Market Research: Foundations to Practice · Analyzing and Interpreting Data
Segmentation Analysis
An aggregate finding can mask sharply different patterns within specific subgroups, sometimes patterns that matter more to the business decision than the overall average ever could. This chapter covers how to find them responsibly.
An aggregate, whole-sample result can mask meaningfully different patterns within specific population subgroups; a new feature might show only lukewarm interest across the overall sample while generating genuinely strong enthusiasm within one specific, identifiable customer segment, an important, actionable pattern the aggregate topline number alone completely hides from view.
Key Takeaways
- An aggregate result can mask a strong pattern within a specific subgroup, an important signal the overall topline number completely hides.
- A priori segmentation defines segments before data collection based on existing theory; post hoc segmentation explores data after the fact for patterns.
- The multiple comparisons problem means testing many subgroup cuts increases the chance of a false-positive finding purely by chance.
- A statistically real segment still needs to be practically actionable and targetable in the real world to actually be useful for a business decision.