Scenario 7: Your task is to ascertain level of knowledge of a certain topic (say, safe sex) among the 13-17 year-olds. This could be measured before and after some educational intervention. There are no unique identifiers for these people and no register is comprehensive enough to give you a representative sample. What do you settle for?

1) School enrolment lists will clearly identify gender, age, socio-economic group (parental income, area of residence, race, etc) of individuals but will obviously exclude 13-17 year-olds who have left school or never enrolled.

2) Unemployment registers will provide similar profile for the school leavers but will also miss out on those teenagers who choose not to register when unemployed.

3) Hospital and other patient registers may identify that other proportion of teenagers who don’t study or work or may not live at home.

4) As all of the above only represent a sector of your community of interest, a population based survey would be best. However, trying to identify households with teenagers may consume time and your budget quickly.

WHATEVER YOUR CHOICE (DEPENDING ON THE TOPIC AND TARGET SUB-GROUP) ENSURE YOU INCLUDE A CONTROL GROUP IF YOU ARE EVALUATING THE IMPACT OF AN EDUCATIONAL INTERVENTION.