Soft subjects suffer low esteem

Scientists and the public share doubts about research, but findings in social science are least respected, say Richard Wiseman and Caroline Watt

A few weeks ago we conducted a survey of delegates and speakers at the British Association Annual Festival of Science to see how much confidence they have in science. The results were surprising and worrying.

We first asked 100 members of the public to indicate their "confidence in the validity of scientists' findings and conclusions" in the physical, biological and social sciences by circling a percentage between zero and 100. Overall, the public indicated a 70 per cent confidence for physical sciences, 62 per cent for biological and 48 per cent for social science. The figures suggest that there is an underlying doubt about findings in the physical and biological sciences and downright scepticism for the social sciences.

Next we presented the same questions to 100 scientists at the festival. This group had 79 per cent confidence in the physical sciences, 69 per cent in the biological sciences and a surprisingly low 44 per cent in the social sciences.

This low figure is not the result of those involved in physical and biological sciences being sceptical about the social sciences - the social scientists we surveyed had only 51 per cent confidence in their own discipline. In short, scientists were slightly more confident than the public in the physical and biological sciences, but less confident than the public about the social sciences. None of the confidence ratings is particularly high. The figures could be taken to indicate that scientists have a lack of confidence in about a quarter of findings in the physical and biological sciences, and are sceptical of more than half of findings in the social sciences.

Finally, we surveyed 100 "science communicators", including journalists, teachers and professional science communicators. They showed the same pattern as the other groups and had levels of confidence similar to those of the scientists.

The generally low confidence in science may be because people realise that science is a far from perfect activity. Any result may be contaminated by methodological bias, inappropriate statistical procedures or fraud. Also, the public is often exposed to contradictory findings and advice from experts in the media, which may lower public confidence in science.

But why such low confidence in social science? One possibility is that this reflects an understanding of the inherent unpredictability of human behaviour compared with physical systems. Although experimental methods may be carefully applied in behavioural research, the participants may respond in more complex ways, thus reducing the consistency of experimental findings. This means that "higher power" studies with more participants or data points are required to detect the existence of an effect compared with the other sciences. One recent review paper has found that social science research is plagued by low statistical power, making it difficult to replicate findings. Another possibility is that the results may be an honest admission of flaws in social science research, or indeed throughout the sciences, which social scientists are more willing than others to admit to.

Finally, we were able to poll 100 school and college and 100 university students. Combining these groups with our existing data from the professional scientists allowed us to examine what happens to confidence in each of the three areas of science as people progressed from being school pupils to university students and then professional scientists. Divergent trends appeared for the three disciplines. Confidence in physical sciences rose from 70 per cent to 79 per cent with increased academic involvement. Confidence held stable at about 69 per cent for biological sciences. For the social sciences, however, confidence fell from 53 to 44 per cent.

Perhaps the increasing confidence in physical sciences is a result of working within relatively well-elaborated theoretical systems in combination with the benefits of high status and associated funding advantages. For the social sciences, confidence may drop as an understanding grows of the complexities and difficulties of this kind of research.

The only good news is that the survey would count as social science, and so it seems likely that only about 50 per cent of people will take the findings seriously.

Richard Wiseman is professor of psychology at the University of Hertfordshire. Caroline Watt is a research fellow in the psychology department, Edinburgh University.