Comparing the Centers of Two Groups
In Why Comparing Distributions Needs All Four Features, you learned that center is only one part of a comparison of distributions. This tutorial focuses on making that part clear and accurate: comparing two groups’ medians or means for the same quantitative variable.
A center comparison should do more than list two numbers. It should identify the groups and variable, name the measure of center, include the units, and say which group has the higher center and by how much. For example, “Group B’s median wait is 5 minutes longer than Group A’s” gives the direction and size of the difference in context.
The median and mean describe center in different ways. As discussed in Choosing Mean or Median to Describe Center, the median is often useful for skewed distributions or distributions with outliers, while the mean is often useful for approximately symmetric distributions without strong outliers. A comparison should use a measure that represents the distributions sensibly, rather than selecting whichever measure makes one group look higher.
A Reliable Way to Write the Comparison
Before comparing numbers, check that both summaries describe the same quantitative variable and use the same units. A median of 30 minutes cannot be directly compared with a mean of 35 dollars: they refer to different variables and units. Even when units match, compare medians with medians or means with means unless you have a clear reason to discuss both measures separately.
Choose a subtraction order that makes the direction easy to explain. One useful convention is to subtract Group A’s center from Group B’s center:
If the result is positive, Group B’s center is higher by that amount. If the result is negative, Group A’s center is higher by the absolute value of the difference. You can also state the difference without a signed number, as long as your sentence makes the direction clear.
Name both groups, the quantitative variable, and its units.
Use shape and unusual values to guide whether the median or mean is more informative, as in earlier tutorials on choosing summary statistics.
Subtract the centers in a stated order and keep the result in the variable’s units.
Say which group has the higher center and by how much. Make clear whether you compared medians or means.
A positive difference does not mean every observation in that group is higher. It describes the relative locations of the two centers, not how much the distributions overlap or how their spreads compare. As emphasized in the previous tutorial, those questions require looking at other features of the distributions.
Worked Example: Comparing Appointment-Wait Medians
Worked Example: Comparing Appointment-Wait Medians
A fictional clinic team summarizes appointment waits for two groups of patients. The median wait is 30 minutes for Group A and 35 minutes for Group B. Compare the centers.
Identify. The groups are Group A and Group B. The quantitative variable is appointment-wait time, measured in minutes. Both summaries are medians, so they describe center using the same measure and units.
Calculate. Subtract Group A’s median from Group B’s median:
The positive result means Group B’s median is higher. Reversing the order gives \(30-35=-5\) minutes, which expresses the same comparison: Group A’s median is 5 minutes lower.
Compare in context. The median appointment wait for Group B is 5 minutes longer than the median appointment wait for Group A. In other words, the central wait time is higher for Group B by 5 minutes.
This sentence reports a comparison of centers, not a complete comparison of the distributions. It does not establish that every patient in Group B waited longer than every patient in Group A, or even that most individual waits in Group B exceeded those in Group A. To assess the broader comparison, we would also need information about the groups’ shapes, spreads, and unusual values.
Comparing Means
The same structure works when the appropriate measure is the mean. The sample mean, written \(\bar{x}\), is the arithmetic average: add the observations and divide by the number of observations. As in Computing the Mean and Interpreting It, its units are the same as the original data.
When comparing means, report the difference in the variable’s units and interpret it as a difference between the groups’ averages. Avoid changing the interpretation into a claim about every individual. A mean is influenced by every observation, so inspect the distributions before relying on it, especially if one group has a strong skew or an extreme value.
Worked Example: Comparing Average Quiz Scores
Worked Example: Comparing Average Quiz Scores
Two fictional groups of students completed the same quiz. Their scores, in points, are listed below. For this example, the values are approximately balanced around their centers, with no strong outliers apparent.
| Group A (points) | Group B (points) |
|---|---|
| 68, 72, 76, 80, 84 | 71, 75, 79, 83, 87 |
Compare the groups’ mean quiz scores.
Find each mean. Group A’s scores sum to \(68+72+76+80+84=380\). There are 5 scores, so:
Group B’s scores sum to \(71+75+79+83+87=395\). There are also 5 scores, so:
Calculate the difference. Subtract Group A’s mean from Group B’s mean:
Compare in context. Group B’s mean quiz score is 3 points higher than Group A’s mean quiz score. This describes the difference between the groups’ average scores in these fictional data. It does not say that every student in Group B scored more than every student in Group A.
When the Mean and Median Tell Different Stories
The mean and median can give different comparisons when the distributions have different shapes or unusual values. This is not a calculation error. The two measures respond differently to the data: the median depends on the ordered positions of observations, while the mean uses every numerical value. As covered in Why Right Skew Pulls the Mean Above the Median, a relatively large value in a right tail can pull the mean upward.
If both measures are reported, explain which is more useful for describing center and why. Do not present one measure as “the right answer” in every situation. A careful comparison links the choice of center to the distribution’s shape and unusual features, then keeps the interpretation consistent with that choice.
Worked Example: A High Wait Affects the Mean
Worked Example: Comparing Skewed Delivery Times
A fictional business records delivery times, in minutes, for two groups of orders. The times in each group are ordered:
| Group A (minutes) | Group B (minutes) |
|---|---|
| 10, 12, 14, 16, 18 | 8, 10, 12, 15, 45 |
Compare the groups using both the median and the mean, and decide which measure gives a more useful description of center here.
Compare medians. With five observations in each group, the median is the third value. Thus, Group A’s median is 14 minutes and Group B’s median is 12 minutes. The difference, taking Group B minus Group A, is:
The negative result means Group A’s median is 2 minutes higher than Group B’s. In context, the central delivery time by median is 2 minutes longer for Group A.
Compare means. Group A’s total is \(10+12+14+16+18=70\), so its mean is \(70/5=14\) minutes. Group B’s total is \(8+10+12+15+45=90\), so its mean is \(90/5=18\) minutes. The mean difference is:
By the means, Group B is 4 minutes higher. The median comparison points in the other direction: Group A is 2 minutes higher. Group B’s value of 45 minutes is far above its other times and creates a long upper tail, so the mean is pulled upward. For describing a typical delivery time in these groups, the medians are more informative than the means.
Conclude in context. Group A’s median delivery time is 2 minutes longer than Group B’s median, while Group B’s mean is 4 minutes longer than Group A’s mean. Because Group B has a very high delivery time relative to its other observations, the median is the more representative center for this comparison. The difference between the two comparisons is a reason to inspect the distributions, not a reason to choose whichever statistic supports a preferred conclusion.
Common Mistakes and AP Exam Tips
- Listing two centers without comparing them. “Group A’s median is 30 minutes and Group B’s is 35 minutes” gives useful values, but does not state the difference. Add: “Group B’s median is 5 minutes higher.”
- Leaving the direction unclear. “The medians differ by 5” does not tell the reader which group has the higher median. State both the amount and the direction.
- Omitting the variable or units. “Group B is 5 higher” is incomplete. Say, for example, “Group B’s median wait is 5 minutes longer.”
- Comparing unlike centers without explanation. A mean for one group and a median for the other are different measures. For a straightforward comparison, use the same suitable measure for both. If you discuss both measures, label each one and explain why they differ.
- Claiming more than a center comparison supports. A higher median does not mean all observations in that group are higher. Do not infer that one group’s values are more consistent, less spread out, or non-overlapping without evidence about spread and the distributions.
- Choosing a measure just because it gives a desired result. Inspect shape and unusual values, then choose a measure that represents center sensibly. As in Choosing Appropriate Summary Statistics, the choice should follow the data, not the conclusion you hope to make.
- Generalizing beyond the data. A descriptive comparison of observed groups does not by itself establish a cause or justify a claim about a larger population. Keep the conclusion tied to the groups represented by the summaries.
For a strong AP response, state the measure, name both groups and the quantitative variable, report the difference with units, and clearly identify which group has the higher center. If the shape makes one measure more appropriate, briefly explain that choice. Keep the statement focused on center; compare spread, shape, and unusual features separately when asked for a full comparison.
Check Your Understanding
For each item, write a comparative statement that names the measure of center and interprets the difference in context.
- Group A’s median weekly exercise time is 120 minutes; Group B’s is 105 minutes. Which group has the higher median, and by how many minutes?
- Two groups’ mean plant heights are 18 centimeters and 21 centimeters. Write a sentence comparing their average heights, identifying which group has the higher mean.
- Group X’s median commute time is 24 minutes and Group Y’s is 29 minutes. Calculate Group Y’s median minus Group X’s median and interpret the result.
- A distribution has a strong right tail and one unusually high observation. Which measure of center is often more useful for comparison, the mean or median? Briefly explain why.
- One group has a higher median than another. Does that establish that every observation in the first group exceeds every observation in the second? Explain what the median comparison does establish.