Qlarity Access Blog

How to Recognize and Prevent Bias in Research

Written by Communications for Research | Jul 12, 2019 7:49:00 PM

Businesses conduct market research in order to learn more about customer preferences and market behavior. The validity of any conclusions drawn necessarily depends on how well these businesses (and/or the firms they hire) recognize their own assumptions, preferences and prejudices and how hard they work to mitigate them as they select samples, develop questions and interpret results. Here are a few tips for preventing bias in research:

 

Acknowledge It Exists

The varied experiences each of us have as we live our lives undoubtedly colors the way we think and act. It’s a widely researched notion, with cultural bias and other implicit biases contributing to individual worldviews that affect everything from the food we prefer to the subjects we choose study. As humans, our pasts help shape our futures. In this way, market researchers, no matter how well-trained, are still vulnerable to unintentional inclinations that can (and often do) impact research design and research findings. The first step, therefore, is simply acknowledging that bias in research exists. Only then can you begin to recognize and prevent it.

 

Consider Your Source

In college, you may have been admonished to “always consider the source.” Credible sources yield credible facts. Similarly, market research results are only as good as the people to whom you choose to talk. Pick the wrong respondents, and you can’t reliably apply their feedback to the larger market on hand. You must carefully consider and then define your target sample so that the answers you get reflect those of the right population.

 

Watch Your Words

Another type of bias in research occurs when researchers ask the wrong type of questions, such as:

  1. Leading - “How hard is it to open our packaging?”

This type of question “leads” your respondent to answer in a way that confirms your own point of view. In this case, it’s believing that your packaging IS hard to open.

  1. Loaded - “What website do you use to purchase our products?”

With loaded questions, you essentially force your respondent to answer in a certain way. What if your respondent buys your products at a local retail shop?           

  1. Absolute - “Do you always work from home?”

Absolute questions demand respondents choose only one answer, “yes” or “no.” This limits respondent honesty as rarely do people “always” or “never” do “all” or “none” of anything!

  1. Double-Barreled - “Is our product X easy to assemble and use?”

Never ask respondents to answer two questions at once. Instead, structure questions so that only one subject is addressed at a time.

 

With each of the above examples, researchers have made assumptions about their respondents and/or limited the type of responses that their respondents can make. Either scenario diminishes the capacity for respondents to give forthright and comprehensive answers, thus negating the efficacy of the research results as a whole. Better questions would include:

  1. “How would you describe the efficiency of our packaging?”
  2. “How do you purchase our products?”
  3. “On average, how many hours a week do you work at a location other than your home?”
  4. “Please rate the ease or difficulty of assembling product X,” and then “Please rate the ease or difficulty of using product X.”

 

Of course, to further safeguard against research bias, you also want to use language that resonates with your audience, offer exhaustive and accurate answers when dealing with multiple choice questions and allow for conditional branching and/or opting out when applicable. All of these things help to minimize respondent confusion and encourage participation in the research project, ensuring that you get the best answers possible.

 

Master Your Technique

Too frequently, companies rely only a research methodology in which they are skilled or with which they have prior experience. This poses a problem since certain methods of data collection are better suited for certain types of populations. Similarly, many choose to rely on statistical measures that they may not fully comprehend or that can not be properly applied to the data collected. To limit bias, consider the best way to reach the most people and verify that any statistical analysis used is appropriate for the type and quantity of data available.

 

Want to Learn More?

Even though bias exists, skilled researchers know how and where to look for it. If you need assistance identifying and preventing bias in your own research project, contact our team at Communications for Research (CFR). We have over 20 years experience crafting research that garners actionable results.