Confidence Intervals for Pairwise Comparisons of Average Marginal Effects

[Guest]

Hi, when using the following commands:

asdoc regress lnincome i.race age age_squared i.region i.highestqualification i.fulltime_parttime i.occupation i.marital race##c.dependentchildren i.wave, cluster(pidp), if sex==2

asdoc margins race , dydx(dependentchildren) pwcompare

When showing the output of the pwcompare command, only confidence intervals, but not p-values are shown. Therefore, can this be done with p-values, to show if the co-efficients of the pairwise comparisons of the margins from the interaction term are statistically significant? If so how would the second line of code be adapted?

Thanks

[Attaullah Shah]

Please post a sample of your data using dataex (from SSC) and the actual code you used to estimate the marginal effects. This will give us some ideas of what problem are you facing and what possible solution can be articulated to solve that problem.

[Godsaviour Christopher]

I have the same issue, how can I know whether the contrast is significant or not?

[Clyde Schechter]

Re-running the -margins- command as -margins female, dydx(muslim) pwcompare(effects)- will expand the results to include a p-value.