Testing a single mediator model involving binary dependent variable in Stata (Version 17)

[@camillabjelland6947]

Dr. Crowson, thank you for your videos. Just to clarify, my IV is binary and so in the code I can just use the variable without the i. prefix?

[@omidmahdieh7882]

Thanks. informative and interesting. Could you please, share a video demonstration about how to handle "control variables" in research?

[@avraam13]

Thank you Mike for your great video! A quick question please what if i have more than one mediator (two or three mediators)? For example in my model I have one IV = continuous, three Mediators = Continuous and one DV=binary? Thank you

[@Sheng-m8n]

Great! Like this video! Do you have any examples showing Stata handles more than 1 mediators?

[@heathersky8218]

hi i'm sorry if this sounds stupid but, the result we want from this is Indirect Effect right? So if 0 is out of confidence interval, it means that Political does mediate the relationship right?
Also, do you treat confounder the same with covariate?
thank you so much in advance

[@user-eo4tn5ms4y]

Hi Dr Crowson, in the first example here, path a is a linear regression coefficient whereas path b is a logistic regression coefficient and so both of the regression coefficients have different interpretations and scales (change in a unit vs change in log-odds). Is it still acceptable to multiply path a and b to obtain the indirect effect? I am struggling to find an answer to this, your perspective would be very much appreciated. V

[@3minuticolprof]

I have the same doubt, and cant find an answer

[@sqworld9420]

Hello
I have a model where the independent variables are binary and the mediator and dependent variables are continues is there a do file for that ?

[@camillabjelland6947]

Did you figure out the answer? He said above that it cannot be treated as a factor variable as in "i." before. My code is working fine without the "i." that is my independent variable but I want to make sure I am doing it correctly.

[@lisakramer1105]

Do you have a video on how this is done in SPSS? I'm testing a moderated mediation model with a binary outcome variable

[@quetran8098]

Instead of bootstrap, with the same model, could you please guide how to handle BRR for complex survey data in STATA? Does earlier STATA versions (not 17) work? Thanks for your channel!

[@elizabethbaumler8371]

This is a great video, such a time saver. I downloaded the do file and am getting an error just after : *Path a coefficient
. di e(b)[1,1]
invalid syntax
r(198);
Do you have any idea what might be the issue? It runs the model but stops at this line. I did not alter anything other than adding the variable names in. Thanks for any help. This will be very useful for me if I can get it working.

[@elizabethbaumler8371]

Update, I was running Stata 15, updated version, now it work great! Thanks!

[@samuelwakuma8088]

what if both the outcome and mediator are binary?

[@mikecrowson2462]

Hi Samuel, right now the syntax is just set up for a binary outcome and a continuous mediator (in the manner by Hayes with the Process macro). I hope to put something out there soon to address the model you are asking about. Cheers!

[@mdhasibreza5161]

@@mikecrowson2462 Dr Crowson, thanks a lot for helping the community through your lectures. I am trying to use ANCOVA model in the Latent Change Score specification and estimate the mediation effect (ref: Valente and MacKinnon, 2017). However, I could not figure out how to do that in STATA. I have both the pretest-posttest measures of the outcome variable and the mediator from the treatment and control households. All the variables are binary. Could you please give me any idea regradig this?

[@sqworld9420]

what did you do how did you solve the problem please?
i have the IV binary data and the mediator and dependent variables are continues
@@mdhasibreza5161

[@ElizabethBodunde]

@@mikecrowson2462 Hi Mike, I am just wondering if you have been able to put together something regarding this., thank you.