§4.2
Small Multiples for Heterogeneity
A national average is often the first honest summary and the first dangerous shortcut. It tells managers whether a pattern exists somewhere in the business. It does not tell them whether the pattern is broad enough to act on everywhere. Small multiples solve that problem by making the same comparison repeatedly, one market at a time.
The executive question: is the national pattern broad-based or region-specific?
The soup case has four census regions: East, Midwest, South, and West. A single national chart says Progresso raises price into a seasonal demand trough. The managerial question is whether that pattern is national or whether one region is doing most of the work.
Figure 1 uses small multiples with the same scales across regions. The same-scale choice matters. If every panel had its own y-axis, each region would look equally volatile. A common y-axis makes the differences in level and movement visible.
Same-scale small multiples make regional levels comparable
Metric-major layout: every panel shares one share axis, so the East’s much higher Progresso share is obvious — a level difference a national average would hide.
Figure 1 gives managers two lessons at once. First, the price seasonality is not a single-market artifact. Second, the share level differs sharply by region. The East is a much stronger Progresso market than the South or Midwest. A national pricing memo that ignores that fact would turn a visual average into an operational mistake.
Small multiples as model preparation
Small multiples also prepare students for later regression. A regression with one coefficient asks for one summary relationship. A small-multiple chart asks whether that one summary is likely to be stable across groups.
Figure 2 previews the later pricing chapter by plotting log(Progresso volume) against log(Progresso price), separately by region. The downward trend line is an elasticity-style visual: a one percent increase in price is associated with a percent change in volume. We are not yet claiming a causal elasticity. We are teaching students how to see the slope before they see the equation.
Log price–volume slope by region
A downward log-log slope previews price elasticity. The fitted line and slope are descriptive, not yet causal.
Figure 3 turns those visual slopes into coefficient intervals. This is a bridge view: still descriptive, but already moving toward statistical graphics.
| Region | Slope | 95% interval | R-squared |
|---|---|---|---|
| East | -1.74 | -1.81 to -1.68 | 0.51 |
| Midwest | -3.02 | -3.10 to -2.93 | 0.44 |
| South | -2.19 | -2.25 to -2.12 | 0.31 |
| West | -2.10 | -2.16 to -2.05 | 0.46 |
Figure 3 should not replace Figure 2. The table makes the slopes precise; the scatterplots show whether the precision comes from a clean pattern or from compressing a messy one. Managers need both views before asking whether a national elasticity estimate is useful.
Concept check
These three questions span the role of an explicit baseline as a comparison anchor, the same-scale rule that keeps small-multiple panels honest, and the judgment of when small multiples earn their space versus when a national average suffices.
- 1.A brand manager circulates a slide reading "Progresso non-winter price: 3.42 dollars" as evidence that pricing is too aggressive in the off-season. A colleague objects that the number, standing alone, cannot support that conclusion. What is the strongest version of the colleague's objection?
- 2.You build a four-panel small-multiple of soup volume by region. To "let each region breathe," a teammate sets each panel's y-axis to its own min and max so every region fills its frame. Why does the chapter treat this as a defect rather than a polish?
- 3.A VP asks whether the company's countercyclical pricing pattern "holds everywhere or just in a few markets." An analyst answers with the single national average price-by-month line. When does the national line suffice, and when do small multiples earn their space here?