| Message/Author |
|
|
| Eric M. posted on Monday, January 29, 2018 - 12:30 pm
|
|
|
Hello. I've seen similar post on this board. I apologize for the redundancy. However, I am still not clear. I have am using MPlus 8. To obtained descriptives statistics I have run Type = general and included sampstat for output. In my output I see that a correlation matrix is provided under ESTIMATED SAMPLE STATISTICS
1) Is there a way to obtain p-values for the significance of these correlations? I would use SPSS results, however, due to some missing data and how the missing data seems to be handled between the two produces slightly different correlation coefficients. Thank you! |
|
|
|
|
Using an example of 5 variables, specify an unrestricted model in the Model command:
y1-y5 with y1-y5;
and request a standardized solution. |
|
|
| Eric M. posted on Wednesday, January 31, 2018 - 11:02 am
|
|
|
| Well, that was extremely simple. Thank you. |
|
|
|
|
Hello Dr. Muthen,
I simply want to report a point biserial correlation matrix.
However, I couldn't get it work using y1-y5 with y1-y5 approach.
I could get a covariance matrix along with se using type=basic along with estimator = WLSMV. However, I am not sure how to transform it to a correlation matrix with p value.
Is there a way to get point biserial correlation matrix in Mplus? I am using a 100 MI of LCA class assignment correlated with my predictors at bivariate level, so I am hoping I can do this all in one piece within Mplus.
Is there a way I can make it happen?
Thank you for helping! |
|
|
|
|
| The approach of using y1-y5 with y1-y5 should work and you would look at the standardized solution. If you have problems with this, send output and data to Support along with your license number. |
|
|
|
|
| Greetings, thank you very much for the information. I am also using the WLSMV estimator and the simple strategy described here worked well (see the post from Dr. Muthen, January 29, 2018). I just wanted to add for users who may want to do the same that I first tried to list all my variables in the model statement (y1 y2 y3 y4 y5 WITH y1 y2 y3 y4 y5) and that did not work, but when I used the hyphenated statement (y1 - y5 with y1 - y5), it worked fine. Thanks again. |
|
|
| Back to top |
