When a data set contains two categorical variables, the main goal is to compare group percentages carefully and decide whether differences in distributions suggest a meaningful association.

What comparing categorical variables means

With two categorical variables, the most useful summary statistics are usually proportions or percentages within categories. These statistics show how one variable is distributed for different groups of the other variable.

For AP Statistics, the key idea is not to compare groups by raw totals alone. Instead, compare how the categories are divided up within each group. That lets you judge whether the pattern stays about the same from group to group or changes noticeably.

If the distributions are very similar across groups, the variables do not appear to be associated. If the distributions are clearly different, the variables do appear to be associated.

Using percentages to compare distributions

The best way to compare categorical variables is usually to examine conditional distributions.

Segmented (stacked) bar charts display conditional distributions by showing each group as a 100% bar partitioned into category percentages. Differences in segment sizes across the bars indicate that the response distribution changes by group, which is visual evidence of association between the two categorical variables. Source

These are the percentages for one variable calculated separately within each category of the other variable.

Counts can be misleading when groups are different sizes. A larger group often has larger counts simply because there are more observations in it. Percentages adjust for that difference, so they make the comparison fair.

When comparing distributions, focus on the same response categories across groups. Ask questions like:

• Is the proportion in each category about the same across groups?

• Are some groups much more likely to fall into one category than others?

• Do the differences show a consistent pattern?

Suppose one variable has three categories. A good comparison checks all three categories, not just the most common one. That matters because the full distribution shows how the categories fit together. If one group has a higher percentage in one category, it must have lower percentages somewhere else.

The strength of an apparent association is judged by how different the percentages are. Very small differences may suggest little or no association. Larger, more consistent differences suggest a clearer association.

Deciding whether variables appear to be associated

There is no single summary number in this topic that automatically tells you whether two categorical variables are associated. Instead, the decision comes from comparing the distributions carefully.

Look for these features:

• Similarity: If corresponding percentages are close across groups, the variables do not appear to be associated.

• Difference: If several categories have noticeably different percentages, the variables appear to be associated.

• Consistency: If the same general pattern shows up across the categories, the evidence of association is stronger.

• Context: A difference that matters in one setting may be less important in another.

It is also important to compare the entire distribution.

A mosaic plot represents a two-way (or multi-way) table using rectangles whose areas are proportional to cell frequencies, so the full distribution is visible at once. When the partitions differ noticeably across groups, the plot visually communicates association because the conditional distributions are not the same. Source

A student might notice one category with a gap between groups, but AP Statistics expects a judgment based on the overall pattern, not one isolated value.

Sometimes you can compare the variables in either direction. For example, you might compare the distribution of variable A within categories of variable B, or the distribution of variable B within categories of variable A. The better choice depends on the question being asked. Use the comparison that gives the clearest answer in context.

How to describe the comparison in words

On AP questions, strong responses are clear, specific, and tied to the context. Good statistical writing usually does three things:

• identifies the groups being compared

• states relevant percentages or proportions

• explains whether those differences suggest an association

Useful sentence patterns include:

• “The distribution of _____ differs across categories of _____.”

• “The proportion in category _____ is higher for _____ than for _____.”

• “Because the conditional distributions are different, there appears to be an association between _____ and _____.”

Avoid vague comments such as “the numbers are different” or “the graph looks uneven.” Those statements do not explain how the distributions compare.

Also, be careful with language. Association describes a relationship in the data. It does not prove that one variable causes the other.

Common mistakes to avoid

Students often lose accuracy by making one of these errors:

• comparing counts instead of percentages when group sizes differ

• switching between different sets of percentages without noticing

• discussing only one category instead of the whole distribution

• giving numbers without explaining what they mean in context

• treating an observed association as proof of causation

• deciding too quickly from tiny differences that may not matter much

A strong comparison stays focused on percentages, the full distributions, and a clear contextual statement about whether the variables appear to be associated.