Know-How Exchange : Research/Metrics

I'd suggest starting with a simpler survey to gauge response and only invite those that spent sufficient time on the survey to do the in-depth one (pre-qualify the respondents). It's quite possible that those that are completing it quickly aren't the target research audience.

Consider offering a prize for those with most complete survey. Give the respondents a reason to care about the survey.

If you are already using the time taken to fill in a survey, also look at the other end of the scale – someone who has taken a day to complete details without being forced to log off and back on is unlikely to be giving your question much attention.

I would also look at what I term “Parity” questions (After parity bits which remove errors in data transmission)

These are questions designed to spot that the answers being given are either random or made-up. They might reveal contrary views on the same point for example. They differ from red-herring questions by degree of association i.e. they don’t just test go / no-go on whether a respondent is making things up.

This is not so simple as it sounds and some marketing giants have invested heavily in the science behind such questions, spending over $500,000 on both devising and testing a survey, but I have put the thesis to companies which look at the psychology and human reaction to research issues such as Corr Willbourn in the UK and I believe that smaller scale studies are equally practical for a fraction of this budget – if they are well targeted and researched on scientific behavioural marketing models.

I chose Corr Willbourn because they rather intelligently carried out impartial market research into the acceptability of GM in the food chain and did some excellent work to avoid the results being too weighted by vested opinion holders both from the pro and anti GM lobby.

The practice is akin to the more controversial method of using multiple, possibly conflicting questions to conduct a psychometric test on a candidate’s suitability to a given role.

Best wishes

Steve Alker
Xspirt

Hi Christian,

Another possibility, is to include questions where you are asking the respondent to enter a sequence of numbers or letter (or to choose the sequence from a set of multiple choice questions) a few times in the survey. This is usually something that can help to make sure they are being attentive.

One other thought - 15-20 minutes is a bit on the long end of surveys - you might be getting people going through quickly because they are getting fatigue and just trying to go through it quickly. The number of pages seems fine, but the number of minutes for completion might be a little excessive.