Demographic and Health Surveys Methodology - Measure DHS
Demographic and Health Surveys Methodology - Measure DHS
Demographic and Health Surveys Methodology - Measure DHS
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
CalculationInitial sample weights are produced by the <strong>DHS</strong> sampler using the sample selection probabilities of eachhousehold <strong>and</strong> the response rates for households <strong>and</strong> for individuals. The initial weights are thenst<strong>and</strong>ardized by dividing each weight by the average of the initial weights (equal to the sum of the initialweight divided by the sum of the number of cases) so that the sum of the st<strong>and</strong>ardized weights equals thesum of the cases over the entire sample. The st<strong>and</strong>ardization is done separately for each weight.H<strong>and</strong>ling of Missing Values—Not applicableApplicationSample weights are calculated to six decimals but are presented in the st<strong>and</strong>ard recode files without thedecimal point. They need to be divided by 1,000,000 before use to approximate the number of cases.In tabulation programs, sampling weights need to be applied through the use of special comm<strong>and</strong>s.Examples:Notes <strong>and</strong> Considerationsa) In SPSS using the WEIGHT comm<strong>and</strong> with the weight variable:COMPUTE rweight = V005/1000000WEIGHT by rweight.b) In ISSA using the weight parameterrweight = V005/1000000x = xtab(table1, rweight).1. The sum of the sample weights only equals the number of cases for the entire sample <strong>and</strong> not forsubgroups such as urban <strong>and</strong> rural areas.2. Where there are no differential probabilities, weights may not be calculated since weights basedjust on response rates usually make little difference in results.3. Use of sample weights is appropriate when representative levels of statistics are desired, such aspercentages, means, <strong>and</strong> medians.4. Use of sample weights is inappropriate for estimating relationships, such as regression <strong>and</strong>correlation coefficients.5. Use of sample weights biases estimates of confidence intervals in most statistical packages sincethe number of weighted cases is taken to produce the confidence interval instead of the truenumber of observations. For oversampled areas or groups, use of the sample weights willdrastically overestimate sampling variances <strong>and</strong> confidence intervals for those groups.Guide to <strong>DHS</strong> Statistics 14 September 2003