Generates a plain-language interpretation of an exponentiated regression
coefficient: the percent change in the expected count (or rate, if an offset
is present), with a 95% Wald confidence interval. For zero-inflated models,
the component argument selects the count or zero-inflation part.
A note is appended whenever the coefficient is not discernibly different
from zero (p > 0.05).
Arguments
- model
A fitted model returned by
poissonGLM(),negbinGLM(),tweedieGLM(),zeroinflPoissonGLM(),zeroinflNegbinGLM(),zeroinflTweedieGLM(), orcountGLM(). ForcountGLM, the best-fitting model is used automatically.- predictor
Character; the exact term name as it appears in the coefficient table (e.g.
"pctnonwhite10","EPAregion2"). Optional when the model has exactly one non-intercept term, in which case that term is used.- component
For zero-inflated models:
"count"(default) for the count component, or"zero"for the zero-inflation component. Ignored for Poisson and negative binomial models.