SURVEY MYTHS DEBUNKED - People Do Not Participate In Online Surveys
It’s true, isn’t it? There are just too many online surveys going on. Chances are, there’s one in your inbox on a regular basis. And when did you last see a website that did not have a section asking you to give feedback through a brief survey?
So you can imagine most people think it’s almost impossible to get a good response rate.
But it’s not. Marketers, business planners, academics, and government agencies routinely use online surveys, and they get lots of responses from all sorts of respondents. Even from the typically busy types like doctors and stock exchange brokers.
So what’s up with that?
Why It Seems Like Online Surveys Get Few Respondents
One reason why online surveys get a bad reputation is because there are, in fact, too many of them happening at any one time. But – and this is where the confusion comes in – these are usually just one type of online surveys. They are what is known in the trade as surveys that rely on the self-selecting sampling method. They are surveys open to anyone who feels like participating, and they are usually only about interacting with customers, not necessarily getting concise research results.
Another reason is that online surveys are the most visible type of survey method. As Internet access increases globally, online surveys are becoming more and more common. But the challenge of not getting responses from all targeted respondents is not unique to online surveys. It’s a problem faced by all types of survey methods, like phone, face-to-face interviews, and focus groups. In fact, it’s a standard hazard of any survey, and there are even scientific techniques to anticipate and deal with it.
Which brings us to the third reason: online surveys seem unpopular with respondents because most people only get to see the type of online surveys that are not made to perform well. In other words, there are lots of other online surveys taking place that have high response rates. It’s just that those surveys are not all over the place.
How To Get Higher Response Rates
Even before online surveys came along, professional researchers had identified the problem of surveys not getting enough responses. The professional term for the problem is non-response bias. It means that everyone conducting a survey has to factor in the possibility that some of the people targeted in your sample may refuse to respond, and that the survey results could be seriously compromised if most non-respondents share a particular perspective. As an example as a bad result due to a non-response bias consider the following: if your survey is about the quality of your product and most people who do not like your product fail to respond, you might be misled to think that your product is better perceived than it actually is.
So, naturally, experts in research surveys have designed measures to minimize non-response bias. A standard technique is to do proper sampling as explained in our article on The Effect of Sample Size On a Survey's Results. In short, good sampling greatly reduces the number of your non-response bias because it targets not just fewer people, but also people who are more likely to respond to the survey.
Another approach is to give respondents an incentive to respond to the survey. This may be anything from cash incentives, to discounts on online purchases or gift certificates from major online retailers such as Amazon. Of course, online surveys with an incentive component are not mass-emailed to a million respondents. They are more targeted and limited in exposure, which is easy with proper survey sampling.
What it all comes down to is that professionally designed online surveys get high rates of response. They rarely get 100% response rates, although it is important to note that few surveys – online or offline – achieve that feat. But properly done online surveys do get response rates that are no worse than those of other offline survey methods.
What To Do When You Have a Low Response Rate
But what can you do if your online survey has a high non-response bias? Not to worry, there are four techniques of ensuring non-responses do not compromise the quality of your survey:
First, surprisingly, is to do nothing. But this option is only valid for certain types of surveys, such as pilot surveys that are meant to help in the planning of a larger, more elaborate survey, or in surveys that, by their nature, are harder to restrict to a set sample of respondents, such as customer feedback or exploratory surveys, because for those kinds you would usually prefer to have a lot of different opinions, and sometimes you just cannot guess who has which opinions.
Option two is to compare the number and types of those who responded against the ones that didn’t respondent. A careful analysis of the two groups – on the criteria set during sampling such as age, area of residence, income and education levels - can reveal the reasons why some people responded and others did not. It may also give an indication of what perspective the non-respondents hold, which can then be factored in to the surveys to limit the non-response bias.
Third is to compare the number and types of respondents who responded quickly and those that took much longer. Again the comparison is made on the criteria set at the sampling stage. The logic here (which is still evolving in research survey studies) is that there usually is a co-relation between those who respond late and those that fail to respond at all. Apparently, those who respond late tend to be closer in perspective to the non-respondents.
Finally, a follow-up on the non-respondents can greatly improve the response rate by having the respondents answer the survey after all. Sometimes it just takes a couple of reminders to get them to do it.
As is so often the case, in the end it all depends on a couple of factors such as the target of your survey, the type of survey, and the thoroughness with which you want to survey. If you’d like to check what would work best for you, let us know and we can help.
Is there something about surveys that you think is confusing or often done wrong? Give us a shout in the comments and we will do our best to help clear it up!
Photo credit: Matt Cornock (thanks, Matt)
Feb 07, 2014