rfid ~ species + site + length + (1|id), family = binomial( link = "logit"), data = data)

This the output:

Fixed effects:
            Estimate Std. Error z value Pr(>|z|)   
(Intercept) -1.55910    0.59693  -2.612  0.00901 **
speciesTUTI  1.37388    0.66570   2.064  0.03903 * 
speciesWBNU  1.16029    0.76006   1.527  0.12687   
siteL3      -0.98732    0.63769  -1.548  0.12156   
siteYB2     -0.01193    0.74703  -0.016  0.98726   
length       0.25954    0.11648   2.228  0.02587 *

How do I go about interpreting what the output? I've done this for count data, but not binary data (Yes/No).

Rachael
  • 81
  • 1
  • 9

1 Answers1

2

The fixed-effects coefficients have the interpretation of log odds ratios (and log odds for the intercept). For example, the interpretation for the coefficient of speciesTUTI is the log odds ratio for the level TUTI versus the reference level for the factor variable species.

However, note that the inclusion of the random intercepts terms complicates to a degree this interpretation. Namely, these log odds ratios are conditional on the random effects. For more on this, check this post.

Dimitris Rizopoulos
  • 17,519
  • 2
  • 16
  • 37