Experimenter differences in cognitive correlates of paranormal belief and in PSI

Journal of Parapsychology, The, Dec, 2002 by Caroline Watt, Richard Wiseman

Experimenter effects, that is, experimenters' influence on their research participants and subsequently on the outcome of their research, have a long history in psychology (e.g., Harris & Rosenthal, 1985; Rosenthal, 1967, 1976, 1990; Rosenthal & Rubin, 1978). Many different factors may play a role in experimenter effects, such as expectancy and belief (e.g., Luborsky et al., 1999) and personality (e.g., Cooper & Hazelrigg, 1988; Hazelrigg, Cooper, & Strathman, 1991).

The present study builds on this work by exploring the possible existence of experimenter effects within two important areas: (a) the psychology of belief in the paranormal and (b) psi research. The experiment is made up of two strands, and the background to each aspect of the study is described in turn.

COGNITIVE CORRELATES OF BELIEF IN THE PARANORMAL

Some psychologists have examined whether belief in the paranormal correlates with performance on various cognitive tasks, including critical thinking, reasoning skills, and IQ (see reviews by French, 1992; Irwin, 1991, 1993, 1999). Although findings are mixed, the overall trend is that people who disbelieve in the paranormal tend to outperform believers on these tasks (Blackmore & Troscianko, 1985; Roberts & Seager, 1999; Wiezerbiki, 1985), and Irwin named this the cognitive deficits hypothesis. However, many of these studies have been conducted by experimenters who are skeptical about the existence of the paranormal (e.g., Alcock & Otis, 1980), and some experimenters who are more open to the notion of ESP have failed to replicate these results (e.g., Irwin, 1991; Roe, 1999). This has led some researchers (e.g., Irwin, 1991; Smith, Foster, & Stovin, 1998) to hypothesise that these inconsistent results may be due, at least in part, to the experimental context influencing participants' reports concerning their level of belief in the paranormal and/or their performance on cognitive tasks.

Research investigating how the belief--cognitive ability relationship may be influenced by experimental context has had mixed results. Merla-Ramos (2000) found that paranormal believers tended to have poorer syllogistic reasoning ability than disbelievers, but only for syllogisms that included a paranormal or religious content. No belief--reasoning correlation was found for syllogisms that had a neutral content. Irwin (1991) used a similar syllogistic reasoning task, including neutral, proparanormal, and antiparanormal syllogisms. Overall he found a nonsignificant syllogisms--belief correlation, and the content of the syllogisms did not affect the correlation. Smith et al. (1998) varied the context in which participants completed a paranormal belief questionnaire and did an intelligence test (Raven's Advanced Progressive Matrices), with some participants being given an ESP-supportive context and some an ESP-unsupportive context. This study found that participants expressed greater paranormal belief in the ES P-supportive condition. Overall a negative correlation was found between belief and performance on the matrices task. However, contrary to prediction, context did not appear to affect this correlation.

What are we to make of this inconsistent body of findings? Many of these experiments may differ in the kinds of participant population they are drawing upon, in the kinds of procedures they use, and in the experimental setting. In addition, all of these studies used a single experimenter, and it is possible that differences in the experimenter's characteristics, from one study to another, may be a factor in producing inconsistent results. For instance, experimenter--participant dynamics may vary depending on the gender of the experimenter, his or her personality, his or her beliefs about the paranormal, or a host of other possible factors. With such a complex web of factors, it is difficult to begin to untangle them to answer the question of which may be important in eliciting certain patterns of results.

The present study attempts to take a first step towards untangling the web by asking a very simple question: When one keeps constant the participant population, experimental setting, and procedures but uses two different experimenters, is it possible for the two experimenters to obtain different patterns of results? Stated another way, the present study investigates the proposition that inconsistent findings in the study of cognitive correlates of paranormal belief may be attributed to (as yet unidentified) differences between experimenters.

In the present study, two different experimenters (CW and RW) carry out a joint experiment examining whether belief in the paranormal correlates with performance on two cognitive tasks. To build on the previous literature reviewed earlier, the two tasks selected to assess cognitive ability are a syllogistic reasoning task (as used by Irwin, 1991) and Raven's Progressive Matrices (as used by Smith et al., 1998). To facilitate comparisons with these earlier studies, we also use the same belief questionnaire as used by Irwin and Smith et al., and we use correlational analyses as they did. Hypothesis 1 is therefore that the experimenters will differ significantly in the correlation each finds between cognitive ability and paranormal belief.