AP Statistics · Unit 4 · Inference for Quantitative Data: Means
4.6 Differences in sample means
Calculate the mean and standard deviation of the difference of two sample means, describe its sampling distribution, and check the conditions for a normal approximation.
1/5The difference of two sample means varies from sample to sample
- The difference of two sample means varies from sample to sample
- The mean of the difference equals the difference of the means
- The spread of the difference combines the spread of both samples
- The sampling distribution is approximately normal under the right conditions
- Checking conditions justifies the normal approximation
- Check your recall
- Practice: 10 questions
Part 1 of 5: The difference of two sample means varies from sample to sample
Suppose you want to compare average commute times in two cities. You take a random sample from each city and subtract one sample mean from the other. A second pair of samples gives a different difference. Each group has its own mean, standard deviation and sample size, so we label them 1 and 2.
The pattern of those differences is the sampling distribution of the difference. It describes every value that can take across repeated pairs of samples. To describe it, you need its center, spread and shape.
Notation for two groups
- Quantity
- Population mean
- Group 1
- Group 2
- Quantity
- Population standard deviation
- Group 1
- Group 2
- Quantity
- Sample size
- Group 1
- Group 2
- Quantity
- Sample mean
- Group 1
- Group 2
- Quantity
- Sample standard deviation
- Group 1
- Group 2
Key term
sampling distribution of the difference
The sampling distribution of the difference is the distribution of over all possible pairs of independent random samples of sizes and .
Part 2 of 5: The mean of the difference equals the difference of the means
Now consider where those differences are centered. Each sample mean is centered at its own population mean. So the differences are centered at the gap between the two population means.
In symbols, . Suppose group 1 has a true mean of 80 and group 2 has a true mean of 75. Over many pairs of samples, the differences average 5. The statistic is a point estimate of .
Key term
point estimate
A point estimate is a single value computed from sample data and used to estimate a population parameter.

Quick check
One question on the part you just read
Easy
Show answer and explanation
Answer: A · Worked explanation
Part 3 of 5: The spread of the difference combines the spread of both samples
Subtracting one sample mean from another does not cancel their variability. Each mean adds its own randomness to the difference.
When the two groups are independent samples, the variances of the two sample means add. This gives . Variances add, but standard deviations do not.
In practice, and are usually unknown. Replace them with and to get the standard error of the difference, .
Worked example
Mean and standard deviation of a difference
- Mean: .
- Group 1 part: .
- Group 2 part: .
- Standard deviation: .
Answer: The mean is 5 points and the standard deviation is about 2.83 points.
Key terms · tap one for its definition
independent samples
Independent samples are two samples in which the individuals chosen for one have no influence on the individuals chosen for the other.
standard error of the difference
The standard error of the difference is , the estimate of the standard deviation of when population standard deviations are unknown.
Part 4 of 5: The sampling distribution is approximately normal under the right conditions
We now know the center and the spread. The remaining question is shape. When the conditions in the next section hold, the sampling distribution of is approximately normal.
- Shape: approximately normal.
- Center: .
- Spread: .
A normal model therefore describes the difference. Values near the center are common, and values far from it are rare. About 95% of differences fall within 2 standard deviations of the mean.
Shape depends on the data. Normal populations give a normal difference for any sample sizes. Large samples give an approximately normal difference even when the populations are skewed.

Quick check
One question on the part you just read
Easy
Show answer and explanation
Answer: A · Worked explanation
Part 5 of 5: Checking conditions justifies the normal approximation
A normal model is only an approximation, and it needs support from the study design and the data.
- Random: each sample is random, or treatments were randomly assigned.
- Independent groups: the two samples do not influence each other, and neither is paired.
- 10% condition: each sample is less than 10% of its own population when sampling without replacement.
- Normal or large: each population is approximately normal, each , or each sample has no strong skewness or outliers.
The shape check applies to both groups separately. One large sample does not rescue a small, strongly skewed one.
Check your recall[13 marks]
5 written questions. Answer each one in your head or on paper, then open the mark scheme.
- 1.Calculate[3 marks]
Calculate the mean and standard deviation of when , , , , and .
Show mark schemeHide mark scheme
Model answer
The mean is . The standard deviation is .
- 1 mark for finding the mean as
- 1 mark for substituting correctly into the standard deviation formula
- 1 mark for obtaining
- 2.Describe[3 marks]
Describe the sampling distribution of when the conditions are met.
Show mark schemeHide mark scheme
Model answer
It is approximately normal in shape. Its center is . Its standard deviation is .
- 1 mark for stating approximately normal shape
- 1 mark for stating the center as
- 1 mark for stating the spread formula
- 3.Describe[3 marks]
Describe the conditions that support using a normal model for .
Show mark schemeHide mark scheme
Model answer
Each sample must be random or randomly assigned, and the two samples must be independent. Each sample must be less than 10% of its population when sampling without replacement. Each population must be approximately normal, or each sample size at least 30, or each sample free of strong skewness and outliers.
- 1 mark for stating random selection or assignment and independent groups
- 1 mark for stating the 10% condition for each sample
- 1 mark for stating the normal or large sample condition for each group
- 4.Explain[2 marks]
Explain why the standard deviation of is not found by subtracting the two standard deviations of the sample means.
Show mark schemeHide mark scheme
Model answer
Each sample mean contributes its own randomness. For independent samples the variances add, so the combined spread is larger than either one alone.
- 1 mark for stating that variances of independent sample means add
- 1 mark for linking this to a combined spread at least as large as either one alone
- 5.Justify[2 marks]
Group 1 has values with no strong skewness. Group 2 has values that are strongly skewed with an outlier. Justify whether a normal model is appropriate for .
Show mark schemeHide mark scheme
Model answer
A normal model is not appropriate. The condition must hold for each group, and group 2 is small and shows strong skewness and an outlier.
- 1 mark for concluding that the normal model is not appropriate
- 1 mark for citing group 2 being small, skewed and having an outlier, since the check applies to each group

Practice this topic
10 quick-fire multiple-choice questions, easy to hard. Tap an answer to see if you're right and the worked explanation. (The written questions with mark schemes are under “Check your recall” above.)
0 right0 of 10 answered
Includes the 2 quick checks in the note above.
EasyIdentify[1 mark]
Show answer and explanation
Answer: A · Worked explanation
MediumCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
MediumPredict[1 mark]
Show answer and explanation
Answer: A · Worked explanation
MediumDetermine[1 mark]
Show answer and explanation
Answer: A · Worked explanation
MediumCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardCalculate[2 marks]
Show answer and explanation
Answer: A · Worked explanation
HardIdentify[2 marks]
Show answer and explanation
Answer: A · Worked explanation