Study: Re-use of respondents for tracking surveys in the modern world

Monday, 8 August 2005

By Jens N. Laugesen, Research Manager, Zapera.com A/S

In the old days, samples were drawn on the entire population why there was only little concern that the same respondent would participate in two consecutive tracking surveys on the same product. The simple reason is that the population is normally considered large why the probability of reaching the same person twice in the same tracking is considered very small.

In modern online market research access panels replace the population as sample basis. An access panel is a previously recruited pool of people who have agreed to take part in surveys by filling in a background questionnaire. The background information allows the researcher to sample respondents for various surveys. As access panels are smaller than the entire population the probability of reaching the same respondent in two consecutive tracking surveys is larger.

Thus one concern related to data quality in modern research is the re-use of respondents in tracking surveys – i.e. how long should an exclusion period be before a respondent qualifies to participate in the same tracking again.

No correlation found

A recent study on the subject has been carried out on an established tracking survey that Zapera operates.

The set-up is a weekly tracking in Denmark. Three different samples were tested;

1. a normal sample of 200 new respondents
2. a control sample of 100 respondents that participated between two and five months ago
3. a control sample of 100 respondents that participated less than two months ago

No demographical differences between the groups were present.

Two areas were investigated;

1. Helped awareness (i.e. “Do you know or have you heard about X?”)
2. Evaluation (i.e. “How would you evaluate X?”).

The result is very clear.

There is no evidence that a respondent’s answers are correlated between two tracking surveys.

The results of the normal sample of new respondents was statistically tested against the results of the two control samples; both as two separate groups and as one common group. No difference was detected on either measure. This means that even though you answered the same questions a couple of months ago your answers today are not correlated to your previous ones.

Other studies give similar evidence

In the Nordic area Zapera made a similar study three years ago with similar conclusions.

Another recent study was carried out in Great Britain. The brand, advertising and media awareness for travel companies were questioned. Four different samples were interviewed; a sample of new respondents that was compared with samples of respondents participating in the tracking respectively one, three and five months ago. The survey gave no evidence that prior exposure to the survey has made the respondents more aware of the subject three months after.

If you are interested in learning more about the results please contact Statistical Manager Jakob Preisler