Need to study Web survey addicts

16/May/2001

The event, entitled "The Challenge of the Internet", was held at Latimer, just outside London, by the Association for Survey Computing (ASC) in association with the Market Research Society, Social Research Association and Offfice for National Statistics. Poynter's keynote speech, "A Guide to Best Practice in Online Quantitative Research", covered a great deal of ground.

He pointed out that last year about 5% of market research in Western Europe and the US was conducted via the Internet. This was expected to rise to about 8% this year and reach 50% by 2004.

At the same time there was increasing concern about "professional" or "high-frequency" respondents. In the UK there was evidence that the average housewife interviewed for a telephone omnibus survey does 11 surveys a year. Anecdotally "we hear that 50% of all responses in the US come from just 4% of the population."

The problem could be much worse with online interviewing. "This is because there is evidence that some people are seeking out surveys to complete. There are a large number of websites (such as www.surveys4money.com, www.doughstreet.com, and www.rewardsites.com) that tell surfers where to find surveys from which they can earn money or win prizes.

"If you run a survey with a nice, attractive incentive and all people have to do is to go to a specific location to complete the survey, you will get some people who are there just for the incentive."

The problem was intensified by growing reliance on panels and e-mail lists. "Different panels have differing policies on how often they interview people, for AOL/Opinion Place and the Millward Brown IntelliQuest Tech Panel it is once every two weeks, for SSI it can be up to twice a week.

"However, none of the panel and e-mail list companies can control how many other panels and lists their members are on."

A survey by SSI had found that almost 90% of those who replied wanted to do at least one survey a week, with over 60% wanting to do more than that. Another investigation had revealed that 46% of ACNielsen/BASES panellists wanted to do surveys more frequently.

Two questions of particular concern were:

*Do people who respond to more surveys become more sensitised to certain issues and therefore less representative of the rest of the population?
*Do people tell the truth, particularly if they are being motivated by incentives?

Poynter set out the following typology of Internet respondents.

Fraudulents
These are people who complete a large number of surveys and tend to give incorrect answers, either to maximise their earnings or out of malice. When researchers have looked for these people, for example by searching panels for inconsistent responses, they have appeared to be few in number.

Professionals
High-frequency respondents who are motivated by incentives but tend to be truthful.

Volunteerists
High-frequency respondents who are motivated either by a desire to help, or because they find surveys fun! In a study carried out by SSI in March 2000 it was found that about 50% of respondents were Professionals and 50% were Volunteerists.

Rejectionists
People who do not want to do surveys. Some people are Rejectionists for all media. Others would reject an online interview but might take part in a store exit survey. There is evidence that the number of Rejectionists is steadily growing and also that, when talked into doing interviews, they are likely to be less honest.

Occasionals
The people researchers would like to reach more often. If asked to do a survey at the right time, on the right topic, in the right way, they will comply. Rare in panels and email-lists, they can be reached by techniques such as website pop-ups.

As part of the research that neded to be done to determine the reliability of different types of respondent, Poynter recommended that survey questionnaires should ask "In the last six months how many surveys have you taken part in?" and "What is the most important reason you complete surveys? Opportunity to influence decision makers? Rewards and Prizes? Fun? Interesting topics? To help the interviewer?"

Among many other points made by Poynter was that incentives were increasingly common and necessary on the Web since many respondents had to pay for their connect time. The consensus was that for most forms of interview, incentives increased response rates.

"However, small incentives and prizes that are not of universal appeal may produce no beneficial results."

Among the many types of incentive, including prize draws, Amazon coupons and Air Miles, one that appeared to meet with frequent success in business-to-business research was the promise to give the respondent an abridged copy of the survey results.

© Copyright Market Research News