Are you introducing a new product on the market? Work on client segmentation? Or maybe you plan important changes to existing products? In all those tasks online survey research can be more than helpful.
In an ideal world, you would ask each person (e.g. client) in your population about opinions, preferences, or behavior. As your target group may be very large and dispersed you may not be able to do so. Luckily it is not necessary. You can randomly select from a population certain number of people and conduct a survey on a selected sample.
Sounds good, but how many people do I need to start my online research? We will help you to answer this question.
Estimate population size
If you don’t know how to start to use the Internet or go to the statistical office website. In each county, there is a statistical office that can provide you (usually free of charge) with general information about the economy and population.
Decide how accurate survey results you need
Because you won’t ask every person in your target population about his/her opinion, your survey results will contain some statistical error. You need to decide how big error you are willing to accept. Most of the researchers decide for a 5% error, but you can use any value from the range 1-10%. Remember, the smaller error the more accurate results you will get.
Calculate sample size
Now you can calculate your sample size. To make this task easier we created a table for three margin errors.
| Margin error | |||
| Population size | 1% | 5% | 10% |
| 500 | 475 | 217 | 81 |
| 1.000 | 906 | 278 | 88 |
| 5.000 | 3288 | 357 | 94 |
| 10.000 | 4899 | 370 | 95 |
| 50.000 | 8057 | 381 | 96 |
| 100.000 | 8762 | 383 | 96 |
| 500.000 | 9423 | 384 | 96 |
| 1.000.000 | 9513 | 384 | 96 |
| 5.000.000 | 9586 | 384 | 96 |
Please treat these values as a guideline. You can round up your sample size to the nearest tens, hundreds, or thousands.
Do you wonder how we have calculated values in the table above? Below you will find a formula (for finite population) and explanations that will allow you to calculate the sample size for any population.
Formula
n = P*(1-P) / (e^2/z^2) / (P*(1-P)/N)
Explanations
n – sample size, in other words, how many responses you need to collect
N – size of the population
P – estimated proportion in the population (if you don’t know what to enter use 50%)
e – margin error (in the table above those are values 1%, 5%, and 10%)
Z – Z value, which is connected to the selected confidence level. Most of the researchers use a 95% confidence level. The Z value of 95% is equal to 1.96.
Estimate response rate
Remember that your sample is people who actually completed surveys, not people who received an invitation to participate in a survey. Those numbers can be significantly different. Therefore it is important to estimate response rate and take it into account as you plan your survey.
The response rate is a relation between the number of people that completed survey and number of sent invitations. The good result in online survey research is 20-30%.
NOTE. If you are using the SurveyLab survey panel, you will pay only for delivered responses. This means that we will take care of proper response rate and collection of a requested number of responses, so you can relax.
Estimate how many people do you need for your market research
Now you have all dates to calculate how many people you need for your survey research. You need to divide sample size by response rate. The number you will get is the number of survey invitations you need to send in order to collect a proper number of responses.
sample size / response rate
Example. If your target population counts 500000 people and you are willing to accept 5% statistical error, then for your research you will need 384 / 20% = 1920 people. You can round up this number to 2000 people.