design effect (deff)

gianni — Mon, 09 Jan 2012 22:07:15 GMT

Since I got a sample and then calculate the respective desired estimates (total and proporconal)
How to calculate the corresponding design effect (deff)?

design effect (deff)

mfisher — Tue, 10 Jan 2012 20:31:03 GMT

Proc surveyfreq will calculate the design effect for proportions by using the DEFF option in the TABLES statement.

Mark

design effect (deff)

gianni — Tue, 10 Jan 2012 21:43:03 GMT

Hello, thanks for your answer, but

this is my problem, I have a database of lessons presidential, with variables, candidate 1, candidate 2, Total_ votes, Strata.
And what I want is to calculate the estimators of reason, or proportion between a candidate and the total votes, and estimate of the total votes cast, their corresponding variances and their corresponding deff.
My program is as follows:

DATA PREP2006;

RUN;

*STRATIFIED SAMPLING;

PROC SORT DATA=PREP2006 OUT=ORDER_PREP2006; BY STRATA;

RUN;

PROC SURVEYSELECT DATA=ORDER_PREP2006

OUT=WORK.SAMPLING METHOD=SRS N=3;

STRATA STRATA;

RUN;

PROC PRINT DATA=SAMPLING (OBS=10);

RUN;

*proportion between a candidate and the total votes;

PROC SURVEYMEANS DATA=SAMPLING;

RATIO CAN21 CAND2 / TOTAL_VOTES ;

RUN;

*ESTIMATED TOTAL VOTES ^T;

PROC SURVEYMEANS DATA=SAMPLING SUM VARSUM STD RANGE NMISS CVSUM CLSUM;

VAR TOTAL_VOTOS;

WEIGHT SAMPLINGWEIGHT;

RUN;

*THE FOLLOWING PROC ME BACK THE DEFF, BUT NOT ON THE ESTIMATOR TOTAL OF TOTAL VOTES;

PROC SURVEYFREQ data=SAMPLING;

TABLE TOTAL_VOTES / DEFF;

strata ESTRATOS;

RUN;

HOW TO GET the following estimates?

* estimates of population means and totals

* estimates of population proportions

* standard errors

* confidence limits

* deff For each ESTIMATOR

* ratio estimates

topic design effect (deff) in SAS Procedures

design effect (deff)

design effect (deff)

design effect (deff)