Counts and rates answer different questions. A state can add 100,000 voters while its turnout rate declines if the eligible population grows faster. A candidate can receive more votes but a smaller vote share if total participation expands. Percentages may not sum to 100 when categories overlap, values are rounded, or respondents may select more than one answer. Before interpreting, reconstruct the numerator and denominator.

Axes and visual encodings deserve equal attention. Evidence display 41C begins its vertical axis at 58 percent rather than zero. The point estimates remain 61 and 63 percent, but the two-point gap occupies a large share of the plotting height. The whiskers add another fact: each estimate has two points of displayed sampling uncertainty, so the intervals overlap. Overlap does not prove equality, and a formal comparison may use more information, but the display does not justify a dramatic or causal claim.

Percentage-point change and percent change are separate calculations. A rise from 48 percent to 52 percent is four percentage points. Relative to the starting level, the increase is four divided by 48, or about 8.3 percent. A large relative increase from a tiny base may still describe very few people, so raw numbers provide useful context.

Time series introduce starting dates, intervals, smoothing, and revisions. Selecting two convenient endpoints can hide reversals between them. A moving average reduces short-term noise but can delay a turning point. Missing observations should be marked rather than connected as if measured. Error bars, confidence bands, sample sizes, and data-quality notes show uncertainty; they are not decoration. First describe the largest pattern the display clearly supports. Then calculate comparable quantities, consider uncertainty, and only afterward evaluate an explanation.

Video lesson: Maps for Social Studies | GED, HiSET, and TASC Prep