Cvent Survey

People tend to make calculating the sample size for their online web survey much more difficult than it needs to be. A quick Google search offers a wide range of survey sample size calculators that will eliminate having to do the calculations by hand. There are a few factors you need to know before using the calculator:

• Size of your total population. Your total population is the group for which you want to survey, but of course you'll only be surveying a smaller group of the total population. An easy way to think of the total population is with client surveys. The number of clients you have is the population, you'll probably choose not to survey all of them and need to pick a sampling method (surveying all of them would be a census survey method).

• Confidence interval. This is the plus or minus number reported in survey reports. The confidence interval means that the total population falls within plus or minus your survey findings. Typically, ±5 is used. When reporting a survey finding where 68% of survey respondents said they would purchase your product again, you can be sure that between 63% and 73% of the entire population would have said they would purchase again.

• Confidence level. This represents the percentage of your total population that lies within your confidence interval. Typically, 95% accuracy is accepted for sample sizes. When you put this factor with the one above, you get a statement that looks something like this: You're 95% certain of ± 5% accuracy in your survey results.

If you're using a tool to calculate sample size, you can use these standards above or decide is how accurate you want to be. Some people can accept ±10% some need it to be ±3%. Your confidence interval should depend on what you plan to do with the data, how important are the decisions being based off your survey report?

The number of survey respondents you need is not linear to the size of your total population. As the size of your total population increases, the needed sample size increases at a smaller rate. Here's a brief example (sticking with 95% certain of ±5% accuracy):


You see how it begins to level off? The idea is that you want your sample to be representative of the entire population so you can be confident using the data to make decisions. Keep in mind, however, if you plan to split your sample into segments to do analysis (for example gender or age) you need to consider those segments separately when determining sample size. Otherwise, you may end up with enough females to represent the female segment of your total population but not enough males.