R: Tidying methods for ordinal logistic regression models

tidy.polr {broom}

R Documentation

Tidying methods for ordinal logistic regression models

Description

These methods tidy the coefficients of ordinal logistic regression models generated by ordinal::clm() or ordinal::clmm() of the ordinal package, MASS::polr() of the MASS packge, or survey::svyolr() of the survey package.

Usage

## S3 method for class 'polr'
tidy(
  x,
  conf.int = FALSE,
  conf.level = 0.95,
  exponentiate = FALSE,
  quick = FALSE,
  ...
)

## S3 method for class 'polr'
glance(x, ...)

## S3 method for class 'polr'
augment(
  x,
  data = stats::model.frame(x),
  newdata,
  type.predict = c("probs", "class"),
  ...
)

## S3 method for class 'clm'
tidy(
  x,
  conf.int = FALSE,
  conf.level = 0.95,
  exponentiate = FALSE,
  quick = FALSE,
  conf.type = c("profile", "Wald"),
  ...
)

## S3 method for class 'clmm'
tidy(
  x,
  conf.int = FALSE,
  conf.level = 0.95,
  exponentiate = FALSE,
  quick = FALSE,
  conf.type = c("profile", "Wald"),
  ...
)

## S3 method for class 'clm'
glance(x, ...)

## S3 method for class 'clmm'
glance(x, ...)

## S3 method for class 'clm'
augment(
  x,
  data = stats::model.frame(x),
  newdata,
  type.predict = c("prob", "class"),
  ...
)

## S3 method for class 'svyolr'
tidy(
  x,
  conf.int = FALSE,
  conf.level = 0.95,
  exponentiate = FALSE,
  quick = FALSE,
  ...
)

## S3 method for class 'svyolr'
glance(x, ...)

Arguments

`x`	a model of class `clm`, `clmm`, `polr` or `svyolr`
`conf.int`	whether to include a confidence interval
`conf.level`	confidence level of the interval, used only if `conf.int=TRUE`
`exponentiate`	whether to exponentiate the coefficient estimates and confidence intervals (typical for ordinal logistic regression)
`quick`	whether to compute a smaller and faster version, containing only the term, estimate and coefficient_type columns
`...`	extra arguments
`data`	original data, defaults to the extracting it from the model
`newdata`	if provided, performs predictions on the new data
`type.predict`	type of prediction to compute for a CLM; passed on to `ordinal::predict.clm()` or `predict.polr`
`conf.type`	the type of confidence interval (see `ordinal::confint.clm()`)

Value

tidy.clm, tidy.clmm, tidy.polr and tidy.svyolr return one row for each coefficient at each level of the response variable, with six columns:

`term`	term in the model
`estimate`	estimated coefficient
`std.error`	standard error
`statistic`	z-statistic
`p.value`	two-sided p-value
`coefficient_type`	type of coefficient, see `ordinal::clm()`

If conf.int=TRUE, it also includes columns for conf.low and

glance.clm, glance.clmm, glance.polr and glance.svyolr return a one-row data.frame with the columns:

`edf`	the effective degrees of freedom
`logLik`	the data's log-likelihood under the model
`AIC`	the Akaike Information Criterion
`BIC`	the Bayesian Information Criterion
`df.residual`	residual degrees of freedom

augment.clm and augment.polr returns one row for each observation, with additional columns added to the original data:

`.fitted`	fitted values of model
`.se.fit`	standard errors of fitted values

augment is not supportted for ordinal::clmm() and survey::svyolr() models.

All tidying methods return a data.frame without rownames. The structure depends on the method chosen.

Examples

if (require(ordinal)){
  clm_mod <- clm(rating ~ temp * contact, data = wine)
  tidy(clm_mod)
  tidy(clm_mod, conf.int = TRUE)
  tidy(clm_mod, conf.int = TRUE, conf.type = "Wald", exponentiate = TRUE)
  glance(clm_mod)
  augment(clm_mod)

  clm_mod2 <- clm(rating ~ temp, nominal = ~ contact, data = wine)
  tidy(clm_mod2)

  clmm_mod <- clmm(rating ~ temp + contact + (1 | judge), data = wine)
  tidy(clmm_mod)
  glance(clmm_mod)
}
if (require(MASS)) {
  polr_mod <- polr(Sat ~ Infl + Type + Cont, weights = Freq, data = housing)
  tidy(polr_mod, exponentiate = TRUE, conf.int = TRUE)
  glance(polr_mod)
  augment(polr_mod, type.predict = "class")
}

[Package broom version 0.5.5 Index]