Slice and Dice Your Data: Crosstabs 101

Any insights professional looking for some flexibility in analysis can make use of crosstabs.

Often presented as a sheet or grid of data, they offer access to descriptive statistics and help mixed-methods researchers imbue some quant into their open-ended data.

Crosstabs afford—as the name suggests—a crossing of variables for trendspotting, theme aligning, and frequency noting. They can be robust and support large-scale conclusions or be scrappier, injecting some numbers into largely qualitative outputs. However you use them, they're an important tool in an insight professional's methodological kit.

Crosstabs are a great way to add quant to your research repertoire without "being" a quant scholar per say. The crosstab presentation of data can unlock new insights, as it visually organizes the data in a different way than you may have collected it.

Putting data—even open-ended questions—into a grid or matrix can shift the cognitive vantage point for a researcher, unlocking new perspectives and ways to contextualize and frame "what's there" in the data.

Below we walk through what they are, how you might leverage them, and the situations they're a good—and better—fit.

What is a crosstab?

Cross-tabulation (also known as crosstabs, xtabs, crossbreaks, contingency tables, and a few other names) are matrix tables that show relationships between multiple variables within your data. A “variable” can mean any categorical element at all in your data, including:

• Demographic information (e.g. gender, race/ethnicity, or age)
• Pre-determined segments or personas (e.g. power users, light users, potential customers, etc.)
• Answers to a specific question in your survey (e.g. High vs. low NPS scores)

Crosstabs slice and dice your data by these variables and allows you to compare counts and frequencies from your survey across different groups at once.

Making and using crosstabs

Crosstab variables can take the form of any categorical data or metadata in your survey close-ended questions (single- or multiple-select), rank questions, demographic information, segment / persona groups, or tags applied to qualitative data.

The key decision you need to make is what categories you want to compare against. These will turn into your columns in your chart. The data you actually want to compare within those categories will be your rows - this will normally be answers to close-ended questions, though it could easily also be tags you’ve applied to open-ended answers.

The process of actually generating crosstabs can be accomplished in a lot of different ways. Excel and other spreadsheet programs can create crosstabs from raw data using pivot tables. Statistical programs such as SPSS, R, and Stata all have crosstab capabilities. If you’re on dscout, you can export raw data to turn into crosstabs, or dscout can export crosstabs for you.

Once you have them created, they will usually look something like this:

This table shows the frequency of answers to each individual question choice overall (in the left column) and then broken down by individual segments in subsequent columns.

In this example, our team ran a project on generational approaches to the workplace. This table shows a comparison between four generation segments in regard to what they think a “normal” amount of time is to spend with one company.

In this crosstab, I observed that half of the Baby Boomers in our study said that more than 10 years was a normal tenure, compared to 0% of scouts in the Millennial and Gen Z groups. Conversely, younger people tend to think that 1-2 years or 3-5 years is a normal tenure, and only 6% of Baby Boomers selected either of those answers. That’s a big generational difference in expectations!

Often, the findings I uncover are insights in and of themselves. For a large-scale study, I may validate the differences I uncover with statistical testing.

But with qualitative research, these frequencies are often less stand-alone observation and more like an important guide. I think of surprising frequencies as blinking neon signposts for my qualitative data that say, “Cool findings here!”

In the case of the study above, having seen that expectations for tenure at companies is extremely different between generations, I can now ask myself, “Why does that difference occur?” In answer, I would filter to the specific segment and question in mind and read through the answers, exploring the nuance in what drives that group to answer so differently than the others.

In this case, I would probably start by filtering to the respondents Baby Boomer segment who answered, “More than 10 years” and read through their open-ended responses about what they value in a workplace. I’d also look through younger generations who answered 1-2 years and see if I can uncover what has shifted between older and younger generations.

Why use crosstabs?

For qualitative researchers, crosstabs may seem like an unnecessary step, since numbers aren’t usually our primary interest. But if you can get the hang of creating and reading them, they can be an efficient tool in your analysis toolbox.

First off, they’re easily the best way to answer comparative questions about data, such as:

• Do people of different genders have different reactions to using our product?
• Do older people interact with our website differently than younger people?
• Do people who use our product once a month have different preferences than people who use it more often?
• Does the way someone browses before shopping impact which prototype of our app that they prefer?
• If someone answered that they like the taste of our product, are they also more likely to rate it as healthy for them?

Crosstabs are particularly well-suited for research with built-in segments, such as personas research or jobs-to-be-done—you can quickly scan your quantitative results across all segments at once to zero in on the questions that you need.

However, I personally make crosstabs for every project I run, even if it’s purely generative and exploratory with no top-down segments. I do this especially when I don’t have any pre-planned analytical directions. There’s a couple of reasons why:

It’s a fast way to get acquainted with my data

Qualitative surveys can be extremely overwhelming. The amount of time it would take to dig through all the nuances and permutations of the data we get back is usually more than I have.

Using close-ended questions lightens the load, but only by so much. Scrolling through bar graphs can still be a bit of a slog, especially if I don’t know what I’m looking for. Getting all that data in one birds-eye view helps me to orient myself faster than any other data view.

It generates leads

I don’t always have the time to scrub through every single open-ended response I get back looking for interesting patterns. If I can use a crosstab to find the interesting patterns first, I can use that as a guide for which particular answers to look at.

You don’t know what you don’t know

I think this is the true value that crosstabs give you. If my data is exploratory, I often just don’t know what I’m looking for. If I didn’t slice and dice my data in these ways and instead relied on aggregate numbers, I could miss out on some truly rich insights.

Crosstabs are a mainstay of quantitative research and are an effective way to organize your dtat for clearer analysis. To learn more about using crosstabs on dscout or have questions about starting your own project, connect with our team.