Post Hoc Power Analysis with Categorical PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Joe Anson posted on Sunday, April 20, 2008 - 2:32 pm
I need to conduct a post hoc power analysis for a model that includes both categorical exogenous and categorical endogenous variables.

The model has two categorical (dichotomous) independent variables, one continuous endogenous variable, and one categorical endogenous (outcome) variable.

Only one (of four) paths in the hypothesized model was found to be significant, though another approached significance.

I want to conduct a post hoc power analysis to determine whether sample size was sufficient for this model.
 Linda K. Muthen posted on Monday, April 21, 2008 - 8:11 am
See the following paper for a way to assess the power of one parameter:

Muthén, L.K. & Muthén, B.O. (2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 4, 599-620.
 Joe Anson posted on Tuesday, April 22, 2008 - 10:29 am
Thanks! I read the article you cited and I ran the analysis following the instructions in the User Manual. However, in my output, the values for %Sig. Coeff. are all 0.000 for the paths and 1.000 for intercepts and residual variances. I don't understand this, because % Sig. Coeff. is supposed to represent power. I re-ran the analysis with sample size 1000 and got the same result.

Any idea what I'm doing wrong?
 Linda K. Muthen posted on Tuesday, April 22, 2008 - 11:05 am
Please send your output and license number to support@statmodel.com.
 Angela Wolff posted on Monday, February 16, 2009 - 3:49 pm
Hi, I am trying to conduct a post hoc power analysis using categorical variables for the exogenous, endogenous and mediator variables. I have used threshold values of the exogenous variables. I carried out the first step discussed in the user's guide to save the dat file. When I proceeded to step 2, I received an error message that the analysis would not work due to the thresholds, so I removed the thesholds. However, now when I ran analysis I received an error message in that the data file was too long (and I'm assuming now does not match the dat file I created because I removed the thresholds). I am seeking direction in terms of how to conduct the post hoc power analysis with categorical variables using thresholds. Thanks,
Angela
 Linda K. Muthen posted on Monday, February 16, 2009 - 4:58 pm
Please send your files and license number to support@statmodel.com.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: