Behavioural Surveillance Surveys - The Wisdom of Whores
Behavioural Surveillance Surveys - The Wisdom of Whores
Behavioural Surveillance Surveys - The Wisdom of Whores
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Table 6 provides a “lookup” table based<br />
upon the above formula to permit final sample<br />
sizes to be determined without having to<br />
perform calculations. <strong>The</strong> table provides<br />
sample sizes needed to measure changes in<br />
behavioral indicators <strong>of</strong> magnitude 10 and 15<br />
percentage points for different initial values<br />
<strong>of</strong> a given indicator, as well as for different<br />
combinations <strong>of</strong> significance and power.<br />
Note that some indicators include two<br />
behavioral dimensions. For example the<br />
proportion <strong>of</strong> all students who had sex with<br />
a sex worker in the last year (dimension one)<br />
who used a condom the last time they had sex<br />
with a sex worker (dimension two). In these<br />
cases a further step is needed in calculating<br />
the sample size. In this case, the first step in<br />
calculating the sample size required would be<br />
to determine how many students would be<br />
needed to measure a change in the proportion<br />
who used a condom during an encounter with<br />
a sex worker during the previous year. For<br />
this step, proceed exactly as in the examples<br />
given in Figure 4. Say the number comes to<br />
200. Does that mean the final sample size will<br />
be 200? No, because not all <strong>of</strong> the students<br />
sampled will have had sex with a sex worker<br />
in the last year, and this indicator takes only<br />
the ones that have had sex with a sex worker<br />
into account in the denominator. So it is<br />
necessary to estimate how many students<br />
would be needed in the overall sample, in<br />
order to capture 200 who have had sex with<br />
a sex worker in the last year.<br />
Computationally, the procedure is simple -<br />
one merely divides the required sample size<br />
calculated as described above by the estimated<br />
proportion <strong>of</strong> the sub-population with the<br />
required “qualifying” behavior. For example,<br />
if 40 percent <strong>of</strong> male vocational students in a<br />
given setting are thought to have had sex with<br />
a sex worker in the last year, it would be<br />
necessary to interview n=500 (=200/.4) students<br />
in order to find n=200 subjects who had had<br />
sex with a sex worker in the last year, to allow<br />
for the indicator <strong>of</strong> condom use to be calculated.<br />
Examples <strong>of</strong> this type <strong>of</strong> calculation are given<br />
in Figure 5.<br />
<strong>The</strong> difficult part <strong>of</strong> this procedure is,<br />
<strong>of</strong> course, knowing what proportion <strong>of</strong> the<br />
total population engages in the behavior <strong>of</strong><br />
interest. Here, other surveys or anecdotal<br />
information might be consulted for guidance.<br />
If resources permit, a small pilot survey might<br />
be conducted to better inform sample size<br />
calculations for the main survey effort.<br />
As there may be considerable uncertainty<br />
concerning these parameters, the general<br />
guidance is to err toward under-estimating<br />
the proportion engaging in a given behavior,<br />
as this will ensure a sufficient sample size for<br />
the main survey effort. For example, if it were<br />
thought that between 20% and 30% <strong>of</strong> students<br />
typically engage in sex with sex workers on<br />
an annual basis in a given setting, the 20%<br />
figure should be used in determining sample<br />
size requirements for BSS.<br />
B EHAV I OR A L S U R V EI L L A NC E SURV EY S CHAPTER 4<br />
49