A biased sample is one that does not represent the characteristics of the larger population from which it was collected. This phenomenon is also known as sampling error.
In qualitative research, there is risk of “selection bias,” meaning that the individuals in the sample differ in some way from those in the population of interest (Catalogue of Bias, 2017). For example, let’s say you wanted to study the eating habits of nurses in the COVID-19 pandemic. You might put up flyers advertising your study and wait for people to volunteer. However, those who see the flyers and choose to participate may be more health-conscious than those who opt out. Therefore, the information that you collect from this sample would be skewed, as you would only be interviewing people who are already concerned with healthy eating.
There is also the possibility for response bias in survey and interview research. Response bias simply means that participants are answering questions inaccurately, whether consciously or unconsciously. Reasons for this include desire to conform to social norms, attempting to predict the “right” answer, or simply misunderstanding the phrasing of the question (The Decision Lab, n.d.).
For example, if your survey asks college students whether or not they have ever cheated on an exam, it is possible that participants will answer “no” even if they have cheated. In order to counter this, you could be explicit about the anonymity of the survey responses.
The clip below elaborates on some other important considerations when it comes to bias in sampling such as non-response bias and underrepresentation.
Ethics Check!
Researchers tend to make generalized claims about human nature based only on samples from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies. In their article “The weirdest people in the world?” Joseph Henrich and his colleagues note that “WEIRD” societies are actually among one of the least representative populations (Henrich et al., 2010). If you are looking to make a claim about college students, then a list of students on a college campus would be the perfect place to start. However, if you are looking to generalize your results across a different population, looking beyond the college campus will create far more validity in your results.
