sampling weights


I am using a national dataset for my analysis. I am confused with the sampling weight because my understanding of sampling weights is to compensate disproportionate stratification, such as gender, SES, educational level, area, etc. However, if I use a variable, such as frequent vs infrequent exercise, without incorporating gender, SES, area variables, do I still need to take sampling weights into account?

Any comments are highly valued.

Thank you.


Phineas Packard
Yes. You should use the sampling weights for all analyses as the weights aim to give estimates which are more consistent with a truely representative sample.
Hi Lazar,
Thanks for the prompt reply. Please forgive me for not having sufficient stat knowledge. What should I do to take care of sampling weights? Should I recalculate another set of value or does estimation method (Mplus) take care of that problem?
If I divided number of times exercise per week into two groups to analyze a set of attitudinal variables rated on 5-point scale (treated as continuous), should I use the sampling weights for the two groups or the attitudinal variables?


Any comments are appreciated.