glmer in R: summary and tab_model giving different values

Question

I have been running glmer in R and using summary to extract values for write up:

model.CERT=glmer(certain2 ~ cuecong2 + hmaxCS*rotcat2 + (1|ParticipantPrivateID), data=data,family=binomial(link = "logit"));

certain2=categorical predictor (labelled factor, order=TRUE); cuecong2=binary predictor;
hmaxCS=continuous predictor;
rotcat2=categorical predictor (labelled factor, order=TRUE)

I've just started using tab_model to created nicer tables and the "estimates" are different. I'm finding it hard to work out what is being reported in the two different approaches. and which one to use?

sjPlot::tab_model(model.CERT,
                  show.re.var= TRUE,
                  show.stat = TRUE,
                  show.se = TRUE,
                  show.p = TRUE,
                   p.style = "stars",
                  digits = 3,
                    string.se = "se",
                   pred.labels =c("(Intercept)", "Cue", "EdgeDis","TexRot.L","TexRot.Q","EdgeDis*TexRot.L","EdgeDis*TexRot.Q"),
                   dv.labels= "Certainty")

Produces enter image description here

whilst

summary(model.CERT)

produces

                 Estimate Std. Error z value Pr(>|z|)    
(Intercept)       1.35113    0.15210   8.883  < 2e-16 ***
cuecong2incong   -0.33386    0.08696  -3.839 0.000123 ***
hmaxCS           -0.39722    0.04549  -8.731  < 2e-16 ***
rotcat2.L         0.03428    0.09227   0.372 0.710225    
rotcat2.Q        -0.01933    0.06313  -0.306 0.759496    
hmaxCS:rotcat2.L  0.13963    0.09169   1.523 0.127781    
hmaxCS:rotcat2.Q  0.14973    0.06310   2.373 0.017656 *  

I have updated R, sjplot and lmerTest today and the problem persists. Are the estimates given by summary not odds ratios?

Thank you.

From searching 'summary' possibly uses Satterthwaite and 'tab_model' is using Wald? Whici is the correct stat to be reporting?

Answer 1

After some more reading the tab_model odds ratio can be calculated from the summary estimate using exp(). Summary value is the raw coefficient.

Answer 2

Alternatively, not using the nicely formatted tab_model() function, there is always the option to use the summary() function and manually calculate the CI

See maybe this thread.