Getting Valuable Insights From Small Data Sets

work hard Data Analytics Statistics Information Business Technology.

By Kevin Buth, M.S.

Let’s face it. Not every organization has the time and budget for months-long studies that survey thousands of people. But that doesn’t mean there aren’t creative ways to uncover new insights about your customers and business. And it certainly doesn’t take research off the table.

Just like other areas of business, research, analysis, and planning can be done on a variety of scales, with projects tailored to your organization’s budget, schedule, and workload capacity. What’s more, you can still get the game-changing ideas without the time and investment of big research studies.

In some situations, in fact, smaller sets of data can be more beneficial to problem solving from a business perspective.

Getting big insights from small studies. Sounds challenging, right?

Here are five key ideas that prove you can get solid insights from smaller sets of data:

1. Usefulness doesn’t necessitate high sample sizes.

Let’s say you want to study a large population of customers. However, your budget only allows for 50 responses. The margin of error will be higher than if you were able to get 400 responses, but you can still get solid directional data from just 50.

There are tools and methods an analyst can employ to tease out useful information, including:

  • Statistical tests (i.e., two sample t-test, Fisher’s exact test)
  • Resampling techniques (i.e., bootstrap method, jackknife method) 
  • Ranking techniques (i.e., Mann-Whitney U test)

By employing the techniques above, you can significantly strengthen confidence in the data, even with a small sample size.

2. Small doesn’t mean weak.

Saying your sample is “small” is a relative term. For example, if you can get 50 responses from a population of 500, that data is much stronger than if your population is, say, 5,000.

Using effective screening and segmentation can have an impact on the quality of results. There may be a narrowly defined group within your population that drives key decisions. By carefully asking the right screening questions and segmenting your responses, you can arrive at key insights that can be highly valuable for business decision making, even with a small sample size.

3. Qualitative studies are easier done small.

We’ve all answered surveys with open-ended questions. That data is qualitative because, no matter how many responses you have, every answer will be unique.

Analyzing qualitative data is hard work. However, small, qualitative studies can be a great jumping off point for further study or a source for new and innovative ideas—like a crowd-sourced brainstorm session. There are times when you “don’t know what you don’t know.” A small qualitative study can uncover highly valuable information that can then be tested and validated in other ways.

4. When in doubt, shoot for 380.

Without getting too deep into the math, we can say that 380 is considered a benchmark number of responses when conducting quantitative research such as surveys and polls. This allows for high levels of confidence when studying populations of 10,000 or more. If you work with stakeholders who tend to “argue with the research,” a sample size of 380 will get the statistical validity you need to back up the results with confidence.

5. What, how, and who you ask questions.

Analyzing and interpreting data is only one step in the entire research process. To get started properly, surveys and their individual questions need to be designed in a way that get you the answers to the questions you’re asking.

Be hyper-cautious about avoiding leading questions, otherwise the research may provide the answers you want but won’t match reality. In cases like this, even a mountain of data can prove useless.

Also keep watch for questions that inadvertently ask two things at once. For example: “Do you drive your vehicle to go sightseeing and fishing? (Yes/No)” If the respondent does go sightseeing, but is not a fisher, there is no correct answer. This sort of issue can usually be remedied by limiting the use of the word “and” in your questions. Be clear and concise in your questions to get clear and effective results.

If creating and coordinating an entire study in-house is proving to be a challenge, or if you have some uncertainty about how to ask the right questions, getting help from a third-party analyst can make all the difference in getting bias-free, valid results even with small samples.

Using the techniques above will help you to get the greatest value out of your research. Now get out and ask some questions. There are insights there to be found!

About Kevin Buth, M.S.

When it comes to analyzing stats, Kevin Buth is your guy. He’s a graduate of North Dakota State University, where he received both his bachelor’s and master’s degrees in statistics. Since graduation, Buth has been with Prime46 as a research analyst. He prides himself on doing things right the first time and is especially great at complex data analysis.

In order to achieve the best results, Buth uses a variety of platforms and tools to create surveys that gather the most important and relevant data. Buth has worked with a diverse set of clients across industries such as manufacturing, health care, education, professional services, and more.

Sign up for email updates.

"*" indicates required fields

By signing up, you agree to receive emails from Prime46. Unsubscribe at any time by clicking on the unsubscribe link at the bottom of our emails. Questions?