Have You Ever?
Have you ever wondered why some studies show results that just aren’t consistent with reality? This can happen for many reasons, and one of them is when the subjects in the study don’t actually represent the target population. Take a study about the trials of a lung cancer vaccine. The vaccine is intended for lung cancer patients but in this study, the participants consist of healthy individuals. The researcher continues to test the healthy participants and finds that none of them have lung cancer after taking the vaccine. Therefore, the researcher labels the vaccine appropriate for lung cancer patients.
Since the vaccine was only tested on healthy, non-cancer patients, there is no evidence that suggests that the vaccine can be beneficial to lung cancer patients. This issue with the researcher’s approach was that lung cancer patients were not recruited for the study and instead, the trials were performed on healthy individuals. The healthy individuals don’t represent the experiences of lung cancer patients, therefore the study isn’t representative of them. This idea is known as selection bias.
Selection bias occurs before a study is even conducted. It is a bias that arises in the process of selecting people, or subjects, for the study itself. It consists of different subtypes of biases, such as sampling bias and attrition, to name a few. Most of these biases can be avoided if there are preventative measures taken before the results of a study. If selection bias is avoided, it can result in studies having more accurate findings that are consistent and representative of the target population of a given study.
Definition of Selection Bias
Selection bias is a phenomenon that occurs when research subjects in studies are chosen improperly or nonrandomly, which creates a sample that isn’t representative of the actual target population of the study.
How It Works
Selection biases were identified after reviewing the findings of studies that showed inaccuracy and non-representativeness of the population that the study was based on. They’re usually labeled as limitations of a study. These biases don’t only occur in psychological studies because studies and experiments in any field are subject to selection biases and inaccuracies.
One type of selection bias is known as sampling bias. This is when the sample of research participants is collected nonrandomly, giving some members of a population a higher or lower probability of being chosen than others. For instance, take a study about high school students’ opinions on the books they finished reading for their English class at the end of the academic school year. The researchers choose a local high school and go to the different sections of AP Literature and AP Language & Composition classes. In these classes, they survey the students and ask them to rate the books they completed reading during the school year. However, the researchers did not survey the students in the honors or non-AP English classes. This limits the response by more than half of the English classes at the high school. They chose their participants non randomly, and this gives the students in non-AP classes a lower probability or no chance of being represented in the study.
Another type of selection bias is known as self-selection bias. This mostly occurs when people volunteer themselves as subjects of a study, leading them to create a nonrandom, biased sample. Take the same study about high school English students rating English books from above. However, this time around, the survey isn’t mandatory across the English classes, it’s optional. This gives students who enjoyed reading the books a higher chance of partaking in the survey. Students who didn’t feel as strongly about the books probably didn’t care and neglected to complete the survey. The survey will show a biased sample of book lovers due to the researchers allowing self-selection bias to occur.
Many people think selection biases are only an issue when it comes to research experiments and nonexperimental studies. However, they have a direct impact on the very groups of people that those experiments and studies are researching. If the study itself is not accurately selecting participants that reflect the topic and target population, the study will produce results that can potentially skew data and undermine or exaggerate the experiences of many groups of people.
For instance, if a study was conducted with the research question, “Do African-Americans live longer than white Americans?”, our target populations to compare are African-Americans and white Americans. The recruiters, intentionally or unintentionally, used methods that created barriers for more African-American participants to join the study. Because of selection bias, we now don’t have a representative study. This has major repercussions for the results as it will produce invalid findings for the study and potentially lead to conclusions about the two groups that are inconsistent with reality, just because of access issues creating inequality in the study. This goes to show the impact that selection bias in studies can have in the real world. It is important to combat such biases from occurring in the first place.
In order to minimize the threat of selection bias, it is necessary to create a random means of recruiting participants so that one method over another means of recruitment doesn’t give certain people a higher or lower chance of being recruited in the study. The random means should give everyone an equal chance of being recruited. For instance, for the study above, say the researchers only visited senior home centers to interview and collect participants. The senior homes consisted of a majority of white Americans, which is why the sample ended up having more white American participants than African-American ones. In order to prevent this bias, the researchers could generate a larger list of random places from which to recruit participants. This list could consist of religious and worship centers, community centers, public parks, and more, giving more people than just those in senior homes a chance to participate in the study and to equally represent the intended target population. Combatting biases is the first step to finding the truth.