Delivering the perfect online survey. Does it exist or is it as mythical as dragons, fairies or the Cubs winning the World Series?
We can speculate, but as researchers we can also ensure we provide the best possible surveys for respondents. There are many methods to do this, much of it shared on qSample’s blog. One key way is to warrant that an online survey comprises certain aspects that make it valid.
According to Russell Renka, political science professor at Southern Missouri University, there are three aspects to a valid survey/poll. To exclude any of these, in Renka’s view, means the data will be dangerously flawed.
Here they are:
- The questions asked must be clear, written in neutral language and provide a range of answers to choose from.
This might be basic for marketers, but often unclear or even subconsciously-driven messages can tilt the survey towards flawed data. For example, a client’s survey utilizing our veterinarian panel once asked: “Do you think microchipping dogs might lead to a safer life for the animal?”
The problem was the obscure “might” in the question. The researcher was in essence vacillating. Therefore, the answer to the question would be compromised. (We made sure to bring this to his attention, of course).
In other words, be clear and concise and short. Furthermore, as we’ve promoted, include opt-out questions like “Don’t Know,” “Not Sure” or “Undecided.”
Your panel will thank you with richer data and not microchip your research with failure.
- The respondents must be randomly selected.
Yes, more Captain Obvious information, it seems. Still, marketers often penny pinch and end up drowning in river sample (in fact, Renka denounces a market research company in the cited article, the chief reason he detailed the three aspects of valid online surveys).
Detail must be placed on a questionnaire, indeed, but it also must be placed on the quality of the sampling. Beyond a good online survey provider, double-check the persona of your respondents.
We consider qSample a quality sample provider, of course. Yet there are other companies who specialize in niche panels. They will all take care of your efforts, even if it’s slightly more strain on the proverbial budget.
- The sample involved must be large enough to keep the margin of error fairly small, about 5 percent.
According to Renka, that should be at least 400 respondents. At 300 respondents, the margin of error grows to 5.6%. Obviously, certain respondent demographics and certain budgets necessitate a sample to be under 400 respondents. But all things being equal, keep your respondent numbers at a healthy size.
There are other aspects (or safeguards) to quality good data in online research, and these may include:
– Quality of questionnaire
– Quality of project execution
– Quality of analysis
We agree with Renka in the end that these three simple aspects can go a long way in making sure your market research doesn’t get more complicated that it needs to be. It may not be perfection, but it’s probably closer to it than the Cubs winning the World Series.