A Few More Words About Bias In Research
It is important to understand that it is relatively easy to find some form of bias in research; therefore, it is technically correct to refer to reducing bias in research rather than eliminating bias altogether. The issue is not that bias in research is present rather, the issue with bias is when it is unidentified, undeclared or otherwise not considered or taken into account in the overall conclusions a research study.
In some cases, bias in research is referred to as random, and it is possible to demonstrate the bias does not impact the overall results. Conversely, bias may also be systemic and can largely impact study results and conclusions. Importantly, bias may occur in research in many different forms.
Recall from our previous post: Influence and Bias in Research, we briefly introduced the concept of bias and defined bias as “inclination or prejudice for or against one person or group” (1) and important to research, specifically in statistical analysis bias is referred to as “a systematic distortion of a statistical result due to a factor not allowed for in its deviation”(2).
What Can Bias Look Like?
One such example of bias in research may result from the study design itself. Consider data from a survey that collected the data in only one language among a population that has a significant number of other language speaking individuals. The research in this example may contain bias as it did not allow results from a majority of the population.
Another study design issue relates to our previously discussed ABC drug experiment [See ABC drug study example], where we tested a new pain reliever drug among patients recovering from knee replacement surgery. If our study process had allowed patients in the experimental group taking our new drug to also use an additional type of pain reliever at the same time, we would have introduced bias as the additional or non-ABC drug could be responsible for reducing pain versus the ABC drug we were investigating. A second, unaccounted for pain reliever drug during the duration of our study is referred to as a confounder.
Beyond study design issues, bias in research may be introduced in other ways such as through a conflict of interest. If a researcher had a financial interest in a company investigating a new product, there may be a conflict of interest and particular safeguards may be required to ensure there is no undue influence from the researcher’s circumstances in the research study. A third type of bias can occur surrounding the sharing and dissemination of the study results. For example, if a research study’s funding source (an organization, corporation or individual) had influence over how the study’s results were shared such a circumstance could influence where and what research information regarding the project was shared, which may introduce a possible source of bias.
Identify Bias in Research
A well respected international professor of sociology, at Western University in London, Ontario Canada, Dr. Anton Allahar was once asked by a student what to do about bias in research and he responded: “always declare it”. Since bias exists in research in one form or another, the goal cannot simply be to avoid all bias. Rather, once bias is declared and factored into the overall conclusions of a research study, bias fails to have an impact that is not considered in the overall conclusions. Dr. Allahar’s advice is excellent and a guiding principle. As researchers, we must first declare the bias in our own research and call-out the impact of bias across all research information. Importantly, we must always demonstrate whether or not any identified bias is likely to have an impact on the overall study results.
There are a number of different forms of bias and we’ve simply discussed three different ways bias may be introduced into a research study. One the actual study design itself, two a conflict of interest and finally how or where research study results are published and shared. In future posts, we will delve more specifically into the issue of bias in research and how it can impact research results but for now, the important piece to remember is that all research, regardless of what type of research or approach is used can contain some degree of bias. The goal is to minimize the amount of bias as much as possible in the processes of our research before we begin our study and to examine any possible bias when stating our overall conclusions.
Thankfully, the majority of researchers and research activities across the globe are deeply committed to ethical and unbiased research. However, our job as colleagues, peers and critical consumers of research information is to ensure research maintains ethical and principled activities and processes. Such standards are important for both the well-being of individuals involved in research projects and to ensure research is as reliable and valid as possible across all of our shared research knowledge. Remember, bias may be systemic and seriously undermine the quality of a research project such as only publishing selected results. Conversely, bias may be the result of random error where it is possible to demonstrate that the bias is minimal and not particularly influential.
Continuing The Development of Our Critical Consumer of Research Information Skills
If you have followed the discussion during our introductory posts here at SNJ Associates, we hope you have sharpened your critical consumer skills and gained more insight into answering the question What is Research? and the importance of the systematic process including the scientific method [See Characteristics of Research and Parts of the Scientific Method]. We are always bringing the focus back to thinking about how researchers designed and what processes were used to create the research information we are considering. Now we’ve added the step of specifically considering possible bias in the research information that you are using and investigate for yourself if such bias could possible impact the overall study results.
References
1, 2. “Bias.” Oxford Dictionaries Language Matters. Oxford University Press. Web. 5 Jan. 2016.