Message/Author 


Hello, I am novice in using Mplus. I'm trying to analyse my data from a longitudinal study with three phases using Latent Growth Curve. I want to analyse the course of a dependent variable included in the three moments and also want that the xaxis of the graph will contain the ages of subjects (variables for age were 3: subjects’ age on first phase, on second phase and on third phase = edat1 edat2 edat3). In addition we want to include a control variable throughout all analyses (a quantitative variable) and not really know how to add this control variable in the analyses (concretely the nse variable) We also want to analyse the trajectory of the subjects (on dependent variable) by sex; what will be the correct way to add this function? Is the Grouping function? Syntax is included in another message; can you tell me which parts are incorrect, considering our goal? 


variable: names = sexe edat1 edat2 edat3 nse; names = f11 f12 f13; grouping = sexe (1=male 2=female); usevariables=sexe f11 f12 f13; model: i s  f11@0 f12@1 f13@2; %WITHIN% s  f11 ON edat1; edat1 ON f11@0; s  f12 ON edat2; edat2 ON f12@1; s  f13 ON edat3; edat3 ON f13@2; analysis: type = random; plot: type=plot1; series = f11 (0) f12 (1) f13 (2); 


I think what you want is Example 6.12. You don't need a multilevel model unless you have clustering other than the repeated measures. The edat variables can be timevarying covariates as shown in the example. 

Núria V M posted on Friday, May 16, 2014  7:59 am



Thank you for your response. I followed the example 6.12 but the syntax fails. What happens? INPUT INSTRUCTIONS Title: Factor 1 SCARED; Data: File is FACTOR1_SCA_3FASES.txt; variable: names are sexe edat1 edat2 edat3 nse; names are f11 f12 f13; grouping = sexe (1=male 2=female); tscores = f11 f12 f13; analysis: type = random; model: i s  edat1 edat2 edat3 AT f11 f12 f13; st  edat1 ON f11; st  edat2 ON f12; st  edat3 ON f13; i s st ON nse; *** ERROR in MODEL command Time score variable F11 is used incorrectly in the model. 


You have names are twice: variable: names are sexe edat1 edat2 edat3 nse; names are f11 f12 f13; 

Núria V M posted on Wednesday, May 28, 2014  8:00 am



Finally we have performed a LINEAR GROWTH MODEL FOR A CONTINUOUS OUTCOME with our data. We indicate in the model that in the first phase the average of participants’ age is 10, in the second phase is 11 and in the third phase is 13 (from the first to the second phase a year passed and from the second to the third phase two years passed). The output is correct, but when we create the graph we only observed the possibility to obtain a curve for each subject. How could we create a single curve considering the dependent variable average score of all subjects ? (We want that the average of the dependent variable appear at the yaxis) We are interested in graphs with the trajectory of our dependent variable along three phases and on the other hand we want to observe the differences between the group of boys and the group of girls . We know that it is necessary to use the grouping function, but the graph does not represent at the same time the trajectory of the two groups. This is our sintax variable: names = f11 f12 f13 nse; analysis: type = general; model: i s  f11@10 f12@11 f13@13; i s ON nse; plot: type=plot1; series=f11(10) f12(11) f13(13); Thanks in advance! 


Use the PLOT2 setting and look at Estimated Means. 

Núria V M posted on Friday, May 30, 2014  7:48 am



When I perform my syntax (see below), two messages appear and I don't understand it properly variable: names = sexe nse f11 f12 f13; Grouping is sexe (1=male 2=female); analysis: type = general; estimator = mlr; model: i s  f11@10 f12@11 f13@13; i s ON nse; plot: type=plot3; series=f11(10) f12(11) f13(13); output: TECH4; 1)WARNING: THE LATENT VARIABLE COVARIANCE MATRIX (PSI) IS NOT POSITIVE DEFINITE. THIS COULD INDICATE A NEGATIVE VARIANCE/RESIDUAL VARIANCE FOR A LATENT VARIABLE, A CORRELATION GREATER OR EQUAL TO ONE BETWEEN TWO LATENT VARIABLES, OR A LINEAR DEPENDENCY AMONG MORE THAN TWO LATENT VARIABLES. CHECK THE TECH4 OUTPUT FOR MORE INFORMATION. PROBLEM INVOLVING VARIABLE I.  There is something that does not work? Is correct the data that appear in this output? 2)The chisquare value for MLM, MLMV, MLR, ULSMV, WLSM and WLSMV cannot be used for chisquare difference testing in the regular way. MLM, MLR and WLSM chisquare difference testing is described on the Mplus website. MLMV, WLSMV, and ULSMV difference testing is done using the DIFFTEST option.  Whether we have a longitudinal study with three measurement times, should we apply the SatorraBentler scaled chisquare difference test? How to apply? I am very grateful! 


1. You need to look at TECH4 and the results to see the problem with i. This message means the results are not valid. 2. You must use the strategy outlined on the website when you have MLR. 

Núria V M posted on Monday, June 02, 2014  1:01 am



Can you facilitate me a link where I could consult regarding the MLR function? 


http://www.statmodel.com/chidiff.shtml 

Núria V M posted on Tuesday, June 03, 2014  8:22 am



Considering this syntax, variable: names = sexe nse f21 f22 f23; usevariables are sexe nse f21 f22 f23; Grouping is sexe (1=male 2=female); analysis: type = general; estimator = mlr; model: i s  f21@10 f22@11 f23@13; i s ON nse; plot: type=plot2; series=f21(10) f22(11) f23(13); output: TECH4; Which would be the syntax for the alternative model in order to compute a chisquare test for nested models with the MLR estimator? 


First, you want to change your time scores to have zero somewhere, for instance i s  f21@0 f22@1 f23@3; Second, Mplus defaults to a chisquare test against an H1 which is an unrestricted mean and covariance structure. 


hello i have aproblem would you please guide me? i want to do a simple LGM, I have 86 individuals that half of them are men and the other half are wemon, but when i write my program in this way and run it, the program run truly but only 50 of my observation is read!!! can you help me in this matter? TITLE: this is an example of a linear growth model for a continuous outcome DATA: FILE= lgm.dat; VARIABLE: NAMES ARE id y1y5 sex country; USEVARIABLES ARE y1y5; IDVARIABLE IS id; missing=.; Analysis: ESTIMATOR=MLR MODEL: i s  y1@0 y2@1 y3@2 y4@3 y5@4; 


i forgot to say that when i delete "id gender country" of NAMES and USEVARIABLES line it run truly and my observation report truly 86 how can i do to solve my problem? 


The number of variable names must be the same as the number of columns in the data set and the name must refer to the information in the column. It sounds like you had more names than columns. 


Thanks for your response but I Have exactly 8 coulmns and i use 8 names too, I should say that 6 observations have 1 and two missings in y1 and y2, Do you think this missings make this problem? 


Please send the output, data, and your license number to support@statmodel.com. 

Back to top 