Message/Author |
|
Tess Yanisch posted on Tuesday, December 13, 2016 - 12:04 pm
|
|
|
Hello. I'm trying to get graphs of estimated means in a latent class analysis. My output says they are not available: PLOT INFORMATION The following plots are available: Histograms of sample values Scatterplots (sample values) Sample proportions and estimated probabilities I do not know why; I've done this approach before with a different dataset and gotten them. Relevant syntax: ANALYSIS: Type= mixture ; Starts = 100 25; Plot: type=plot3; series is V5252 (1) V5253 (2) V5254 (3) V5255 (4) V5256 (5) [and so on to 17 items] ; Thank you! |
|
|
Perhaps your variables are categorical in which case estimated probabilities are provided instead of estimated means. |
|
Tess Yanisch posted on Wednesday, December 14, 2016 - 8:52 am
|
|
|
Yes, that's it! Thanks. Is there a way to get a graph with estimated probabilities like the first one shown here? http://www.ats.ucla.edu/stat/mplus/seminars/IntroMplus/lca.htm My variables are 1-5 scales of agreement, so while I could arguably treat them as continuous, I would rather not. |
|
|
Yes, that's the graph those estimated probabilities show. You can show them for the highest category or sums of some of the highest categories. |
|
Tess Yanisch posted on Wednesday, December 14, 2016 - 12:55 pm
|
|
|
Got it! I was trying to find it through the Scatterplot menus. Thank you very much! |
|
Tess Yanisch posted on Wednesday, December 14, 2016 - 1:07 pm
|
|
|
Another question, sorry. I have an odd thing happening, and I'm not sure if I'm doing something wrong methodologically or if I'm interpreting the graph incorrectly. Some of my items are on a 1-5 scale; others are actually 1-6. On the plot of estimated probability for 6, I see some classes with very high probabilities on items where 6 is not actually an option. In fact, the estimated probabilities on these items are higher than for some of the items where it is! Am I misunderstanding what "category 6" is? |
|
|
Please send the output and gh5 files to Support along with your license number so we can look at what you are seeing. |
|
Back to top |