Message/Author 

Ali posted on Thursday, December 10, 2015  6:13 am



I generated the data with two latent classes,and two classes's latent trait are from the normal distribution. Th data are generated from the graded response model with 3 point. I run the mixture factor, but it gave the message "One or more pairs of ordered thresholds are not increasing in Class 2. Check your starting values. Problem with the following pairs: U1$1 (1.233) and U1$2 (0.595) U4$1 (1.158) and U4$2 (0.374) U5$1 (0.890) and U5$2 (0.410) U7$1 (0.603) and U7$2 (0.100) U8$1 (0.138) and U8$2 (0.278) U9$1 (0.228) and U9$2 (0.304) U10$1 (0.747) and U10$2 (0.395) Is it possible to run the mixture factor model with likert points? 


Mplus does not use the syntax U1$1 (1.233) Maybe you are trying to give starting values? If so, they need to be increasing. 

Ali posted on Friday, December 11, 2015  1:20 am



I use Mplus 7.11 version.Here is my codes: VARIABLE: NAMES ARE ID u1u10 member; USEVARIABLES ARE u1u10; CATEGORICAL = u1u10; CLASSES = c(2); auxiliary= member; ANALYSIS:TYPE = MIXTURE; ALGORITHM=INTEGRATION; STARTS = 100 20; MODEL: %OVERALL% f BY u1* u9u10; %c#1% f@1; [u1$1u10$1]; %c#2% f@1; f; [u1$1u10$1]; OUTPUT: TECH1 TECH8; After I run mplus, it showed me the message:Check your starting values. Problem with the following pairs: 


I don't think that model is identified. You need to fix the factor means at zero. See UG examples. 

Ali posted on Sunday, December 13, 2015  12:36 pm



I fix the factor means as zero and change another data set. Th codes are MODEL: %OVERALL% f BY u1u10; [f@0]; %c#1% f BY u1@1 u2u10; f; [u1$1u10$1]; %c#2% f BY u1@1 u2u10; f; [u1$1u10$1]; However, it still showed :*** ERROR in MODEL command One or more pairs of ordered thresholds are not increasing in Class 2. Check your starting values. Problem with the following pairs: U2$1 (0.195) and U2$2 (0.070) U3$1 (0.728) and U3$2 (0.331) U5$1 (0.982) and U5$2 (0.257) U6$1 (0.636) and U6$2 (0.487) U7$1 (0.392) and U7$2 (0.294) U9$1 (0.580) and U9$2 (0.164) U10$1 (0.542) and U10$2 (0.418) I didn't use any starting values for any thresholds. I simulated data from Graded response model with 3category for each item 

Back to top 