Topic 1.4 Representing a Categorical Variable with Graphs: choose, construct, and interpret bar graphs and other displays of a single categorical variable, and describe the distribution of categories.

What this topic is asking

The College Board (Topic 1.4) wants you to display a single categorical variable with an appropriate graph, principally a bar graph (in frequency or relative frequency form), and sometimes a pie chart, and to describe the distribution and spot misleading displays.

The bar graph

The gap is the visual signal that distinguishes a bar graph from a histogram. In a histogram (for quantitative data) the bars touch, because they cover adjacent intervals of a number line. In a bar graph the bars are separated, because "apple" and "banana" are not adjacent points on any scale; their left-to-right order is arbitrary and you may sort them however is clearest, often from tallest to shortest.

Frequency versus relative frequency bars

A frequency bar graph uses counts on the vertical axis; a relative frequency bar graph uses proportions or percentages. The shape is identical; only the axis labels change. Relative frequency is preferred when comparing two groups of different sizes, for exactly the reason it matters in tables: it puts both groups on a per-total scale so that a difference in sample size does not distort the picture.

Pie charts and their limits

A pie chart shows each category as a slice proportional to its relative frequency, so the whole circle represents $100\%$. Pie charts read well when there are few categories and you want to emphasize parts of a whole, but they make it hard to compare slices of similar size, and they cannot show counts directly. For most exam purposes a bar graph is the safer, more readable choice, and you should be ready to say so if asked which display is better.

Describing a categorical distribution

Describing a categorical display is not the same as describing a quantitative one. There is no center, spread, or shape in the quantitative sense, because the categories have no numerical order. Instead, a good description names the modal (most common) category, the least common category, and gives a sense of how the relative frequencies compare, all in context. For instance: "Type O is the most common blood type in this sample at $45\%$, with A close behind at $40\%$; B and AB are much rarer, at $11\%$ and $4\%$." Resist the temptation to talk about "skew" or "outliers," which are quantitative ideas; the bars could be reordered without changing the data, so those words do not apply. The exam specifically rewards descriptions tied to the context and the actual proportions, rather than generic phrases.

Misleading displays

A favorite exam target is the truncated axis. A bar graph whose vertical axis starts at, say, $0.30$ instead of $0$ makes the bars' visible heights non-proportional to their values, so a small real difference looks dramatic. Other manipulations include using pictures of different widths (so area, not height, conveys size) or omitting some categories. When you are asked to critique a display, the cleanest answer names the specific feature (for example "the axis does not start at zero") and explains the effect ("this exaggerates the difference between categories"). Being able to both build an honest graph and diagnose a dishonest one is exactly the dual skill Topic 1.4 trains, and it recurs whenever the exam shows you a graphic and asks whether it is appropriate or fair.

Try this

Q1. State one visual feature that distinguishes a bar graph from a histogram. [1 point]

• Cue. Bar-graph bars are separated by gaps (categorical); histogram bars touch (adjacent quantitative intervals).

Q2. A bar graph's vertical axis starts at $40$ rather than $0$. Explain the effect. [2 points]

• Cue. The visible bar heights are no longer proportional to their values, so differences between categories are exaggerated, misleading the reader.