Kernel smoother and weights

[Maria Ventura]

Hi,

I am producing a plot using the Stata command lpolyci. I am using survey data, so I usually utilize probability weights for all the analysis and graphs. I am a bit confused about whether I should use weights (frequency weights in this case, rounding up the pweights) with lpolyci. They would be used as Kernel weights but would it correspond to the usual way we include survey weights?

Thanks,

Maria

[FernandoRios]

Hi Maria
You can still use weights (i would say aweights rather than fweights) with lpoly. Based on the manual, the way lpoly constructs the confidence intervals is by using the robust "sandwich" formula, so it would be equivalent to using "pweights" on the rest of your analysis/
HTH