| Message/Author |
|
|
|
|
I am performing a LPA with 8 continuous indicators. I have found the best fitting primary model (6 class) and now I want to test for equivalency of the profiles between (1) males and females; (2) age; and (3) race/ethnicity.
I found this article which helped me with the coding and selection of the ru3step approach
https://www.tandfonline.com/doi/abs/10.1080/10705511.2014.915181
(1) I want to make sure that I am selecting the correct statistical approach (dustep3)
(2) If that is the correct approach, do you know of any resources on how to interpret the findings? I can only find simulations and not instructions on how to interpret. I think the section that follows "EQUALITY TESTS OF MEANS ACROSS CLASSES USING THE 3-STEP PROCEDURE
WITH 5 DEGREE(S) OF FREEDOM FOR THE OVERALL TEST" is indicating that classes are sig different between males and females according to the Overall test Chi Square test =340.949 p= 0.000 (and then details regarding which specific classes are different follows), but I am not sure this is correct since I do not have any documentation to support this interpretation.
Thanks in advance for your help! |
|
|
|
|
(1) We recommend DCAT - see Web note 21's last table.
(2) Your understanding is correct. Perhaps one of our papers under Papers, LCA would have a thorough interpretation - we thought the output was transparent. |
|
|
|
|
Thank you so much Dr. Muthen. Does DCAT do the 3 step approach? Or only a single step? If no, would DU3STEP still produce accurate estimates since the variable is only coded as 0 or 1?
Thanks in advance,
Jillian |
|
|
|
|
| DCAT does the 3-step. |
|
|
|
|
Hello again,
Thank you for your clarification. By using the DU3step or DCAT approach, does this:
(1) adjust the original LPA model for the auxiliary variables (ie auxiliary variables become covariates)?
Or (2) simply provide you with mean differences or differences in proportions across classes for the auxiliary variable?
For example, if I were to include gender as an auxiliary variable (using DCAT) would this mean my final LPA model is “adjusted” for gender? Or is my LPA unaffected by the inclusion of gender as an auxiliary variable and rather just telling me significant differences in proportions of males and females across the model?
I am wondering at what point do I know if I need to stratify by gender prior to determining profiles?
Thanks in advance,
Jillian |
|
|
|
|
| Interpretation (2) is the correct one. |
|
|
| Back to top |
