Sample Size Calculator

Scientifically determine the ideal number of respondents for your survey to ensure statistically significant results for your market research, academic study, or customer feedback campaigns.

Population Size (N)

The total number of people in the group you are trying to study. If larger than 20,000, you can leave this blank.

Confidence Level (Z) (%)

How sure you want to be that your data is accurate. 95% is the global industry standard.

Margin of Error (e) (%)

The range of error you are willing to accept. 5% is standard (e.g. results are +/- 5% of reality).

Required Sample Size

385

Completed responses needed for statistical validity

The Science: Cochran’s Formula

To scientifically determine sample size, we use Cochran’s Formula. This equation balances precision with feasibility, telling you exactly how many data points you need to represent a larger group without wasting resources.

n =

Z² • p(1-p) e²

Z: Z-Score (Confidence Level)

p: Population Proportion (0.5)

e: Margin of Error (Decimal)

n: Sample Size

Note on Finite Population: If the calculated sample size (n) exceeds 5% of your total population (N), our calculator automatically applies a second formula called "Finite Population Correction" to reduce the number, saving you time and money.

Understanding the Components

Sample size calculation isn't magic; it is a balance of four statistical levers. Understanding these will help you design better surveys.

Confidence Level

How confident do you want to be? A 95% confidence level means if you ran the survey 100 times, the results would match the true population 95 times. Higher confidence requires a much larger sample.

Margin of Error

The "wiggle room" in your data. If 50% of respondents say "Yes" with a 5% margin of error, the true reality is between 45% and 55%. Smaller margins require exponentially more data.

Standard of Deviation

This represents the expected variance. Since we usually don’t know the results before we survey, we use 0.5 (50%). This is the most conservative estimate, ensuring your sample size is sufficient for any outcome.

Population Size

The total number of people in the group you are studying. Surprisingly, population size matters very little once it exceeds 20,000 people. A sample of 385 is valid for both a city and a nation.

What is a "good" sample size?

385

The "Magic Number" for Large Populations

For almost any population greater than 100,000 (whether it is New York City or the entire United States), the math stabilizes at 385 respondents. This provides the industry standard of 95% Confidence with a 5% Margin of Error.

However, "good" depends on your goals. Here are common benchmarks used by professionals:

Research Goal	Typical Accuracy	Sample Needed
High-Stakes (Medical/Gov)	99% Conf, 2% Margin	4,160
Standard market research	95% Conf, 5% Margin	385
Quick Customer Feedback	90% Conf, 10% Margin	68
Rough Internal Poll	80% Conf, 10% Margin	41

Real-World Calculation Scenarios

Seeing the math in action helps clarify the decisions you need to make. Here are four detailed examples from different industries.

1. Corporate HR Survey

An HR Director at a company of 500 people wants to survey employees about a new hybrid work policy. Accuracy is important, but the population is small.

Population (N) 500

Confidence 95%

Error Margin 5%

Sample Needed 217

2. National Product Launch

A brand manager is launching a product in a country of 50 million people. They need reliable data to convince investors.

Population (N) 50,000,000

Confidence 95%

Error Margin 5%

Sample Needed 385

3. Clinical Health Study

A medical research team is testing a new protocol on 10,000 patients. They cannot afford a 5% error margin; precision is life-or-death.

Population (N) 10,000

Confidence 99%

Error Margin 2%

Sample Needed 3,700

4. Political Pulse Poll

A local campaign needs a quick "gut check" on a candidate in a district of 100,000. Budget is zero, speed is key.

Population (N) 100,000

Confidence 80%

Error Margin 10%

Sample Needed 41

Common Sampling Pitfalls

Even if your mathematical calculation is perfect, your survey can still fail if you fall into these traps.

1. Selection Bias

If you need 385 people to represent a country, but you only interview people in the capital city, your data is worthless. Your sample must be RANDOM and distributed across your target demographics.

2. Ignoring Sub-Groups

If you want to compare "Men vs. Women", you need a statistically significant sample for EACH group, not just the total. 385 total respondents might only give you 40 men if your sampling is skewed, making comparison impossible.

Expert FAQ

What if I can't get enough responses?

If you cannot reach your target sample size, your survey is not necessarily invalid, but it is less precise. You simply have to accept a higher Margin of Error. For example, if you aimed for 400 responses (5% margin) but only got 100, your margin of error increases to roughly 10%. You can still identify broad trends, but avoid making high-stakes decisions based on small percentage differences.

Why don't I need millions of responses for a large country?

This is due to the law of diminishing returns in statistics. Imagine a massive pot of soup. You don't need to eat the whole pot to know how it tastes; one spoonful (a sample) is enough, provided the soup is well-stirred (randomized). Once your "spoonful" is big enough (385 people), adding more doesn't significantly change the "taste" (the data accuracy).

How do I calculate response rate vs. sample size?

The calculator gives you the number of COMPLETED surveys needed. To know how many invitations to send, divide the Sample Size by your expected Response Rate. Example: You need 400 completed surveys. You expect a 10% response rate. Math: 400 / 0.10 = 4,000. You must send 4,000 invitations.

Does the survey method (Online vs. Phone) matter?

Mathematically, the formula is the same. However, different methods introduce different biases. Online surveys might exclude older demographics, while phone surveys might exclude younger ones. To ensure your sample is valid, you must ensure your collection method gives every member of the population an equal chance of being selected.

Can I simply survey 10% of my population?

This is a common misconception. For small populations (e.g., 500 people), 10% (50 people) might be too small, yielding a high error margin. For large populations (e.g., 1 million), 10% (100,000 people) is wastefully expensive and unnecessary. It is always better to use the calculator than a flat percentage.

What is a Z-Score?

A Z-score is a statistical value that indicates how many standard deviations a data point is from the mean. In survey sampling, it essentially converts your desired "Confidence Level" into a number the formula can use. For 95% confidence, the Z-score is 1.96.