Ordered Logit Model

a) SAS/Stata Tasks.

i) Use proc logistic in SAS and ologit in Stata to regress bmiC on the the female dummy and the age and exfreqwR variables.

ii) In Stata, use the brant command to test proportional odds assumption.

(The test of proportional odds is already in the default SAS output).

iii) In Stata, re-display the results showing odds ratios with the or option.

(Odds ratios are already in the default SAS output).

iv) In Stata, use the prchange, fromto command to request predicted probabilities, discrete change, and marginal effects.

b) Write-Up Tasks.

i) What do you conclude about the proportional odds assumption, based on the output from SAS and Stata?

ii) Use the factor change in odds approach to interpret the effect of being female. Calculate the odds ratios by hand (write the formula out and calculate the odds ratios with your calculator rather than relying on SAS and Stata) and interpret the result.

iii) Based on the predicted probabilities shown in the prchange output, show how to calculate and interpret in words the discrete change between men and women, while holding age and frequency of exercise constant at their overall sample means. Interpret the results.