Feature Releases: Stat Assist - Identify statistically significant signal in open-ended responses

Uncover what truly matters in open-ended survey data

We’re thrilled to introduce Stat Assist, a powerful feature set - now live in the Fathom analytics interface.

Stat Assist helps you surface statistically significant differences across the distribution of open-ended themes — empowering you to quickly spot patterns that can drive smarter strategies and faster decision-making.

Powered by comprehensive statistics and layered on top of Fathom’s high-quality thematic coding, Stat Assist supports transparent, rigorous analysis of open-ended data at scale.

Analyze open-ended theme segments with ease & confidence

• Identify key segments with statistically significant differences in theme distribution, based on closed-ended responses or demographic attributes

• Spot meaningful insights visually with highlighted themes on interactive charts, so you can analyze results instantly

• Access robust, detailed statistics in an intuitive dashboard, built for transparency and in-depth review

• Simplify reporting with summary insights that make crafting compelling presentations easy

When Stat Assist is enabled, three things happen:

1. Segments with statistically significant difference are highlighted in the drop down menu

2. Themes are highlighted in the interactive chart to show where the variance is occuring 

3. Insights open in the Stat Assist drawer, giving you a detailed breakdown of that segment and its key variations.

Statistical rigour in qual at scale is hard, so we’ve research and tested dozens of approaches to define a powerful solution

Stat Assist uses three powerful metrics—Cramér's V, Odds Ratio, and Prevalence—to give you comprehensive insight into the relationship between themes and segments in your data.

Here’s how these tests were chosen and why they’re the best fit for qualitative analysis.

1. Suitability for Categorical Data

When analyzing open-ended survey responses, which are typically qualitative and categorical, many standard statistical tests like t-tests or ANOVA aren’t suitable. They assume a normal distribution or require numerical data, which isn’t applicable for themes and segments in feedback data.

Instead, Stat Assist leverages Cramér's V and Odds Ratio —two metrics well-suited for assessing relationships in categorical data. These metrics - combined with Prevalence - help ensure insights derived from Fathom are reliable, designed specifically for the types of responses market research professionals encounter, and strategically useful.

2. Comprehensive Insights in Three Key Areas

• Is there a significant relationship?
  Cramér's V quickly identifies whether there’s a meaningful association between a theme (like "sustainability") and a segment (such as Gen Z respondents), giving you a big-picture view of what matters in your data.

• How are different segments driving it?
  The Odds Ratio goes a step further by measuring the variability in the likelihood of respondents mentioning a particular theme across different categories of a segment (e.g., different age groups or customer types) This insight adds both direction and depth to your analysis, helping you see not just that a relationship exists, but how it's influenced by specific segments.

• Is the theme prevalent enough to matter?
  Prevalence ensures you’re not just finding patterns but also determining whether they’re relevant. This metric reveals how common a theme is within a particular segment, ensuring that your analysis leads to practical, data-driven actions.

3. Interpretability and Actionability

Stat Assist’s metrics are fully transparent in the Fathom platform, and easily interpretable for market researchers and customer insights teams. By focusing on clear, actionable metrics, Stat Assist empowers researchers with the rigour and clarity needed to be separate signal from noise, and focus on the insights that truly matter.

Rigorous stats with AI support, but no AI shortcuts

Unlike large language models that focus solely on summarizing and generating text, Stat Assist provides rigorous statistical analysis directly within Fathom. In other words, we never ask an LLM to compute stats!

Stat Assist computes real, rigorous statistics within Fathom. With Stat Assist, you see each stat clearly and can interpret results with confidence, knowing the software ensures accuracy every step of the way.

And then yes, of course, we leverage AI to assist in sensemaking and reporting - making the analysis process smooth and delightful.

Want to see Stat Assist in action? Request a demo!

We’ve worked tirelessly to deliver this breakthrough in open-ended analysis and would love to chat, geek out on methodology and answer all your questions!