Reading a Cross-Tabulation

Reading a cross-tabulation means finding the relationship between two variables in a table by comparing percentages within categories, never raw cell counts.

A table crosses education against voting, with rows for degree and no degree, columns for voted and did not. The cells hold 600, 200, 300, 400. Reading cells alone suggests degree-holders vote most, since 600 is the largest number. But the row totals differ: 800 degree-holders, 700 without. Convert within rows. Among degree-holders, 600 of 800 voted — 75 percent. Among those without, 300 of 700 voted — 43 percent. Now the comparison is honest, because each group is measured against its own size. The rule is to percentage in the direction of the causal claim: if education is treated as the independent variable, percentage across education categories and compare down the outcome column. Percentaging the wrong way produces a real number that answers a question nobody asked.