Your report on online market research (MW last week) gave a generally balanced view of the issues. It did not, however, identify one major concern – one that is widely overlooked when the merits and shortcomings of online surveys are debated.
Almost all online research is conducted using so-called “panels” of respondents. In contrast to what happens in most “conventional” research, these individuals effectively opt in, even though the better controlled panels seek to obtain their members from a variety of sources. It’s true you can balance the panel by various demographic characteristics, but only within the self-selecting group you start with.
Thus there are two stages of possible bias in the samples selected using online panels – the first due to the fact that perhaps only 60 per cent of people have internet access; and the possibly much greater one caused by the fact that only a small proportion of this 60 per cent are contactable by being members of a panel (or, in some cases, several panels!). We need to know much more about the extent of this second bias and how operators deal with it.
Given that online panels are certain to increase in importance, research users should be encouraged to challenge operators to demonstrate how they set about minimising the effects of this problem.
RSMB Audience Research