Are larger sample sizes more accurate?

When it comes to your survey it’s nearly always impossible, or at least impractical, to collect data from all the people or items that you’re interested in.

Yet you still want to be able to get accurate insights that are representative of the audience you’re targeting. This is where the value of sampling and how to determine sample size come to the fore.

What is sampling and why does sample size matter?

Essentially, when you need to survey a large population of respondents, you’ll be interested in the entire group. However, because it’s not realistically possible to get answers from everyone, you’ll need to take a random sample of individuals, who will be representative of that population as a whole.

At this point if you’re to maximise your chances of obtaining accurate and statistically significant results, you’ll need to work out the right sample size for your survey.

If your sample is too small, you could risk including a disproportionate number of individuals leading to anomalies that skew your results. Yet, on the other hand while a much larger sample may give you much greater accuracy, the extra time and expense may mean the benefits don’t outweigh the costs. So, it’s crucial that you get the right sample size for your needs.

How to determine sample size

Before determining the sample size for your survey through calculation, there are some vital definitions you need to become familiar with, if you’re to get the level of accuracy you need.

Population size

The first thing you need to do, is work out how many people you want to talk to in total. To find this out, you need to be clear about who does and doesn’t fit into your group.

For example, if you want to know about cat owners, you’ll probably want to include everyone who has at some point owned at least one cat. However, if you’re only interested in current cat owners, you may want to exclude those who previously owned a cat. Don’t worry if you’re unable to calculate the exact number, as it’s not uncommon to have an estimated range.

Margin of error (confidence interval)

While some errors are inevitable, you’ll have in your mind the level of accuracy your research requires and therefore the margin of error you’re willing to accept. Typically referred to as the confidence interval, it’s usually displayed as a plus or minus number that sits beside a percentage figure.

For example, a margin of error is typically displayed during the reporting of political polls, such as:

“59% of voters believe John Smith would be a good Prime Minister, with a margin of error of +/- 5%.”

The larger your margin of error, the less likely it is that your study will be representative of the general population

Confidence level

The confidence level is concerned with ensuring that your margin of error is set at the right level.

Essentially, what this means is that if you repeated a study and used the range each time to make your calculations, the actual value should lie inside the same ranges 95% of the time.

The most commonly used confidence levels are 90%, 95% and 99% confident.

Sample size

Finally, the sample size itself is the number of people you need to estimate an accurate analysis of the whole population. If people don’t respond to your survey or research, you’ll have to increase your sample size in order to gauge a precise calculation.

Ways of calculating sample size

When it comes to how to determine the sample size for a survey, it’s useful to have some examples of how to do this to draw on. So, here’s some methods to help you.

Run a census

If you have a relatively small target population, you might decide to carry out a census instead of using a sample.

If we’re talking about 100 people or below, then this would be relatively simple to keep track of. So, you could create a census and create a statistically robust study.

However, for research involving thousands of people or more, it would be impossible to survey everyone, making this option unsuitable at this stage.

Use a similar study

Given the sheer number of studies around, the chances of finding one similar to your own is quite high, particularly if you use a research tool like the best academic research databases.

In such a scenario, you can use their sample size to create your own. However, you might want to bear in mind that if those researchers didn’t do their calculations correctly, it could impact your own study’s credibility.

Manually calculate your sample size

If you have strong maths skills, you might decide to calculate sample size manually.

There are a number of formulas you can refer to for this including Cochran and Slovin’s.

However, calculating your sample size is challenging, and even statisticians can struggle with these manual calculations. So, for accuracy, we would recommend you using a calculator instead.

Try using a sample size calculator

When it comes to determining your sample size, the sample size calculator is one of the simplest ways to do this.

The good news is that we have a sample size calculator to help you with this. By allowing you to input the data for your population size, confidence level and margin of error, it will enable you to calculate your sample size with the required accuracy that you need in just a few short simple steps.

Concluding thoughts

We hope after reading this you’ll realise the importance of getting the right sample size and have some methods you can use to quickly identify the response figure you need.

We appreciate there’s many other factors that can influence the quality of data you end up with, from your survey’s structure and wording of questions to your choice of distribution channels and more. However, if you can get your sample size right, it should give you the best start in terms of the correct pathway to obtaining the insights you need to achieve.

Get the right survey plan to deliver the insights you need

From your survey creation to your data collection and analysis. While it’s important to understand how to obtain the right survey sample size, you also need the right tools to deliver great surveys.