Reading a Crosstab

A crosstab is read by finding the percentage base first: know what each cell divides by before comparing anything.

Take a table of turnout by age from a survey of 1,000 adults: rows are age groups, and each row shows percent voting. Row one — ages 18 to 29: 50 percent. Row two — 45 to 64: 70 percent. Step one, the base: each percentage divides by its own row's respondents, not by the whole sample — 50 percent of the young respondents, not of everyone. Step two, the comparison: 70 minus 50 is a twenty-point turnout gap by age, matching Gov 206's pattern. Step three, the trap: if the young row holds 200 respondents and the older row 400, the counts of voters are 100 and 280 — comparing raw counts across rows of different sizes manufactures a bigger gap than exists. Step four, the caveat: survey turnout runs higher than official turnout, because respondents overreport voting — the Gov 173 lesson in table form.