Wednesday, 23 January 2008

Just how representative are the people who volunteer for psychology experiments?

People who volunteer for psychology experiments are more stable and outgoing than those who don't - a finding that has wide-ranging implications for the integrity of psychological research.

Jan-Erik Lonnqvist and colleagues in Finland approached dozens of military officers who had completed mandatory personality tests three years earlier as part of the army recruitment process. The researchers mailed 158 of these men a survey on values and, as an incentive, told them they'd be given feedback on how their answers compared to the general population.

According to their earlier personality test results, the 61 officers who returned the new survey were lower in the Big Five personality dimension of Neuroticism and higher in Conscientiousness, and they also showed a tendency to be higher in Extraversion and Agreeableness, than did the officers who didn't take part in the new survey.

In a second study, the researchers used data gathered as part of a larger epidemiological survey. Siblings from 15 families assessed the personality of their brothers and sisters and were also asked to volunteer for further neuropsychological tests and interviews. Consistent with the study of military officers, the sibling ratings revealed that the 55 participants who volunteered for the further stage of the study scored higher on Conscientiousness, Extraversion and Agreeableness and lower on Neuroticism, than did the 29 who declined to participate further.

The researchers said their findings have important implications for psychology research. For example, it's been shown that people, like those choosing to volunteer, who are lower in Neuroticism, are more likely to show a positive response to drug treatments for depression and panic disorder. Also, when someone completes a personality test, for example as part of a job application process, the idea is that their score is compared against a population average, but this study suggests the personality averages (or norms) that have been calculated for tests are likely to be skewed because they're based on the scores of volunteers.

To combat these problems when recruiting participants, Lonnqvist and colleagues said researchers should "make the research as attractive as possible to potential volunteers" and "attempt to evaluate the representativeness of the volunteer sample against the relevant population on the variables of interest."

Lonnqvist, J-E., Paunonen, S., Verkasalo, M., Leikas, S., Annamari, T-H. & Lonnqvist, J. (2007). Personality characteristics of research volunteers. European Journal of Personality, 21, 1017-1030.

Post written by Christian Jarrett (@psych_writer) for the BPS Research Digest.


Rana said...

I agree withe basic hypothesis, that we should attempt to evaluate the representativeness of the volunteer sample against the relevant population on the variables of interest.

Bu they also said "make the research as attractive as possible to potential volunteers"
But surely that is just going to attract a different class of volunteer? We need to have more experiments where the subjects do not know they are being tested ... Not easy at all. Perhaps post-experiment consent, perhaps non-intrusive tests and depersonalised results. Of course I don't have the answer, but it suggests more emphasis on statistical analysis of actual history and less emphasis on volunteering.

Anonymous said...

I doubt we will easily go back to 'deception' methods where an experiement could commence without prior informed consent.

Post a Comment

Note: only a member of this blog may post a comment.