Weighted smoothing of estimated undercount
This paper is about statistical regression models for estimating undercount. Surrounding it is the highly contentious issue of whether the fitted models should then be used to adjust the census counts. I shall say very little about that here; my goal is to embed previous undercount regression models in a general setting. Estimation of model parameters and prediction of undercount will be implemented for the general model, although two special cases (heteroskedastic and homoskedastic) will be featured.