Surveying too few people wastes a study before it starts, since a sample that's too small can't reliably represent the population it's meant to describe. Surveying far more people than necessary wastes time and budget on responses that add little additional statistical precision beyond a certain point.

The **sample size calculator** solves for the exact number of responses needed using Cochran's formula, the standard method behind virtually every published survey's stated sample size. This article covers exactly how the formula works, how a finite population changes the required number, and where researchers most often misjudge what size sample they actually need.

What Is a Sample Size Calculator and Why Does It Matter

A **sample size calculator** determines the minimum number of survey responses or study participants needed for a result to be statistically reliable, based on your desired confidence level, acceptable margin of error, and an estimated proportion. Students and academic researchers use it to justify a study's methodology, while market analysts and public health researchers use it to plan real surveys and clinical studies with defensible sample sizes. The core method: required sample size grows with a higher confidence level and shrinks with a larger acceptable margin of error. Getting it wrong, especially skipping the finite population correction for a small target group, means oversampling a small population or undersampling a large one.

What You Need Before You Start

Have this information ready before you enter anything into the calculator.

- Your desired confidence level, typically 90, 95, or 99 percent, reflecting how sure you want to be that your sample reflects the true population.

- Your acceptable margin of error, typically expressed as a percentage, such as plus or minus 5 percent.

- An estimated proportion, if you have one from prior research, or the conservative default of 50 percent if you don't.

- Your total population size, if you're surveying a specific, countable group rather than an effectively unlimited population.

- Whether your study is estimating a proportion, like a percentage who prefer something, or requires a different sample size approach, such as comparing two group means.

- Your expected response rate, if you need to know how many people to invite rather than just how many completed responses you need.

How to Use This Sample Size Calculator Step by Step

Follow these steps to get an accurate required sample size.

1. Choose your desired confidence level: 90, 95, or 99 percent are the most common choices.

2. Enter your acceptable margin of error as a percentage, such as 5 percent.

3. Enter your estimated proportion if you have one from a prior study, or leave it at the conservative default of 50 percent.

4. Enter your total population size if it's a known, finite number rather than effectively unlimited.

5. Click Calculate to see your required sample size, with the finite population correction automatically applied if a population size was entered.

6. Divide your required sample size by your expected response rate to estimate how many people you actually need to invite.

7. Reconsider your margin of error or confidence level if the resulting sample size is larger than what your budget or timeline can support.

The Sample Size Calculator Formula Explained Simply

Sample size calculation balances three competing factors: how confident you want to be in your result, how much error you're willing to accept, and how much natural variability exists in what you're measuring. A tighter margin of error or a higher confidence level both require more responses, while an expected proportion far from 50 percent, if genuinely known in advance, can allow for a somewhat smaller sample.

n0 = Z squared times p times (1 - p) / e squared

Finite population correction: n = n0 / (1 + (n0 - 1) / N)

- n0: the initial required sample size, assuming an effectively unlimited population.

- Z: the z-score corresponding to the chosen confidence level, 1.645 for 90 percent, 1.96 for 95 percent, and 2.576 for 99 percent.

- p: the estimated proportion of the population expected to have the characteristic being studied, using 0.5 as the conservative default when unknown.

- e: the acceptable margin of error, expressed as a decimal, such as 0.05 for 5 percent.

- N: the total population size, used only when the population is a known, finite number.

- n: the final, corrected sample size after adjusting for a finite population.

The formula works because p times (1 minus p) reaches its mathematical maximum at p equals 0.5, which is exactly why using 0.5 as a default, when the true proportion is unknown, produces the most conservative, largest possible required sample size, guaranteeing the study won't be undersampled regardless of the true underlying proportion. This is the widely accepted method known as Cochran's formula, published by William G. Cochran in "Sampling Techniques" in 1977, and it remains the standard reference formula cited throughout academic, market research, and clinical study methodology sections.

Sample Size Calculator Worked Examples

Example1:Beginner Case - Planning a Customer Satisfaction Survey for an Unlimited Population

A market researcher wants to survey customers nationally about satisfaction with a product, using a 95 percent confidence level and a 4 percent margin of error, with no prior proportion estimate available.

1. Identify the values: Z equals 1.96 for 95 percent confidence, p equals 0.5, the conservative default, e equals 0.04.

2. Apply the formula: n0 equals 1.96 squared times 0.5 times (1 minus 0.5), divided by 0.04 squared.

3. Calculate the numerator: 1.96 squared equals 3.8416, times 0.5 times 0.5, which is 3.8416 times 0.25, equaling 0.9604.

4. Calculate: 0.9604 divided by 0.0016 equals 600.25, rounding up to 601 responses.

The researcher needs approximately 601 completed responses for a nationally representative result at this confidence level and margin of error, and since the population of potential customers is large enough to be treated as effectively unlimited, no finite population correction is needed.

Example2:Advanced Case - Determining Sample Size for a Clinical Study With a Known Complication Rate

A clinical researcher is planning a post-market follow-up study and needs the required sample size using a 95 percent confidence level, a 5 percent margin of error, and a prior literature estimate that 56 percent of similar procedures experience a specific complication.

1. Identify the values: Z equals 1.96, p equals 0.56, based on the literature estimate, e equals 0.05.

2. Apply the formula: n0 equals 1.96 squared times 0.56 times (1 minus 0.56), divided by 0.05 squared.

3. Calculate the numerator: 1.96 squared equals 3.8416, times 0.56 times 0.44, which is 3.8416 times 0.2464, equaling approximately 0.9466.

4. Calculate: 0.9466 divided by 0.0025 equals approximately 378.6, rounding up to 379 participants.

This initial calculation calls for 379 participants, and since this specific study's actual eligible population is a small, defined group of only 80 patients, the researcher must next apply the finite population correction, which brings the truly required sample down substantially, since sampling from a small, known population requires proportionally fewer responses than an unlimited-population calculation would suggest.

Example3:Edge Case - Applying the Finite Population Correction to a Very Small Target Group

Continuing directly from the advanced example, the clinical researcher applies the finite population correction using the actual population of 80 patients, since surveying or enrolling anywhere close to the initial 379 figure is impossible with only 80 people to draw from.

1. Identify the values for the correction: n0 equals 379, N equals 80.

2. Apply the finite population correction formula: n equals n0 divided by (1 plus (n0 minus 1) divided by N).

3. Calculate the denominator: (379 minus 1) divided by 80 equals 378 divided by 80, which equals 4.725, plus 1, equaling 5.725.

4. Calculate: 379 divided by 5.725 equals approximately 66.2, rounding up to 67 participants.

This dramatic reduction, from an initial 379 down to approximately 67, isn't an error; it correctly reflects that when a population is small and fully known, a much smaller fraction of it needs to be sampled to achieve the same statistical confidence, since the finite correction accounts for the fact that sampling a larger share of a small population inherently reduces uncertainty faster than the unlimited-population formula alone would suggest.

Example4:Regional Variant - Comparing Cochran's Formula to the Simplified Slovin's Formula Commonly Used in Thesis Research

A graduate student researching thesis methodology encounters Slovin's formula in several published papers and wants to understand how it relates to the Cochran's formula-based calculator being used for their own study.

1. Recognize that Slovin's formula, n equals N divided by (1 plus N times e squared), is mathematically a special case of Cochran's formula that fixes the confidence level at 95 percent and the proportion estimate at exactly 0.5.

2. Apply Slovin's formula to a population of 5,000 with a 5 percent margin of error: n equals 5,000 divided by (1 plus 5,000 times 0.05 squared), which is 5,000 divided by (1 plus 5,000 times 0.0025), equaling 5,000 divided by (1 plus 12.5), or 5,000 divided by 13.5.

3. Calculate: 5,000 divided by 13.5 equals approximately 370.4, rounding up to 371.

4. Apply full Cochran's formula with the finite correction to the same population and margin, explicitly using Z equals 1.96 and p equals 0.5: this produces the identical result, approximately 371, confirming Slovin's formula is simply a shorthand version of Cochran's under these specific fixed assumptions.

This confirms that Slovin's and Yamane's formulas, both widely cited in social science and thesis research, aren't separate statistical methods at all; they're simplified shortcuts for Cochran's formula that only work correctly when the 95 percent confidence level and 0.5 proportion assumptions actually match the researcher's real study design, and any study using a different confidence level or a known proportion estimate should use the full Cochran's formula directly instead.

Sample Size Calculator Results What the Numbers Mean

A larger required sample size generally means a study demanding more resources, time, and participant recruitment effort, and this number grows specifically when a tighter margin of error or a higher confidence level is required. A required sample size that comes out larger than your entire actual population is a signal to apply the finite population correction, as shown in the edge case above, rather than assuming the initial, unlimited-population calculation is the final answer.

A required sample size based on the conservative 0.5 proportion default represents the largest possible sample needed regardless of the true, unknown proportion, which is why it's considered the safe choice when no prior estimate exists; using an actual prior estimate, as in the advanced clinical example, can meaningfully reduce the required number when that estimate is genuinely reliable.

The most common misunderstanding is assuming a bigger sample is always simply better, without weighing the diminishing statistical returns against the real cost of recruiting each additional participant. Beyond a certain point, adding more responses narrows the margin of error only marginally, while the cost of recruiting each additional participant stays the same or increases.

How Your Inputs Change the Result

Confidence level has a significant effect on required sample size, and this relationship isn't linear; moving from 95 to 99 percent confidence increases the z-score from 1.96 to 2.576, and since the z-score is squared in the formula, this seemingly modest jump increases the required sample size by roughly 73 percent for the same margin of error and proportion.

Margin of error has an inverse-squared relationship with sample size, meaning cutting the margin of error in half doesn't just double the required sample; it quadruples it, since the margin of error term is squared in the denominator. This is why tightening precision from a 5 percent to a 2.5 percent margin of error can dramatically increase a study's required participant count and cost.

Population size only matters once it's small enough for the finite population correction to meaningfully reduce the initial calculation, as shown in the edge case above. For populations in the tens of thousands or more, the correction has a negligible effect, but for populations in the hundreds or low thousands, it can cut the required sample size substantially.

Confidence Level | Z-score | Relative Sample Size Needed (same margin and proportion)

90 percent | 1.645 | Smallest required sample

95 percent | 1.96 | About 42 percent larger than 90 percent

99 percent | 2.576 | About 73 percent larger than 95 percent

Sample Size Calculator Common Mistakes to Avoid

- Forgetting to apply the finite population correction when surveying a small, known population, leading to a wildly oversized and unrealistic required sample size.

- Using an overly optimistic proportion estimate without solid prior data, when the conservative 0.5 default is the safer choice absent a genuinely reliable prior estimate.

- Confusing margin of error with confidence level, since these represent two entirely different, independently adjustable parameters in the formula.

- Assuming Slovin's or Yamane's formula applies to a study using a confidence level other than 95 percent or a proportion estimate other than 0.5, when these simplified formulas only work correctly under those specific fixed assumptions.

- Treating "sample size" and "number of people to invite" as the same number, forgetting to divide the required sample size by an expected response rate to determine actual invitation volume needed.

- Assuming a larger sample is always worth the added cost, without weighing the diminishing statistical returns of precision against the real expense of recruiting additional participants.

- Using this proportion-based formula for a study actually comparing two group means or requiring a power analysis, which needs a different, related but distinct statistical approach.

- Rounding the required sample size down instead of up, when standard practice always rounds a fractional required sample size up to the next whole participant.

Key Terms Explained

Confidence level

Confidence level is the probability that your sample's result falls within the stated margin of error of the true population value, commonly set at 90, 95, or 99 percent.

Margin of error

Margin of error is the maximum amount your sample's result might differ from the true population value, commonly expressed as a plus-or-minus percentage.

Z-score

A z-score is a value from the standard normal distribution corresponding to a specific confidence level, used directly in the sample size formula.

Proportion estimate

Proportion estimate is your best guess, or a prior study's finding, for what percentage of the population has the characteristic you're measuring.

Finite population correction

Finite population correction is an adjustment applied when sampling from a small, known population, reducing the required sample size compared to an unlimited-population calculation.

Statistical significance

Statistical significance means a result is unlikely to have occurred by random chance alone, supported by an appropriately sized and properly drawn sample.

Response rate

Response rate is the percentage of invited participants who actually complete a survey, used to calculate how many people to invite based on a required completed sample size.

Study power

Study power is the probability that a study will correctly detect a real effect or difference if one genuinely exists, a related but distinct concept from the proportion-based sample size formula covered here.

How This Calculator Works Methodology

This calculator applies Cochran's formula, n0 equals Z squared times p times (1 minus p), divided by e squared, originally published by William G. Cochran in "Sampling Techniques" in 1977, and standard throughout academic, market research, and clinical study methodology. When a finite population size is provided, the correction n equals n0 divided by (1 plus (n0 minus 1) divided by N) is applied automatically. Z-scores default to the standard values, 1.645, 1.96, and 2.576 for 90, 95, and 99 percent confidence respectively. Required sample sizes are always rounded up to the next whole participant. This calculator is designed for proportion-based sample size estimation and does not directly calculate sample sizes for comparing means between groups or for formal statistical power analysis, which require related but distinct formulas.

Frequently Asked Questions About Sample Size Calculator

1. How do you calculate sample size?

Use Cochran's formula: n0 equals Z squared times p times (1 minus p), divided by e squared, where Z is the z-score for your confidence level, p is your estimated proportion, and e is your margin of error. If surveying a small, known population, apply the finite population correction afterward.

2. How accurate is a sample size calculator?

It's exact for the standard formula once your confidence level, margin of error, and proportion estimate are entered correctly. The actual real-world reliability of your study also depends on random sampling being done properly and an adequate response rate, which the formula itself doesn't guarantee.

3. What is a reliable sample size for most studies?

For large or unlimited populations, a sample size of approximately 384 to 400 is commonly sufficient for a 95 percent confidence level with a 5 percent margin of error. Smaller or high-stakes studies, like clinical trials, may require a different, often larger, sample depending on the specific proportion and precision needed.

4. What sample size is considered too small?

Fewer than 30 responses is generally considered too small to provide statistically reliable results for most studies, particularly when sampling from a large population, since such a small sample carries a high risk of a wide margin of error and low confidence in the results.

5. How does population size affect required sample size?

For large populations, population size itself has little effect on the required sample size beyond a certain point. For smaller, known populations, the finite population correction meaningfully reduces the required sample, since a larger proportion of a small population can be sampled to reach the same confidence level.

6. What is the difference between Cochran's formula and Slovin's formula?

Slovin's formula is a simplified special case of Cochran's formula that fixes the confidence level at 95 percent and the proportion estimate at 0.5. Cochran's formula is more flexible, allowing any confidence level and any known or estimated proportion to be specified directly.

7. What is a typical sample size calculation people check with this tool?

Students and academic researchers most often check sample size requirements to justify a thesis or dissertation's survey methodology. Market researchers and public health professionals more often check required sample sizes for customer surveys or clinical study planning, frequently needing the finite population correction for a defined target group.

8. Should I always use the maximum possible sample size for the most reliable results?

Not necessarily. Beyond the sample size that meets your required confidence level and margin of error, additional responses provide only marginal further precision while adding real recruitment cost and time, so matching your sample size to your actual precision needs is generally the more efficient approach.

Related Calculators Worth Bookmarking

- Margin of Error Calculator: use this to find the margin of error resulting from a specific sample size, the reverse calculation from this tool.

- Confidence Interval Calculator: helpful for calculating the actual confidence interval around a result once your survey data is collected.

- Statistical Power Calculator: relevant for studies comparing group means or effects, requiring a different sample size approach than proportion estimation.

- Z-Score Calculator: useful for finding the exact z-score for a confidence level not covered by the standard 90, 95, or 99 percent defaults.

- Sampling Error Calculator: pairs with sample size planning to estimate the error associated with a specific completed sample.

- Survey Response Rate Calculator: useful for converting a required completed sample size into the number of invitations needed to send.

- A/B Test Sample Size Calculator: relevant for studies comparing two variants rather than estimating a single population proportion.

- Standard Deviation Calculator: useful groundwork for studies requiring a mean-comparison sample size formula rather than the proportion-based method here.

References and Further Reading

1. Cochran WG. Sampling Techniques, 3rd Edition. John Wiley and Sons, 1977.

2. Verma JP, Verma P. Determining Sample Size and Power in Research Studies: A Manual for Researchers. Springer Nature, 2020.

3. Israel GD. Determining Sample Size. University of Florida IFAS Extension, PEOD6.

Important Disclaimer

This calculator is for educational and informational purposes and provides a standard statistical estimate. For formal academic, clinical, or regulatory research, consult a qualified statistician to confirm the appropriate sample size methodology for your specific study design.