HL_test: Hosmer-Lemeshow Goodness of Fit Test

In lbreg: Log-Binomial Regression with Constrained Optimization

Description

The HL decile-of-risk test. Validity of the test assumes that the number of covariate patterns is close to the number of observations which is violated when many observations have the same covariate pattern and several ties will impact the required ordering and grouping (by deciles) of observations. This is less likely when there is at least one continuous covariate. Not valid for grouped data.

Usage

 HL_test(object, g = 10) 

Arguments

object	object of class 'lbreg'.
g	number of groups

Value

A list with elements

X2	HL statistic
pvalue	p-value for the test from Chi Squared with df = g-2

Author(s)

Bernardo B. Andrade

References

Hosmer D W, Lemeshow S 2000. Applied Logistic Regression. New York, USA: John Wiley and Sons.

See Also

lbreg

Examples

require(lbreg)

# data preparation
data(PCS)
w <- PCS
w <- w[,-1]
w$race <- factor(w$race)
w$dpros <- factor(w$dpros)
w$dcaps <- factor(w$dcaps)

fm <- lbreg(tumor ~ ., data=w)

HL_test(fm)

Example output

   The Hosmer-Lemeshow Goodness-of-Fit Test 
$X2
[1] 28.89441

$pvalue
[1] 0.0003309147

$g
[1] 10

lbreg documentation built on Dec. 26, 2019, 5:06 p.m.