PLOT3 adjusted by covariates PreviousNext
Mplus Discussion > Growth Modeling of Longitudinal Data >
 Anonymous posted on Monday, February 14, 2005 - 8:35 am
Dear Prof.,

I would like to use covariates (i.e., aget6 aggpphy6) in plot to show adjusted mean curves. But I even could not get the window mentioned at page 471 in Mplus user guide. Actually, there are few options (10) in my Graph button though PLOT3 is used compared to the list given at page 465 for TYPE=PLOT3 where there are 18 options. What's wrong with my program.


The following is my program:

title: Girl-LTA;
data: file is c:\Mplus\Janice\data\JMplus2.dat;
variable: names are
stid sext6 aget6 aget7 aget8
physic6 physic7 physic8 mental6 mental7 mental8
aggpphy6 aggpsoc6 aggpto6 aggrphy6 aggrsoc6 aggrto6 aggto6 bully6
vicpphy6 vicpsoc6 vicpto6 vicrphy6 vicrsoc6 vicrto6 victo6 bulvic6
shper6 shvic6
aggpphy7 aggpsoc7 aggpto7 aggrphy7 aggrsoc7 aggrto7 aggto7 bully7
vicpphy7 vicpsoc7 vicpto7 vicrphy7 vicrsoc7 vicrto7 victo7 bulvic7
shper7 shvic7
aggpphy8 aggpsoc8 aggpto8 aggrphy8 aggrsoc8 aggrto8 aggto8 bully8
vicpphy8 vicpsoc8 vicpto8 vicrphy8 vicrsoc8 vicrto8 victo8 bulvic8
shper8 shvic8;
Auxiliary = aget6 aggpphy6;
useobservations are sext6==2;!girls

usevariables are physic6 physic7 physic8 mental6 mental7 mental8;

classes= c1 (2) c2 (2) c3 (2);
missing is .;
Analysis: Type=mixture; !estimator=wlsmv; !parameterization=theta;
Model: %overall%
C3#1 on C2#1;
C2#1 on C1#1;
!C3#1 on C1#1;

physic6 with physic7;
physic6 with physic8;
physic7 with physic8;
mental6 with mental7;
mental6 with mental8;
mental7 with mental8;

Model c1:
[physic6 mental6];
[physic6 mental6];
Model c2:
[physic7 mental7];
[physic7 mental7];
Model c3:
[physic8 mental8];
[physic8 mental8];
plot: type is plot3;
series is physic6-physic8 (*)| mental6-mental8 (*) ;

output: tech1; tech8; !residual; standardized; !modindices; patterns;
Savedata:file is GMMCov;
 Linda K. Muthen posted on Monday, February 14, 2005 - 9:54 am
If you send your input, output, and data to, we can take a look at it.
 Christian M. Connell posted on Thursday, May 05, 2005 - 8:41 am
I am having a similar problem running a two-part (semicontinuous) growth model (Ex. 6.616 of Mplus manual) with a covariate.

Relatedly, is it possible to treat the binary and continuous portions of the two-part model as separate series and plot both curves?

 Thuy Nguyen posted on Thursday, May 05, 2005 - 11:27 am
The estimated means and estimated probabilities plots may not be available for models with a covariate that require numerical integration.

To find out if this is the case, request the output option RESIDUAL. If the model estimated means cannot be computed, there will be a note in the RESIDUAL output section as to why. Consequently, plots will not be available.

If it is still not clear why you are not getting plots, please send your files to
 Mplususer  posted on Thursday, February 07, 2008 - 4:41 pm
Dear Prof.,
I used a Multiple Indicator Growth model for my data and tried to output a plot for estimated factor means vs estimated individual factor scores across time using Plot3 but failed.

I thought I could get a plot analogous to that of usual growth models, i.e., "estimated means and observed individual values".

Take 6.14 as an example, I wrote
TYPE = plot1 plot2 plot3;
series = f1-f3(s);
and only obtained histograms, scatterplots and estimated means.

Shall I save the factor scores and draw the plot in other software or my syntax is inappropriate.
 Linda K. Muthen posted on Thursday, February 07, 2008 - 4:49 pm
You can get an estimated mean plot in Version 5.
 Mplususer  posted on Thursday, February 07, 2008 - 5:22 pm
Sorry, I think I didn't describe my question clearly.
Yes, I am using Version 5 and got estimated mean trajectory plot but just couldn't get a plot where we could see both individual trajectories based on their factor scores across time and the estimated factor mean trajectory. For usual growth models, we could see observed individual trajectories vs. an estimated mean trajectory in mplus. For this multiple indicator growth model, I just couldn't get a similar plot. Still take 6.14 as an example,
f1 BY y11
y21-y31 (1-2);
f2 BY y12
y22-y32 (1-2);
f3 BY y13
y23-y33 (1-2);
[y11 y12 y13] (3);
[y21 y22 y23] (4);
[y31 y32 y33] (5);
i s | f1@0 f2@1 f3@2;
TYPE = plot1 plot2 plot3;
series = f1-f3(s);

Should I save the factor scores across time for each individual and draw the plot in other software or my syntax is wrong.
 Linda K. Muthen posted on Thursday, February 07, 2008 - 5:44 pm
If you want to plot the factor scores, you would need to save them and then you could plot them in Mplus or another program. I think it is more important to look at the growth curves for each observed indicator over time. If the shape of the growth curve, linear or quadratic, for example, is not the same for each indicator, then multiple indicator growth does not make sense.
 Mplususer  posted on Thursday, February 07, 2008 - 8:09 pm
Thanks for your instruction and suggestion!
 Mplususer  posted on Friday, February 08, 2008 - 7:54 am
If the individual growth curves generally exhibit three types of quadratic shapes for each indicator while these indicators are invariant across time, can I proceed with multiple indicator growth mixture models? If I can, can I just use the plot of factor mean trajectory vs. individual factor score trajectories instead of several plots for observed indicators? If I can't, what would you recommend?
 Linda K. Muthen posted on Friday, February 08, 2008 - 8:14 am
If the indicators of the factor exhibit the same type of growth, then using a multiple indicator growth model would be acceptable as long as measurement invariance of the factors across time has been established. You say all indicators have quadratic growth so this should be okay. If not, you would need to look at the observed variables separately.
 Mplususer  posted on Friday, February 08, 2008 - 8:51 am
OK. Thanks!
 Gareth posted on Wednesday, June 08, 2011 - 5:11 am
Is it possible to plot estimated growth curves for different levels of a covariate, in addition to the observed data curves, on the same plot? How is this achieved?
 Bengt O. Muthen posted on Wednesday, June 08, 2011 - 10:28 am
Use the plot menu option "adjusted estimated means and individual observed curves"
 Gareth posted on Thursday, June 09, 2011 - 1:14 am
I can only see 'Adjusted estimated means' and 'Observed individual values' in the list, as two separate options. Where can I find 'Adjusted estimated means and individual observed curves'?
 Linda K. Muthen posted on Thursday, June 09, 2011 - 2:15 pm
lease send the output and your license number to so we can see the exact situation.
 Scott D. Martin posted on Thursday, August 22, 2013 - 1:45 pm

I have a quick question about plot3. When I select "adjusted estimated means" from the "view graphs" button I am not getting a variable list. I would like to view estimates of the outcome by different levels of the covariate, but none are coming up on the list to select.

Any suggestions?
 Linda K. Muthen posted on Friday, August 23, 2013 - 10:52 am
When the “Covariate values for adjusted estimated means” appear, you must first name a covariate set. In the top right corner, enter a name for the covariate set next to the button “Name covariate set” and then click on that button. After you name a covariate set, the “Covariate values” section will be populated with the list of covariates for you to select.
 Scott D. Martin posted on Monday, August 26, 2013 - 6:31 am
Dr. Muthen,

Thank you for that reminder, that is exactly what I needed. I appreciate your help!

 Eric Knowles posted on Thursday, April 23, 2015 - 10:20 am
I have a growth model with observations at four time-points (e1980, e1990, e2000, e2010). I also have three time-invariant covariates (ent, ess, and wid). I'd like get plots of the growth curve at various levels of the covariates but I can't find "adjusted estimated means" in the Plots menu. Do I need to specify additional options in my code (see below)? Any guidance is greatly appreciated! I'm using Mplus 7.

Stata2Mplus conversion for wide file for Pia.dta
File is wide file for Pia.dat ;
Names are
e1980 e1990 e2000 e2010 ent_1 ent_2 ent_3 ent_4 ent_5 ess_1 ess_2 ess_3
ess_4 ess_5 ess_6 ess_7 ess_8 ess_9 solid_1 solid_2 satis_1 satis_2 centr_1
centr_2 centr_3 selfd_1 selfd_2 cf_1 cf_2;
Usevariables are
e1980-e2010 ent_1-ent_5 ess_1-ess_9 solid_1 solid_2 satis_1 satis_2
centr_1-centr_3 selfd_1 selfd_2;
Missing are all (-9999) ;
i s | e1980@0 e1990@1 e2000@2 e2010@3;
ent BY ent_1-ent_5;
ess BY ess_1-ess_9;
wid BY solid_1 solid_2 satis_1 satis_2 centr_1-centr_3 selfd_1 selfd_2;
i s ON ent;
i s ON ess;
i s ON wid;
ess WITH ent wid;
ent WITH wid;
Type is Plot3;
Series = e1980 e1990 e2000 e2010 (*);
 Linda K. Muthen posted on Thursday, April 23, 2015 - 12:20 pm
Please send the output and your license number to
 Arthur "Trey" Andrews III, Ph.D. posted on Thursday, July 27, 2017 - 11:33 am
I'm estimating a piecewise GCM and I'm wanting to overlay estimated and actual means with individual responses or growth curves. I haven't been able to do this with the plots available for my model (e.g., it always produces separate plots for each growth function). Is there a way to do this while plotting both growth functions in the same graph in Mplus? Perhaps I'm missing it in the available Mplus resources. Thank you!
 Xu, Man posted on Friday, August 24, 2018 - 5:56 am
Dear Dr. Muthen,

I think the answer is no, but just to be sure: if I have factors based on multiple indicators at each wave. When I estimate growth model based on those factors with multiple indicators, can I make plot 3 with the growth curve of these factors in Mplus?

Thank you!

 Bengt O. Muthen posted on Monday, August 27, 2018 - 1:49 pm
Try it.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message