The thesis topic is on sexual behaviour among truck drivers where questions were asked to assess whether they were practicing safe or unsafe sex.
Further no time period was specified in the questionnaire for assessing sexual behaviour like "did you in the last one year visit prostitutes" instead question were asked did you ever visit CSW and ever use condoms.
So in this context my query is can i still compute prevalence of safe or unsafe sex?
Thankyou
Normally, it is impossible to calculate the prevalence of any health event without the defined population.But in such study, out of the total number of participants( target population) included in the study, the prevalence of those follows safe or unsafe sex can be calculated in a cross sectional study.
Going by the definition prevalence ie 'existing number of truck drivers involved in safe or unsafe sex' you can calculate prevalence by simply counting the number (in each category) that gave positive responses to your questions. However, the problem is that you cannot extrapolate (generalize) your findings based on a non probability sampling technique like convenience sampling which you proposed to use in recruiting the drivers. It is possible to calculate (estimate) the prevalence (even prevalence rate using total sample size as denominator) of health events from a sample without knowledge of total number of the subjects that constitute the target population, as long as the sampling method is such that gives every unit in the population an equal and independent chance of being selected (probability sampling technique). So I suggest you adopt the probability sampling method (preferably Systematic or Cluster) to recruit the subjects. In the alternative, interview all the truck drivers in the area if you have the resources and the go ahead to determine the prevalence. Hope this is is helpful.
Yes. You can assess the prevalence of HIV in truck drivers. The sexual behaviour surveys can determine the HIV and sexual infections prevalence in truck drivers recruted through convenience sampling.
You might want to take care of the validity of your study.
External validity: already covered by previous comments, which is that you might not be able to generalize your findings with acceptable levels of uncertainty. Actually, in a convenience cross-sectional survey, you may be able to describe with some limitations the frequency of sex with commercial sex workers and condom use.
Internal Validity: you may want to be as specific as possible in order to reach adequate conclussions on the topic you asked even if it might be referred to the sample you studied regardless of whether you want to extrapolate the results or not (as expressed under external validity).
One of the most common problems is that because of rapid assessments sometimes there are threats to validity and thus your results might be limited.
You may want to review the objectives of the thesis and critically assess what is what is needed as input according to the hypothesis posed in the beginning. If you want to make the central point the frequency (we might argue if this would be true prevalence or just frequency of report) this might not be the optimal design.
Finally, even if you decided to have a probability sampling method (as simple as random, clustered or any other type of sampling) or not, the time when behaviors occur and the intensity of those need to be very clear. For example, it might be difficult to reach a conclusion if some respondents say yes, sex with commercial sex workers and that might be referred to behaviours 30-50 years ago than someone who is reporting behavior from the last month or year, there are age-differences that you might want to take into account, besides temporality, pre-AIDS or current, last sexual encounter, etc.
Besides, you also might want to weight the responses by each time a sex act occurred. It might not be enough protection if a condom was used once over a number of sex encounters even with the same person/day.
Ashish: The answer to your question is yes, you could ‘compute’ prevalence but only for the population observed (see previous comments) and only with strong caveats about the sampling methods. This is very limited!
If you have any other information about those questioned (age, marital status, ethnic origin, religion, etc), and equivalent data are available elsewhere for other defined groups of truck drivers, then you could show how representative (or not) of truck drivers in general your sample was and comparisons could be made. Alternatively you might take a different group for comparison, say shop-owners or factory or farm workers, matched for age, etc.
I would suggest that you regard the data you have gathered as ‘hypothesis-forming’ and use them to design a prospective study.
There is a rich literature regarding truck drivers and the spread of HIV in Africa during the 1980s which might help you formulate your study and identify a helpful control group.
David B