Message/Author |
|
|
Hello, I am planning to conduct an LCA and LTA using a mix of binary, categorical, and continuous indicators. Do you have any recommendations regarding resources for how to specify the input, as well as how to interpret the output? Thank you in advance. |
|
|
See UG ex7.11 for syntax. The interpretation is in line with regression analysis with those types of outcomes. We describe such regressions in Topic 2 as well as some more in Topics 5 and 6 of our video-taped short courses from Baltimore and Berlin. I can't point to application papers I'm afraid - others? |
|
|
Hello, thank you for your response. I will look at those resources. Related to entering binary data, do the response options have to be 0's and 1's, or can they be 1's and 2's? Or can the response options for some of the variables be 0's and 1's while the response options for other variables be 1's and 2's? Thank you again. |
|
|
They can be 0/1 or 1/2 or a combination. They are change to 0/1 before analysis. |
|
|
Hello, thank you for your response. So they need to be changed to 0/1 before analysis, correct? |
|
|
You don't have to make any change to your data. The program changes to 0/1 internally. |
|
|
Ah, thank you! |
|
|
Sorry, one more question -- I am assuming that in the "results in probability scale" output, category 1 represents the 0's and category 2 represents the 1's, right? Thank you again. |
|
|
Right. |
|
Gizem Samdan posted on Wednesday, November 06, 2019 - 4:36 am
|
|
|
I have ordinal and continuous indicators to identify classes of an emotional climate. The indicators are for example numbers of positive comments(POS). Ordinal variables are for example relationship (B) with three categories. N=92. I get warnings like:THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES MAY NOT BE TRUSTWORTHY FOR SOME PARAMETERS DUE TO A NON-POSITIVE DEFINITE FIRST-ORDER DERIVATIVE PRODUCT MATRIX. THIS MAY BE DUE TO THE STARTINGVALUES BUT MAY ALSO BE AN INDICATION OF MODEL NONIDENTIFICATION. THE CONDITION NUMBER IS -0.725D-18. PROBLEM INVOLVING THE FOLLOWING PARAMETER: Parameter 12, %CLASS#2%: [ NEGQ ] Do I have to change something? Are there assumptions due to the distribution of the continuous variables? Would you say I should categorize the continuous Variables? I would not like to do that because of information loss. data: file = Speechsamples.dat; variable: names = ID IS B TV S EM W EOI POS POSQ NEG NEGQ ZKB ZKC; auxiliary = ID; usevariables = IS B W POS POSQ NEG NEGQ ZKB ZKC; categorical = IS B W ZKB ZKC; classes = class(2); analysis: type = mixture; starts = 500 50; stiterations = 50; Thank you for your answer! |
|
|
We need to see your full output to say - send to Support@statmodel.com along with your license number. |
|
Back to top |