Friday, January 04, 2008

Would you answer this question honestly?

Synovate In:fact June 2005
Interesting survey done on the honesty of survey respondents (and also indirectly, a test on how accurate surveys are)...

---------------------------------------------------------------------------

Do people respond to survey questions honestly? This is a critical question to anyone who commissions market research. After all, what use are surveys if a lot of the respondents are great big fibbers?

Synovate recently set out to test the honesty of 2,640 survey panellists in the US, Germany and Thailand.


Hook 'em up to the polygraph?

It's not much use to simply ask someone, "Are you telling the truth?"

Fortunately, market researchers have developed a range of devious ways to ferret out honest answers. The technique used in this case is called "projective questioning".

This involves asking subjects how they think other people will react to a particular situation. The answer actually reveals the subject's likely behaviour.


I cannot tell a lie

In general, Synovate found that US respondents are more likely to provide truthful answers than Germans and Thais.

We learned this by asking neutral projective questions. For example, "How likely is it that most panellists would share answers honestly about the number of research surveys they have taken part in over the previous three months?"

Of American respondents, 82% indicated we could expect honest answers to this question, versus 68% of Germans and 62% of Thais.


Benchmarking honesty

That ratio may be regarded as the standard cultural difference between the honesty of answers from Thais, Germans and Americans.

Of course, the survey topic under discussion has a big influence over the degree of honesty that can be expected. In general, our study identifies three types of issues: Safe Issues, Careful Issues and No-go Issues.


No Pinocchio nose grows

SAFE ISSUES are those that elicit an honest response from most respondents most of the time. Surveys about daily activities such as television viewing and shopping can be considered safe.

But regional differences often apply. While a person's average weekly amount of exercise is generally a safe survey issue everywhere, honest answers are significantly more likely from Thais (78%) than from Americans (54%) and Germans (51%). This also applies to questions about religious worship.

On the other hand, the Americans and Germans are more likely to be truthful about impulse shopping than the Thais. And Americans are especially honest about whether or not they hang out at discount stores.


Economical with the truth

CAREFUL ISSUES are topics that moderately elicit an honest response. For example, when it comes to personal finances (except for discussions about charitable contributions) an equal number of people would provide honest answers as not.

In surveys about the consumption of stimulants, alcohol and illegal drugs, the honesty of answers depends on the social acceptability of the substance under discussion. So 58% of Americans would be honest about cigarettes, 37% about alcohol and just 14% about illegal drugs. The Thais are more frank about alcohol and drugs, the Germans, less.

There's also much country-based diversification about ethical issues such as sex education, abortion and gun ownership. While 75% of Americans will offer honest opinions about sex education in schools, you'll get frankness on this subject from only 62% of Germans and 55% of Thais.


Liar, liar, questionnaire on fire

NO-GO ISSUES are the unmentionables of survey topics.

Synovate learned that at least 60% of all our respondents would lie about sexual relationships, especially when it comes to taboo subjects like marital infidelity and sexual dysfunction.

Issues such as sexual orientation and the quality of marriage may appear safe, but in fact would receive only 44% honest answers in the US and as little as 23% in Germany.


Lies, damned lies, and surveys

OK, some respondents don't tell the truth — does this mean that surveys are inaccurate? Thankfully no. Studies like this one allow the honesty of responses to be calibrated against topic and demographic.

You may ask, "But can that calibration be accurate if the respondents don't answer survey calibration questions honestly?"

Actually, the degree of inaccuracy is insignificant — marketers can rely on the results of a carefully designed and calibrated-for-honesty survey.

No comments: