Home›Calculators›Statistics›Margin of Error Calculator

Margin of Error Calculator

Statistics

Calculate the margin of error for a survey or sample estimate in seconds. Enter confidence level, standard deviation, and sample size to get the exact margin instantly.

01,000,000
11,000,000

Margin of Error

2.94
Z-Score Used
1.96
Lower Adjustment
-2.94
Upper Adjustment
2.94

This calculator computes your Margin of Error, Z-Score Used, Lower Adjustment, Upper Adjustment from the values you enter.

Inputs
Confidence LevelStandard DeviationSample Size (n)
Outputs
Margin of ErrorZ-Score UsedLower AdjustmentUpper Adjustment

What is a Margin of Error?

The Margin of Error Calculator computes the ± precision range around a sample estimate, based on your chosen confidence level, standard deviation, and sample size. It answers the question: "given my sample, how far off could my estimate realistically be from the true population value?"

Margin of error is the single most-cited (and most-misunderstood) statistic in survey research and polling. This calculator computes it directly from three inputs, showing the exact z-score used and the resulting standard error, so you can verify and explain the number with confidence.

Once you have your margin of error, pair it with your sample mean using the Confidence Interval Calculator to get the full lower and upper bounds of your estimate.

How to use this Margin of Error calculator

  1. Select your confidence level — 90%, 95% (most common in research), or 99%.

  2. Enter your standard deviation — from your sample data, a prior study, or a reasonable estimate of variability.

  3. Enter your sample size — the number of observations you collected or plan to collect.

  4. Read the margin of error — this is the ± precision of your estimate at the chosen confidence level.

  5. Build a full confidence interval — take your sample mean and this margin of error to the Confidence Interval Calculator for the complete lower and upper bounds.

  6. Plan ahead if the margin is too wide — if the resulting margin of error is larger than you need, use the Sample Size Calculator to find out how many more observations would tighten it.

Formula & Methodology

Margin of error:
E = z × (σ / √n)

Variable definitions:
- z — z-score for the chosen confidence level (1.645 for 90%, 1.96 for 95%, 2.576 for 99%)
- σ — standard deviation
- n — sample size

Worked example:

A satisfaction survey of 150 customers has a standard deviation of 12 (on a 100-point scale), at 95% confidence.

Step 1 — Standard error: 12 / √150 = 12 / 12.247 ≈ 0.9798

Step 2 — Margin of error: 1.96 × 0.9798 ≈ 1.92

If the sample mean satisfaction score was 78, the result would be reported as 78 ± 1.92, or a confidence interval of approximately [76.08, 79.92].

Assumption: This formula assumes the sampling distribution is approximately normal, which holds for sample sizes of roughly 30 or more, or when the underlying population itself is close to normally distributed.

Frequently Asked Questions

Margin of error is the ± amount of uncertainty around a sample-based estimate, reflecting how much the true population value might differ from your sample mean due to random sampling variation. A poll reporting 52% support with a ±3% margin of error means the true support level is likely between 49% and 55%. Margin of error matters because a single point estimate without it can mislead readers into thinking the number is more precise than it actually is.
Margin of error = z × (standard deviation / √sample size). First find the z-score for your confidence level (1.96 for 95%), then compute the standard error by dividing your standard deviation by the square root of your sample size, then multiply the two together. For example, with σ = 15, n = 100, and 95% confidence: standard error = 15/10 = 1.5, margin of error = 1.96 × 1.5 = 2.94.
Margin of error is proportional to 1/√n, so it shrinks as sample size increases, but at a diminishing rate. Quadrupling your sample size (e.g., from 100 to 400) halves the margin of error, because √4 = 2. This diminishing-returns relationship is why researchers eventually hit a point where collecting more data isn't cost-effective relative to the small further reduction in margin of error.
±5% is a commonly accepted standard for general business and market research surveys. ±3% or tighter is used for higher-stakes contexts like national political polling. ±10% is sometimes tolerated for quick, low-budget exploratory research. The 'right' margin of error depends entirely on how much precision your decision requires — a go/no-go product launch decision may tolerate more uncertainty than a regulatory submission.
Standard deviation (σ) measures the spread of individual data points around the mean. Standard error (σ/√n) measures the expected spread of sample means around the true population mean if you repeated the sampling process many times — it's always smaller than the standard deviation for n > 1. Margin of error is the standard error multiplied by a z-score for your chosen confidence level, giving the final ± range added to your point estimate.
Yes — margin of error scales directly with the z-score of your chosen confidence level. Moving from 90% confidence (z = 1.645) to 99% confidence (z = 2.576) increases the margin of error by about 56%, because you need a wider range to be more certain the true value falls inside it, holding sample size and standard deviation fixed.
The confidence interval is the sample mean plus and minus the margin of error: [mean − E, mean + E]. Margin of error is just the ± half-width of that interval. If you want the full interval bounds rather than just the ± value, use the [Confidence Interval Calculator](/confidence-interval-calculator/), which performs this exact calculation and returns both bounds directly.
Pollsters report a candidate's support percentage alongside a margin of error (commonly ±3% for national polls) to signal how much the reported number could differ from the true population sentiment due to sampling variability alone. If two candidates are shown at 51% and 49% with a ±3% margin of error, the race is statistically a toss-up because the true values could plausibly overlap — this is why headlines like 'statistical tie' appear even when reported numbers differ.
Margin of error can be reduced slightly by accepting a lower confidence level (e.g., reporting at 90% instead of 99% confidence), which reduces the z-score multiplier. It can also be reduced by improving measurement precision to lower the underlying standard deviation (e.g., better survey instrument design), but the most reliable way to meaningfully shrink margin of error is to increase sample size, which you can plan for using the [Sample Size Calculator](/sample-size-calculator/).
Higher standard deviation directly increases margin of error, since it appears in the numerator of the formula. A dataset with wildly varying responses (like income, which spans a huge range) will produce a larger margin of error than a dataset with tightly clustered responses (like satisfaction ratings on a 1-5 scale), even at the same sample size and confidence level.
No — margin of error depends on the variability (standard deviation) of the specific variable being measured, which can differ from question to question within the same survey. A survey's commonly reported 'overall margin of error' typically applies to the main outcome variable (like a yes/no proportion question) and assumes maximum variability; other questions with different response distributions may have a smaller or larger true margin of error.
Also known as
margin of errorMOE calculatorsurvey error marginsampling error calculatorpoll margin calculator