Latent profile analysis in Mplus

#Mplus #statistics #Mplusforbeginners #LPA #LPA #mixture #latentclass #geiser #quantfish #statisticstutorials

Get Dr. Geiser's FREE INTRO TO MIXTURE MODELING WITH MPLUS MINI-COURSE:

FREE MPLUS FOR BEGINNERS COURSE:

LATENT PROFILE ANALYSIS WITH MPLUS COURSE:

LATENT CLASS ANALYSIS WITH MPLUS COURSE:

Dr. Geiser's BOOKS:

ON-DEMAND MPLUS COURSES:

GET IN TOUCH with DR. GEISER:

ABOUT THIS INSTRUCTOR:
Dr. Christian Geiser is a quantitative psychologist, author of two books on Mplus, and a leader in the development of latent variable techniques for complex data. With his accessible books and sought-after workshops, he has helped thousands of researchers and students around the world to achieve their analytic goals.

ABOUT QUANTFISH:

Some affiliate links may be included at no additional cost to purchasers.

[piquica]

Dr. Geiser, thank you so much for this video. I am using LPA for my dissertation and your playlist of LPA videos has been immensely demystifying. I very much appreciate you taking the time to offer these free resources!

[kenjisisland]

Hi Dr. Geiser, I wanted to say thank you for your videos. For years I had been interested in latent class/profile analysis and after watching your video I was able to perform the analysis and have meaningful, interpretable results that meet all the desiderata. So thank you very much for making these valuable instructional videos! Great channel, all the topics are excellent.

[deborahhillcone]

Kia ora Dr Geiser, thank you so much for these very helpful videos.

[brianstran]

Fantastic and informative video! Looking forward to the next LPA video.

[bianca_hbr]

Hi Dr. Geiser, thanks so much for all your hard work. Do you think it is possible to use the latent classes derived from an LCA to predict the latent profiles derived from an LPA using the same sample of people? Or would this run into a problem as the same participant would be in one group of LCA which is then predicting that same participant's membership in another class in an LPA? I’m struggling to find any examples of this in the literature but think it might be useful

[winnielam]

Hi Dr Geiser, Thanks for providing this tutorial. Which video are you referring to for the output? Many thanks.

[ganoniem]

Hi,

I am conducting a LPA for my master's thesis. I decided on a 4-class solution based on the fit indices and theory. Another aim of my study is to examine how class membership predicted a participants’ continuous outcome measure and whether gender moderates this. In my analysis plan I wrote that this continuous measure will standardised and added to the LPA model as distal outcome of each class. At the same time, gender will be added to the LPA model as a covariate on the distal outcome, to investigate its potential moderating effect.
My supervisor told me I could use an automated three-step approach (du3step) was taken, and perform Wald tests to compare class differences on the distal outcome and to examine the moderating role of gender.

I keep running into errors however and I am not sure whether this is the correct way to research my research questions. Could you perhaps help me?

[Chloe-sqqb]

Dear Christian, thanks a lot for this valuable video (and all the others)! Do you know if it is possible to to extract the profiles from latent variables (so, including the measurement model in the analysis) in LPA? Or do we necessarily have to use composite scores? I believe that FMAs allow to include a latent variable (when, for instance, all indicators belong to a common factor model) but I don't think the latent variable itself is used to extract the profiles. So is it even possible? Thanks!

[yuanzhang]

Hi Dr. Geiser, Thank you for the useful video. I have a question related to missing data. How do you handle input variables with missing data. Do you do list-wise delete or multiple imputation on the input variables? Is there any articles you recommend? Thank you.

[charlie_iphang]

Hi Sir, may I know how to add covariates as predictors in LPA in Mplus? Thank you very much!

[rachelzambrano]

Hi Dr. Geiser, thanks for the helpful tutorial. I keep getting an error message about the non-positive definite first-order derivative product matrix. I have adjusted the variables in my model to adjust for misspecification and employed troubleshooting techniques described by Muthen to no avail. Do you have any suggestions on how to deal with this error message?

[acitrix]

Can you walk us through Monte Carlo simulations in Mplus? An LPA example would be amazing!

[diana-jx]

hello. We are conducting LPA analysis with Mplus. I would like to inquire about a problem related to missing value handling. First, when classifying profiles, missing values were used as FIML. Next, when covariates, which are predictive factors that affect profile classification, were introduced and FIML was used to handle missing values, all missing values were listwise, significantly reducing the sample size. Therefore, I try to replace the missing values with MI, but an error continues to occur. Any help would be greatly appreciated.

[maravillalongo]

Hi Dr. Geiser, Thank you for this tutorial it has been really helpful! I have a follow up question to your comment to Hannah. I have a 5-profile model in which the best loglikelihood value was initially not replicated. I have increased the random starts to 500 and iterations to 50 and it was replicated. To ensure this was correct I increased the random starts to 1000 in which the output stated it was replicated as well. However, when I look at the actual values it is not the same best loglikelihood value that is replicated across the models with different random starts. I was wondering if you meant that the same best loglikelihood value should be replicated when increasing random starts, or that the output should reflect that the best loglikelihood value was replicated in general. Any guidance is appreciated! Thank you!

[siljebardstu]

Dear Dr. Geiser, thanks for all your videos! I am running the same analysis as this video, but would like to see whether the profiles differ based on gender. Do you have a video or syntax for Multigroup LPA that works for continuous variables (mean scale scores)? I have wathced your multigroup LCA videok, but you use categorical item indicators and not mean scale scores, so the syntax may not apply. Thank you! Best, SIlje

[hannahperkinsstark]

Hi Dr. Geiser, thanks so much for providing this tutorial! Quick question: how does one decide on the number of starts & iterations to use? I'm using LPA to study organizational commitment profiles and based on previously published studies, people have used anywhere between 1, 000 to 10, 000 starts. I also haven't been able to find any suggestions for selecting an optimal number of iterations. Is there a method for selecting these values? Any guidance would be appreciated!

[daniellin]

Thank you very much! Great video!
Is it normal that mplus usually take a long time to generate the output file for LPA? like 30 minutes or sometimes even one hour?

[matthewsavage]

Hi Dr. Geiser, Thank you for this video it has been extremely useful! Quick question relating to the progression from Latent Class Analysis to Latent Transition Analysis, how much more complex does the syntax become once you start adding longitudinal data to the model? Additionally, do you have any plans to provide any guidance as to how to conduct LTA in MPlus in the future? I for one would find this very useful. Thanks!