Dr. Sven van de Wetering is an Associate Professor of Psychology at the University of the Fraser Valley, Canada. His research interests are in “Conservation Psychology, lay conceptions of evil, relationships between personality variables and political attitudes.” In a 4-part interview series, we explore the philosophical foundations of psychology.
Scott Douglas Jacobsen: What is the epistemology underlying statistics in psychology? Where does psychology begin to find its statistical limits?
Dr. Sven van de Wetering: I think the more or less explicit epistemological assumption underlying the use of statistics in psychology comes right out of Skinner and his notion that the human organism can be thought of as a locus of variables. In other words, human cognitive, emotional, and behavioural propensities can be meaningfully studied as dimensions that can be expressed numerically, as can environmental events likely to influence those propensities. Furthermore, the task of psychology is conceived of as being to figure out ways of measuring those underlying variables and of inferring how they influence one another. We depart from Skinner, though, in rejecting his absurd claim that one can explain all variability, that the concept of error variance is meaningless. Because error variance is a fundamental feature of the complexity of human organisms, and the even more complex environment in which they operate, inferential statistics then become an important tool to separate incorrect hypotheses from correct ones. Also important in all this is the assumption that human beings are very good at finding patterns in any sort of data, including pure noise, and that safeguards are needed to prevent us from inferring patterns where none exist. Human beings are seen as very fallible creatures, and inferential statistics are seen as safeguards against that fallibility.
“We depart from Skinner, though, in rejecting his absurd claim that one can explain all variability, that the concept of error variance is meaningless.”
SJ: What are some of the most embarrassing examples of statistical over-extension in psychology studies ?
Dr Sven van: I’m not sure, I routinely get embarrassed by over- or misapplication of statistics, but I do sometimes think people don’t know what inferential statistics means. Two patterns frequently bother me, though I can’t think of particular examples off the top of my head. One is people who conduct a study with a small sample size, fail to find a statistically reliable difference between treatment groups, and then blithely proclaim that the null hypothesis is true, as if the study’s lack of statistical power is some sort of virtue. The second pattern is almost the opposite of the first: people who conduct studies with enormous sample sizes, find a statistically reliable difference between groups, and then trumpet the finding as an important one. They don’t bother to report effect sizes, probably because to do so would be to acknowledge that the effect they have found, though statistically reliable, is too small to have a lot of real-world significance.
SJ: We did some preliminary work in an interesting area, environmental psychology. You have an expertise in political psychology. How can statistical knowledge about political psychology influence knowledge around issues of environmental psychology, e.g. climate change denial – as opposed to scepticism?
Dr Sven van: Many people who are very concerned about anthropogenic climate change are baffled by the large numbers of people who deny that human actions are having an appreciable effect on the Earth’s climate. The scientific evidence appears to be so overwhelming to those who accept it (not that most of them have read much of it) that the only explanation that they can fathom for climate change denial-ism is that it is rooted in sheer ignorance of the scientific facts. Statistically, though, scientific ignorance does not appear to be a major factor in climate change denial-ism, given that the correlation between belief in anthropogenic climate change and general scientific literacy is close to zero. Instead, we find an extremely strong correlation between belief in anthropogenic climate change and measures of ideology. In the US, people who strongly identify with the Republican Party or who self-identify as very right-wing are very likely to deny that human actions are responsible for changes in climate, regardless of how much they know about science in general or climate science in particular.
SJ: The statistical approaches often come in conjunction with “folk psychology.” So, some Folk psychological explanations for a phenomenon exist, then they either become supported or not through scientific studies. Why is this the basis of lots of research? How is it weak? How is it robust?
Dr Sven van: We use folk psychology as a heuristic because we don’t really have standardised procedures for hypothesis generation. If we don’t have a formal theory that acts as a source of research hypotheses, then informal theories (i.e. folk psychology) are the next best thing. The primary strength and primary weakness of folk psychological theories are the same, namely that they are fairly easy for us to understand with our limited cognitive apparatuses. This is a strength because theory is always under-determined by data, so if multiple theories are possible, we might as well go with the ones that are easy to understand. This is a weakness because there is no a priori reason to believe that true theories of human psychological functioning are easily comprehensible. An example of this is connectionist modelling of human cognition. Connectionism has some pretty substantial explanatory successes to its credit, but has not caught on as well as might be expected just because it is so absurdly non-intuitive that nobody really has a good gut sense of what connectionist models are actually asserting.
Jacobsen: Thank you for your time, Sven – pleasure as always.
Read Q & A of Session 1 with Dr Sven van de Wetering here
Read Q & A of Session 2 with Dr Sven van de Wetering here