Dear Colleagues
I am wondering why PLS can't handle non-recursive path models whereas covariance based SEM can do the same.
Please let me know
Thanks and regards
Sanjit
why PLS can't handle non-recursive models
why PLS can't handle non-recursive models
I want to use the SMART PLS for analyzing data for my dissertation.
- Diogenes
- PLS Super-Expert
- Posts: 899
- Joined: Sat Oct 15, 2005 5:13 pm
- Real name and title:
- Location: São Paulo - BRAZIL
- Contact:
Hi Sanjit,
In covariance based we have:
1) Model --> equations --> covariance matrix written as a function of the parameters of the model
2) Data set --> sample covariances
3) We minimize the difference between 1 and 2.
In this case beta12 is a different parameter from beta21, they don’t minimize the residuals in each equation.
In PLS based we have:
1) Model --> Equations --> LV scores --> Regressions (measurement and structural)
In this case the betas minimize the residuals in each equation (regression).
I am not sure if the non-recursive models aren’t implemented in the estimation procedure for convergence problems.
I hope this help.
Bido
In covariance based we have:
1) Model --> equations --> covariance matrix written as a function of the parameters of the model
2) Data set --> sample covariances
3) We minimize the difference between 1 and 2.
In this case beta12 is a different parameter from beta21, they don’t minimize the residuals in each equation.
In PLS based we have:
1) Model --> Equations --> LV scores --> Regressions (measurement and structural)
In this case the betas minimize the residuals in each equation (regression).
I am not sure if the non-recursive models aren’t implemented in the estimation procedure for convergence problems.
I hope this help.
Bido