Test for Proportional Odds in the Proportional-Odds Logistic-Regression Model

The poTest function implements tests proposed by Brant (1990) for proportional odds for logistic models fit by the polr function in the MASS package.

Usage

poTest(model, ...)
# S3 method for class 'polr'
poTest(model, ...)
# S3 method for class 'poTest'
print(x, digits=3, ...)

Arguments

model: a proptional-odds logit model fit by polr.
x: an object produced by poTest.
digits: number of significant digits to print.
...: ignored.

Value

poTest returns an object meant to be printed showing the results of the tests.

References

R. Brant, "Assessing proportionality in the proportional odds model for ordinal logistic regression." Biometrics 46: 1171–1178, 1990.

Author

John Fox jfox@mcmaster.ca

Examples

if (require("MASS")){
    .W <- Womenlf
    .W$partic <- factor(.W$partic, levels=c("not.work", "parttime", "fulltime"))
    poTest(polr(partic ~ hincome + children + region, data=.W))
}
#> Loading required package: MASS
#> 
#> Tests for Proportional Odds
#> polr(formula = partic ~ hincome + children + region, data = .W)
#> 
#>                 b[polr] b[>not.work] b[>parttime] Chisquare df Pr(>Chisq)    
#> Overall                                               26.47  6    0.00018 ***
#> hincome         -0.0569      -0.0453      -0.1011      5.73  1    0.01666 *  
#> childrenpresent -2.0099      -1.6043      -2.7266     17.90  1    2.3e-05 ***
#> regionBC         0.1530       0.3420      -0.7228      2.69  1    0.10114    
#> regionOntario    0.2687       0.1878       0.0620      0.07  1    0.78563    
#> regionPrairie    0.4797       0.4719       0.3501      0.05  1    0.82760    
#> regionQuebec    -0.0711      -0.1731      -0.2974      0.07  1    0.79252    
#> ---
#> Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1