September 25, 2020

Are All Outliers Bad Research Data?

When we were children in our science classes, many of us conducted experiments where we had to record the results of multiple events or the same event repeated. After we finished recording all of the experiment results, we found that some of the data points didn’t line up with the others. These are outliers, which can have serious consequences on the final product of your study if not properly taken into account. These points can easily become bad research data that affect your project from top to bottom, but not always. 

Bad research data is any data that has been falsified, corrupted, dishonest, or misrepresentative of the sample group you studied. This makes only some outliers bad research data. For instance, if you poll people on their need for a specific product, and the majority of people agree except one, this is an outlier, but not bad research data. Having one person disagree is accurate to the sample size and the population. This gives the researchers important expectations on how successful the product they’re preparing can be.

Knowing when to ignore outliers and when to study them can help you save what data you’ve collected and avoid bad research data in the future. Sometimes, ignoring outliers turns what was originally good and accurate data into bad research data.

How Outliers Hurt Data

There’s one main reason to believe that outliers hurt data. It’s so important because it can have ramifications on your projects and your company’s revenue. They skew results. 

Depending on what kind of measurements and graphs you use for your data, an outlier will skew your trends. If you’re basing a product’s marketability or operational success on data with an outlier or two, you’re making inaccurate assumptions. This can lead to your team making decisions that don’t benefit your product’s success as much as they should. In some ways, it even damages a product’s ability to succeed at all.

But say you catch the outlier and remove it, so it doesn’t skew your data. You now have a smaller sample size that is not as accurate to your sample population as it should be. This can leave you without key information or believe something that’s not true about your results.

Whether you catch the outliers or don’t, they can severely affect the success of your research. This can snowball into problems with your distribution, manufacturing, pricing, and/or whether or not your project even works. So after recognizing this, why would an outlier not be considered bad research data more often than not?

How Outliers Help Data

There are several ways of recording and judging data that will make outliers appear dangerous or misleading. When you see outliers as representative of a possible endgame, you can see how they can be an accurate representation of your data.

An outlier can show you a part of your target audience or target results that won’t meet your expectations. When you need to test an element against multiple temperatures in an experiment, one temperature can cause the element to react differently than the rest. If this is a consistent outlier, it wasn’t a mistake but a discovery the rest of the experiment needs to account for. In this instance, the outlier isn’t bad research data, despite the fact that it doesn’t match the trend. With this information, you can pivot and take into account the potential risks as you continue your research.  

Even when the outlier isn’t consistent and may not be emblematic of the population, this tells you key information that makes the outlier worth studying. It shows you where you make measurement errors and possibly why you are getting outliers. Often, you may find that there are variables that you didn’t account for that are causing outliers. This allows you to reduce their effect or possibly eliminate them for the rest of the research and future research.

Avoid Bad Research Data With Focus Forward

We understand the difference between bad research data and outliers. Some outliers are bad research data, but not all, and we help researchers find out which their data is. We can help you succeed with your project and find the most success in your own research. Contact Focus Forward for more information on our research services.