There’s nothing more deterring in research than finding out the data you’ve gathered is inaccurate or not what you expected. We define quality data as not only accurate or trustworthy, but also relevant— data can be reliable but still not mean anything to a business. Improving data quality increases the reliability of insights, reduces the cost of re-fielding and saves on potential time lost. Lucky for us, when it comes to survey data there are many tactics we can incorporate into the process to avoid poor data quality.
1. Use Technology to Your Advantage
Whether it’s qualitative or quantitative in nature, technology does have its perks to ensuring data quality. More often quantitative survey solutions and panel providers already have some form of framework put in place in the back end to help avoid repeat or poor quality respondents, but they could be taken further. For example, in quant research, survey parameters can be put in place to avoid speeders and top or bottom box respondents. In online qualitative methods, researchers can ensure the respondent is who they say they are through Facebook profile integration and minimum response lengths to make sure they’re providing true qualitative responses.
2. Incorporate Consumer Language
While a company may think the explanation of their technology makes sense to them, consumers could explain it in a very different way. Incorporating consumer language from past research or conducting qualitative research to define how consumers speak about a particular topic can be very instrumental in communicating to respondents in a way that makes sense and allows for more relevant feedback.
3. Avoid Bias
Researchers most often account for question and selection bias. Question bias is when the design of a question or the way it’s asked leads respondents to answer in one way or another. Examples of this type of bias include leading or double-barreled questions. Response bias is a bit different in that it happens when respondents that are included in the target sample are actually inaccurate representations of the intended audience. But while avoiding bias in research is already a no-brainer, sometimes it comes up in more areas than we anticipate, like during the questionnaire design or during analysis. Again, technology can help to avoid bias but in order to avoid as much bias as possible, researchers should constantly remind themselves at each step of the research process, when bias can interfere.
4. Incorporate Human Monitoring
As technology integrates itself more and more into the day-to-day tasks of market research, it’s easy for researchers to feel they are no longer needed in certain areas— this can be fatal to data quality. While technology parameters can help filter through some of the muck, humans are still necessary in order to set the level of quality in data. For example, while it can be time-consuming, our researchers review each open-end in quantitative data and thoroughly review qual responses before removing particular respondents. Often this is more subjective than what technology can pick up and depends on the objectives of the research, which is why it’s important a human does it. Soft launches are also a great opportunity for researchers to be able to assess a smaller data set and ensure things are working smoothly before the data becomes too robust to fix errors.
5. Set Standards and Develop Processes
Each of the tips above can be incorporated into the research process in order to help improve data quality. But standards for question development, survey programming, quality assurance, and reporting are also important to make sure that each study is being held to the same level of quality. And more importantly, processes are guaranteed to help reduce the likelihood of an error. To learn more about data quality, and additional techniques that can help, check out the eGuide below on projective techniques.