In article <[EMAIL PROTECTED]>, [EMAIL PROTECTED] says...
> programs such as WESVAR and SUDAAN assume that cases are weighted to represent
> the survey design.
>
> several colleagues have made contradictory suggestions on case weighting.
> I am looking for url's, cites, instructional notes, etc that address case
> weighting. Particularly for two stage cluster sampling with the first stage
> drawn with probability proportional to size.
>
Here's an url:
http://www.fas.harvard.edu/~stats/survey-soft/survey-soft.html
If cases have an unequal probability of being included in the sample,
then you should weight. Inclusion probabilities can be affected by the
sampling design (stratified sampling, cluster sampling) or by
poststratification to variables with no distribution in order to take
sampling bias or non-response into account. In cluster sampling, there
will be an intra-cluster correlation that must be taken into account as
well.
Failure to weight leads to biased estimates, weighting in the usual
manner (as in SPSS and non-survey programs in SAS) leads to standard
errors that are too small. Wesvar, SUDAAN and similar programs can
provide standard errors that take the sampling design into account.
Some references:
Lee, E.S., Forthofer, R.N., Lorimor, R.J. (1989). Analyzing complex
survey data. Beverly Hills: Sage Publications.
- a Sage book, good place to start
Winship, C., Radbill, L. (1994). Sampling weights and regression
analysis. Sociological Methods & Research 23: 230-257.
- a very clear discussion of the issues involved
Johnson, D.R., Elliot, L.A. (1998). Sampling design effects: do they
affect the analyses of data from the National Survey of Families and
Households? Journal of Marriage and the Family 60: 993-1001.
- an example of the impact of taking sampling design into account
Levy, P.S., Lemeshow, S. (1999). Sampling of populations. Methods and
applications. New York: John Wiley & Sons, Inc.
- for general information of sampling designs, there's also a chapter on
variance estimation in complex surveys.
Good luck,
John Hendrickx
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
