Skip to content
Math & Scientific

Statistically Significant Sample Size Calculator

Statistically Significant Sample Size Calculator

Find how many responses you need for results you can trust.

%
%

Your Required Sample Size

Confidence level
Margin of error
Estimated proportion
Unrounded result
Calculation

How many survey responses do you actually need before you can trust the numbers? Collect too few and your results might be nothing more than noise; collect far more than necessary and you waste time and money. This statistically significant sample size calculator gives you the exact number of respondents your study needs. Choose your confidence level, set your margin of error, and enter your best guess at the answer split — the calculator does the rest and shows every step of the math.

What This Calculator Does

This calculator answers one precise question: how large must a sample be so that its results are statistically significant? In plain terms, it tells you the smallest number of people you need to survey so that your findings reflect the real population within your chosen margin of error, at your chosen confidence level. It uses the standard infinite-population formula that researchers, pollsters, and market analysts rely on every day.

You get more than a single number. The result box shows your chosen confidence level with its Z-score, your margin of error, your estimated proportion, the unrounded mathematical result, and the full calculation written out step by step. The headline banner gives you the final answer: the minimum number of respondents to aim for. Because the result is always rounded up, you can be sure the sample truly meets your requirements rather than falling just short.

This tool is designed for large or unknown populations — a city, a customer base, website visitors — where the total population is so big that it does not meaningfully change the math. If you are surveying a small, known group such as a single school or a company of 500 people, the companion survey sample size calculator with its finite population correction will give you a smaller, more accurate number.

How to Use This Calculator

Begin by selecting your confidence level from the dropdown. The 95 percent option is the default because it is the most widely used standard in research — it means that if you repeated your survey 100 times, about 95 of those surveys would capture the true population value within your margin of error. Choose 90 percent if you need a smaller, cheaper sample and can accept more uncertainty, or 99 percent when the decision riding on the results is critical and you want maximum certainty.

Next, enter your margin of error as a percentage between 1 and 20. This is the “plus or minus” range around your result. A 5 percent margin of error on a finding of 60 percent means the true value is likely between 55 and 65 percent. Smaller margins give more precise answers but demand much larger samples, so this setting is where cost and precision are traded off.

Finally, enter your estimated proportion — your best guess at what share of people will give a particular answer. If you have no idea, leave it at 50 percent. That is the safe default because it produces the largest required sample, guaranteeing your study is big enough no matter how the answers split. Press the Calculate button and your required sample size appears instantly, with the full breakdown beneath it.

The calculator validates every input. A margin of error outside 1 to 20 percent, or a proportion outside 1 to 99 percent, triggers a clear error message instead of a misleading result. The Reset button clears the form so you can test different scenarios, which is a great way to feel how each setting changes the required sample.

The Formula Behind Statistical Significance

The calculator uses the classic sample size formula: n equals Z squared times p times (1 minus p), divided by E squared. Written out, that is n = Z² × p(1−p) ÷ E². Each symbol has a concrete meaning, and understanding them helps you choose better inputs.

Z is the Z-score for your confidence level: 1.645 for 90 percent, 1.96 for 95 percent, and 2.576 for 99 percent. It comes from the standard normal distribution and measures how many standard deviations from the average you must go to capture your desired share of possible outcomes. A higher confidence level means a bigger Z, which directly inflates the required sample — certainty is expensive.

The letter p is your estimated proportion as a decimal, and p(1−p) is the variance of a yes-or-no answer. This product is largest when p equals 0.5, which is exactly why 50 percent is the conservative default: it assumes maximum uncertainty in the answers. E is your margin of error as a decimal. Because E is squared in the denominator, halving your margin of error quadruples the sample size — precision gets costly fast.

The final step is rounding up to a whole person. A result of 384.16 means 384 people are not quite enough, so you need 385. The calculator always rounds up, never down, because rounding down would leave your study just short of the significance threshold you asked for.

Worked Examples

Example one is the most common scenario in survey research: 95 percent confidence, 5 percent margin of error, and a 50 percent estimated proportion. Start with the Z-score squared: 1.96 × 1.96 = 3.8416. Next the variance term: 0.5 × (1 − 0.5) = 0.5 × 0.5 = 0.25. Multiply them: 3.8416 × 0.25 = 0.9604. Now the margin of error squared: 0.05 × 0.05 = 0.0025. Divide: 0.9604 ÷ 0.0025 = 384.16. Round up to the next whole person: 385 respondents. This is why you so often see 385 cited as the standard sample for a 95/5 survey — the math lands here every time.

Example two shows a high-precision study: 99 percent confidence, 3 percent margin of error, 30 percent estimated proportion. Square the Z-score: 2.576 × 2.576 = 6.635776. Variance term: 0.3 × (1 − 0.3) = 0.3 × 0.7 = 0.21. Multiply: 6.635776 × 0.21 = 1.39351296. Margin squared: 0.03 × 0.03 = 0.0009. Divide: 1.39351296 ÷ 0.0009 = 1548.347…, which the calculator shows as 1548.35. Round up: 1,549 respondents. Compare this with example one — the jump from 385 to 1,549 shows how brutally the combination of higher confidence and a tighter margin multiplies the required sample.

Example three is a quick, low-cost poll: 90 percent confidence, 10 percent margin of error, 50 percent proportion. Square the Z-score: 1.645 × 1.645 = 2.706025. Variance: 0.5 × 0.5 = 0.25. Multiply: 2.706025 × 0.25 = 0.67650625. Margin squared: 0.1 × 0.1 = 0.01. Divide: 0.67650625 ÷ 0.01 = 67.650625, shown as 67.65. Round up: 68 respondents. This is ideal for an informal internal poll where a rough directional answer is enough and speed matters more than precision.

Example four demonstrates the effect of the proportion input. Keep 95 percent confidence and 5 percent margin of error, but change the estimated proportion to 20 percent. Z squared is still 3.8416. Variance: 0.2 × 0.8 = 0.16. Multiply: 3.8416 × 0.16 = 0.614656. Divide by 0.0025: 0.614656 ÷ 0.0025 = 245.8624, shown as 245.86. Round up: 246 respondents. Knowing the answers will be lopsided (20/80) instead of even (50/50) cuts the required sample from 385 to 246 — but only use a non-50 estimate when you have real evidence for it.

Reading Your Results

The result box walks you through the answer in five rows. “Confidence level” restates your choice with its Z-score so you can confirm the right setting was used. “Margin of error” shows your precision as a plus-or-minus percentage. “Estimated proportion” shows the answer split you assumed. “Unrounded result” shows the raw formula output and the rounded-up integer side by side, so you can see that, for example, 384.16 becomes 385. “Calculation” writes the entire arithmetic in one line for verification.

The green banner states the actionable conclusion, such as “You need at least 385 respondents.” Treat this as a minimum target for completed, usable responses — not invitations sent. If you expect a 20 percent response rate, you must invite five times as many people as the banner number. Always plan your outreach from the completed-responses target, working backward through your expected response rate.

Why 50 Percent Is the Safe Default

The estimated proportion input confuses many first-time users, but the logic is straightforward. The term p(1−p) in the formula measures how spread out the answers might be. It reaches its maximum of 0.25 exactly at p = 0.5, and it shrinks toward zero as p moves toward 0 or 100 percent. A bigger variance term means a bigger sample, so 50 percent always produces the largest — and therefore safest — sample size.

Think of it this way: if you are asking whether people prefer tea or coffee and you have no clue, the answers could split evenly, and even splits are the hardest to pin down precisely. But if you already know from past data that 90 percent prefer coffee, the outcome is nearly certain and you need fewer people to confirm it. Unless you have solid prior evidence of a lopsided split, keep the default at 50 percent and enjoy the safety margin.

There is one caution: do not set a low proportion just to shrink the sample and save money. If your guess is wrong and the true split is closer to 50/50, your study will be underpowered and its conclusions unreliable. The honest approach is to use 50 percent whenever you are genuinely uncertain, and only use a specific estimate when previous research or pilot data backs it up.

Common Mistakes in Sample Size Planning

The biggest mistake is confusing the confidence level with the probability that your result is correct. A 95 percent confidence level does not mean there is a 95 percent chance your finding is true. It means that if you ran the same survey 100 times, about 95 of those runs would produce intervals containing the true value. It is a statement about the method’s reliability, not a grade on your single result.

Another frequent error is demanding a tiny margin of error without budgeting for it. Moving from a 5 percent to a 2 percent margin quadruples the sample — from 385 to over 2,400 at 95 percent confidence. Many projects stall halfway when the team realizes what their precision target actually costs. Test different margins in the calculator before committing to one.

Stopping data collection early “because the results look clear” is a subtler trap. Peeking at results and stopping when they look significant invalidates the very significance you are claiming, because the math assumes the sample size was fixed in advance. Decide your target with the calculator, then collect the full sample before drawing conclusions.

Finally, remember that a big enough sample cannot fix a biased one. Surveying 385 people who all come from your own mailing list tells you about your mailing list, not the general public. Random, representative selection matters as much as sample size — the formula assumes every member of the population had a fair chance of being included.

Frequently Asked Questions

1. What does “statistically significant” actually mean?

It means your result is unlikely to be a fluke of random sampling. A statistically significant finding is one where the observed pattern is strong enough, given your sample size, that it probably reflects the real population rather than chance variation. The sample size formula ensures your study is large enough for that judgment to hold.

2. What is a Z-score and where do 1.96 and 2.576 come from?

A Z-score measures how many standard deviations from the average you must cover to capture a given share of outcomes, based on the normal distribution. For 95 percent confidence you need 1.96 standard deviations; for 99 percent you need 2.576. These are fixed mathematical constants, not opinions, which is why every sample size calculator uses the same values.

3. Should I choose 90, 95, or 99 percent confidence?

Use 95 percent for most research — it is the accepted standard in academia, polling, and market research. Choose 90 percent for quick, low-stakes polls where a smaller sample matters more than certainty. Choose 99 percent for high-stakes decisions, such as medical or safety research, where being wrong is costly. Higher confidence always means a larger sample.

4. What exactly is the margin of error?

It is the plus-or-minus range around your survey result. If 60 percent of respondents say yes with a 5 percent margin of error, the true population value is likely between 55 and 65 percent. Smaller margins mean more precise answers, but they require much larger samples because the margin is squared in the formula.

5. Why is the result always rounded up?

Because 384.16 people cannot complete a survey, and 384 people fall just short of the mathematical requirement. Rounding up to 385 guarantees the sample truly satisfies your chosen confidence level and margin of error. Rounding down would leave the study slightly underpowered, defeating the purpose of the calculation.

6. What should I enter for estimated proportion if I have no idea?

Leave it at 50 percent. That value maximizes the variance term in the formula and therefore produces the largest required sample, which keeps your study safe no matter how the answers actually split. Only use a different value when prior research or pilot data gives you a solid reason to expect a lopsided result.

7. Does a bigger sample always give better results?

Up to a point, yes — larger samples shrink the margin of error. But the gains diminish quickly: quadrupling the sample only halves the margin of error. Beyond a few thousand respondents, extra responses add little precision while adding real cost. The calculator helps you find the sweet spot instead of overspending.

8. Why doesn’t this calculator ask for my population size?

Because for large populations, the total size barely changes the answer. Whether your population is 100,000 or 10 million, the required sample for 95 percent confidence and a 5 percent margin is about 385 either way. Population size only matters for small, known groups — that is what the finite population correction in the survey sample size calculator handles.

9. How many invitations do I need to send to get 385 responses?

Divide 385 by your expected response rate. With a 20 percent response rate, you need about 1,925 invitations (385 ÷ 0.20). With a 10 percent rate, you need 3,850. Always plan from completed responses backward through your response rate, and be conservative — it is better to invite too many than to fall short.

10. Can I stop collecting data early if the results look clear?

No — not if you want to claim statistical significance. The math assumes your sample size was fixed before you looked at the data. Peeking and stopping early when results look good inflates false positives. Set your target with the calculator, collect the full sample, then analyze.

11. Does this work for A/B tests on my website?

The formula gives a reasonable starting estimate for the visitors needed per variation, but proper A/B testing also considers your baseline conversion rate and the minimum improvement you want to detect. Use this calculator for a quick ballpark, then use a dedicated A/B test calculator for the final plan.

12. What is the difference between confidence level and confidence interval?

The confidence level (like 95 percent) is how sure you want to be about your method; the confidence interval is the actual range around your result (like 55 to 65 percent). The margin of error is half the width of that interval. This calculator takes the level and the margin as inputs and returns the sample size that connects them.

13. My survey has twenty questions. Do I need a bigger sample?

Not necessarily for each individual question — the formula applies per question answered by the full sample. But if you plan to slice results into small subgroups (like comparing five age bands), each subgroup needs enough respondents on its own. Size the sample for your smallest important subgroup, not just the total.

14. Can a large sample fix a biased survey?

No. Sample size controls random sampling error, not bias. If your respondents are not representative — say, only your most enthusiastic customers reply — then 10,000 biased responses are just a precisely wrong answer. Random, representative recruitment matters at least as much as hitting the calculated number.

15. Is 385 respondents a magic number?

It is not magic; it is just the answer to the most common setup: 95 percent confidence, 5 percent margin of error, 50 percent proportion. Change any input and the answer changes. Treat 385 as a handy rule of thumb for standard surveys, and use the calculator whenever your situation differs.

Final Thoughts

Statistical significance is not a luxury reserved for academics — it is simply the discipline of collecting enough evidence before believing your own numbers. This calculator turns that discipline into a thirty-second task: pick your confidence, set your margin, and get a target you can defend. Aim for the full number of completed responses, recruit fairly, and resist the urge to peek early. Do that, and your survey results will be something you can actually act on, instead of numbers you merely hope are true.