Sampling and research design

In each of the country projects, the research team will include partners from local academic institutions and other experts who can advise on the research design and sampling strategy that is appropriate to the specific context.

HOW DO YOU CHOOSE WHO SHOULD BE INTERVIEWED?

Developing a sampling strategy for a research study can be a complex task. When the team is ready to consider this, it is recommended that they contact the index coordinator at IPPF and seek their advice. In addition, advice could also be sought from a statistician within a local university or research institute.

SAMPLING

There has been much debate about sampling for this index. Research studies such as these take a representative sample of the population of interest, with the aim of producing results that can be generalized for that population. This means that information collected from a sub-section of a population (the representative sample) can be used to reach conclusions that apply to the population as a whole. This process is called inference.

The problem in our case is that it is not possible to take a representative sample of the whole population of people living with HIV. That would require us to use an objective method of selection where all members of the study population have a definable chance of being part of the sample (what is termed a probability-based method of selection). However, it is very difficult to take a representative sample of people living with HIV given that we don't have a complete list of everyone living with the virus (i.e. the whole population of interest). We are therefore left with two options...

THE SAMPLE SIZE

Finally, it is important to bear in mind the question of sample size. You need to take a sample that is large enough to capture the main features of the population as well as the divergence from the main features - or what is termed variability. But what is large enough? Or to put this more formally, What sample size is required to ensure that the study results can be relied upon to make generalizations?

This is a difficult question to answer because it depends on a number of factors, including how precise you want your findings to be and the variability of the study population with respect to the characteristics of interest. The sample size has nothing to do with the underlying size of the population of interest (a common myth). However, it does depend on whether you want to present results at the international (global) level only or use it to compare different countries, or areas within a particular country. It also depends on whether you want to show national-level results or breakdowns for specific areas or groups of people.

For example, one study in Tanzania sampled three key groups of interest to measure stigma related to HIV: general community members, health care providers, and people living with HIV. The study intended to gather 100 responses from people living with HIV "to fall within the minimum required size on which statistical tests could be meaningfully conducted and within the maximum that the study resources could support". In the end, to ensure gender balance in the sample size, they extended the survey to include 218 respondents since the majority of the first 100 respondents were women. In the Asia Pacific Network of People Living with HIV/AIDS study, there was a sample size of 764 (302 from India, 338 from Thailand, 82 from Philippines and 42 from Indonesia).  These examples highlight that sample size will vary depending on the context and focus of each country study.

Some important things to remember for each individual study using the People Living with HIV Stigma Index include:

  • What is the minimum size required to meaningfully run statistical tests?
  • What is the maximum size that the study resources can support?
  • If balance is important for the study (e.g. gender), do you have an equal number of responses (e.g. from men and women)?
  • Even if the study cannot be representative, do you have a sample that contains a mix of positive people of different ages, sexuality and economic, social and educational backgrounds?

Ultimately, you need to settle on the best and most feasible sampling strategy for your research. The People Living with HIV Stigma Index coordinator at IPPF will be able to inform you about what other countries' teams have done and provide you with the necessary advice for your country.