KHB Method: Obtaining P Value for Mediated Effects

[Oindrila Roy]

I am using the KHB method on STATA to analyze the mediated impacts of religious identity and religiosity via political predispositions of attitudes toward refugees. My model type is ordered logit, and the number of observations in my model is 2, 567.


I am writing to ask if there is a command that I can use to directly compute the p-values from the STATA output with components of difference that is obtained using the “disentangle” command. If not, is it safe to divide the coefficient by the standard error to obtain the z score and the associated p-value? Below is my code and results and any advice on the topic would be helpful:

[ericmelse]

Dear Roy,

The

that you refer to is not included.

Which is not increasing the likelyhood that someone is able to write a response to your question.

[Mike Lacy]

My colleagues and I examined confidence interval performance of -khb- with and ordinal logit model in:

Smith, E.K., Lacy, M.G. and Mayer, A., 2019. Performance simulations for categorical mediation: Analyzing khb estimates of mediation in ordinal regression models. The Stata Journal, 19(4), pp.913-930. See https://journals.sagepub.com/doi/abs...36867X19893638

We found that conventional asymptotic confidence intervals from -khb- worked quite well. We did not examine p-values, nor did we specifically look at the components revealed by the -disentagle- option.