On this concept: Formula

Not sure which lesson to do first? Find your level and get a plan →

Math · Statistics & Probability · Grade 9-12

Margin of Error

The margin of error is the maximum expected gap between a sample statistic and the true parameter; it's exactly half the width of a confidence interval, the '±' part.

E=z∗⋅sn

Quick Answer

The margin of error is the maximum expected gap between a sample statistic and the true parameter; it's exactly half the width of a confidence interval, the '±' part. Use it to say how precise a poll or survey estimate is, as in "support is 52%, give or take 3 points," or to plan how big a sample you need. For a mean, E=z∗sn: it grows with the confidence level and with the spread of the data, and it shrinks as the sample grows, but only in proportion to n.

Why This Matters

News reports call a race "tied" or a candidate "ahead" from poll gaps, and the margin of error is what separates a real lead from sampling noise. It also tells a researcher how many people to survey: cutting the margin in half takes four times the sample. Treating it as protection against every kind of error is the classic misuse; a biased sample can be off by far more than the margin.

Intuitive Explanation

A poll reports 52% support with a margin of error of 3 points. That means the poll's method usually lands within 3 points of the truth, so plausible true values run from 49% to 55%. The 3 is half the width of the confidence interval: the interval is 52±3, which spans 6 points from end to end.

The formula E=z∗sn shows three dials. Raise the confidence level and z∗ grows (1.96 for 95%, 2.576 for 99%), so the margin widens. More variable data, a bigger s, widen it too. A bigger sample shrinks it, but by the square root: going from 400 to 1,600 people halves the margin, because 1600=2400. A hundred times as many people only cuts it by a factor of 10, which is why national polls usually stop at about 1,000 to 1,500 people.

The margin covers only the luck of random sampling. If a phone poll reaches mostly older voters, or the question is worded to push an answer, the result can be off by 10 points while the reported margin still says 3. It is also not the standard deviation of the data: individual answers can differ widely even when the margin on their average is small.

Think of it like this. The tolerance on a machined bolt, "50 mm ± 0.2 mm," works like a poll's "± 3 points": a half-width on either side of a center. The difference is that a poll's margin holds only at its confidence level, about 95% of the time, and it says nothing about bias in how the sample was collected.

Core idea. The margin of error is half the width of a confidence interval — how far the estimate might be from the truth.


When to Use

Use the margin of error when a question asks for the plus-or-minus amount on an estimate, how precise a poll is, or what sample size gives a desired precision. Compute E=z∗sn, or z∗p^(1−p^)n for a proportion. If the question wants the full range of plausible values, give the confidence interval, estimate ±E. If it asks how spread out individual responses are, that is the standard deviation.

Tip. Ask: Am I reporting how far an estimate might be from the truth as a single ± value (half the interval width)?

How to Recognize It

Look at whether the answer is one plus-or-minus number or a full range, and whether you already have the standard error or still need to compute it.

  1. Are you given an interval's two endpoints and asked how precise the estimate is?

    Yes means subtract and halve: an interval from 44% to 50% has a margin of 3 points around a center of 47%. If the question wants the endpoints themselves, give the confidence interval.

  2. Has the standard error already been multiplied by a critical value like z∗=1.96?

    Yes means you have the margin of error. A standard error alone, such as sn=2, is the typical sampling wobble before scaling; at 95% the margin is 1.96×2=3.92.

  3. Is the question asking what sample size is needed to reach a given margin?

    Yes means solve the margin formula for n: for a proportion at 95% with p^=0.5, n=(1.96E)2(0.25), so a 3-point margin needs about 1,068 people. Round up, never down.

  4. Could the error come from who was asked, rather than from random luck in the sample?

    Yes means the margin of error does not cover it: an online poll that reaches only volunteers can be off by 10 points while quoting a margin of 3. The margin measures random sampling variation only, not bias.

Margin of Error compared with Confidence interval, Standard error, and Standard deviation

All four rows measure some kind of spread. Ask what is spreading and whether it has been scaled: individual values (standard deviation), the estimate across samples (standard error), that standard error times z∗ (margin of error), or the whole range of plausible values (confidence interval).

Margin of Error

Meaning
Use this when you need the ± wiggle room on an estimate, or half the width of a confidence interval.
Key test
Am I reporting how far an estimate might be from the truth as a single ± value (half the interval width)?
Formula
E=z∗⋅sn
Example
A poll of n=400 has sample SD s=20. At 95% confidence (z∗=1.96), what is the margin of error?

Confidence interval

Meaning
The full range (both endpoints); the margin is half its width.
Key test
Use when you want the actual lower and upper values, not just the ±.
Formula
xˉ±E
Example
(49%,55%)

Standard error

Meaning
The SD of the statistic; the margin is z∗ times it.
Key test
Use when you need the raw variability of the estimate before scaling by z∗.
Formula
sn
Example
Spread of sample means

Standard deviation

Meaning
Spread of the individual data, not of the estimate.
Key test
Use when describing the data's variability, not the estimate's precision.
Formula
s
Example
How spread the scores are

Formula & Notation

E=z∗⋅sn
E=z∗⋅sn for means; E=z∗⋅p^(1−p^)n for proportions

How to read it: E is the margin of error; the confidence interval is xˉ±E.

Worked Examples

Example 1 — Compute the margin

Easy

A poll of n=400 has sample SD s=20. At 95% confidence (z∗=1.96), what is the margin of error?
Show the steps and the answer
  1. Find the standard error: sn=20400=2020=1.

    The margin describes the estimate's precision, so it scales the spread of sample means, not s itself.

  2. Multiply by the critical value: E=1.96×1=1.96.

    At 95% confidence, estimates fall within 1.96 standard errors of the truth for 95% of samples.

  3. Report it as ±1.96, in the units of the data.

    The margin is a distance on each side of the estimate, not a range by itself.

Answer: E≈±1.96

Takeaway: Margin of error = critical value × standard error.

Example 2 — Asking for the full interval

Standard

With xˉ=70 and the margin E=1.96 above, give the confidence interval.
Show the steps and the answer
  1. Notice that the question asks for the whole interval, not the margin.

    An interval needs two endpoints; the margin is the distance from the center to each one.

  2. Subtract and add the margin: 70−1.96=68.04 and 70+1.96=71.96.

    The interval is centered on the sample mean.

  3. Check the width: 71.96−68.04=3.92=2×1.96.

    The full width is twice the margin, confirming that the margin is half the interval.

Answer: (68.04, 71.96)

Takeaway: Interval = estimate ± margin, and its width is twice the margin.

Example 3 — Find the mistake: full width used as the margin

Application

A news site reports a 95% confidence interval of 41% to 49% for the share of residents who support a new park. Omar writes: "The margin of error is 49−41=8 points, so support could be as low as 45−8=37%." Find the mistake and correct it.
Show the steps and the answer
  1. Find the center of the interval: 41+492=45%.

    The interval is estimate ± margin, so the estimate sits in the middle.

  2. Find the margin as half the width: 49−412=4 points.

    The interval reaches 4 points below 45 and 4 points above it, so the width of 8 counts the margin twice.

  3. Recheck Omar's low end: 45−4=41%, not 37%.

    Using the full width doubled the uncertainty and produced values the poll does not support.

Answer: The margin of error is 4 points, and plausible support runs from 41% to 49%.

Takeaway: The margin is half the interval's width; subtract it once from the center to get the low end.

Where You'll See It

  • Reading a close election poll

    Candidate A polls 48% and Candidate B 45%, each with a margin of error of 3 points. A's plausible range is 45% to 51% and B's is 42% to 48%. The ranges overlap, so the 3-point lead is not clear evidence that A is ahead; the margin on the gap between two candidates is even larger than 3.

  • Planning a survey's sample size

    A school wants average homework time within ±10 minutes at 95% confidence, and a pilot survey shows s=50 minutes. Solving 1.96×50n≤10 gives n≥9.8, so n≥96.04: survey at least 97 students. A margin of ±5 minutes would need at least 385.

  • Why national polls stop near 1,000

    For a proportion near 50%, the 95% margin is about 1n: roughly 3 points for 1,000 people and 1 point for 10,000. Getting from 3 points down to 1 costs ten times the interviews, so most pollsters settle for about 3.

Common Mistakes

  1. The slip: Reporting the full interval width as the margin of error

    Instead: the margin is HALF the width.

  2. The slip: Using the data's SD instead of the standard error

    Instead: the margin is z∗sn, which shrinks with bigger n.

  3. The slip: Thinking margin of error covers bias or bad sampling

    Instead: it only captures random sampling variability, not a flawed survey design.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. A sample of n=81 has s=18. Find the 95% margin of error with z∗=1.96.

    Hint: Find the standard error before multiplying by z∗.

  2. A confidence interval is (46,54). What are the estimate and the margin of error?

    Hint: The margin is measured from the center to one end.

  3. A poll of 1,000 people finds 50% support. What is the 95% margin of error, in percentage points?

    Hint: Use 1.96p^(1−p^)n.

  4. With s=20 and 95% confidence, what sample size gives a margin of error of 2?

    Hint: Solve 1.96⋅20n=2 for n, then round up.

  5. A report gives a mean of 72 with a standard error of 1.5. Is the standard error the 95% margin of error?

    Hint: What multiplies the standard error to make a margin?

  6. An online poll of 5,000 self-selected visitors reports 70% with a margin of error of 1.3 points. Ben says: "The true value must be within 1.3 points of 70%." Correct him.

    Hint: What kind of error does a margin of error measure?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

If a poll shows 51% to 49% with a 3-point margin, is the race a tie?

Not exactly, but the poll cannot call it. Each candidate's figure could be off by 3 points, so 51% could really be 48% and 49% could really be 52%. Also, the margin on the gap between two candidates is roughly twice the stated margin, about 6 points here, because when one figure is too high the other tends to be too low. A 2-point lead sits well inside that.

Why do pollsters plan their sample size as if the result will be 50%?

Because p^(1−p^)\hat{p}(1-\hat{p}) is largest at p^=0.5\hat{p}=0.5, where it equals 0.25, so 50% gives the biggest possible margin for a given nn. Planning with it guarantees the margin will be no larger than promised, whatever the result turns out to be. If the true share were 10%, the product would be only 0.09, and the margin would be smaller.

Why is the margin bigger for small groups inside a poll?

Because each subgroup is a smaller sample. A poll of 1,000 adults might have a margin of about 3 points overall, but if only 150 of them are aged 18 to 24, the margin for that group is about 1.960.25150≈0.081.96\sqrt{\frac{0.25}{150}}\approx0.08, or 8 points. News stories that quote a subgroup's numbers along with the overall margin make the subgroup look far more precise than it is.

Learning Path

Margin of Error

You are here

Next →

Nothing after this yet
You need the Confidence Interval and Standard Deviation first. The margin of error isolates the ± half of the interval and shows how confidence level, variability and sample size control it, including the rule that halving the margin takes four times the data. With it you can plan survey sizes, judge whether a poll lead is real, and read the "margin of sampling error" line under any news poll.

See Also