Terence Hines - Pseudoscience and the Paranormal

Regarding the physiological reasons for acupuncture’s antipain effects, it is commonplace to attribute these effects to the release of endorphins, peptides secreted by the brain that modulate pain perception. However, the research on this issue is unclear. Endorphins may play a role, but other neurotransmitter systems are also probably involved (White 1999).

Does prayer of one person for the health of another (intercessory prayer) result in any improvement in the condition of the prayed-for person? Somewhat surprisingly, empirical research on this question dates back more than 125 years. What was apparently the first study of this question was conducted by Sir Francis Galton, one of the founders of modern statistical analysis (Tankard 1984). Galton (1872) reasoned that members of the clergy prayed more (and were probably prayed for more) than those in medical and legal professions. To the extent that prayer was effective, clergymen should live longer than doctors or lawyers. He found no difference in the life spans of members of these three professions.

More modern studies have provided little additional support for the positive effects of prayer. In their review of the research in this area, Matthews, Conti, and Christ (2000) discuss eight empirical studies of prayer effects. Six found no effects of prayer. One (Green 1993) found that for subjects who had a strong belief that prayer would reduce anxiety, being prayed for reduced their anxiety. The eighth study (Byrd 1988) is often cited as providing supportive evidence for real effects of prayer. In this study, patients suffering from cardiac disease were prayed for. This study suffers from two serious flaws. First, in the original publication it was claimed that the study was done in a double-blind fashion. It wasn’t. Tessman and Tessman (2000) have pointed out that Byrd assessed the degree to which patients had improved after he became aware of whether or not they had been prayed for. Second, Byrd examined thirty-two different variables for prayer effects and found such effects on only six variables. The pattern of results can best be interpreted as a statistical fluke. The more variables one examines, the more likely it is that, just by chance, some type of significant effect will be found.

Conti and Matthews (reported in Matthews, Conti, and Christ 2000) performed a clever study to tease apart the effects of patients’ expectations about prayer effects from any actual prayer effects. They used as subjects twenty-five patients with renal disease. Two groups of patients were told that they would be prayed for. One group was, in fact, prayed for, while the other was not. Two more groups were told that they would not be prayed for. One, contrary to expectation, was prayed for and the other was not. The results showed that whether or not patients were actually prayed for had no effect on outcome measures. However, patients who thought that they were being prayed for improved more than those who did not think they were being prayed for, regardless of whether or not they really were being prayed for.

Another study of prayer effects is that of Harris et al. (1999). These authors used as subjects more than one thousand patients admitted to a coronary care unit (CCU). The prayers for the prayed for group asked specifically for a fast recovery and a lack of complications. There was no difference between the prayed-for and not-prayed-for groups in terms of either the time spent in the CCU or the total time spent in the hospital. Two other, more general measures of recovery were also examined. On one there was a slightly significant difference favoring the prayed-for group. There was no difference on the second general measure which, interestingly, was one of the same measures on which Byrd (1998) claimed to find a benefit for his prayed-for group.

Even if this was all there was to this study, the finding of one barely significant difference out of three would not be a very impressive result. But there is another interesting feature of this study. Humphrey (2000) points out that the Harris et al. (1999) study started with a total of 1,013 patients, of which 484 were assigned to be prayed for and 529 were assigned not to be prayed for. However, because it took a day to begin prayer after a patient was assigned to a group, twenty-three patients who stayed in the CCU for less than twenty-four hours were eliminated from the study. But these eliminations were far from random. Of the twenty-three, eighteen had been assigned to the prayer group and five to the non-prayer group. This is a highly significant difference. Humphrey notes that there are three possible explanations for this effect. The first is that prayer shows “backward causation”, working on patients assigned to the prayer group before anyone actually starts praying for them. A second, more mundane possibility is that assignment to the two groups was not random, with the less severe cases being placed in the prayer group. But there is another reason that a patient can leave the CCU rapidly: He can die. So it is possible that there was a higher immediate death rate among the to-be-prayed-for patients. Humphrey asked Harris about this possibility but, at the time of this writing, Harris has not performed this analysis (Humphrey, personal communication, Nov. 28, 2000).

Cha, Wirth, and Lopo (2001) reported that prayer had a highly significant effect on the success of in-vitro fertilization. The success rate was 50 percent in a group that was (sort of; see below) prayed for, compared to 26 percent in a control group. However, the prayer used was extraordinarily vague—that God’s will be done. With that as the prayer, even a finding in which there were no differences between the groups could be described as supporting the effectiveness of prayer—it just happened that God’s will was for there to be no difference between the groups. And if the sort-of prayed-for group had had a lower success rate, that could have been seen as God’s will as well. In addition, as Flamm (2002) has noted, the prayers weren’t really for the success of the fertilization. The highly complex, even bizarre, experimental design had successive groups of people praying for the success of the prayers of the previous groups of people praying. The obvious way of designing the study would have been to have one group prayed for with an obvious prayer for successful fertilization, and the other not. I suspect this finding is simply one of those statistical flukes that we all run into in our data from time to time (about one in 20 times, to be precise!).

The term collective delusion , also known as mass hysteria, describes a situation in which a significant part of the population of an area, which can be as small as a single building or as large as a nation, becomes convinced that some strange event is taking place for which there is no immediately obvious explanation. The event—sometimes an outbreak of illness occurring in rapid succession among people living or working in the same environment—can be attributed to a wide range of causes. Sometimes paranormal or pseudoscientific causes are proposed and accepted. In many, but not all, cases of collective delusion, the media play an important role in spreading the delusion. See Bartholomew (2001) for an excellent book-length treatment of this topic.

A prototypical case of collective delusion, described by Medalia and Larsen (1958), took place in Seattle, Washington, in March and April of 1954. At first a few—and then more and more—people noticed mysterious tiny pits in their car windshields. Anything that could pit glass, it was reasoned with some justification, could certainly do damage to frail human flesh. Concern grew, as did the number of reports. Explanations for the pits were varied and creative. One held that acid pollution was responsible. Another held that fallout from atomic bomb tests in the Pacific, blown east and falling on Seattle, was causing the damage. On April 15 the mayor of Seattle asked for the assistance of the governor of Washington and the president of the United States. Clearly, Seattle was facing a dangerous situation. Or was it? The pits turned out to have a prosaic explanation: They had been caused by pebbles thrown up from the unpaved roads by cars, which then struck the windshield of any car behind. They had simply not been noticed before. Yet when one person noticed them and pointed them out to someone else—who also had not noticed them, but now found them on his or her car and assumed, incorrectly, that they had never been there before—the stage was set for the collective delusion to appear. Medalia and Larsen (1958, p. 180) note that people in this particular episode came up with “evidence” to support what turned out to be clearly incorrect explanations. In the case of the atomic fallout theory, “many drivers claimed that they found tiny, metallic-looking particles about the size of a pinhead on their windows.” Other examples of collective delusions involve the sudden outbreak of a mysterious illness, the symptoms of which are nonspecific, such as vomiting, headache, shortness of breath, or fainting (Colligan, Pennebaker, and Murphy 1982). Small and Borus (1983) present a detailed study of one such case, in which the victims of a mysterious illness were schoolchildren in a small Massachusetts town. At rehearsal for a concert children grew ill. Later, during the actual concert, more children became ill. The illness was at first attributed to environmental pollution, but, as Small and Borus show, it was actually a case of mass hysteria.