Message/Author |
|
|
Hi! I would like to ask for help with running LPA. I am trying to run the syntax below but I get an error. When I include file is JADP.dat to upload the data, I get an error that the file does not exist. When I try to load the data with file is "C:\Users\LM\Desktop\LPA.dat"; it tells me that the program cannot run a multiple group analysis with mixture command. What I would like to do is to correct the syntax below and get it to work. I would appreciate your help in correcting the syntax. Thank you! title: 2-Class Latent Profile Analysis of Peer Nominations Wave 2 Data; data: file is JADP.dat; file is "C:\Users\LM\Desktop\LPA.dat"; variable: names=x1-x10; classes=c(2); analysis: type=mixture; PLOT: TYPE=PLOT3; SERIES IS x1(1) x2(2) x3(3) x4(4) x5(5) x6(6) x7(7) X8(8) x9(9) x10 (10); model: OUTPUT: TECH1 TECH5 TECH10; SAVEDATA: FILE IS myfile2c.dat; SAVE = CPROBABILITIES; |
|
|
What is the name of the data file? Is it jadp.dat or lpa.dat? |
|
|
It is LPA.dat I just realized the use of both names. Now, I tried to run it based on the lpa.dat which I saved on my desktop. It is still not working. How can I fix it? ** ERROR in DATA command The file specified for the FILE option cannot be found. Check that this file exists: LPA.dat title: 2-Class Latent Profile Analysis of Peer Nominations Wave 2 REAL Data; data: file is LPA.dat; variable: names=x1-x10; classes=c(2); analysis: type=mixture; PLOT: TYPE=PLOT3; SERIES IS x1(1) x2(2) x3(3) x4(4) x5(5) x6(6) x7(7) X8(8) x9(9) x10 (10); model: OUTPUT: TECH1 TECH5 TECH10; PLOT: TYPE=PLOT3; SERIES IS x1(1) x2(2) x3(3) x4(4) x5(5) x6(6) x7(7) X8(8) x9(9) x10 (10); SAVEDATA: FILE IS myfile2c.dat; SAVE = CPROBABILITIES; |
|
|
Search your hard drive for lap.dat to be sure it does not have another extension. If it is not in the directory you are running from, use the full path name. |
|
|
Hello, I am new to Mplus and after spending quite some time reading the user guide (v.7) and various papers, I'm left with some basic questions about model specification. I have seen some examples where the model is specified as: MODEL: %OVERALL% %c#1% [y1–y4*1]; y1-y4; %c#2% [y1–y4*-1]; y1-y4; (from example 7.10 in user guide) and MODEL: %OVERALL% and other examples with no model specified in the input. (And, of course, other specifications, but let's start here.) 1) I saw from the user guide what this syntax stands for, but I don't see in the guide the interpretation or the goal of this syntax. Can you please indicate a resource that explains how to choose the syntax for the goals of your research? 2) From example 7.10, I don't see how to scale this syntax to different class sizes, like 3 or 4 classes. If it helps, I wish to do LPA with 6 continuous dependent variables for 3 or 4 classes (to be determined), and eventually use 2 covariates. Thank you very much for your help. Best, Jessica |
|
|
The type of analysis you want to do determines how to create an input file. If you want to do LPA, follow Example 7.9. LPA is LCA with continuous latent class indicators. The CLASSES option is used to change the number of classes. The number in parentheses refers to the number of classes. We have several courses available on the website. Topic 5 is an introduction to mixture modeling. |
|
Chris Giebe posted on Tuesday, October 29, 2019 - 12:54 am
|
|
|
Hi, I'm trying to run an LPA with 4 cont. indicator variables (means of scales, range 1-5). According to 7.9 this is just like running an LCA but leaving out the Categorical option in the Variable command of my 4 indicators. However, I am getting this: *** ERROR in VARIABLE command CLASSES option not specified. Mixture analysis requires one categorical latent variable. I have tried to specify my indicators as categorical as well and am still getting the same error message. Can you please help? I'm not sure why this isn't working. |
|
Chris Giebe posted on Tuesday, October 29, 2019 - 1:06 am
|
|
|
NVM, I found the missing semicolon. Thanks anyways. |
|
|
Hello, I am running LPA. The AIC, BIC and ABIC values continually decrease with no clear tapering off. The loglikelihood repeats so tech11 and tech14 were requested using the optseed. This was done for 2-7 classes. However, VLMR p-values are non-significant. Nevertheless, the BLRT p-values are all significant from 2-7 classes. I would appreciate any advice on how to select which class solution from this point on. |
|
|
In these situations, you want to look for residual covariances, that is, violations of conditional independence of the variables given the latent classes. You can also try UG ex 7.22 where all residual covariances are included. |
|
|
Thank you I have a couple more questions: How do I request the residual covariances? What am I looking for in the output and what do I do if assumption of conditional independence has been violated? When I run ex 7.22 I get a warning that mplus cannot expand (autonomous - amot* - 1) for the first class Thanks again |
|
|
Q1: Use WITH Q2: If a WITH estimate is significant, the conditional independence is violated so you keep this residual covariance in the model. For your last question, we need to see your full output - send to Support along with your license number. |
|
|
What if all the covariances are significant? Is it problematic to keep them all in the model? Thank you for the offer to send the output but I figured it out the syntax problem. |
|
|
You can keep them all in the model. It won't be an LPA model anymore but that's ok. As you see in the references for UG ex 7.22, the model with all WITH's has a rich history in statistics. |
|
|
Thank you for your prompt reply. Should I then be writing this up as a mixture model as opposed to LPA? |
|
|
Yes. |
|
|
Hello, Thank you for the previous support. I have 23 continuous items. CFA for construct validity gave 5 factors. I saved the factor scores and used them to do LPA. Indices did not point to any solution so I relaxed the condition of local independence and have rerun 2-7 classes. There are several sig covariances which I would expect. Given that the class indicators are factor scores should I be following UG 7.27 (A factor mixture model) rather than UG 7.22 (mixture modelling with continuous variables)? |
|
|
Sounds like you should use UG ex 7.22. |
|
Back to top |