Page 1 of 1
FIMIX PLS
Posted: Wed Mar 14, 2007 11:28 am
by antoniocaballero
I would like to ask how to use the membership probabilities obtained after using FIMIX in order to run a post hoc analysis. I read Hann´s book and it is not clear to me.
From my point of vie there are two choices:
A) weighting each descriptive info of each case by these probabilities and running this for all the classes and compare them.
B) Assign each case to the class with highest probability and calculating the descriptives on the cases assigned to each latent.
Which of these is the right one? Maybe a third one...
Re: FIMIX PLS
Posted: Wed Mar 14, 2007 4:20 pm
by jjsailors
antoniocaballero wrote:I would like to ask how to use the membership probabilities obtained after using FIMIX in order to run a post hoc analysis. I read Hann´s book and it is not clear to me.
From my point of vie there are two choices:
A) weighting each descriptive info of each case by these probabilities and running this for all the classes and compare them.
B) Assign each case to the class with highest probability and calculating the descriptives on the cases assigned to each latent.
Which of these is the right one? Maybe a third one...
If you're interested in theory, then A. Well, almost A. You don't weight the descriptive info, you weight the respondents and then compare the descriptive info across classes.
If it's an applied project, then B.
FIMIX PLS: group comparision on descriptive variables
Posted: Sat Sep 03, 2011 5:14 pm
by gopshal
Sir, I am struggling how to perform:
A) weighting each descriptive info of each case by these probabilities and running this for all the classes and compare them.
B) Assign each case to the class with highest probability and calculating the descriptives on the cases assigned to each latent.
Please tell me in detail. My work is truck on this point. I would be very thankful if you explain the steps of the above methods.
I thank you in advance.