14.03.2015 Views

Behavioural Surveillance Surveys - The Wisdom of Whores

Behavioural Surveillance Surveys - The Wisdom of Whores

Behavioural Surveillance Surveys - The Wisdom of Whores

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

Conventional clusters vs.<br />

time-location clusters<br />

Before determining how PSUs (clusters)<br />

will be selected, a key decision must be made<br />

about whether to use conventional clusters or<br />

time-location clusters. This decision is<br />

based on the nature <strong>of</strong> the population to be<br />

sampled. When members <strong>of</strong> the population<br />

are associated with sites in a “fixed” manner,<br />

then conventional clusters can be used.<br />

Examples <strong>of</strong> “fixed” populations would include<br />

brothel based sex workers who live at the<br />

brothel, police attached to a certain unit or<br />

division, or secondary school students at<br />

particular schools. However, frequently the<br />

populations included in surveillance are not<br />

associated with a site, but rather come and<br />

go freely from the site. Examples <strong>of</strong> these<br />

“floating” populations might include<br />

street-based sex workers, men who have sex<br />

with men in parks (or pick-up points), truck<br />

drivers at halt-points, or injecting drug users<br />

at shooting galleries. For these “floating”<br />

populations, the number you might find at a<br />

particular site will vary from day to day, and<br />

even by time <strong>of</strong> day or by the time <strong>of</strong> the<br />

month. This is not very important if the<br />

individuals at a PSU share the same patterns<br />

<strong>of</strong> behavior regardless <strong>of</strong> when they go to that<br />

site. But if people with different levels <strong>of</strong> risk<br />

visit a site at different times, then the time<br />

selected for sampling may affect the results <strong>of</strong><br />

survey and make it less representative <strong>of</strong> the<br />

population as a whole. An example might be<br />

men who have sex with men and gather at<br />

cruising sites. Men who are found at these<br />

sites mid-afternoon may be disproportionately<br />

more likely to be unemployed than men who<br />

come to the sites in the evenings, and may<br />

therefore be more likely to sell sex for income.<br />

So for this population, sampling mid-afternoon<br />

would probably give higher estimates <strong>of</strong> risk<br />

behavior than sampling in the evening.<br />

It may well give a doubly distorted picture,<br />

in fact, because more men may visit a cruising<br />

site in the evening (after work and a few<br />

drinks) than mid-afternoon. In this case, risk<br />

levels measured in the afternoon would be<br />

representative <strong>of</strong> a smaller proportion <strong>of</strong> the<br />

overall population than risk levels measured<br />

in the evening.<br />

Time location sampling provides a way<br />

around these difficulties. <strong>The</strong> same site may<br />

be included in the sampling frame more than<br />

once, at different times <strong>of</strong> the day or different<br />

days <strong>of</strong> the week, in order to give a more<br />

accurate picture <strong>of</strong> the different levels <strong>of</strong> risk<br />

associated with the site. In other words, the<br />

PSU is defined not as the site alone but as the<br />

site plus the time <strong>of</strong> the day/week/month at<br />

which sampling takes place. <strong>The</strong> same<br />

physical site can therefore become several<br />

PSUs. This system has the added advantage<br />

that it is not necessary to count the total<br />

number <strong>of</strong> individuals associated with a single<br />

location. Only the number <strong>of</strong> individuals in<br />

the sampling time interval need be discussed.<br />

B EHAV I OR A L S U R V EI L L A NC E SURV EY S CHAPTER 4<br />

37

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

Saved successfully!

Ooh no, something went wrong!