In depth Making inferences Population Convenience samples are not representative In statistics, population refers to all the units of analysis you want to study. Example 1 You want to learn how many refugees entering country X by boat are being denied access to from convenience samples of the population. They are biased in unknown ways, as we do not know which population units were left out, or which units were less or more likely to be selected. We can not learn about our population, and we must not overstate the conclusions that we can draw from the sample. health care. The population is all refugees who arrived in country X by boat and were denied access to healthcare. Example 2 You are examining whether commercial flower plants have caused water pollution in a designated region. Here, X is all flower plants in that region. Convenience samples Convenience samples are not representative of the population: they are biased in unknown ways. All convenience samples contain selection bias. In statistical terms, this means we do not fully know how well our sample represents the population. We do not know which parts of the population we are not seeing, how big of a proportion they constitute of the population, or how our sample is different. Conclusions you can draw from convenience samples ›› The number of people you were able to document whose right to health care was violated. ›› Summarise and describe the individuals concerned and medical issues they suffered. ›› Create graphs to visualise the violations you documented. Be sure to clearly identify these graphs as summaries of “reported violations” or “documented violations.” (See this HRDAG report for an example of how to describe convenience sample data and how to annotate graphs of such data.) ›› Describe the state practices you’ve identified that lead to a denial of healthcare access. ›› State where and at what times you have identified these practices. If you have documented these practices across several healthcare facilities throughout country X and/ or repeatedly over time, you can point out this spatial and/or temporal spread. 75