Unit 02.03: The comparison the question implies

Every question implies exactly one comparison, and the comparison chooses the chart.

Five questions, five comparisons

Naming the comparison is the step between the question and the chart type.

The code pairs each question with the comparison it implies.

QUESTIONS = [
    ("is revenue growing?",              "this period against earlier periods"),
    ("which region is largest?",         "regions against each other, one period"),
    ("are we hitting target?",           "actual against target"),
    ("did the promotion work?",          "promoted against comparable unpromoted"),
    ("how is revenue distributed?",      "each unit against the whole"),
]
print(f"{'question':34} the comparison it implies")
for question, comparison in QUESTIONS:
    print(f"{question:34} {comparison}")

print("""
Every question implies exactly one comparison, and the comparison determines
the chart type before any aesthetic decision is made.

"Did the promotion work?" is the one people get wrong: the comparison is not
promoted-before against promoted-after, it is promoted against comparable
unpromoted over the same period, or the trend does the work for you.
""")

"Did the promotion work?" is the one people get wrong. The comparison is not promoted-before against promoted-after - that captures every other change in the period too. It is promoted against comparable unpromoted over the same period.

Once the comparison is named, the chart type usually follows without further debate: two groups over time is a slope chart or two lines, parts of a whole is a stacked bar, and so on.

The mistake this prevents

The mistake is arguing about chart types before naming the comparison. Those arguments are unresolvable, because the participants are each imagining a different comparison and are both right about their own.

Takeaway

Name the comparison the question implies before choosing a chart type. Before-and-after is rarely the right comparison for an intervention.

Mark complete