Digital SAT Math · Problem-Solving and Data Analysis

SAT Sample Statistics and Margin of Error Practice Questions

These questions ask what a random sample lets you conclude about a population, and how to interpret an estimate with a margin of error.

College Board skill name: Inference from sample statistics and margin of error

Questions in the bank
200
Easy / Medium / Hard
60 / 90 / 50
Subtopics
10
Typical time
~1.2 min each

Updated . Questions are original BlitzSAT items written to the College Board's published skill specifications. They are not official SAT questions.

How Sample Statistics and Margin of Error questions appear on the SAT

Short descriptions of surveys or samples, followed by 'Which conclusion is best supported?' or 'Which is a plausible value of the population mean?'

Sample Statistics and Margin of Error is part of the Problem-Solving and Data Analysis domain, which makes up about 15% of the Math section (5–7 questions per test). A calculator is allowed on every Math question, and the built-in Desmos graphing calculator is available in Bluebook.

How to approach Sample Statistics and Margin of Error questions

  1. 1A random sample supports conclusions only about the population it was drawn from.
  2. 2An estimate of 42% with a margin of error of 3% means plausible population values run from 39% to 45%.
  3. 3Larger samples give smaller margins of error, all else equal.
  4. 4To estimate a population count, multiply the sample proportion by the population size.

Common mistakes

  • Extending a conclusion beyond the sampled population, for example from one school to all students.
  • Treating the margin-of-error interval as certain rather than plausible.
  • Assuming that a bigger sample changes the estimate itself instead of its precision.

10 SAT Sample Statistics and Margin of Error practice questions with answers

Pick an answer to check it. Each explanation covers the correct answer and why each of the other choices is wrong. The set runs from easy to hard.

0 / 10 answered
1EasyDistinguish sample statistics from population parameters
In a hypothetical transport-planning survey, planners randomly select 180180 people from all weekday bus riders in a town. Of those selected, 6363 report transferring between buses during their commute. Which quantity is a sample statistic?
Show answer and explanation

Answer: B

A sample statistic describes the selected sample rather than the entire population. The proportion reporting transfers among the selected riders is 63/180=0.3563/180=0.35, so it is a sample statistic.Why the other choices are wrongAThis proportion describes all weekday bus riders in the town, so it is a population parameter.CThis count describes the size of the entire population, not a characteristic calculated from the selected sample.DA mean calculated for all weekday bus riders is a population parameter, not a sample statistic.
2EasyCompare precision of two stated intervals
Two random samples from the same finite collection of numbers produce the following intervals for the collection's mean. Both intervals use the same confidence level.
SampleInterval for the mean
X41 to 4941\text{ to }49
Y43 to 4743\text{ to }47
Which statement correctly compares the precision of the two estimates?
Show answer and explanation

Answer: B

X's interval has width 49−41=849-41=8, while Y's has width 47−43=447-43=4. At the same confidence level, the narrower interval indicates greater precision, so Y's estimate is more precise.Why the other choices are wrongAExtending farther on either side of the estimate means greater uncertainty, not greater precision.CBoth midpoints are 45, but precision depends on the interval's width, not its midpoint.DSample sizes can affect precision, but the reported intervals already provide the information needed to compare it.
3EasyRelate a larger random sample to smaller margin of error
In a hypothetical design project, a team randomly samples 100 tiles from a very large shipment to estimate the percentage whose angles meet a design requirement. Which change would generally reduce the margin of error, keeping the confidence level and sampling method the same?
Show answer and explanation

Answer: C

With the confidence level and sampling method unchanged, a larger random sample generally produces a smaller margin of error. Increasing the sample from 100 tiles to 400 tiles is the only option that increases the amount of sample information.Why the other choices are wrongAReducing the sample size generally increases the margin of error.BA different sample of the same size does not systematically reduce the margin of error.DRecalculating from the same observations adds no sample information and does not reduce the margin of error.
4MediumEvaluate whether a value is plausible within a given interval
In a hypothetical design project, every rectangular plate in a shipment has a length twice its width. A random sample gives an estimated population mean perimeter of 48 centimeters with a margin of error of 3 centimeters. A proposed population mean width is considered plausible if its corresponding population mean perimeter lies within the reported interval. Which proposed population mean width is plausible?
Show answer and explanation

Answer: B

For each plate with width ww, the perimeter is 2(w+2w)=6w2(w+2w)=6w, so the population mean perimeter is six times the population mean width. The perimeter interval is 48±348\pm3, or 45 to 51 centimeters. Dividing by 6 gives a plausible mean-width interval of 7.5 to 8.5 centimeters. Only 8.25 centimeters lies within this interval.Why the other choices are wrongAA mean width of 7.25 centimeters corresponds to a mean perimeter of 6(7.25)=43.56(7.25)=43.5 centimeters, below the interval's lower endpoint of 45 centimeters.CDividing the estimate by 6 but leaving the margin of error unchanged would incorrectly give a width interval of 5 to 11 centimeters. The margin must also be divided by 6; a mean width of 10 centimeters would give a mean perimeter of 60 centimeters.DDividing 48 by 4 treats the plates as squares. Their lengths are twice their widths, so the correct perimeter-to-width factor is 6. A mean width of 12 centimeters would give a mean perimeter of 72 centimeters.
5MediumIdentify the population supported by a random sample
In a hypothetical recreation survey, a center randomly selects 120 people from its complete list of current annual-pass holders. Every selected person responds to a question about whether the center should add an evening exercise class. Some respondents have never attended a class at the center. To which population can the survey results most appropriately be generalized?
Show answer and explanation

Answer: A

The random sample was drawn from the complete list of current annual-pass holders, and everyone selected responded. Therefore, the results can most appropriately be generalized to all current annual-pass holders, regardless of whether they have attended a class.Why the other choices are wrongBClass attendance did not define the sampling population. The sample includes pass holders who have never attended, and it does not necessarily represent class attendees without annual passes.CCommunity residents without annual passes had no chance of being selected, so the sample does not support generalizing to all residents.DRandom selection from the complete pass-holder list supports generalization beyond the respondents to the population from which they were selected.
6MediumConvert a percentage interval to estimated counts
In a hypothetical materials experiment, researchers test a random sample of 200200 seals from a shipment of 2,5002{,}500 seals. They estimate that 28%28\% of the shipment's seals will remain flexible after a heat treatment, with a margin of error of 44 percentage points. Which interval gives the estimated number of seals in the entire shipment that will remain flexible, using the endpoints of this percentage interval?
Show answer and explanation

Answer: A

The percentage interval is 28%−4%=24%28\%-4\%=24\% to 28%+4%=32%28\%+4\%=32\%. Applying these endpoints to the shipment gives 0.24(2,500)=6000.24(2{,}500)=600 and 0.32(2,500)=8000.32(2{,}500)=800. The corresponding estimated count interval is 600600 to 800800 seals.Why the other choices are wrongBThese endpoints come from taking 4%4\% of the estimated count of 700700. The stated margin is 44 percentage points of the shipment, which corresponds to 100100 seals, not 2828 seals.CThese endpoints apply 24%24\% and 32%32\% to the sample size of 200200, rather than to the shipment size of 2,5002{,}500.DThe lower endpoint uses the point estimate, 28%28\% of 2,5002{,}500, instead of subtracting the margin of error first.
7MediumEstimate a population count from a sample proportion
In a hypothetical recreation-center survey, every member belongs to exactly one of two membership groups. A separate random sample is selected from each group, as shown in the table.
Membership groupTotal membersMembers sampledSampled members favoring later hours
Morning3,0008020
Evening1,8004030
Using each sample proportion to estimate support within its own membership group, what is the estimated total number of members who favor later hours?
Show answer and explanation

Answer: D

For morning members, the estimated count is 2080(3,000)=750\frac{20}{80}(3{,}000)=750. For evening members, it is 3040(1,800)=1,350\frac{30}{40}(1{,}800)=1{,}350. Adding these estimates gives 750+1,350=2,100750+1{,}350=2{,}100 members.Why the other choices are wrongAPooling the samples gives 50120(4,800)=2,000\frac{50}{120}(4{,}800)=2{,}000, but the groups were sampled at different rates. Each sample proportion must be applied to its own group.BThis averages the two sample proportions without weighting by group size: 0.25+0.752(4,800)=2,400\frac{0.25+0.75}{2}(4{,}800)=2{,}400. The membership groups have different population sizes.CThis applies the evening group's proportion to all members: 3040(4,800)=3,600\frac{30}{40}(4{,}800)=3{,}600. It ignores the morning group's lower sample proportion.
8HardInterpret a stated margin-of-error interval
In a hypothetical school project, students estimate support for a schedule change at a school with 1,2001{,}200 students. All 300300 seniors respond, and exactly 180180 support the change. A random sample of the other 900900 students produces an estimated support percentage of 52%52\%, with a margin of error of 66 percentage points. The project team converts both endpoints of this interval to counts among the 900900 nonseniors, adds the known senior support count, and divides by 1,2001{,}200. Which conclusion is supported by the resulting schoolwide interval?
Show answer and explanation

Answer: A

The nonsenior interval is 46%46\% to 58%58\%, corresponding to 0.46(900)=4140.46(900)=414 through 0.58(900)=5220.58(900)=522 supporters. Adding the 180180 senior supporters gives 594594 to 702702 supporters schoolwide. Dividing by 1,2001{,}200 gives 49.5%49.5\% to 58.5%58.5\%. This interval includes percentages above 50%50\%, but it also includes percentages that are not a majority.Why the other choices are wrongBThe schoolwide lower endpoint is 594/1,200=49.5%594/1{,}200=49.5\%, so not every value represents support from more than half of all students.CThe schoolwide upper endpoint is 702/1,200=58.5%702/1{,}200=58.5\%. The interval therefore includes values representing majority support.DAlthough 180/300=60%180/300=60\% of seniors support the change, the schoolwide upper endpoint is only 58.5%58.5\%. The seniors' percentage is not the schoolwide percentage.
9HardCompare estimates while respecting overlapping intervals
In a hypothetical environmental survey, a reserve consists of a northern zone and a southern zone. The northern zone has twice the area of the southern zone. Samples provide the following estimates of the zones' area-weighted mean surface temperatures at a particular time.
ZoneEstimated mean (degrees Celsius)Margin of error (degrees Celsius)
Northern426
Southern468
For this question, assume each zone's true mean lies within its estimate plus or minus its margin of error, including the endpoints. A separate measurement establishes that the true area-weighted mean for the entire reserve at that time is exactly 44 degrees Celsius.Let dd be the southern zone's true mean minus the northern zone's true mean, in degrees Celsius. Which inequality describes the complete set of possible values of dd consistent with all this information?
Show answer and explanation

Answer: A

Let NN and SS be the northern and southern true means. The reported intervals give 36≤N≤4836\leq N\leq48 and 38≤S≤5438\leq S\leq54. Because the northern zone has twice the area, (2N+S)/3=44(2N+S)/3=44, so S=132−2NS=132-2N. Substituting into the southern interval gives 38≤132−2N≤5438\leq132-2N\leq54, or 39≤N≤4739\leq N\leq47. This entire interval also satisfies the northern bounds. Therefore, d=S−N=132−3Nd=S-N=132-3N ranges from −9-9 when N=47N=47 to 1515 when N=39N=39, with every intermediate value possible. Thus A is correct. Since the range includes both negative and positive values, either zone could have the higher true mean.Why the other choices are wrongBSubtracting the intervals without using the whole-reserve mean gives a minimum of 38−48=−1038-48=-10 and a maximum of 54−36=1854-36=18. However, those endpoint pairs do not satisfy 2N+S=1322N+S=132, so this range includes impossible differences.CSubtracting corresponding lower endpoints gives 38−36=238-36=2, and subtracting corresponding upper endpoints gives 54−48=654-48=6. The true means are not required to occupy corresponding positions in their intervals, and neither of these endpoint pairs satisfies the whole-reserve mean constraint.DThis is the range for N−SN-S, the northern mean minus the southern mean. The question defines the difference in the opposite order, so the correct endpoints are −9-9 and 1515.
10HardInterpret uncertainty without claiming certainty
In a hypothetical fabrication project, a geometric wall design requires 4,000 triangular panels. The project meets its quality requirement if no more than 240 panels fail a dimensional check. An inspection of a random sample produces the following estimate for the entire set of panels.
QuantityValue
Estimated percentage passing96%
Margin of error3 percentage points
Which interpretation of the estimate and margin of error is most appropriate?
Show answer and explanation

Answer: B

The reported passing-percentage interval is 93% to 99%, so the corresponding failing-percentage interval is 1% to 7%. For 4,000 panels, this gives 0.01(4,000)=400.01(4{,}000)=40 to 0.07(4,000)=2800.07(4{,}000)=280 failing panels. This range includes counts both at or below 240 and above 240. Thus, the estimate alone does not establish that the requirement is met.Why the other choices are wrongAAlthough the point estimate is 0.04(4,000)=1600.04(4{,}000)=160 failing panels, treating that estimate as conclusive ignores the stated uncertainty.CA count of 280 is an endpoint of the estimated range, not an established actual count. The range also includes counts that meet the requirement.DThis incorrectly takes 3% of the point estimate of 160. The margin is 3 percentage points of all 4,000 panels, corresponding to 120 panels.

These are 10 of 200 questions on this skill in the BlitzSAT question bank.

Practice the rest by difficulty, timed or untimed, with your history saved. Free, no credit card.

Practice all 200 questions

What the 200 Sample Statistics and Margin of Error questions cover

The BlitzSAT bank splits this skill into 10 subtopics. Counts are from the current bank.

SubtopicQuestionsEasyMediumHard
Compare estimates while respecting overlapping intervals20695
Compare precision of two stated intervals20695
Convert a percentage interval to estimated counts20695
Distinguish sample statistics from population parameters20695
Estimate a population count from a sample proportion20695
Evaluate whether a value is plausible within a given interval20695
Identify the population supported by a random sample20695
Interpret a stated margin-of-error interval20695
Interpret uncertainty without claiming certainty20695
Relate a larger random sample to smaller margin of error20695

Other Problem-Solving and Data Analysis skills

More Math skills

Keep going