Literature Review (English Writing)

sagepub.com/journalsPermissions.nav DOI: 10.1177/1075547010389819 http://scx.sagepub.com 389819 SCX 33 3 10.1177/107554701 0389819 AllumScience Communication © 2011 SAGE Publications Reprints and permission: http://www.

sagepub.com/journalsPermissions.nav 1University of Essex, Colchester, UK Corresponding Author:

Nick Allum, University of Essex, Wivenhoe Park, Colchester, UK CO4 3SQ Email: [email protected] What Makes Some People Think Astrology Is Scientific?

Nick Allum 1 Abstract Citizens in both North America and Europe are apt to read horoscope columns in newspapers and magazines. While some people read these casu- ally and purely for entertainment, some believe that astrology has scien\ tific status and can provide real insight into events and personality. Using d\ ata from a European survey, this article explores some of the reasons why so\ me people think that astrology is scientific and how astrology is viewed in\ rela- tion to other knowledge-producing practices. Three hypotheses in particu\ lar are tested. The first is that some Europeans lack the necessary scientif\ ic literacy to distinguish science from pseudoscience. The second is that p\ eo- ple are confused about what astrology actually is. The third is derived \ from Adorno’s work on authoritarianism and the occult and postulates that \ those who adhere to authoritarian values are more likely to believe in astrolo\ gical claims. Support is found for all three hypotheses.

Keywords quantitative analysis, public understanding of science, pseudoscience, a\ strology, authoritarianism Introduction Anyone reading this article is likely at some point to have read their horo- scope. Astrology columns are widespread in print media and on the Intern\ et 342 Science Communication 33(3) and have been a staple for a surprisingly long time. One of the earliest\ recorded columnists was 17th-century astrologer William Lilly, who may have predicted the Great Fire of London, albeit 14 years early (Curry, \ 1989).

Webster-Merriam dictionary defines astrology as “divination of the su\ pposed influences of the stars and planets on human affairs and terrestrial eve\ nts by their positions and aspects.” A horoscope, on the other hand, is defined as a “diagram of the relative positions of planets and signs of the zodiac\ at a specific time (as at one’s birth) for use by astrologers in inferri\ ng individual character and personality traits and in foretelling events of a person’\ s life.” The more common understanding of horoscopes is that they are astrologica\ l forecasts, such as those that appear in newspapers. It is this definitio\ n that I use for the rest of the article. Ten years ago, just less than half of A\ mericans read their horoscope at least occasionally in 1999 (National Science Bo\ ard, 2000), and there is little reason to think that the numbers have declin\ ed since then. It is one thing to read an astrology column for amusement or entertainme\ nt but quite another to believe that astrological predictions about events \ or per- sonality will come true. A surprisingly large quantity of scientific res\ earch has been carried out to evaluate the claims of astrology. Less surprisin\ gly, there is really no evidence to support such claims (Blackmore & Seebold, 2001; Carlson, 1985; Eysenck & Nias, 1982). It is therefore cause for c\ oncern if citizens make important life decisions based on entirely unreliable a\ stro- logical predictions. For instance, people may decide for or against a po\ tential marriage partner based on astrological sign, they may make rash financia\ l decisions based on predicted good fortune, and so on. For this reason, i\ n 1984, the Committee for Skeptical Inquiry (at that time the Committee f\ or Scientific Investigation of Claims of the Paranormal) began a campaign \ to persuade American magazines and newspapers to attach a “health warnin\ g” to horoscope columns to indicate that they were to be read for entertain\ ment purposes only. Only around 70 publications out of 1,000 or more in the United States that carried horoscopes at that time agreed to carry such \ a warn- ing. This probably indicates that newspaper proprietors do not want to spoil their readers’ enjoyment by telling them that they should really ignore all the advice given. In any case, since the 1980s, there has been a proliferati\ on of horoscope and astrology-related websites, such as horoscope.com. This attests not only to the popularity of the topic but also to the impossib\ ility of policing the way in which information is presented to the public. But perhaps people do not set any store by astrological predictions and \ a health warning is really not necessary. After all, one does not need to \ believe something is true to be entertained by reading it. However, it appears t\ hat Allum 343 belief in astrological claims is quite widespread, at least in the Unite\ d States.

Losh and colleagues, in a review of 20 years of the U.S. survey data, fo\ und that many Americans believed in astrology, with polls putting the figure\ at around 25% (Losh, Tavani, Njoroge, Wilke, & Mcauley, 2003; National Science Board, 2002).

Astrology and Science Why should the credulity of some sections of the public toward astrology\ be a matter for concern for science communicators? Even if people do believ\ e in astrology, or ghosts and alien abductions for that matter, does this have a bear- ing on people’s understanding of and engagement with science? Again, \ the evidence is that it probably does. For not only do sizeable proportions \ of the American and European public believe in the efficacy of astrology, they \ also believe that it is scientific (National Science Board, 2006). The abil\ ity of citi- zens to distinguish between scientific and pseudoscientific claims is se\ en by many as an important component of scientific literacy. In a social and eco- nomic environment increasingly permeated by science, and the technologic\ al developments that flow from it, citizens require some basic competencies\ in order to meaningfully engage in rational judgments about a whole host of issues. For example, climate change, biofuels, stem cell cloning, synthe\ tic biology are all topics that have acquired, or are quickly gaining, polit\ ical sta- tus, which in turn require societal decisions to be made. In Miller’s\ framework for measuring civic scientific literacy, the rejection of astrology is a\ n empirical criterion for identifying those who are and are not scientifically liter\ ate (Miller, 2004). Surveys in Europe and the United States have tracked p\ ublic beliefs about astrology and science since 1988. In the United States, re\ spon- dents have been asked whether astrology is “very scientific, sort of \ scientific, or not at all scientific.” Around 60% said astrology is not at all sc\ ientific with around 30% saying it is “sort of scientific” in seven surveys betw\ een 1988 and 2001. In 2004, the proportion rejecting astrology rose slightly, to 66% (National Science Board, 2006). In Europe, there appears to be more wi\ de- spread belief that astrology is scientific. In 1992, respondents were as\ ked how scientific they thought astrology was, with a 5-point scale anchored at either end with very scientific or not at all scientific. Only one quarter considered astrology not at all scientific, with another quarter considering it very scientific and the remaining respondents falling somewhere in between (INRA, 1993).

In 2001, the question was asked in a slightly different way, with only t\ wo response options offered— rather scientific or not scientific. Fifty-three per- cent thought astrology was rather scientific (European Commission, 2001a). 344 Science Communication 33(3) The evidence, then, suggests that a sizeable minority of Americans and a\ n even greater proportion of Europeans believe that astrology in some sense “works,” either because it is based on scientific methods or for other reasons.

What might account for this widespread belief? And more interestingly, p\ er- haps, what might explain differences in degree of belief in astrology be\ tween individuals and groups? Most research on astrology as been focused on directly evaluating its claims but there is a stream of research on the psychological foundations of belief in pseudoscience. Integrating much of this work, Lindeman (1998) proposes two key reasons why pseudoscientific claims can appear credible\ .

First, motivation: astrology offers, at first blush, a way to comprehend\ and make sense of a complex and contingent world and also one’s “inner\ life,” personality, luck, and so on. It is attractive for that reason. Second, \ cognition:

it often appears to be consistent with heuristic or “experiential”\ forms of cognitive processing, which makes it believable. In terms of motivation to believe, astrological predictions are very oft\ en favorable. This makes them attractive, even to those who are skeptical o\ f astrology (Dean, 1987; Glick, Gottesman, & Jolton, 1989). The use of a\ strol- ogy to explain personality and, for instance, to understand how to achie\ ve success and control in personal affairs is a generally strong motivation\ for people (Baumeister & Leary, 1995). In terms of the way evidence is cog- nized, there are many well-established findings showing that people tend\ to use information shortcuts that are easy and quick to process that facili\ tate decision-making and judgment (Kahneman, Slovic, & Tversky, 1982). One \ of the clearest forms of heuristic cognition in relation to astrology is\ confir- mation bias (Fiske & Taylor, 1991; Ross & Anderson, 1982). This underl\ ies the “Barnum effect.” Named after the 19th-century showman Phileas T. B\ arnum, whose circus act provided “a little something for everyone,” it refers to the idea that people will believe a statement about their personality that i\ s vague or trivial if they think that it derives from some systematic procedure \ tailored especially for them (Dickson & Kelly, 1985; Furnham & Schofield, 1987; \ Rogers & Soule, 2009; Wyman & Vyse, 2008). For example, the more birth \ detail is used in an astrological prediction or horoscope, the more cred\ ulous people tend to be (Furnham, 1991). However, confirmation bias means that people do not tend to pay attention to other information that might disconfirm the credibility of the predictions. The fact that people tend to assume that the more complex the informatio\ n used as input into astrological readings, the more accurate the predicti\ ons are perhaps points to the informal reasoning that people use in mistaking as\ trol- ogy for science. “It sounds complicated—it must be scientific!”\ Allum 345 The Present Study There is evidence, then, of the general and widespread reasons why pseudo- science, and astrology in particular, might garner public credibility. M\ ost of this evidence is small scale and experimental, with very little based on\ large population surveys. Rather, less is known about the causes and correlate\ s of heterogeneity of belief about astrology. The present study sets out to e\ valuate several potential explanations for variation in the credibility given to astrol- ogy, qua science, by European citizens, using a recent Eurobarometer sur- vey. In doing so, I also examine how astrology is viewed alongside other\ knowledge-generating practices, scientific or otherwise, in order to und\ er- stand where astrology is located in the European public’s representat\ ional field. In the following sections, I briefly outline some putative factor\ s that I consider might account for variation in citizens’ beliefs before describing in more detail the data and methods used for the empirical analysis. The “immunization” hypothesis. From a traditional science communication perspective, it is scientific knowledge, particularly knowledge of the meth- ods of science, that would be expected to “immunize” citizens agai\ nst false belief in pseudoscience. Hence, those who are more scientifically literate, who understand principles of experimentation, the combining of empirical\ evidence with logical inference, and so forth, should be more likely to \ realize that astrology, for all its formalistic presentation, is not consistent \ with the tenets of scientific method. The survey evidence broadly supports this hypothesis, albeit indirectly. In both Europe and the United States, cor\ relates of skepticism about astrology’s “scientificness” tend to be hig\ her levels of education, higher social class, and higher income although there is some\ inconsistency between surveys. In 1992, Eurobarometer surveys showed that more highly educated Europeans were less likely to think that astrology \ is scientific, whereas in 2001, this was not the case (European Commission\ , 2001a). In the United States, education has been a consistent predictor\ . For example, the most recent National Science Foundation data show that whil\ e 84% of college graduates think that astrology is not at all scientific, just 62% of those who only graduated from high school share this belief (Nationa\ l Sci- ence Board, 2008). Education is not, of course, coterminous with scient\ ific knowledge, but it is strongly correlated (Allum, Sturgis, Tabourazi, & \ Brunton- Smith, 2008; Miller, 2004). Income and social class are themselves inva\ ri- ably correlated with education. Examining the relationship between scien\ tific knowledge and belief in astrology as a science net of education, income \ and social class, would be a stronger test of the “immunizing” hypothe\ sis, and one that I test in this article. 346 Science Communication 33(3) What’s in a name? One of the better established findings in survey mea- surement is that responses can be extremely sensitive to the particular form of words used in the asking of the question (Schuman & Presser, 1996).\ In the European surveys reviewed earlier, the English version of the questionnaire uses the word “astrology” as the stimulus object of the item:

People can have different opinions about what is scientific and what is not. I am going to read out a list of subjects. For each one tell me how scientific you think it is by the scale on this card . . . [other subjects] . . . . Astrology.

It is possible that people are unfamiliar with the this term but may be \ more familiar with terms like horoscopes, star signs, sun signs, which, if asked how scientific they are, might elicit different results. In fact, there \ is some evidence on this from the 1992 Eurobarometer survey. Respondents were randomized to two different versions of the list of subjects. One simply\ con- tained the single word, as above, while in the other condition, a brief \ explana- tion of each subject was given. For astrology, the explanation was the following: “. . . that is the study of occult influence of stars, pla\ nets etc. on human affairs.” There was no significant difference in responses betw\ een the two conditions (INRA, 1993). However, in most European languages the s\ uf- fix “ology,” “ologie,” “ologia” connotes an academic f\ ield of study. This may be enough to encourage respondents to think that astrology is indeed\ a science where the use of an alternative term may bring to mind a differe\ nt kind of activity. Another hypothesis that has intuitive plausibility, bu\ t which has not previously been tested, is that many people mistake astrology fo\ r astronomy—a simple semantic confusion. The extent to which this might\ occur is also likely to vary across countries according to the similarit\ y of the two terms in different languages. The stars down to earth. The explanations just outlined for belief, or appar- ent belief, in the scientificness of astrology are, in one form or anoth\ er, based on deficits of understanding and information. There may be other reasons\ why some people more than others place faith in astrological predictions\ .

One of the most interesting social psychological viewpoints on this ques\ tion is found in the work of Theodor Adorno. In 1952-1953, Adorno carried out\ a study of Caroll Righter’s Los Angeles Times astrology column. The fruits of this did not appear in English until published in Telos in 1974 as “The stars down to earth” (Adorno, 1974). In the study, referred to by the aut\ hor as a “content analysis,” Adorno analyses, somewhat haphazardly and selectively, the advice given to readers in the column over a period of several month\ s. He identifies many of the aspects of astrological readings that other psych\ ological Allum 347 research (e.g., Forer, 1949) confirmed were effective in making them c\ on- vincing: the Barnum effect, the tendency to personalize general statemen\ ts and so forth. He is witheringly critical of astrology, dubbing it, with \ the rest of occultism, a “metaphysic of dunces” and suggesting that “a c\ limate of semierudition is the fertile breeding ground for astrology” (Adorno,\ 1994, p. 44). The claim is that it resembles other “irrational creeds” lik\ e racism by offering a shortcut to (erroneous) knowledge, which actually requires \ no intellectual effort or capacity (Dutton, 1995). What is particularly interesting for the present study, though, is the c\ on- nection drawn between astrology (and other forms of popular occultism)\ with authoritarianism, fascism, and modern capitalism. Adorno sees astrology \ as emphasizing conformity and deference to higher authority of some kind.

Nederman and Goulding (1981) sum this up concisely as “Take things \ as they are, since you are fated for them anyway.” Adorno posits an “\ astrologi- cal ideology” that he claims “resembles, in all its major characte\ ristics, the mentality of the ‘high scorers’ of the Authoritarian Personality”\ (Adorno, 1994). The work on “Authoritarian Personality” by Adorno and colleagues has been much criticized since its appearance in 1950 (Adorno, Frenkel- Brunswik, Levinson, & Sanford, 1950; Kirscht & Dillehay, 1967) with par\ - ticular criticism being directed toward the test items in the “F-Scale” (Hyman & Sheatsley, 1954). Nevertheless, it is possible to deduce a reasonably\ clear empirical hypothesis from “stars down to earth.” Those who value c\ onfor- mity, obedience, and tend toward uncritical acceptance of in-group moral authority will be more likely to give credence to the claims of astrolog\ y. Adorno also discusses the relationship of organized religion, or religio\ us belief with astrological belief. He suggests that part of astrology’s\ appeal is that it formalizes the notion of some higher authority at work controlli\ ng life events yet does not come with the explicitly restrictive structure of fo\ rmal religious adherence, churchgoing, and so on. This is part of what, for Adorno, makes astrological belief and capitalist individualism such well\ - suited bedfellows. That is to say that religious belief and astrological\ belief are both consistent with the same authoritarian trait of personality. If\ this is true, one might expect beliefs about astrology and about religion or God\ to be related.

Hypotheses and Questions The foregoing discussion leads to the derivation of the following hypoth\ eses:

Hypothesis 1: The suffix “ology” means that people should tend to rate “astrology” as more scientific than “horoscopes.” 348 Science Communication 33(3) Hypothesis 2a: Because of potential confusion or elision of meaning between “astronomy” and “astrology,” we should expect there \ to be a positive correlation between how people rate the scientificness of these two subjects.

Hypothesis 2b: Assuming Hypothesis 2a to be correct, we should not expect to see the same positive correlation between ratings of horo- scopes and astronomy because the potential for semantic confusion is much less.

Hypothesis 3C: Citizens who are more knowledgeable about science should be less likely to rate astrology as scientific.

Hypothesis 4: Following Adorno’s thesis, we should expect that people who score higher on a measure of authoritarianism will be more likely to rate astrology as being scientific.

In addition to the evaluation of these empirical expectations, there are\ two more general questions that are addressed in the analysis:

Question 1: How is astrology viewed by Europeans in relation to other scientific and nonscientific subjects?

Question 2: How much of the variability in beliefs about astrology across Europe is related to country of citizenship?

Data and Measures Data The data for this study come from the Special Eurobarometer 224 and 225 surveys, “Europeans, Science and Technology,” “Social Values, S\ cience and Technology” (European Commission, 2005a, 2005b). The author was\ part of the team that worked on the design of questionnaire. Both these \ sur- vey modules were fielded as part of the same face-to-face interview to c\ itizens in 25 EU member states during the fall of 2004. Approximately 1,000 respondents were interviewed in each country, using a multistage probabil- ity design. (For more details on the survey methodology, see European Commission, 2005a). The resultant data set contains rich information on\ citizens’ beliefs, attitudes, and knowledge about science and technology, as well as on political and social values. Also embedded within the survey was a split-sample randomized question wording experiment. Allum 349 Measures The key dependent variable, belief in the scientificness of astrology, was measured by asking respondents how scientific they consider each of ten subjects to be, on a scale from 1 to 5 where 1 indicates not at all scientific and 5 indicates very scientific. As part of the list of 10 subjects a randomized half of the sample was asked about “astrology” and the other half \ about “horoscopes.” The other 9 subjects were physics, medicine, astrono\ my, eco- nomics, history, homeopathy, psychology, biology, and mathematics. (The\ exact question wordings and response alternatives for the English question- naire for all the measures employed can be found in European Commission,\ 2005a, 2005b.) Authoritarian values are measured with a single indicator. While this is\ a good way short of ideal in terms of best practice in measurement, the qu\ es- tion employed has what appears to be good face validity. Respondents are\ shown a list of qualities that children should be encouraged to learn. O\ ne of these qualities is “obedience.” Responses to this item are on a 4-\ point scale ranging from not at all important to very important. This battery of items was designed to tap into a range of values, including authoritarian values with the item utilized here. Measuring this construct with a question that essent\ ially taps social conformity accords with recent work in political psychology (Feldman, 2003). Child rearing goals that value obedience have also be\ en shown to be related to authoritarianism (Danso, Hunsberger, & Pratt, 19\ 97).

So while this single indicator is by no means perfect, I consider there \ is suf- ficient reason to consider it adequate for purpose here. Also related to Adorno’s work on authoritarianism is religious belief\ . To capture this, I use an item that asks whether a respondent believes in “\ God,” a “spirit,” or neither. From this categorical variable, two dummy \ variables have been derived, indicating belief in God or belief in a spirit (both\ vs. no belief). I also use a dummy variable indicating whether or not the resp\ ondent is Catholic. Around 80% of the sample are Christians of some type, with \ Catholics making up around 50%. Distinguishing between Catholics and all\ others is a reasonable way of capturing the major source of religious de\ nomi- nation absence of any particular hypotheses about denomination-based dif\ - ferences in beliefs about astrology. 1 Knowledge about science is measured in a number of ways in the survey and in the analysis I use three separate indicators. Two of these tap in\ to respon- dents’ understanding of scientific process and method. An understandi\ ng of 350 Science Communication 33(3) method is arguably central to being able to distinguish between scientific and pseudoscientific claims. People were asked “what does it mean to stud\ y some- thing scientifically?” The verbatim responses were coded into one of \ several mutually exclusive categories, based on what was said (in other words, \ not precoded). Typically this question has been a very good predictor of at\ titudes and beliefs about science with the mention of hypothesis testing and exp\ eri- mentation as the critical component. Here I use this as an indicator of \ greater scientific understanding along with another indicator based on the mentioning of “measurement” in response to the question. The third indicator \ is a sum- mated scale of correct responses to a series of 13 true/false quiz items\ that tap textbook type knowledge about scientific facts. For example, “the sun\ goes around the earth” and “lasers work by focusing sound waves.” Th\ ese items have long been used in public understanding of science surveys, having b\ een developed originally by Jon Miller, John Durant, and colleagues (Durant\ , Evans, & Thomas, 1989; Miller, 1998). For the purposes of this analysis\ , “don’t know” responses are coded with a zero, the same way as incorrect true/ false answers. The scale has reasonable internal consistency with a Cron\ bach’s alpha coefficient of .72. A range of other background characteristics were measured in the survey and used in the analysis. Respondent age was coded in bands: 16 to 24, 2\ 5 to 39, 40 to 54, 55+ . Occupational status was measured with a dummy vari- able contrasting white collar and management occupations with all oth- ers. This is a necessarily crude indicator as it is based on standardized Eurobarometer occupational coding that needs to be comparable across European states. However, as it is only being used as a control variable\ in the analyses that follow, it is not critical to obtain a more fine-grain\ ed esti- mate of the effects of occupational status on beliefs about astrology.

Education is measured, again quite bluntly, with a variable that indicat\ es whether or not the respondent left full-time education after the age of \ 20.

This broadly distinguishes the graduate population, and it is this disti\ nction that has in previous research been shown to be the most diagnostic of di\ ffer- ences in attitudes and beliefs about science and technology (e.g., Mill\ er, Pardo, & Niwa, 1997). Finally, the type of area in which the respondent\ lives is captured with a variable that indicates residence in a large town ver\ sus other types of area. Typically one might expect urban populations to hav\ e different cultural and political orientations to rural and provincial po\ pula- tions, net of education and occupational differences; hence, this variab\ le is used as a control. Allum 351 Results Question Wording Experiment Figure 1 shows the response distributions for the two experimental condi\ - tions. In one condition, respondents were asked how scientific they thou\ ght astrology was and in the other how scientific did they think horoscopes \ were.

Quite clearly, many more Europeans think astrology scientific than horo- scopes. 57% think that horoscopes are “not at all scientific,” whi\ le only 24% believe the same about astrology. About one quarter of the sample believ\ e astrology to be “very scientific,” while only 7% think that horosc\ opes are “very scientific.” The difference in distributions is highly signi\ ficant, χ 2 = 3,400 (4 df), p < .001, n = 23,473). Astrology is clearly viewed as greatly more scientifically credible than horoscopes and this is in line with the exp\ ecta- tions set out in Hypothesis 1. Another point to note is that the there i\ s a great deal more heterogeneity in beliefs about astrology than about horoscopes\ , 26 1718 1524 7 712 17 57 0 10 20 30 40 50 60 70 80 90 100 Very scientific Not at all scientific Per cent Astrology Horoscopes Figure 1. How scientific is astrology/are horoscopes? 352 Science Communication 33(3) with both “very scientific” and “not at all scientific” each\ attracting one quarter of all respondents.

The Relationship Between Astrology, Horoscopes, and Other Subjects Figure 2 shows the distribution of European beliefs about the scientific\ ness of all 11 subjects included in the questionnaire (including both horoscopes and astrology). The chart is ranked in descending order according to th\ e percentage of people thinking each subject is “very scientific.” A\ s can be seen, the list includes a range of more or less scientific subjects, inc\ luding both horoscopes and astrology, along with homeopathy, as examples of pseudoscience. Medicine is viewed as the most scientific subject, followed closely by physics and then biology. Horoscopes are least likely to be thought “\ very scientific,” followed by homeopathy, history, and economics (17%, 18\ %, and 19%, respectively). In most respects, this is not a surprising resu\ lt. The natural sciences are at the top with social and behavioral subjects lowe\ r down. What is somewhat surprising, although in line with other surveys 7 17 18 19 2632 51 5354 66 71 7 21 17 22 17 24 22 2324 20 20 12 29 26 29 18 2415 14 15 9 6 17 18 19 17 15 127 6 43 2 57 15 19 14 24 85 5 22 2 01 02030405060708090100 Horoscopes Homeopath y Histor y Economics Astrology Psychology Astronom y Mathematics Biology Ph ysics Medicine Ve ry scientific Not at all scientific Figure 2. “Scientificness” of 11 activities (%) Allum 353 (European Commission, 2001b), is that astrology is considered to be more scientific than economics and only just less so than is psychology. Figure 2 suggests some ambivalence about the scientific status of astrol\ - ogy if not homeopathy and horoscopes. Hypothesis 3 predicts that those w\ ho are more scientifically literate will be less likely to regard astrology\ as scien- tific. In more general terms, one might expect a higher degree of unders\ tand- ing of science to confer greater ability to discriminate between science\ and pseudoscience in general. Later in the article I shall return to this hy\ pothesis in a multivariate setting but it is interesting to see what happens when the result in Figure 2 is stratified by scientific knowledge. The number of \ correct answers given by each respondent was used as an indicator of science literacy that runs from 0 to 13. Figure 3 plots the percentage of Europeans selec\ ting “very scientific” or the scale point below at each successive leve\ l of scientific knowledge. Also plotted out of interest is the percentage agreement with\ another the statement in the questionnaire that arguably taps into super\ sti- tious beliefs “some numbers are lucky for some people.” The plot shows that there is a noticeably steep negative gradient with s\ ci- entific knowledge for belief in the pseudosciences homeopathy, astrology\ , and in lucky numbers. For instance, between 50% and 60% of citizens scor\ - ing 3 out of 13 on the knowledge quiz believe in the scientific credibil\ ity of homeopathy and astrology, and believe in lucky numbers. That percentage \ drops to just 15% to 25% for those that obtain a maximum score on the knowledge quiz. In contrast, for both medicine and psychology, which, re\ call, are quite a long way apart in the rank ordering shown in Figure 2, there\ is little or no difference between the beliefs of the better or worse infor\ med citizen.

(The lines for physics, biology, and mathematics are similar to that for medi- cine and are not shown here, in order to keep the graph intelligible.) A further exploration of the basis on which Europeans make judgments about the scientificness of these subjects was carried out using factor \ analy- sis. Two separate analyses, one for each split half of the sample, were \ per- formed using maximum likelihood estimation and oblique rotation. Table 1\ shows the factor loadings for the three-factor model that resulted, base\ d on an examination of the scree plot and substantive interpretability. An obliq\ uely rotated solution was preferred because I assume that there are individua\ l dif- ferences in the propensity to agree or disagree that any subject is scie\ ntific, as well as a tendency to discriminate between groups of similar subjects. T\ his carries the implication that all factors are likely to be correlated to \ some degree, given that “how scientific . . .” is in the stem of all th\ e questions. In any case, the assumption of correlated factors is weaker than assuming t\ hat they are orthogonal to each other, so is preferred here. Figure 3. Beliefs about “scientificness” versus science knowledge 354 Allum 355 I have labeled the three factors “hard science,” “soft science”\ and, for want of a better term, “new-age science.” The highest factor loadi\ ng for “hard science” is physics at 0.88, while the “soft science” includes \ economics and history, anchored by economics, at 0.71. Homeopathy, astrology, and psy- chology form a “new-age science” group, although psychology has a \ weak cross-loading with “soft science.” The reason for the “new age” label is that the three subjects that load on this factor are the kind of subjects that people might expect to see in the self-help, pop-psychology, or new age therapy\ sec- tions of bookstores. The representation of psychology for the European p\ ub- lic is perhaps less about, for instance, cognitive neuroscience and more\ about self-help for overcoming depression. In Table 2, the rotated solution for the sample that were asked about ho\ ro- scopes is presented. The main difference is that only two factors are ne\ eded to describe the cognitive structure organizing responses to the 10 subje\ cts.

There is no separate “new-age science” dimension, with horoscopes,\ home- opathy, and psychology joining history and economics to form a “soft \ sci- ence” group. The “hard science” subjects remain the same, with \ physics again anchoring the factor with the highest loading, at 0.78.

Table 1. Three Factors Underlying Perceptions of Scientificness (Astrology Condition) Rotated Factor Pattern Loadings (Total Variance Explained 42%) “Hard Science” (28%) “Soft Science” (9%) “New Age Science” (4%) Physics 0.88−0.07−0.12 Biology 0.610.000.04 Medicine 0.58−0.01 0.12 Astronomy 0.49−0.02 0.11 Mathematics 0.490.29−0.12 Economics 0.020.710.01 History −0.030.620.06 Homeopathy 0.000.070.62 Astrology 0.03−0.02 0.44 Psychology 0.110.320.37 Note: Oblique rotation, maximum likelihood. 356 Science Communication 33(3) The results seen thus far indicate fairly unequivocally that while horo- scopes and astrology are to all intents and purposes the same thing, at \ least in so far as the casual engagement of the average citizen is concerned, the\ two terms have rather different connotations. Astrology probably “sounds”\ more scientific than horoscopes; it is consequently evaluated as being more scien- tific and is viewed in a more similar way to psychology and homeopathy t\ han are horoscopes. Having elaborated a description of how astrology is per- ceived, the following section turns to the question of what might underl\ ie the variation in these perceptions among citizens and across European states\ .

Variation in Beliefs Between Citizens The final part of the analysis is a multivariate investigation of social\ and psychological factors that might influence individuals’ propensity to\ believe that astrology is a scientific subject. To do this, I use a variant of a\ n ordinary least squares multiple regression model. This analysis is primarily to examine individual level factors associate\ d with beliefs about astrology. However, there is also likely to be heterogeneity of beliefs between the 25 European countries, even after taking into acc\ ount individual characteristics. Modeling this situation calls for country to\ be Table 2. Two Factors Underlying Perceptions of Scientificness (Horoscopes Condition) Rotated Factor Pattern Loadings (Total Variance Explained 36%) “Hard Science” (26% Variance) “Soft Science” (10% Variance) Physics 0.78−0.11 Biology 0.630.05 Medicine 0.600.03 Mathematics 0.520.17 Astronomy 0.520.09 History 0.120.55 Economics 0.170.52 Psychology 0.260.50 Horoscopes −0.30 0.48 Homeopathy 0.080.43 Note: Oblique rotation, maximum likelihood. Allum 357 included as either a fixed or random effect. The first approach essentia\ lly gives each country its own dummy variable and regards each as a unique entity, so to speak. The second approach treats the countries in the dat\ a set as a random sample of potential countries that could have been included and\ estimates a single mean and variance for a continuous random variable that captures the heterogeneity across countries (Raudenbusch & Bryk, 2001)\ .

The model parameters can then be used to derive an estimate of each coun\ try’s location in the distribution of this variable, a feature that would be useful for the present investigation. If the individual level effects are uncorrelated with this country random effect, the random effects estimator is unbiased and preferred over the \ fixed effects one, because it is more efficient. If this assumption is not met\ , the fixed effects approach is appropriate. The result of a Hausman test (Ha\ usman, 1978), which tests for the presence of this correlation, was highly non\ signifi- cant (χ 2 = 5.86, 17 df, p = .99), so the final model presented here uses the random effects estimator. 2 Table 3 presents the estimates for the model predicting beliefs about astrology. Higher scores on the dependent variable indicate stronger bel\ ief that astrology is scientific. The included predictors account for 17% of\ the variance in beliefs. Turning first to Hypothesis 2a, the expectation was that there should be some correlation between beliefs about astronomy and beliefs about astrology because of semantic confusion. This is indeed the case h\ ere.

The coefficient for astronomy is .35 with a very small standard error. N\ et of all other modeled influences, the more likely it is that citizens believe astron- omy is scientific, the more likely they are to think that astrology is a\ lso sci- entific. It appears that the two subjects are not always well differentiated in European public imagination. In order to counter potential objections to\ this conclusion, I have included an additional variable (“all scientific”\ ) that is calculated as the mean of respondents’ belief scores on all the other\ subjects except for horoscopes and homeopathy. Without this control, it could be argued that the correlation between astronomy and astrology is because o\ f individual differences in the propensity to express the view that anything is scientific. The inclusion of this variable, though, does not eliminate t\ he posi- tive coefficient for astronomy. Hypothesis 2b, if supported, would corroborate further this interpretation.

I would not expect to see the same positive relationship between astrono\ my and horoscopes, precisely because they do not sound similar. To test thi\ s, I fitted the same model using the other split half of the sample, designat\ ing horoscopes as the dependent variable. The pattern of results is very clo\ se to that in Table 3 but, crucially, the coefficient for astronomy is very sm\ all, at 358 Science Communication 33(3) .05 (SE = .01; Z = 4.23; p < .01), compared with the estimate in the astrology model. A formal test that the coefficients for astronomy in the two mode\ ls are different was also conducted by refitting the two models as a single equ\ ation and adding an interaction between experimental condition (horoscopes or astrology samples) and astronomy. This term was, as expected, statistic\ ally significant and confirms that the difference in the effect of astronomy in each of the experimental conditions is unlikely to be because of chance.\ What this comparison between astrology and horoscopes conditions also suggests is that, apart from the semantic confusion between astronomy and astrol - ogy, Europeans seem to use the same basis for evaluation of both astrology and horoscopes, given that the same predictors have the same effects reg\ ard- less of which name is used to identify the activity. Hypothesis 3 concerns the relationship of scientific knowledge or literacy with perceptions of astrology. The expectation is that those who are bet\ ter Table 3. Random Effects Regression Estimates (Individual Characteristics), N = 11,622 B SE Zp Age, years (reference category 15-24) 25-39 −.12.04 −2.66 .01 40-54 −.14.04 −3.34<.01 55+ −.22.04 −5.22<.01 Female .06.03 2.18.01 Higher education −.19.03 −5.75<.01 Professional or management occupation −.08 .04 −1.89 .06 Large town or city dweller −.10.03 −3.04<.01 Religious belief Catholic .13.04 3.70<.01 God .08.04 2.02.04 Spirit .12.04 3.21<.01 Right wing political orientation .01.004 2.26.02 Science knowledge Quiz score −.09.01 −17.42 <.01 Mentions hypothesis testing −.25.06 −4.55<.01 Mentions measurement −.21.05 −4.52<.01 Authoritarian .22.02 10.36<.01 Astronomy scientific .32.01 26.80<.01 All scientific .24.02 12.41<.01 Intercept 1.18.14 8.67<.01 Allum 359 endowed with civic scientific literacy will be better able to distinguis\ h science from pseudoscience. Therefore, I expect positive coefficients for the kn\ owl- edge measures. Looking at Table 3, one can see that this is indeed the c\ ase.

The coefficients for all three knowledge variables are negative and stat\ isti- cally significant, meaning that the more knowledgeable a person is, the \ less scientific they believe that astrology is. To get a sense of the magnitu\ de of the effects, if one compared a citizen who scored at the mean on the quiz an\ d mentioned neither hypothesis testing nor measurement in their open-ended\ answer, their expected rating of astrology would be one point higher on the 5-point scale (in the “more scientific” direction) than a Europe\ an mentioning both measurement and hypothesis testing and who scored at the maximum on\ the quiz, with all other variables held constant. Particularly interesti\ ng is the fact that science knowledge has an effect even after controlling for education. Hypothesis 4 addressed in the analysis considers the relationship of authoritarian personality type with beliefs about astrology. The coeffic\ ient for the authoritarianism question is positive, at .22, and highly signif\ icant.

For each one-point increase on the “importance of obedience” quest\ ion the model predicts just less than a one-quarter point increase in the scient\ ificness of astrology rating. So, controlling for all the other covariates, the p\ redicted difference in belief about astrology between Europeans who do not think \ obedience is at all important to teach to children and those who think i\ t very important is just slightly more than one scale point. Here, then, is empirical support for Adorno’s linking of authoritarianism and openness to pseu\ dosci- ence. Note again that this relationship is robust to the full range of o\ ther controls in the model, in particular age, education and conservative or right wing political orientation. People who report believing in God or in a “spirit of some kind” a\ re more likely to think astrology is scientific. Catholics are also more likely \ to express this view. These results are again in line with what one might expect fr\ om Adorno’s account of the appeal of astrology to those who have a prope\ nsity to defer to higher authority of different kinds, including religion. I h\ ave no clear explanation for why Catholics should be more likely to be credulou\ s of astrology, but it should be borne in mind that this finding of course co\ ntrols for country as well as for individual characteristics. This means that r\ eligious denomination here is not simply a proxy for divergent beliefs of citizen\ s from predominantly Catholic and non-Catholic European states. The coefficients for the other sociodemographic variables are worth not- ing. Women are slightly more likely than men to think astrology is scien\ tific, as are people who self-identify as being politically on the right. Those\ who live in large urban areas, who are better educated and in high status 360 Science Communication 33(3) occupations are all less likely to accord astrology scientific credibili\ ty. A rather interesting finding is that older people tend to be less credulous of as\ trology.

It is the youngest age group, 15 to 24 years, that regards astrology as \ most scientific. This invites speculation as to whether it is a lifecycle or \ a cohort effect we are seeing here. Do people become more skeptical as they age? Or are younger generations in general less skeptical than their parents? Th\ is is something for further research to establish.

Variation in Beliefs Between Countries Table 4 shows the variance components estimates for both individual and country levels. This can be thought of as the proportion of unexplained,\ residual variation between individuals and between countries. The first \ two columns of the table show these variance estimates and the corresponding\ percentages for a model where only the intercept is fitted. In other wor\ ds, it simply partitions total variation into within-group and between-group (indi- vidual and country). As little as 13% of the variation in beliefs about\ astrol- ogy is systematically related to country, while the remainder is because\ of interindividual variation. When the full model, with all the independent\ vari- ables, is estimated, the proportion due to country drops by around on th\ ird, to 8%. This indicates that some of the apparently systematic country var\ ia- tion is due to compositional differences in populations on the individua\ l characteristics entered into the model. In general terms, it would appea\ r that the social-psychological factors that influence beliefs about astrology \ and science across Europe are broadly common to citizens from all countries.\ Nevertheless, this is not to say that there is no systematic difference between countries at all. One way of exploring this is to obtain estimat\ es of the unobserved country-level random effect variable and to compare esti- mates across countries. These estimates are also known as “empirical \ Bayes estimates” (Raudenbusch & Bryk, 2001). Figure 4 shows standardized \ ran- dom effects estimates for each of 25 European member states. These can b\ e Table 4. Variance Components for Intercept-Only and Full Models Intercept-Only Model Full Model Variance (SE) Percentage of Total Variance Variance (SE) Percentage of Total Variance Individual-level residual 2.08 (0.03)871.80 92 Country residual 0.27 (0.08)130.14 8 Allum 361 thought of as unexplained country level residuals, expressed in standard\ devi- ations from the mean (zero), after taking into account all the individ\ ual level variables in the model. There is a surprisingly clear pattern to the var\ iation.

Controlling for the individual characteristics of citizens, all the form\ er Eastern bloc countries are more accepting of astrology as being scientif\ ic than the model would predict, while the majority of Western European sta\ tes have more skeptical citizens than would be expected given the individual attr\ i- butes of their citizens.

Conclusion Europeans have a range of beliefs about astrology and its status as a sc\ ientific or quasi-scientific subject. The evidence from the experiment and from t\ he observational data show that there is considerable blurring over what th\ e –2 –1.5 –1– 0.5 00 .511 .52 Fi nlan d Gr eec e Fr an ce Be lgiu m UK Sw eden It aly Ne ther lands Lu xemb ourg Aust ria Po rtug al Ge rmany H ung ary De nm ark Ir el an d Sl ov en ia Cy prus Malt a Sp ain La tvia Cz ech Re publ ic Po lan d Es toni a Lit hua nia Sl ov ak ia Standa rdised ra nd om effec t es tima tes Figure 4. Standardized random effect estimates for 25 European countries 362 Science Communication 33(3) term means. There is a clear confusion between astronomy and astrology. \ There are also widely differing evaluations of horoscopes as compared wi\ th astrology in terms of their scientific credibility, even though one migh\ t consider the two as functionally equivalent to most intents and purposes.

Astrology is regarded as more scientific than horoscopes but interesting\ ly, the correlates of credulity in both of these are patterned in exactly th\ e same way. In other words, the same types of citizen are likely to regard astr\ ology and horoscopes as more scientific. The only major difference between regression models predicting beliefs was that the effect of astronomy belief was far greater for astrology than for horoscopes. 3 Perhaps previous research on science literacy that has gauged the credulousness of citizens about \ astrol- ogy has led to rather more pessimistic conclusions than are really warranted. However, science literacy clearly makes a difference. Net of a range of other potential confounding influences, the better is one’s understan\ ding of scientific terms and concepts, as well as factual knowledge of science, \ the better one is able to distinguish science from pseudoscience. This immun\ iz- ing effect of scientific knowledge is perhaps not a surprising finding, \ but one that underlines the utility of these survey indicators in distinguishing\ between modes of citizen reasoning about science and confirms the importance of \ scientific literacy in helping Europeans make informed judgments about t\ he validity of pseudoscientific claims. It is interesting too that this for\ mal sci- ence knowledge makes a difference over and above the effect of education\ more generally. The implication is that being well educated and having a\ col- lege degree is not enough on its own necessarily to allow citizens to distin- guish pseudoscience from science. While this study has not focused on he\ alth and consumer choices, it is quite possible that the model presented here\ holds across a range of pseudoscientific domains in which citizens are confronted with the need to make informed choices. The result that authoritarian-type values are associated with greater cr\ edu- lity toward astrology is fascinating and in line with Adorno’s prediction.

Whether it is because of a general propensity to defer to any kind of au\ thority, or whether it is bound up with antirational culture in the way that Adorno hypothesized is open to question. Perhaps linked to this finding are the\ sys - tematic country variations. Former Eastern bloc states seem to be more accepting of astrology and more likely to consider it scientific. This h\ olds true even conditioning on science literacy, religion, education, politic\ al ori- entation, and values. Citizens of these countries have a recent history \ of state and civil society being organized along authoritarian lines, and this cultural norm may be reflected in the readier acceptance of astrology in Eastern \ Europe, over and above that due to variation in individual personality traits. Allum 363 This investigation has examined some of the correlates and putative causes of different beliefs about astrology and its relationship with science. \ One notable omission in the set of correlates examined here is the absence o\ f any communication variables, such as media use and attentiveness. This was d\ ic- tated by the lack of suitable items in the Eurobarometer survey and it remains an interesting question as to what the effect of attention to science co\ verage in the media, and of regular attention to personal horoscopes, might be \ on beliefs about astrology. As a final note, it should probably be borne in mind that for most peopl\ e, reading a horoscope is a leisure activity and not one of central importance either. Nevertheless, by understanding how Europeans differ in their percep- tions of astrology and horoscopes, it is possible to gain some insight i\ nto the bases of how citizens evaluate scientific and pseudoscientific claims mo\ re generally.

Declaration of Conflicting Interests The author declared no conflicts of interest with respect to the authors\ hip and/or publication of this article.

Funding The author received no financial support for the research and/or authorship of this article.

Notes 1. A sensitivity analysis showed that a more detailed breakdown of religiou\ s denom- ination has no impact on the coefficients for other variables in the mod\ el.

2. A 5-point Likert-type scale is not strictly a continuous variable, and t\ herefore ordi- nary least squares may not be appropriate. To assess the sensitivity of \ the results to this, I also ran the analysis with a random effects ordered logit mod\ el (Rabe- Hesketh, Skrondal, & Pickles, 2002). The results of this alternative pa\ rameteriza- tion of the model show no differences in the sign or statistical signifi\ cance of any coefficient except for “belief in God,” which becomes marginally n\ onsignificant.

3. This could be viewed as evidence for semantic confusion, although, as an\ anony- mous referee pointed out, it may also be that astrology conforms on the \ surface to a naive view of what science is and it is this, rather than confusion wi\ th astronomy, that leads to its tendency to be evaluated as more scientific.

References Adorno, T. (1974). The stars down to earth. Telos, 6(Spring) (pp. 13-90). 364 Science Communication 33(3) Adorno, T. (1994). The stars down to earth: The Los Angeles Times astr\ ology column.

In S. Crook (Ed.), The stars down to earth and other essays on the irrational in culture (pp. 46-171). London, England: Routledge.

Adorno, T., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). The authoritarian personality. New York, NY: Harper & Row.

Allum, N., Sturgis, P., Tabourazi, D., & Brunton-Smith, I. (2008). Sci\ ence knowledge and attitudes across cultures: A meta-analysis. Public Understanding of Science 17, 35-54.

Baumeister, R. F., & Leary, M. R. (1995). The need to belong: The desi\ re for inter- personal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497-529.

Blackmore, S., & Seebold, M. (2001). The effect of horoscopes on women\ ’s relation- ships. Correlation, 19, 17-32.

Carlson, S. (1985). A double-blind test of astrology. Nature, 318, 398-399.

Curry, P. (1989). Prophecy and power: Astrology in early modern England: Princeton, NJ: Princeton University Press.

Danso, H., Hunsberger, B., & Pratt, M. (1997). The role of parental re\ ligious funda- mentalism and right-wing authoritarianism in child-rearing goals and pra\ ctices.

Journal for the Scientific Study of Religion, 36, 496-511.

Dean, G. (1987). Does astrology need to be true? Part 2: The answer is\ no. Skeptical Inquirer, XI, 257-273.

Dickson, D. H., & Kelly, I. W. (1985). The “Barnum effect” in pe\ rsonality assessment: A review of the literature. Psychological Reports, 57, 367-382.

Durant, J., Evans, G., & Thomas, P. (1989). The public understanding o\ f science. Nature, 340, 11-14.

Dutton, D. (1995). Theodor Adorno on astrology. Philosophy and Literature, 19, 424-430.

European Commission. (2001a, December). Europeans, science and technology. Retrieved from http://europa.eu.int/comm/public_opinion/archives/eb/ebs_\ 154 _en.pdf European Commission. (2001b). Europeans, science and technology (55.2): SPSS dataset. Retrieved from http://www.data-archive.ac.uk European Commission. (2005a). Special Eurobarometer 224: Europeans, science & technology. Retrieved from http://ec.europa.eu/public_opinion/archives/ebs/ ebs_224_report_en.pdf European Commission. (2005b). Special Eurobarometer 225: Social values, science & technology. Retrieved from http://ec.europa.eu/public_opinion/archives/ebs/ ebs_225_report_en.pdf Eysenck, H. J., & Nias, D. K. B. (1982). Astrology: Science or superstition? London, England: Maurice Temple Smith. Allum 365 Feldman, S. (2003). Enforcing social conformity: A theory of authorita\ rianism. Politi- cal Psychology, 24, 41-74.

Fiske, S. T., & Taylor, S. E. (1991). Social cognition (2nd ed.). New York, NY: McGraw-Hill.

Forer, B. R. (1949). The fallacy of personal validation: A classroom d\ emonstration of gullibility. Journal of Abnormal Psychology, 44, 118-121.

Furnham, A. (1991, January 26). Hooked on horoscopes: We may not be ab\ le to persuade people that astrology and graphology are nonsense. But psycholo\ gy, with the help of a showman, reveals why they are so popular. New Scientist (pp. 33-36).

Furnham, A., & Schofield, S. (1987). Accepting personality test feedba\ ck: A review of the Barnum effect. Current Psychological Reviews and Research, 6, 162-178.

Glick, P., Gottesman, D., & Jolton, J. (1989). The fault is not in the\ stars: Susceptibil- ity of skeptics and believers in astrology to the Barnum effect. Personality and Social Psychology Bulletin, 15, 572-583.

Hausman, J. A. (1978). Specification tests in econometrics. Econometrica, 46, 1251-1271.

Hyman, H., & Sheatsley, P. (1954). The authoritarian personality: A me\ thod- ological critique. In R. Christie & M. Jahoda (Eds.), Studies in the scope and method of the “Authoritarian Personality.” Glencoe, IL: Free Press of Glencoe. (pp. 24-49).

INRA. (1993, December). Europeans, science and technology: Public understanding and attitudes. Retrieved from http://ec.europa.eu/public_opinion/archives/ebs/ ebs_076_en.pdf Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under uncertainty: Heuristics and biases. Cambridge, England: Cambridge University Press.

Kirscht, J. P., & Dillehay, R. (1967). Dimensions of Authoritarianism: A review of research and theory. Lexington: University of Kentucky Press.

Lindeman, M. (1998). Motivation, cognition and pseudoscience. Scandinavian Journal of Psychology, 39, 257-265.

Losh, S., Tavani, C., Njoroge, R., Wilke, R., & Mcauley, M. (2003). Wh\ at does education really do? Skeptical Inquirer, 27, 30-35.

Miller, J. D. (1998). The measurement of scientific literacy. Public Understanding of Science, 7, 203-223.

Miller, J. D. (2004). Public understanding of, and attitudes toward, scientific research: What we know and what we need to know. Public Understanding of Science, 13, 273-294.

Miller, J. D., Pardo, R., & Niwa, F. (1997). Public perceptions of science and technol- ogy: A comparative study of the European Union, the United States, Japan and Canada. Bilbao, Spain: Fundacion BBV. 366 Science Communication 33(3) National Science Board. (2000). Science and engineering indicators: 2000. Arlington, VA: Author.

National Science Board. (2002). Science and engineering indicators: 2002 (NSB-02-1). Arlington, VA: Author.

National Science Board. (2006). Science and engineering indicators: 2006. Arlington, VA: Author.

National Science Board. (2008). Science and engineering indicators: 2008. Arlington, VA: Author.

Nederman, C., & Goulding, J. (1981, Winter). Popular occultism & criti\ cal social theory: Exploring some themes in Adorno’s critique of astrology & the\ occult.

Sociological Analysis: A Journal in the Sociology of Religion, 4, 325-332.

Rabe-Hesketh, S., Skrondal, A., & Pickles, A. (2002). Reliable estimat\ ion of general- ized linear mixed models using adaptive quadrature. Stata Journal, 2, 1-21.

Raudenbusch, S., & Bryk, A. (2001). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: SAGE.

Rogers, P., & Soule, J. (2009). Cross-cultural differences in the acce\ ptance of Barnum profiles supposedly derived from Western versus Chinese astrology. Journal of Cross-Cultural Psychology, 40, 381-399.

Ross, L., & Anderson, C. (1982). Shortcomings in the attribution proce\ ss: On the origins and maintenance of erroneous assessments. In D. Kahneman, P. Slo\ vic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases (Vol. 11, pp. 143-157). Cambridge, MA: Cambridge University Press.

Schuman, H., & Presser, S. (1996). Questions and answers in attitude surveys: Exper- iments on question form, wording, and context. Thousand Oaks, CA: SAGE.

Wyman, A. J., & Vyse, S. (2008). Science versus the stars: A double-bl\ ind test of the validity of the NEO Five-Factor Inventory and computer-generated astrolo\ gical natal charts. Journal of General Psychology, 135, 287-300.

Bio Nick Allum earned his PhD in social psychology at the London School of Economics and Political Science. He is currently senior lecturer in sociology at t\ he University of Essex, U.K. He has published on public understanding of science, risk perception, trust, and survey methodology.