The Data Literacy Cheat Sheet

The Data Literacy Cheat Sheet is the starter map for Charted Truths. It teaches the small set of questions that protect a viewer from most viral chart mistakes: what is the denominator, what is the unit of analysis, is the average hiding the distribution, does the chart show a rate or a count, and does the claim imply causation when the data only supports association.

Use it before believing a chart, sharing a chart, or turning a chart into an argument. The point is not to become cynical. The point is to become harder to fool. A good chart should make the comparison clear, name the source, show the denominator, and include the caveat that changes how far the claim can go.

Mini exercise: take any viral chart and write three sentences. First, what the chart directly shows. Second, what it does not prove. Third, what denominator would change the interpretation. If those three sentences are hard to write, the chart needs a slower read.

1. Name the claim before judging the chart

Most chart mistakes begin before the chart is inspected. The headline says one thing, the visual shows another, and the audience quietly blends them together. Write the claim as a plain sentence first: “This chart shows that X is higher than Y for Z group during T period.” If you cannot fill in X, Y, Z, and T, you are not ready to argue from the chart.

The sentence should be narrow. “Dating apps are broken” is not a chart claim. “In this sample, fewer matches turned into conversations for group A than group B” is a chart claim. Narrow claims are easier to verify and harder to exaggerate.

2. Find the denominator

The denominator is the group underneath the percentage. A chart that says “40 percent” is incomplete until you know 40 percent of what: users, accounts, survey respondents, households, games, arrests, claims, or visits. Changing the denominator can flip the emotional meaning of the same number.

Ask who is missing. A survey of active app users excludes people who quit. A chart of police reports excludes incidents never reported. A chart of average salary can exclude unemployed people. The caveat is not trivia. It tells you how far the claim can travel.

3. Check whether the average hides the shape

Averages are useful when the group is fairly compact. They become dangerous when the group has two peaks, extreme outliers, or a long tail. If one small group has very large values, the average can move while most people experience no change.

When a chart relies on an average, look for a median, percentile range, histogram, box plot, or raw distribution. If the distribution is unavailable, the safest sentence is narrower: “The average moved” rather than “most people experienced this.”

4. Separate counts from rates

Counts answer how many. Rates answer how often relative to the population or opportunity. A city with more people can have more events and still have a lower rate. A rare event can double and remain rare. A popular platform can generate more bad examples simply because it has more total activity.

If the chart shows…	Ask for…	Why it matters
Total events	Events per relevant population	Bigger groups naturally produce bigger counts
Percent change	Starting value and ending value	A huge percentage can come from a tiny base
Ranking	Margin between ranks	Adjacent ranks may be statistically indistinguishable
One vivid example	Base rate	Vivid examples can overwhelm rare-event math

5. Watch for implied causation

Two lines moving together do not prove one caused the other. The pattern may be real and still have the wrong explanation. Ask whether the chart rules out seasonality, selection effects, policy changes, measurement changes, or a third factor driving both lines.

The stronger the causal language, the stronger the evidence needs to be. “Moved with” requires less evidence than “caused.” “May contribute to” requires less than “is responsible for.” If the chart only shows association, keep the sentence associative.

6. Do the pre-share check

Before sharing the chart, write this compact source note:

1. What the chart directly shows.
2. What population or denominator it uses.
3. What caveat would change the interpretation.
4. Whether it shows a count, rate, average, distribution, or relationship.
5. Whether the headline claims more than the chart proves.

If the note makes the chart less exciting, that is useful information. The chart may still be worth sharing, but now it will be shared as evidence instead of bait.