| Message/Author |
|
|
| Anonymous posted on Thursday, September 30, 2004 - 12:51 am
|
|
|
| I have a latent variable (factor) with 4 obseerved indicators. I specify the 4 variables as categorical (7 categories). I say in the model: f1 by y1-y4; I want to use the latent variable also as categorical. How is it possible to do this? |
|
|
|
|
| The factors in factor analysis are always continuous. If you want to work with a categorical latent variable, then you would be in the area of mixture modeling -- latent class analysis. The objectives of the two methods are different. |
|
|
| Anonymous posted on Friday, October 01, 2004 - 12:53 am
|
|
|
| What I wrote with the categorical latent variable was nonsense. I work with a continous one. Is it possible to see it like a z-transformation? Cause when i regard the "thresholds" and the proportions of the categories, these are the same values like the z-values and proportions in a table for a standardized normal variable. |
|
|
|
|
| I think Web Note 4 will answer your questions. In this note, it is clear that y* variables behind the categorical indicators are like z variables. |
|
|
| Back to top |
