Scribe logoScribeNotes & practice
Sign inStart free
  1. Notes
  2. ›AP Statistics
  3. ›3 Inference for Categorical Data: Proportions
  4. ›3.1 Estimators

AP Statistics · Unit 3 · Inference for Categorical Data: Proportions

3.1 Estimators

Distinguish biased from unbiased estimators, and explain why low bias and low variability make a statistic a better estimator.

  • 5 min read
  • 5 parts
  • 10 questions
  • Draft
Start a Quick 5 on this topic→Jump to the 10 questions
1/5A sample statistic gives a point estimate of a parameter
  1. 1A sample statistic gives a point estimate of a parameter
  2. 2An unbiased estimator centers on the parameter it estimates
  3. 3A better estimator has low bias and low variability
  4. 4Several familiar statistics are unbiased, but some are not
  5. 5Bad sampling methods create bias that a larger sample cannot fix
  6. Check your recall
  7. Practice: 10 questions
1

Part 1 of 5: A sample statistic gives a point estimate of a parameter

A city planner wants the mean commute time of all 80,000 workers. She cannot ask everyone, so she surveys 300 random workers.

The mean commute of those 300 workers is a statistic. The mean for all workers is a parameter. The planner uses the statistic as a point estimate, a single value serving as her best guess for the parameter.

A second random survey would give a different estimate. So a point estimate rarely equals the parameter exactly. The useful question is whether the method is reliable over many samples. Statisticians answer it by studying the pattern of all the estimates a method could produce.

Key terms · tap one for its definition

parameter

A parameter is a fixed number that describes a whole population, such as the population mean μ\muμ or proportion ppp.

statistic

A statistic is a number calculated from sample data, such as the sample mean xˉ\bar{x}xˉ or sample proportion p^\hat{p}p^​.

point estimate

A point estimate is a single value of a statistic that is used to estimate a population parameter.

Part 1 done

2

Part 2 of 5: An unbiased estimator centers on the parameter it estimates

To judge a method, picture every possible sample of one size. Calculate the statistic for each sample. The results form the sampling distribution of the statistic.

If this distribution is centered at the parameter, the statistic is an unbiased estimator. Single estimates still miss the target, but they do not miss mostly in one direction. The same idea applies to any statistic, not only the mean. If the center sits away from the parameter, the statistic is a biased estimator.

Figure 1. Each curve is a sampling distribution. Bias shifts the center away from the true value, and variability sets the width.

Common mistake

Unbiased does not mean any single estimate is correct. It describes the center of all possible estimates, not one result.

Key terms · tap one for its definition

sampling distribution

The sampling distribution of a statistic is the distribution of its values over all possible samples of the same size from the population.

unbiased estimator

An unbiased estimator is a statistic whose sampling distribution has a mean equal to the parameter it estimates.

biased estimator

A biased estimator is a statistic whose sampling distribution has a mean that differs from the parameter it estimates.

Part 2 done

Quick check

One question on the part you just read

Easy

Identify the statement that describes an unbiased estimator.
Show answer and explanation

Answer: B · Worked explanation

An unbiased estimator is centered on the parameter: the mean of its sampling distribution equals the parameter. It does not hit the parameter in every sample, because sampling variability still moves single estimates. Small spread describes low variability, which is a separate feature, and sample size alone does not decide bias.
3

Part 3 of 5: A better estimator has low bias and low variability

Being unbiased is not enough on its own. Compare two unbiased estimators whose sampling distributions have standard deviations of 2 and 5. The first produces estimates that stay closer to the parameter.

A good estimator therefore has two features. Its sampling distribution is centered at the parameter, and it has small spread. That spread is the variability of the estimator.

A larger random sample usually lowers variability. For example, quadrupling the sample size halves the standard deviation of a sample mean.

Worked example

Choosing the best estimator

A parameter equals 50. Simulations of three statistics, A, B and C, use many random samples. Their sampling distributions have these means and standard deviations: A has mean 50.1 and SD 2.0, B has mean 49.9 and SD 5.0, C has mean 58.0 and SD 1.0. Decide which is the best estimator.
  1. 1
    Compare each center with 50. A and B are centered very close to 50, so each is roughly unbiased.
  2. 2
    C is centered at 58.0, which is 8.0 above the parameter, so C is biased.
  3. 3
    Compare the variability of A and B: the standard deviation 2.0 is smaller than 5.0.
  4. 4
    C has the smallest spread, but its estimates are consistently too high.

Answer: Statistic A is best: it is nearly unbiased and has the lowest variability among the unbiased options.

Low bias and low variability together mean estimates land close to the parameter most of the time.

Key term

variability

The variability of an estimator is the spread of its sampling distribution, measured by its standard deviation.

Part 3 done

4

Part 4 of 5: Several familiar statistics are unbiased, but some are not

Some statistics have known bias behavior when samples are random. The table lists common ones.

Bias of common statistics under random sampling

StatisticEstimatesBias
Sample mean xˉ\bar{x}xˉPopulation mean μ\muμUnbiased
Sample proportion p^\hat{p}p^​Population proportion pppUnbiased
Sample variance s2s^2s2 (divide by n−1n-1n−1)Population varianceUnbiased
Variance found by dividing by nnnPopulation varianceBiased, too small
Sample maximumPopulation maximumBiased, too small
Statistic
Sample mean xˉ\bar{x}xˉ
Estimates
Population mean μ\muμ
Bias
Unbiased
Statistic
Sample proportion p^\hat{p}p^​
Estimates
Population proportion ppp
Bias
Unbiased
Statistic
Sample variance s2s^2s2 (divide by n−1n-1n−1)
Estimates
Population variance
Bias
Unbiased
Statistic
Variance found by dividing by nnn
Estimates
Population variance
Bias
Biased, too small
Statistic
Sample maximum
Estimates
Population maximum
Bias
Biased, too small

The sample maximum cannot exceed the population maximum, and it usually falls below it. Averages of sample maximums therefore sit too low. These claims assume the samples are random.

The formula for s2s^2s2 divides by n−1n-1n−1 for a similar reason. Dividing by nnn gives values that are too small on average.

Exam tip

To identify bias, compare the center of the sampling distribution with the parameter. State whether estimates tend to run too high or too low.

Part 4 done

Quick check

One question on the part you just read

Easy

Identify the point estimate.
Show answer and explanation

Answer: B · Worked explanation

A point estimate is a single value of a statistic used to estimate a parameter, so the sample proportion 0.46 qualifies. The unknown population proportion is the parameter itself. The distribution of all possible sample proportions is a sampling distribution, and its spread describes variability.
5

Part 5 of 5: Bad sampling methods create bias that a larger sample cannot fix

Bias can also come from how the sample is chosen. A method that systematically favors some members of the population produces estimates that miss in one direction.

Imagine surveying only people at a gym about weekly exercise. Their average is too high for the whole city, however many gym members you survey.

Random selection gives every member a fair chance, so estimates are not pushed in one direction. A well-designed random sample is the first defense against bias.

Common mistake

A larger sample reduces variability but does not remove bias. A biased method stays biased at any sample size.

Note read

Check your recall[12 marks]

6 written questions. Answer each one in your head or on paper, then open the mark scheme.

  1. 1.Describe[2 marks]

    Describe an unbiased estimator.

    Show mark schemeHide mark scheme

    Model answer

    It is a statistic whose sampling distribution has a mean equal to the parameter. On average, its estimates are neither too high nor too low.

    • 1 mark for stating that the mean of its sampling distribution equals the parameter
    • 1 mark for stating that estimates are not systematically too high or too low
  2. 2.Explain[2 marks]

    Explain why an unbiased estimator with large variability can still be a poor estimator.

    Show mark schemeHide mark scheme

    Model answer

    A single estimate can land far from the parameter when the spread is large, even though the center is correct. Low variability keeps estimates close to the parameter.

    • 1 mark for stating that a large spread means single estimates can be far from the parameter
    • 1 mark for linking low variability to estimates staying close to the parameter
  3. 3.Identify[1 mark]

    Identify whether the sample maximum, used to estimate a population maximum, is a biased or an unbiased estimator.

    Show mark schemeHide mark scheme

    Model answer

    It is biased, because it usually falls below the population maximum.

    • 1 mark for identifying the sample maximum as biased (too small on average)
  4. 4.Justify[3 marks]

    A parameter equals 20. Statistic X has a sampling distribution with mean 20.2 and standard deviation 1.1. Statistic Y has mean 24.0 and standard deviation 0.6. Justify which is the better estimator.

    Show mark schemeHide mark scheme

    Model answer

    Statistic X is better. Its mean of 20.2 is very close to 20, so it is nearly unbiased. Y is centered 4.0 above the parameter, so its estimates are consistently too high despite lower variability.

    • 1 mark for choosing statistic X
    • 1 mark for stating that X is centered near the parameter (nearly unbiased)
    • 1 mark for stating that Y is biased even though its variability is lower
  5. 5.Predict[2 marks]

    A survey uses a method that favors some members of the population. Predict the effect on bias and variability of increasing the sample size.

    Show mark schemeHide mark scheme

    Model answer

    The variability will fall, but the bias will remain. A larger sample does not remove a systematic error in how the sample is chosen.

    • 1 mark for predicting that variability decreases
    • 1 mark for predicting that bias remains
  6. 6.Determine[2 marks]

    A parameter equals 25.0. A statistic has a sampling distribution with mean 24.0 and standard deviation 3.0. Determine whether the statistic is an unbiased estimator.

    Show mark schemeHide mark scheme

    Model answer

    It is biased. The mean of its sampling distribution, 24.0, differs from the parameter, 25.0.

    • 1 mark for concluding that the statistic is biased
    • 1 mark for comparing the mean 24.0 of the sampling distribution with the parameter 25.0

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.

3.

Easy·Identify[1 mark]

Identify the estimator that is unbiased under random sampling.
Show answer and explanation

Answer: C · Worked explanation

The sample proportion p^\hat{p}p^​ is unbiased for ppp when the sample is random. The sample maximum and the variance divided by nnn both tend to be too small. A sample of volunteers uses a method that favors some members, so its estimates are pushed in one direction.
4.

Medium·Determine[1 mark]

A parameter equals 30.0. Over many random samples, a statistic has a sampling distribution with mean 33.0 and standard deviation 1.5. Determine whether it is an unbiased estimator.
Show answer and explanation

Answer: C · Worked explanation

Bias depends on the center of the sampling distribution, and 33.0 is 3.0 above the parameter 30.0, so the statistic is biased. Random samples and a small spread do not guarantee unbiasedness. A standard deviation of 0 is not needed for an unbiased statistic, since unbiased estimators still vary.
5.

Medium·Determine[1 mark]

Two unbiased estimators of the same parameter have sampling distributions with standard deviations 0.8 and 2.4. Determine which is better and why.
Show answer and explanation

Answer: D · Worked explanation

Both estimators are centered on the parameter, so they have the same bias, which is none. The one with standard deviation 0.8 has lower variability, so its estimates typically fall closer to the parameter. The claim of smaller bias is wrong because the bias is equal. A larger spread makes estimates less reliable.
6.

Medium·Predict[1 mark]

A website survey uses only volunteers, and its estimates run too high. Predict the effect of increasing the volunteers from 2,000 to 20,000.
Show answer and explanation

Answer: A · Worked explanation

A larger sample lowers variability, so the estimates cluster more tightly. The method still favors volunteers, so the clustering happens around a value that is too high. Bias comes from the sampling method, not from the sample size, so it does not shrink with more volunteers.
7.

Medium·Calculate[1 mark]

The standard deviation of a sample mean is 8 for random samples of size 36. Calculate its standard deviation for samples of size 144 from the same population.
Show answer and explanation

Answer: B · Worked explanation

The sample size is multiplied by 4, so the standard deviation is divided by 4=2\sqrt{4}=24​=2, giving 82=4\frac{8}{2}=428​=4. The value 2 would divide by 4 and forget the square root. The value 16 doubles the spread, which goes the wrong way, and 8 ignores the larger sample.
8.

Hard·Determine[2 marks]

A parameter equals 40.0. Statistic A has a sampling distribution with mean 40.0 and standard deviation 1.0. Statistic B has mean 43.0 and standard deviation 0.9. A student says B is better because its standard deviation is smaller. Determine the best response.
Show answer and explanation

Answer: C · Worked explanation

Spread is only one feature. A is centered on the parameter and has similar variability, while B is centered 3.0 too high, so its estimates are consistently wrong in one direction. The claim that A has the larger bias is false: A has none. Low variability cannot rescue a biased estimator.
9.

Hard·Identify[2 marks]

Identify the correct statement about estimators.
Show answer and explanation

Answer: D · Worked explanation

Unbiased describes the center of all possible estimates, so one sample can still give a value far away. Low variability can occur around the wrong center, so it does not guarantee unbiasedness. A biased estimator may run too low, as the sample maximum does. No estimator stays within one standard deviation every time.
10.

Hard·Determine[2 marks]

A researcher takes random samples of 20 packages and records the heaviest in each sample, hoping to estimate the heaviest package in a large shipment, which weighs 15.0 kg. Across many samples, the maximums average 13.4 kg. Determine what this shows.
Show answer and explanation

Answer: B · Worked explanation

The mean of the sampling distribution is 13.4 kg, which is 1.6 kg below the parameter 15.0 kg, so the statistic is biased low. This matches the idea that the sample maximum cannot exceed the population maximum. Random sampling alone does not make a statistic unbiased, and being close is not the same as being equal.

Keep going

Next topic3.2 Sample proportionsPrevious topic2.12 Sampling distributions and the central limit theorem

More in Unit 3 · Inference for Categorical Data: Proportions

  1. 3.1EstimatorsNotes · 10 questions
  2. 3.2Sample proportionsNotes · 9 questions
  3. 3.3Confidence interval for a proportionNotes · 10 questions
  4. 3.4Claims from a proportion intervalNotes · 10 questions
  5. 3.5Setting up a test for a proportionNotes · 10 questions
  6. 3.6p-valuesNotes · 10 questions
  7. 3.7Carrying out a test for a proportionNotes · 10 questions
  8. 3.8Errors in testsNotes · 10 questions
  9. 3.9Differences in sample proportionsNotes · 10 questions

All AP Statistics topics →

Draft: written by Scribe and checked automatically, not yet by a subject reviewer.

Practice questions are original, written by Scribe and checked automatically for correctness.

Topic codes follow the official numbering. Not endorsed by College Board or any exam board.

On this page

0/5 read

  1. 1A sample statistic gives a point estimate of a parameter
  2. 2An unbiased estimator centers on the parameter it estimates
  3. 3A better estimator has low bias and low variability
  4. 4Several familiar statistics are unbiased, but some are not
  5. 5Bad sampling methods create bias that a larger sample cannot fix
  6. Check your recall
  7. Practice: 10 questions
All coursesScribeHow it worksPricingPrivacyTerms

Notes and questions are written by Scribe in our own words. Topic codes follow the official course numbering so you can find your place. Not produced or endorsed by the IB, College Board, AQA, OCR, Pearson Edexcel or any exam board.

© 2026 Scribe