Part II · Chapter 4

Comparison, Uncertainty, and Dashboards

Every chart is a comparison — name the baseline, show the spread, and end on a decision, not a pattern.

This chapter argues that a chart's first decision is its baseline, not its color, working throughout with Progresso scanner data — monthly sales across two thousand-plus stores from 2001 to 2006. It moves from January-indexed seasonality to region-by-region small multiples, to confidence intervals that describe store-month variation rather than causal effects, to log-log scatterplots that preview an elasticity-style slope before any regression appears. The payoff is a discipline for dashboards — monitor, diagnose, decide — that ends on the next test to run rather than an attractive pattern. A Bean & Basket capstone then chains grain, joins, metrics, and reshaping into a one-page board memo with its causal limits stated out loud.

Topics covered

baseline and index choicewinter / non-winter seasonal splitssmall multiples with shared axesregional heterogeneityconfidence intervals for managerspanel coverage and unbalanced datalog transforms and elasticity intuitionmonitor–diagnose–decide dashboardsdescriptive vs. causal claims

In this chapter

  1. 4.1Baselines, Indexes, and BenchmarksShows how baseline choice — January, a competitor, a region — decides which business pattern a soup-sales chart makes visible first.
  2. 4.2Small Multiples for HeterogeneityUses same-scale regional panels to test whether the national countercyclical pricing pattern is broad-based or driven by one market.
  3. 4.3Uncertainty for ManagersTeaches managers to ask what an interval varies over and which decision it changes, since precision is not identification.
  4. 4.4Statistical Charts Before StatisticsBridges visuals to estimation: skewed volume motivates logs, a log-log scatter previews a −2.46 elasticity, seasonality reveals the confound.
  5. 4.5Dashboard Decision SystemsRedesigns the soup dashboard into a monitor–diagnose–decide system that ends by naming the next pricing test, not a verdict.
  6. 4.6Visual Decision Brief StudioSequences the Progresso soup visuals — indexes, small multiples, uncertainty, dashboards — into a one-page executive pricing brief that ends in a decision and names its causal limits.

Interactive studios

Featured data stories

Interactive D3 pieces from the gallery that put this chapter’s chart ideas to work — each opens in a new tab.