Applying post-stratification weights in data

[crmelchior]

Hello,

I am using seminr to analyse data from a survey with PLS-SEM. However, I would like to apply some post-stratification weights to each observation (survey answers). I generated the weights using the anesrake library, so I already have a weight factor for each observation.
Please note that post-stratification weights is an entirely different concept than the path weights, correlation weights, regression weights, etc, that are currently supported by seminr.

I was able to perform simple statistical analysis with the weighted data using the wpct function from the weights package and the survey package, however it is not possible to gather the weighted data to use in seminr.

I was also able to apply the weights in the lavaan package using the sampling.weights parameter of the sem function, however I prefer to use PLS-SEM instead of the factor-based SEM that lavaan does.

Is there any way to apply the post-stratification weights in the data and perform PLS-SEM using the seminr package?

[soumyaray]

Greetings,

I'm afraid we don't incorporate sample weighting into PLS for seminr yet. It could be a next priority in fact. Is this close to what you have in mind? https://www.sciencedirect.com/science/article/abs/pii/S0263237316300688

[crmelchior]

Hello @soumyaray,

Thanks for the precise answer.
Yes, I believe that might be exactly what I need.

Just so I can adjust my expectations, do you have an idea of a time frame for this feature's implementation?
Depending on your timeline I could still try to use it in my current research. Otherwise I will probably use SmartPLS for this model.

Thanks a lot.

[soumyaray]

Dear @crmelchior I'm afraid I cannot commit to a timeline at the moment -- the implementation would not take long, but thoroughly testing it and then undergoing CRAN's rigorous submission process are all time consuming. We might also want to bundle other fixes at the same time. Regrettably, I would suggest going with SmartPLS for the interim. I will keep this issue open and maintain updates here in this regard.