# st: Re: How to weight two combined survey data sets that each have their own particular strata weights?

 From "Mark Veazie" <[email protected]> To <[email protected]> Subject st: Re: How to weight two combined survey data sets that each have their own particular strata weights? Date Mon, 20 Dec 2004 17:14:38 -0700

```Rick,

When combining repeated national health surveys, like the National Health
Interview Survey or the Behavioral Risk Factor Surveillance System, my
colleagues and I  typically "weight the weights."  Say you want to combine
survey A with survey B with sample sizes n1 and n2 respectively.  Simply
multiply the original weights in survey A by n1/(n1+n2) to obtain the
revised weights.  Similarly for survey B, multiple the original weights by
n2/(n1+n2).  If you sampled large clusters (PSU's like neighborhood or
postal region) that could have been the same between the two surveys, then
you also need to generate and use new Primary Sampling Units as each unique
combination of original psu and year. This is the simplest way I have found
to combine surveys.  This approach, its limitations, and many more
complicated approaches are found in Chapter 8 (Analysis Using Multiple
Surveys) in "Analysis of Health Surveys" written by two NCI Biostatiscians
Korn EL and Graubard BI  in 1999.  Publisher is John Wiley and Sons, Inc,
New York.

Hope this simple response helps.  mark

----- Original Message -----
From: "Richard Tardanico" <[email protected]>
To: <[email protected]>
Sent: Monday, December 20, 2004 12:28 PM
Subject: st: How to weight two combined survey data sets that each have
their own particular strata weights?

> Greetings,
>
> As part of a project on issues of social inequality and basic services
> in Latin American cities, a household survey was carried out in two
> cities (N1=1200 households, N2=1426 households).  Each sample was
> stratified according to the particular socioeconomic composition of its
> city (i.e. according to census-estimated proportions of "lower," "lower
> middle," "middle," and "upper middle" households in each city's
> population).  I want to analyze the combined data.  While of course I
> need to weight the combined data set to compensate for the unequal sizes
> of N1 and N2, how do I weight the combined data in view of the fact that
> each sample has its own particular stratification weights?
>
> Thanks,
> Rick
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```

• References: