04.11.2016 Views

Bangladesh 2014

8ln3LMTjp

8ln3LMTjp

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

ah<br />

M<br />

M<br />

hi<br />

hi<br />

Let b<br />

hi<br />

be the proportion of households in the selected cluster compared to the total number of households<br />

in EA i in stratum h if the EA is segmented, otherwiseb<br />

= 1. Then the probability of selecting cluster i in<br />

the sample is:<br />

a<br />

M<br />

M<br />

h hi<br />

P 1hi = ×<br />

hi<br />

b<br />

hi<br />

Let L<br />

hi<br />

be the number of households listed in the household listing operation in cluster i in stratum h, let<br />

g<br />

hi<br />

be the number of households selected in the cluster. The second stage’s selection probability for each<br />

household in the cluster is calculated as follows:<br />

hi<br />

P =<br />

2 hi<br />

g<br />

L<br />

hi<br />

hi<br />

The overall selection probability of each household in cluster i of stratum h is therefore the production of the<br />

two stages selection probabilities:<br />

Phi<br />

= P1 hi<br />

× P2<br />

hi<br />

The sampling weight for each household in cluster i of stratum h is the inverse of its overall selection<br />

probability:<br />

W = 1/<br />

hi<br />

P hi<br />

A spreadsheet containing all sampling parameters and selection probabilities will be prepared to<br />

facilitate the calculation of the design weight. Design weight will be adjusted for household non-response<br />

and as well as for individual non-response to get the sampling weights for households, for women and men<br />

surveys respectively. The differences of the household sampling weight and the individual sampling weights<br />

are introduced by individual non-response. The final sampling weights will be normalized in order to give<br />

the total number of un-weighted cases equal to the total number of weighted cases at national level, for both<br />

household weight and individual weight, respectively. The normalized weights are relative weights which<br />

are valid for estimating means, proportions and ratios, but not valid for estimating population totals and for<br />

pooled data.<br />

Appendix A • 217

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!