How accurate is your data?
Nearly everyone who’s been involved in an online survey project will have been asked this question at one time or another. This is a particular issue when dealing with results that are very close, or contentious, or challenging to the interests and assumptions of the people to whom the data is being presented.
This need to be able to qualify the likely accuracy of the answers that are being inferred from datasets led to statisticians creating standard measures of how accurate any given result is likely to be, and how confident we can be of that fact.
If you want to calculate your sample size, check out our sample size calculator
How is it calculated?
Let’s look at the numbers that are fed into it. Click on the highlighted areas below for more details:
Margin of error = z-score x ( p ( 1 - p ) / n ) x ( N - n) / ( N - 1)
What the results mean
The two numbers, confidence level and margin of error, need to be considered together.
If you’re trying to make decisions or predictions based on your survey data, basically you want to make sure that your numbers are different by more than the margin of error. If you’ve got a data set that’s split 53-47 and the margin of error is 5%, then drawing a solid conclusion based on your data could be risky.
What to do when the answer is inside the margin of error
The only real way to reduce your margin of error is to increase your sample size. It may just be a matter of time, and that you need to run your survey for a longer period to collect more responses.
If you sent your survey via email, then you can try using email reminders to encourage more people to take part – varying the message used may help with this. Promoting the survey via new channels such as SMS or a Social Media link will help you reach more people. Consumer Panels offer the opportunity to tap into targeted potential respondents outside your own databases.