Top Menu

Employee Satisfaction Surveys: How to Remove Biases and Collect Proper Data

Wednesday, January 18, 2017

If your organization conducts employee satisfaction surveys, the quality of the resulting data is key to their success. That's why we wanted to take a look at a common issue that comes up in any type of survey or poll, and one that you might have even experienced yourself - non-response bias. 

Even if it's not something you've heard of before, it could still be hurting your surveys. With that in mind let's find out what it is, when it occurs, and the consequences it can have for your employee engagement survey results.

What is non-response bias?

As the name suggests, it occurs when people don't respond and there's a marked difference between the non-responses and the people who do complete the survey. For example, you could end up with only satisfied employees take part, or only dissatisfied employees take part. Either way, the data you'll end up with certainly won't be good data, will it?

The consequences of non-response bias

The consequence of non-response bias is, in simple terms, useless data. And useless data means your survey has been a waste of time, money, and resources. It's also dangerous if those at the top levels of an organization are led to believe their workforce is one way or the other - it means incorrect conclusions will be drawn and bad, even potentially dangerous, decisions will be made.            

What causes it?

There can be a number of reasons that employees don't respond to surveys, perhaps they're on long-term sick leave, or a failure in the technology used means that the request doesn't reach them or it's hard to complete, but the main reason is quite simply refusal to take part. An important thing to realize about employee satisfaction surveys is that the more dissatisfied an employee is, the less likely they are to want to get involved in surveys or initiatives run by their employers. 

What you can do to minimize non-response bias

First of all you need to decide whether a particular non-response bias is going to cause issues, because a non-response bias doesn't necessarily mean bad data, it depends on the type of bias. For example, we know that if the non-response bias is that satisfied employees are more likely to participate in your employee satisfaction survey than dissatisfied employees, it's something that's definitely going to be an issue. If, however, the non-response bias is that employees who wear glasses are more likely to participate, it's unlikely to have any impact on your data. 

So your aim is to make sure every respondent (or at least the majority, you're always going to get a few that don't respond for unforeseen reasons) completes the survey. How? Make it easy for them to physically complete the survey by using an automated and customizable online system like Primalogik 360, and test the process of sending out invitations to take part. Automated reminders are also a good way to ensure more participation. Perhaps offer incentives to employees who complete the survey, like entry into a prize draw, and be flexible with the time allowed to complete it.