[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: ROC with complex data sets

From   Steven Joel Hirsch Samuels <>
Subject   Re: st: ROC with complex data sets
Date   Mon, 20 Aug 2007 12:02:27 -0400


Short answer to your question:
Round the pweight variable to the nearest integer and use the resulting variable as an "fweight".

Long answer: The ROC programs will not take into account other aspects of the survey (stratification, clustering, multiple stages). Therefore confidence intervals and p-values quoted in the output will be wrong. Also, the ROC analysis will not be much good if your model is not good: so you are still obligated to build a good model and check its fit. For example, if there are strong interactions, you should include them in your model. Use Stata's survey features: - svyset- your data and build your model with Stata's -svy- logistic command. You may also want to learn about Stata's -zinb- command, which can demonstrate failure of the ordinary logistic model.

Read the survey documentation to understand how the weighting was done. If any part of the weighting or stratification involved your outcome variable, you should probably run an unweighted analysis.


On Aug 20, 2007, at 8:34 AM, Taylor, Sharonda A. wrote:

I am very new to research. I am attempting to do ROC analysis using a complex survey that uses pweights. Stata 9.2 uses frequency weights. I understand that pweights and fweights are not interchangeable. Are there any downloads that will allow me to run the analysis with the pweights? Please advise.
18 Cantine's Island
Saugerties, NY 12477
Phone: 845-246-0774
EFax: 208-498-7441

*   For searches and help try:

© Copyright 1996–2023 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index