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

Re: st: creating sample weights

From   "Stas Kolenikov" <>
Subject   Re: st: creating sample weights
Date   Sat, 11 Aug 2007 11:55:53 -0500

On 8/11/07, Janelle Knox <> wrote:
> I will give post-stratification some consideration.  I think you are
> is just that I have an over-representation of one gender,
> and certain age groups in the data, and would like to correct for
> this.  The data does not have any weighting built in.

Ask again your data provider how the data were collected, and how they
would suggest you correct for over-representation given their sampling
procedure. These days, there are no straight simple random samples
anywhere. If you had a RDD phone survey of US population, then you
have 60-40 female-male, and ages 50+ overrepresented. If you are
willing to interpret different fractions of age by gender (and
probably by race) groups as due to different response mechanisms (it
is easier to find somebody at home if they are retired, and women are
probably more talkative than men overall), then you can interpret
those differences as arising from different, although implicit,
probabilities of selection, and you'll be fine with weights, too. RDDs
are in fact clustered surveys, although arguably larger effects are
due to interviewers, but they are crossed with geography due to the
logistics of the field work.

Stas Kolenikov, also found at
Small print: Please do not reply to my Gmail address as I don't check
it regularly.
*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index