Comparing Two Populations
Learn to compare two data sets using means and mean absolute deviation, and decide when a difference between two groups is actually meaningful.
What you'll do in this lesson
A voice-first session with the Crimsora tutor on Comparing Two Populations, then targeted practice and FRQs — with the tutor adapting to where you get stuck.
What this lesson covers
Two seventh-grade classes both take the same quiz. Class A averages 82 and Class B averages 85. Is Class B really better at the material, or is that 3-point gap just ordinary wobble in the data? You cannot answer that from the means alone. You need to know how spread out each group is.
This lesson gives you a tool for that decision: compare the difference between the two means to the mean absolute deviation (MAD) of the data. If the gap between centers is several times larger than the typical distance data values sit from their own center, the difference is meaningful — the two groups really do look different. If the gap is smaller than the typical spread, the groups overlap so much that the difference could easily be noise. By the end you will be able to look at two dot plots, do a quick division, and defend a conclusion in words.
This lesson gives you a tool for that decision: compare the difference between the two means to the mean absolute deviation (MAD) of the data. If the gap between centers is several times larger than the typical distance data values sit from their own center, the difference is meaningful — the two groups really do look different. If the gap is smaller than the typical spread, the groups overlap so much that the difference could easily be noise. By the end you will be able to look at two dot plots, do a quick division, and defend a conclusion in words.
Why the Means Alone Are Not Enough
Imagine two basketball teams that both average 40 points per game. Team A scores 38, 39, 40, 41, 42. Team B scores 12, 25, 40, 55, 68. Same center, wildly different stories. Team A is dependable; Team B is unpredictable. Any honest comparison of two groups has to describe both center and variability.
Now flip the situation. Two groups have different means, and you want to know whether that difference means anything. Consider heights of two plant groups:
The gap between means is 3 cm. Within each group, values sit only about 1.2 cm from their own mean on average. Every fertilized plant except the shortest beats every unfertilized plant except the tallest — the two groups barely overlap. That 3 cm gap is large compared with the internal spread, so it is meaningful.
But suppose the groups had been 5, 20, 39 and 2, 17, 38. Both means shift by 3 again, yet the values are scattered over 35 cm. Here the 3 cm gap disappears inside the noise. Same difference in means, completely different conclusion. The spread is what decides.
This is why students who only compute two means and announce which is bigger have not finished the problem. The comparison is not complete until the difference is measured against the variability.
Now flip the situation. Two groups have different means, and you want to know whether that difference means anything. Consider heights of two plant groups:
| Group | Values (cm) | Mean |
|---|---|---|
| Fertilizer | 20, 21, 22, 23, 24 | 22 |
| No fertilizer | 17, 18, 19, 20, 21 | 19 |
But suppose the groups had been 5, 20, 39 and 2, 17, 38. Both means shift by 3 again, yet the values are scattered over 35 cm. Here the 3 cm gap disappears inside the noise. Same difference in means, completely different conclusion. The spread is what decides.
This is why students who only compute two means and announce which is bigger have not finished the problem. The comparison is not complete until the difference is measured against the variability.
Mean Absolute Deviation as the Ruler
The mean absolute deviation (MAD) measures typical distance from the mean. To find it, subtract the mean from each value, take the absolute value of each result, and average those distances:For the data 20, 21, 22, 23, 24 with mean 22, the distances are 2, 1, 0, 1, 2. Their sum is 6, and . So MAD cm: a typical plant in that group is about 1.2 cm from its group's mean.
MAD becomes a measuring stick. Instead of asking "is 3 cm big?" — which has no answer without context — you ask "how many MADs is 3 cm?" That ratio is what you report:With a difference of 3 and MAD of 1.2, the ratio is . The centers are two and a half typical deviations apart.
When the two groups have different MADs, use the larger one, or average the two — either is acceptable in an informal comparison, as long as you say what you did. Using the larger MAD is the more cautious choice, because it makes the ratio smaller and the claim harder to support.
A frequent slip is forgetting the absolute value and getting a MAD of zero, since positive and negative deviations always cancel. If your MAD comes out as 0 and the data are not all identical, you skipped the absolute value bars.
MAD becomes a measuring stick. Instead of asking "is 3 cm big?" — which has no answer without context — you ask "how many MADs is 3 cm?" That ratio is what you report:With a difference of 3 and MAD of 1.2, the ratio is . The centers are two and a half typical deviations apart.
When the two groups have different MADs, use the larger one, or average the two — either is acceptable in an informal comparison, as long as you say what you did. Using the larger MAD is the more cautious choice, because it makes the ratio smaller and the claim harder to support.
A frequent slip is forgetting the absolute value and getting a MAD of zero, since positive and negative deviations always cancel. If your MAD comes out as 0 and the data are not all identical, you skipped the absolute value bars.
Judging the Ratio: When Is a Difference Meaningful?
Here is the informal rule used in Grade 7 work:
A ratio of 2 or more is the usual dividing line for calling a difference meaningful. Think about what that means visually: if two dot plots are separated by twice the typical deviation, most of the dots in the higher group sit above most of the dots in the lower group. You could almost guess which group a value came from just by looking at it.
When the ratio is under 1, the two dot plots sit almost on top of each other. Knowing a value tells you nothing about which group it came from, so claiming one group is "better" is not supported.
Two cautions. First, the ratio is a judgment aid, not a law of nature — a ratio of 1.8 deserves the phrase "somewhat separated," not a confident yes or no. Second, this reasoning only supports a claim about the wider populations if the samples were randomly chosen. If you compare volunteers from the school track team with a random group of students, no ratio can fix that flaw in how the data were gathered.
Write your conclusion in a full sentence with the numbers in it: "The means differ by 3 cm, which is 2.5 times the MAD of 1.2 cm, so the fertilizer group is meaningfully taller."
| Ratio of difference to MAD | What it suggests |
|---|---|
| Less than about 1 | Difference is small; groups overlap heavily; not meaningful |
| About 2 | Noticeable separation; difference is likely meaningful |
| 3 or more | Strong separation; clearly meaningful difference |
When the ratio is under 1, the two dot plots sit almost on top of each other. Knowing a value tells you nothing about which group it came from, so claiming one group is "better" is not supported.
Two cautions. First, the ratio is a judgment aid, not a law of nature — a ratio of 1.8 deserves the phrase "somewhat separated," not a confident yes or no. Second, this reasoning only supports a claim about the wider populations if the samples were randomly chosen. If you compare volunteers from the school track team with a random group of students, no ratio can fix that flaw in how the data were gathered.
Write your conclusion in a full sentence with the numbers in it: "The means differ by 3 cm, which is 2.5 times the MAD of 1.2 cm, so the fertilizer group is meaningfully taller."
Reading Two Dot Plots Without Computing
Often you are handed two dot plots and asked for a quick judgment. You can estimate a great deal before doing any arithmetic.
First find each center by eye — where the dots balance. Then look at the overlap. If the two plots share almost the whole range of values, the difference in centers is small relative to spread. If they barely touch, the difference is large relative to spread.
A reliable check: locate the highest value in the lower group and the lowest value in the upper group. If the upper group's minimum is above the lower group's maximum, the groups are completely separated and the difference is definitely meaningful. If the upper group's minimum is near the lower group's center, the overlap is heavy.
Spread also shows up in shape. Dots clumped tightly around one spot mean a small MAD; dots stretched thinly across the axis mean a large MAD. Two plots with the same range can still have different MADs if one has a tight cluster plus a single far-out point.
Where students go wrong here: judging by the tallest stack of dots instead of the balance point, or comparing only the maximums. A group whose maximum is higher does not necessarily have a higher center. Another common error is treating any visible gap between the highest and lowest dots as proof of a difference — a single unusual value can create a gap without shifting the center at all. Always describe center and spread together, then compare them.
First find each center by eye — where the dots balance. Then look at the overlap. If the two plots share almost the whole range of values, the difference in centers is small relative to spread. If they barely touch, the difference is large relative to spread.
A reliable check: locate the highest value in the lower group and the lowest value in the upper group. If the upper group's minimum is above the lower group's maximum, the groups are completely separated and the difference is definitely meaningful. If the upper group's minimum is near the lower group's center, the overlap is heavy.
Spread also shows up in shape. Dots clumped tightly around one spot mean a small MAD; dots stretched thinly across the axis mean a large MAD. Two plots with the same range can still have different MADs if one has a tight cluster plus a single far-out point.
Where students go wrong here: judging by the tallest stack of dots instead of the balance point, or comparing only the maximums. A group whose maximum is higher does not necessarily have a higher center. Another common error is treating any visible gap between the highest and lowest dots as proof of a difference — a single unusual value can create a gap without shifting the center at all. Always describe center and spread together, then compare them.
Key terms
- Measure of center.
- A single number summarizing where a data set's values are located, such as the mean or the median.
- Variability.
- How spread out the values in a data set are; measured by mean absolute deviation, range, or interquartile range.
- Mean absolute deviation (MAD).
- The average distance of the data values from the mean, found with .
- Meaningful difference.
- A gap between two group means that is large compared with the variability within the groups — generally about 2 MADs or more.
- Overlap.
- The range of values that appears in both data sets. Heavy overlap signals that a difference between centers is small relative to spread.
- Difference-to-MAD ratio.
- The difference between two means divided by the mean absolute deviation; tells how many typical deviations apart the centers are.
- Dot plot.
- A graph placing one dot above the number line for each data value, useful for seeing center, spread, and overlap at a glance.
Worked example
Two after-school clubs recorded how many minutes members spent reading one evening. Chess Club: 22, 26, 30, 34, 38. Book Club: 40, 44, 48, 52, 56. Compare the two groups using center and variability, and decide whether the difference is meaningful.
Step 1: Find each mean. Chess Club: , and minutes. Book Club: , and minutes.
Step 2: Find each MAD. Chess Club distances from 30: 8, 4, 0, 4, 8. Sum is 24, so minutes. Book Club distances from 48: 8, 4, 0, 4, 8. Sum is 24, so minutes as well. The two groups have identical variability.
Step 3: Find the difference in means. minutes.
Step 4: Compare the difference to the MAD.Step 5: Interpret. The centers are 3.75 MADs apart, well above the guideline of about 2. Check the overlap as confirmation: the Chess Club maximum is 38 and the Book Club minimum is 40, so the data sets do not overlap at all.
Conclusion: The Book Club mean of 48 minutes is 18 minutes higher than the Chess Club mean of 30 minutes, which is 3.75 times the MAD of 4.8 minutes. That is a meaningful difference — Book Club members read noticeably longer that evening.
Step 2: Find each MAD. Chess Club distances from 30: 8, 4, 0, 4, 8. Sum is 24, so minutes. Book Club distances from 48: 8, 4, 0, 4, 8. Sum is 24, so minutes as well. The two groups have identical variability.
Step 3: Find the difference in means. minutes.
Step 4: Compare the difference to the MAD.Step 5: Interpret. The centers are 3.75 MADs apart, well above the guideline of about 2. Check the overlap as confirmation: the Chess Club maximum is 38 and the Book Club minimum is 40, so the data sets do not overlap at all.
Conclusion: The Book Club mean of 48 minutes is 18 minutes higher than the Chess Club mean of 30 minutes, which is 3.75 times the MAD of 4.8 minutes. That is a meaningful difference — Book Club members read noticeably longer that evening.
Practice questions
Group X has a mean of 62 and a MAD of 9. Group Y has a mean of 68 and a MAD of 9. What is the best conclusion about the difference between the two groups?
- The difference is meaningful, because 68 is greater than 62.
- The difference is not clearly meaningful, because the gap of 6 is smaller than the MAD of 9.
- The difference is meaningful, because both groups have the same MAD.
- No conclusion is possible without knowing the medians.
Answer: The difference is not clearly meaningful, because the gap of 6 is smaller than the MAD of 9.
Divide the difference by the MAD: . The centers are less than one typical deviation apart, so the two data sets overlap heavily and a value of, say, 65 could easily come from either group. Simply noting that one mean is larger, as the first choice does, ignores variability entirely — the whole point of this lesson. Equal MADs make the comparison easier but say nothing about whether the gap is big.
Two soccer teams record goals per game. Team Red: 1, 2, 2, 3, 2 (mean 2). Team Blue: 0, 6, 1, 5, 3 (mean 3). Find each MAD, compute the difference-to-MAD ratio using the larger MAD, and explain what you can and cannot conclude.
Answer: Team Red MAD = 0.4; Team Blue MAD = 2.0; ratio = 1 ÷ 2.0 = 0.5, so the difference is not meaningful.
Red's distances from 2 are 1, 0, 0, 1, 0, summing to 2, so . Blue's distances from 3 are 3, 3, 2, 2, 0, summing to 10, so . The difference in means is goal. Using the larger MAD, — half of one typical deviation. Team Blue's scores swing from 0 to 6, completely covering Red's range of 1 to 3, so the 1-goal gap in means is buried inside Blue's variability. You can say Blue averaged one more goal per game and was far less consistent, but you cannot claim Blue is meaningfully better at scoring.
Explain why using the larger of two MADs is a more cautious choice than using the smaller one when you judge whether a difference in means is meaningful.
Answer: A larger MAD in the denominator makes the ratio smaller, so it takes a bigger gap in means before you call the difference meaningful.
The ratio is . Since the difference stays the same, increasing the denominator decreases the ratio. If the difference is 6 and the two MADs are 2 and 4, using 2 gives a ratio of 3 (looks meaningful) while using 4 gives 1.5 (looks uncertain). Choosing the larger MAD sets the higher standard, so any conclusion you reach is supported even under the more demanding measurement of spread.
FAQ
- How big does the difference-to-MAD ratio need to be?
- In Grade 7 the common guideline is about 2 or more for a meaningful difference. A ratio under 1 means the groups overlap too much to distinguish, and 3 or more indicates strong separation. Around 1.5 to 2 the honest answer is "somewhat separated," and you should say so rather than force a yes or no.
- What if the two data sets have different MADs?
- Either use the larger MAD or average the two, and state which you did. Using the larger one is more cautious, since it makes the ratio smaller and requires a bigger gap before you claim a real difference. Also mention the different MADs in your written comparison — unequal spread is itself an important feature of the two groups.
- Can I compare medians and interquartile range instead?
- Yes. When data have outliers or a strongly skewed shape, comparing medians against the interquartile ranges is the better choice, since the mean and MAD both get pulled by extreme values. The reasoning is identical: ask whether the gap between the centers is large compared with the spread inside each group.
- Does a meaningful difference prove one thing caused the other?
- No. It shows the two groups really do differ on the measurement you took. Cause requires a carefully designed comparison in which the groups were alike except for one factor. Fertilized plants growing taller than unfertilized plants in the same conditions supports a cause; students in two different classes differing on a quiz could reflect many other factors.
Learn this with a teacher, not a page
The Crimsora tutor teaches Comparing Two Populations live — explaining on a whiteboard, asking you questions, and adapting to where you get stuck.