Sample Size Calculator
Required Sample Size
—
Confidence Z-Score
—
With Finite Population Correction
—
How Sample Size Calculation Works
Sample size calculation is a statistical method that determines the minimum number of observations or survey responses needed to draw reliable conclusions about a larger population. The concept rests on the fundamental principle of statistical inference: a properly selected random sample can represent a population with quantifiable precision. According to the U.S. Census Bureau, even the American Community Survey uses sampling rather than complete enumeration for most of its data, demonstrating that well-designed samples can produce highly accurate population estimates.
This calculator implements Cochran's formula, the most widely used method for determining sample size for surveys measuring proportions. The formula accounts for three key variables: your desired confidence level (how certain you want to be), margin of error (how precise you want your results), and expected proportion (your best guess of the result). For populations under 10,000, the finite population correction (FPC) is applied to reduce the required sample, since sampling a larger fraction of a small population inherently provides more precision.
The Sample Size Formula
Cochran's formula for sample size is: n = Z^2 x p x (1 - p) / e^2, where Z is the Z-score corresponding to your confidence level, p is the expected proportion (use 0.5 for maximum variability), and e is the margin of error as a decimal. For finite populations, the corrected formula is: n_adj = n / (1 + (n - 1) / N), where N is the population size.
Worked example: You want to survey customer satisfaction with 95% confidence and 5% margin of error. Z = 1.96 for 95% confidence. Using p = 0.50 (unknown proportion). n = (1.96)^2 x 0.50 x 0.50 / (0.05)^2 = 3.8416 x 0.25 / 0.0025 = 385 responses. If your customer base is only 2,000 people, apply FPC: n_adj = 385 / (1 + 384/2000) = 385 / 1.192 = 323 responses. The FPC saved you 62 surveys while maintaining the same statistical precision.
Key Terms You Should Know
Confidence level is the probability that the true population value falls within your margin of error. A 95% confidence level means that if you repeated the survey 100 times, approximately 95 of those surveys would produce results within the stated margin of error. Margin of error (also called confidence interval half-width) is the plus-or-minus range around your survey result; a 5% margin means your result of 60% actually represents 55-65%. Z-score is the number of standard deviations from the mean corresponding to your confidence level: 1.645 for 90%, 1.96 for 95%, and 2.576 for 99%. Expected proportion is your best estimate of the survey outcome; 50% produces the largest (most conservative) sample size because variance is maximized at p = 0.50. Finite population correction (FPC) reduces the required sample when the population is small, because sampling a larger fraction of the population inherently provides more information. Response rate is the percentage of people surveyed who actually complete the survey, typically 10-30% for email surveys, so you must send far more invitations than your required sample size.
Required Sample Sizes by Confidence Level and Margin of Error
The following table shows the required sample size for an infinite population (over 100,000) at different combinations of confidence level and margin of error, assuming p = 0.50 (most conservative). These numbers represent the minimum completed responses needed, not the number of survey invitations to send.
| Margin of Error | 90% Confidence | 95% Confidence | 99% Confidence |
|---|---|---|---|
| 1% | 6,765 | 9,604 | 16,587 |
| 2% | 1,691 | 2,401 | 4,147 |
| 3% | 752 | 1,068 | 1,843 |
| 5% | 271 | 385 | 664 |
| 7% | 138 | 196 | 339 |
| 10% | 68 | 97 | 166 |
Notice that halving the margin of error roughly quadruples the required sample size. Going from 5% to 2.5% margin of error requires approximately four times as many responses. This is because precision scales with the square root of sample size, not linearly. For most business decisions, a 5% margin of error at 95% confidence (385 responses) provides sufficient precision while remaining achievable. Academic research and political polling often target 2-3% margins, requiring 1,000+ responses.
Practical Examples
Example 1: Customer Satisfaction Survey. A company with 50,000 customers wants to measure satisfaction with 95% confidence and 3% margin of error. Using Cochran's formula: n = 1.96^2 x 0.5 x 0.5 / 0.03^2 = 1,068 responses needed. With a typical email survey response rate of 15%, they should email approximately 7,120 customers to collect 1,068 completed surveys.
Example 2: Small School Election Poll. A school of 800 students wants to predict an election outcome with 95% confidence and 5% margin of error. Base calculation: 385 responses. With FPC: 385 / (1 + 384/800) = 260 responses needed. That is about one-third of the student body. Use our confidence interval calculator to analyze results once collected.
Example 3: Medical Clinical Trial. A pharmaceutical company expects a treatment success rate of 70% (p = 0.70) and wants to detect this with 99% confidence and 2% margin of error. n = 2.576^2 x 0.70 x 0.30 / 0.02^2 = 6.636 x 0.21 / 0.0004 = 3,484 patients needed. Clinical trials often use even larger samples to detect differences between treatment and control groups, which requires different formulas (power analysis) available in our t-test calculator.
Tips and Strategies for Effective Sampling
- Always plan for non-response. Collect 20-50% more responses than the minimum calculated sample size to account for incomplete surveys, disqualified responses, and data cleaning. Email surveys typically have 10-30% response rates, phone surveys 5-15%, and in-person surveys 60-80%.
- Use 50% as the expected proportion when uncertain. This produces the largest (most conservative) sample size. If you have prior data suggesting the true proportion is far from 50%, you can safely use a smaller sample. A proportion of 10% or 90% requires only 40% as many responses as 50%.
- Random sampling is critical. No sample size formula can correct for a biased sample. Ensure every member of your target population has an equal chance of being selected. Convenience samples (surveying only people who happen to be available) produce unreliable results regardless of size.
- Consider stratified sampling for subgroup analysis. If you need to analyze results by subgroups (age, region, department), each subgroup needs its own adequate sample size. A survey of 385 people split into 5 age groups gives only 77 per group, which has a 11% margin of error -- often too imprecise for actionable insights.
- Match precision to decision impact. A $100 million business decision warrants higher precision (1-2% margin of error, 1,000+ responses) than an informal pulse check (10% margin, 68 responses). Use our regression calculator for more complex statistical analyses beyond simple proportions.
Frequently Asked Questions
What is a good sample size for a survey?
For most surveys targeting a large population (over 10,000), 385 completed responses provide 95% confidence with a 5% margin of error, which is the standard benchmark for business and market research surveys. For academic research or high-stakes decisions, 1,000 or more responses are preferred to achieve tighter margins of error (around 3%). For smaller populations, the finite population correction reduces the required sample. The key is matching precision to the importance of the decision the data will inform.
What does margin of error mean?
Margin of error is the plus-or-minus range around a survey result that defines the interval likely to contain the true population value. If your survey finds 60% approval with a 5% margin of error at 95% confidence, you can say with 95% certainty that the true population approval is between 55% and 65%. Smaller margins of error require larger samples but give more precise, actionable results. A 1% margin requires roughly 25 times as many responses as a 5% margin.
What confidence level should I use?
The 95% confidence level is the standard for most business surveys, market research, and social science studies. Academic research, clinical trials, and regulatory decisions often require 99% confidence for more rigorous evidence. Preliminary or exploratory research may use 90% confidence to reduce the required sample size during early investigation. Each step up in confidence requires a larger sample: moving from 95% to 99% confidence increases the required sample by approximately 72% while keeping the margin of error constant.
Does population size matter for sample size?
For populations over 10,000 to 20,000, population size has minimal effect on the required sample. A survey of a city of 100,000 and a survey of a nation of 330 million both require approximately 385 responses for 95% confidence with 5% margin of error. However, for smaller populations, the finite population correction significantly reduces the needed sample. Surveying a company of 500 employees requires only about 217 responses rather than 385, because sampling 77% of a small group inherently captures more information about that group.
How do I account for non-response in my sample size?
You should always inflate your target sample by the inverse of your expected response rate. If you need 385 completed responses and expect a 20% response rate, you need to send the survey to 385 / 0.20 = 1,925 people. Email surveys typically see 10-30% response rates, phone surveys 5-15%, and in-person or intercepted surveys 60-80%. Incentives like gift cards or donations can boost response rates by 10-20 percentage points. Non-response bias can also skew results if certain groups are less likely to respond.
What is the difference between sample size for proportions and means?
This calculator determines sample size for proportion-based questions (yes/no, approve/disapprove, percentage outcomes). For continuous measurements like average income, weight, or test scores, a different formula applies: n = (Z x sigma / e)^2, where sigma is the population standard deviation and e is the desired margin of error in the same units. Estimating sigma requires pilot data or prior studies. Use our confidence interval calculator to work with continuous data after collecting your sample.