summaryExp | R Documentation |
For a glm
or glmer
model fit using the logit or log link function, displays
a model summary with the exponentiation function applied to the fixed effect
point estimates. Thus, estimates of effects on the log odds in a
logistic model become estimates of effects on the odds, which are
usually more interpretable. Similarly, estimates of effects on the log
counts in a Poisson regression become estimate of effcts on the
counts to facilitate interpretation.
summaryExp(model, confidence = 0.95)
model |
|
confidence |
desired level of confidence for the confidence interval. |
Note that once the coefficients have been exponeniated, the estimates now represent multiplicative effects. For instance, a variable associated with an odds ratio of 1.25 in a logistic regression means the variable is associated with 1.25 TIMES the odds ofsuccess. (This is different from regular linear regression, where the effects are additive.)
Consequently, a factor that has no effect on the response variable would have an estimate of 1 rather than 0. (Multiplying something by 1 leaves it unchanged, of course.)
95 effects are multiplicative, the CIs will NOT be symmetric about the point estimate on the original scale. Thus, it would be preferable to report the lower- and upper-bound of the CI (which is the current APA style for confidence intervals anyway) rather than its width.
model summary including exponentiated fixed effect estimates and confidence intervals.
data(VerbAgg, package='lme4') model1 <- glmer(r2 ~ Gender + btype+ (1|id) + (1|item), family=binomial, data=VerbAgg) summaryExp(model1)
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.