To answer this question, we need to find the variable that measures whether a household has to fetch their water and see how that variable is coded. To do this, we first type
lookfor fetch
or
lookfor water
to see that the variable we want is wfetch
To investigate how this variable is coded, we typed:
tab wfetch
and we see that there are four reported values, but no labels. We then turned to the survey where there were only two values reported -1 (have to fetch water) and -2 (don't have to fetch water). This is sort of confusing. To straighten out matters, we looked at the variable wdist_c which measures how far one has to go to fetch water. By typing:
list wfetch wdist_c
we verified that when wfetch was -2, no distance was reported and when wfetch was 1, wdist_c recorded the distance needed to walk to fetch the water. We concluded that a value for wfetch of 1 meant the household had to fetch their water and a value of -2 meant they did not.
For the measurement of number of children we use variable ceb (children ever born) which can be found by typing
lookfor children
We can answer the question by typing:
means ceb if wfetch == 1
means ceb if wfetch == -2
We find that households that fetch their water report having an average of 2.38 children ever born versus 1.95 for households that do not have to fetch their own water.
Does this mean that fetching water causes households to have more children? Of course not. We need to be careful to differentiate between correlation and causation.