Email to John Ray
Change, Superstitions, Job Security, Beaten Children, and Liberal Authoritarian Religions: Additional Comments on "Political Conservatism as Motivated Social Cognition", by Jost, John T.; Glaser, Jack; Kruglanski, Arie W.; Sulloway, Frank J.
From: James Lindgren
July 27, 2003
The more I read the paper on "Political Conservatism" by Jost, Glaser, Kruglanski & Sulloway, the more puzzled I get. I have read some of the cited background papers and even explored some of their variables, but many of my concerns might be somewhat answered by reading more of these background papers in depth, which I am just beginning to work through. My early comments depend a lot on how the evidence is presented in the Jost group's paper. Remember, this is an email, rather than a work of scholarship. These comments are quite tentative. Nonetheless, I do report on some better data than is used by Jost to try to establish some of his arguments _ indeed, questions asked of actual self-reported conservatives representative of the general public, not available undergraduates or people considered to be conservatives because they scored highly on an authoritarianism scale or a dogmatism scale.
My first impressions of this literature are that its practitioners bring considerable statistical and modeling sophistication to their work, but a surprising way of framing questions and inferring conclusions from evidence. It appears that the subfield might have gotten off track early on by assuming that authoritarianism and conservatism were more or less the same thing (when most 20th century authoritarian regimes were left-wing) and then developing measures that moved together but didn't match what actual, typical conservatives in America think and feel.
Much too little attention in this subfield has been paid to the problem of generalizing from atypical student populations. I remember what psychologist and master survey designer Norman Bradburn once said to me. I suggested to him that perhaps representativeness wasn't much of an issue in a study that I was designing, since it was an experiment. He replied: "Of course you don't need a representative sample, as long as you don't plan on generalizing your results." I got the message. I am not suggesting that one shouldn't use casual samples (I have), just that here the results of so many studies done on casual samples cited by the Jost group seem at odds with large, carefully designed representative studies of the general public.
I expect that much good will come of the shocked reaction to the Jost review. At a minimum, the subfield will be opened to new databases for testing its pet theories. Much more might be accomplished if the links between authoritarianism and conservatism turn out to be not what has been assumed for so many years by most, but not all, of the subfield. To an outsider completely new to this subfield of psychology, some of the main claims look highly implausible and the evidence for them exceedingly thin.
1. WHERE THE JOST GROUP IS AT LEAST PARTLY CORRECT
Let me begin with a few positive comments on Jost's paper. In checking several more representative data sources than most of Jost's studies, I did find important parts of Jost's story to be borne out. Conservatives are indeed considerably more moralistic and more religious than liberals on many issues. Conservatives as parents do indeed favor spanking and teaching children to obey, though (contrary to Jost's claims) the survey evidence tends to undercut Jost's claims that conservatives themselves were raised in more punitive environments. Liberals, rather than conservatives, report higher levels of being beaten as children (see Part 4 below), as fits liberals' generally slightly higher level of reported social and psychological problems.
As to blind deference to authority, in General Social Survey data, conservatives are not more likely than liberals to say that people should obey church teachings, but conservatives are indeed more likely to say that people should obey the law without exception. Even here, however, the connection between political ideology and full deference to law is a very weak one, hardly the kind of effect that could give rise to a defining characteristic of conservatism, let alone a conflation of authoritarianism and conservatism. I will deal with this at the end (Part 7).
Last, Jost is correct that conservatives in America do tend to be more tolerant about inequality than liberals, though even here the story is a complex one, since conservatives tend to favor equality of opportunity and liberals tend to favor equality of results. When these conflict, my impression is that conservatives tend to go strongly for the former, while liberals sometimes split on the issue. Now to some basic problems with the Jost group's paper . . .
2. RESISTANCE TO CHANGE
According to the Jost paper, one of the two defining characteristics of conservatives is resistance to change (the other is that conservatives are tolerant of inequality). You would expect to find actual evidence in the paper that conservatives are resistant to change, whether change comes from the right or the left. If conservatives favor right-wing change and oppose left-wing change, then they are not generally resistant to change. After all, psychologists often think it crucial to balance positive and negative items in a scale; you would think that they would take the much more necessary step of balancing left-wing and right-wing change items before concluding that a defining characteristic of conservatism is resistance to change.
As evidence for resistance to change, Jost cites a 1981 paper by Conover and Feldman: "Consistent with this notion, Conover and Feldman (1981) found that the primary basis for self-definitions of liberals and conservatives has to do with acceptance of, versus resistance to, change . . . ."
I have read the Conover/Feldman paper and it shows nothing of the kind. In one model, using 1976 NES data, it shows significant results for an index of economic issues and an index of social issues. I went back to the 1976 NES codebook and found that of the 7 questions in these indices, 6 involved proposed conservative changes as the highest conservative response:
favoring raising marijuana penalties;
favoring a flat tax;
favoring prohibiting all abortions (contrary to Roe v. Wade);
favoring women in the home, not business (contrary to substantial female workforce participation);
favoring individual, not governmental, payment of medical expenses (contrary to then-existing Medicaid); and
favoring people getting ahead on their own, not government ensuring jobs and a good standard of living (contrary to existing jobs programs, welfare, and taxation policies).
In other words, Conover and Feldman's data show that conservatives tend to favor right-wing change, though no mention of this is made in the text of the paper (there are some other oddities that will have to await a fuller scholarly analysis). For myself, I wouldn't take the conservative endpoint on any of these questions, but on some I wouldn't prefer the liberal endpoint either. The only one of the 7 index issues where conservatives were opposed to change from the then-existing status quo was the passage of the ERA -- a left-wing change. There is no evidence presented in the Conover/Feldman data of a conservative general resistance to change from both sides of the political spectrum.
Conover and Feldman also coded answers to an open-ended question in the 1978 NES and found that the second most important idea defining conservatism mentioned for self-identified conservatives was change, while for liberals change was the most important idea. They did not code the attitude toward change, so it's impossible to tell from these data whether conservatives defined themselves as pro-change or anti-change. The most often mentioned idea for conservatives' self-definition was "Fiscal Policies," not change (let alone resistance to change), as the Jost paper incorrectly contends. There is a little more that is arguably relevant in the Conover/Feldman paper, but nothing establishing a conservative preference against both left-wing and right-wing reform.
Jost's claim that "Conover and Feldman . . . found that the primary basis for self-definitions of liberals and conservatives has to do with acceptance of, versus resistance to, change" is flatly false. If there is evidence that actual, typical conservatives are resistant to right-wing change, Jost failed to mention it in his review article, or I missed it. It seems the most obvious of interpretive errors to confuse opposition to left-wing reform for opposition for reform. Given the evidence that conservatives favor right-wing change and the lack of evidence that they don't, I don't see how any one would conclude that conservatives are resistant to change. Perhaps the evidence lies elsewhere.
3. JOB SECURITY AND RESISTANCE TO CHANGE.
The Jost group argues that this supposed resistance to change carries over to the workplace. To support this claim about the workplace, they cite a study of 155 US undergraduate and graduate students (Atieh et al.) and a study of 478 workers from former East Germany (Fay & Frese). According to Jost's account, the latter study seemed to be dealing with authoritarianism, a difficult concept to identify with conservatism in a former left wing authoritarian state. The Fay & Frese abstract concludes: "Specifically, conservatives appear to be reluctant to take responsibility and show less personal initiative at work. Further, conservatives seem less ready to change at work, less interested in work innovation, make fewer attempts to introduce innovations at work, show less active career planning, and are less oriented toward growth and challenge."
Once again, some of these things have been studied, not with casual samples of available subjects, but with carefully drawn samples representative of the noninstitutionalized US population. Over the years, the GSS has asked over 16,000 respondents to rank order five attributes they would prefer in a job. Both liberals and conservatives rank the same attribute first -- the "work [is] important and gives a feeling of accomplishment." But conservatives are significantly more likely to rank this attribute higher than liberals do, suggesting a greater commitment among conservatives to accomplishment.
In second place for liberals is a "high income" job, while for conservatives the second spot is "chances for advancement." This suggests, contrary to the much smaller East German study of authoritarianism that the Jost group relies on, that conservatives are the ones who look forward to change and promotions and embrace change; liberals instead want the security of a high income. In fourth place is "no danger of being fired," with no significant differences between liberals and conservatives. In last place for both groups is "working hours are short, lots of free time," with liberals again giving higher rankings to short working hours.
This is very strong evidence over many years and 16,000 subjects that, as to job security, liberals and conservatives are similar, but when ranking job attributes conservatives value more highly chances for advancement and doing important work and a feeling of accomplishment. Compared to conservatives, liberals express crasser, more conventional values, a high income and short working hours. This fits with other data, such as that showing that conservatives have more control over their jobs at work, again contrary to the German study. The best evidence that has been brought forward to date suggests that the results from the former East Germany are not generalizable to the US. Or perhaps the measures that the authors used confused conservatism with left-wing authoritarianism; their abstract inauspiciously begins by saying that they are examining "conservatism (defined as rooted in a general intolerance of uncertainty) . . . [emphasis added]."
4. WERE CONSERVATIVES BEATEN AS CHILDREN?
Jost provides exceedingly meager evidence for his suggestion that conservatives were treated harshly as children. In the GSS over the years (16,437 respondents to this question), liberals are more likely to report having been beaten as a child (31.8% for extreme liberals to 23.4% for extreme conservatives). That fits with other problems related to social disorganization that liberals report more than conservatives _ liberals are more likely to report being arrested, taking drugs (such as cocaine and marijuana), paying for sex or being paid for sex, seeing pornographic movies in the last year, drinking, getting drunk, having a drinking problem, hanging out in bars, smoking, never having been married, if ever married cheating on their spouses, if ever married being divorced, experiencing and being treated for mental problems, and having been unemployed for more than 2 weeks at some time. These are not directly on point, but they are consistent with the background of social disorganization that liberals tend to report somewhat more often than conservatives. Given this background, the reports of more childhood beatings for liberals are an extremely plausible part of a consistent picture. Also fitting is the higher satisfaction that conservatives report for their jobs, marriages, families, health, friendships, and the places they live.
In addition, John Ray points out in a 1976 article that Milgram and his associate found that those who defied the experimenters' orders to comply (the "rebels") actually had less close relationships with their fathers and were more likely to be treated punitively as children. While there is good evidence that conservative parents favor spanking more than liberal parents, the evidence that actual conservatives representative of the general population were more likely to have been beaten or treated punitively as children is not supported by the weak evidence offered in the text of the Jost article and would directly contradict the self-reports of thousands of representative liberals that they were beaten as children more often than conservatives.
5. SUPERSTITIOUS LIBERALS
In reviewing other people's work, Jost twice associates conservatives with superstition, e.g.: "Conservative attitudinal responses to these sources of uncertainty include superstition . . . ." Further, Jost makes extensive use of Wilson's C-scale, which codes superstition as conservative.
Once again, there are actual data on how many liberals and conservatives believe in superstitions. In 1996, Gallup asked a battery of questions on superstitions, and on almost every one, liberals were more likely to believe in superstitions (most appeared to be significant, though significance information was not present in the Roper poll database I used).
September 3, 1996 Gallup Poll (n=1,000)
Data Source: Roper Poll Database
Percentage Not Believing in "Superstitions":
ESP................................Lib 21, Con 31
Telepathy.......................Lib 34, Con 40
Haunted Houses ...........Lib 44, Con 53
Clairvoyance..................Lib 38, Con 50
Astrology........................Lib 42, Con 56
Ghosts...........................Lib 41, Con 58
Reincarnation.................Lib 43, Con 63
Commun. w Dead...........Lib 48, Con 61
Telekinesis.....................Lib 49, Con 58
Witches..........................Lib 62, Con 73
Channelling....................Lib 61, Con 66
Deja vu..........................Lib 17, Con 28
Devil................................Lib 44, Con 29
Possession by Devil.......Lib 50, Con 36
Angels.............................Lib 20, Con 14
Then Gallup asked about experiences.
Percentage answering YES to having experienced the following:
In touch with dead............Lib 19, Con 16
Consulted psychic............Lib 21, Con 12
Been in Haunted House..Lib 18, Con 16
Talked with Devil.............Lib 6, Con 7
Been with Ghost..............Lib 10, Con 9
Previous Life...................Lib 13, Con 5
Read Horoscope Reg......Lib 20, Con 15
Deja vu............................Lib 68, Con 55
Gallup also asked whether people were superstitious about a black cat, a ladder, the number 13, breaking a mirror, speaking ill of a person making it come true, and knocking on wood. Each time, liberals reported being more superstitious, though only the latter was probably significant.
Except for devils and angels, this is a remarkably consistent pattern. Otherwise, liberals were more likely to believe in a wide range of superstitions. On the same day, Gallup asked about UFOs and by a margin of 56% to 39%, liberals believe that UFOs have visited the earth. Liberals reported acting on superstitions as well_consulting a fortune-teller or psychic (21% to 12% for conservatives) and reading astrology columns (20% to 15% for conservatives). Last, by a margin of 13% to 5%, liberals sometimes thought that they were here on earth in a previous life.
Nor is this the only such poll. A 1998 CBS poll found that 43% of liberals thought that some people can communicate with the spirits of dead people, compared to only 29% of conservatives. A 1994 Gallup poll found that liberals believed in Astrology (29% to 20%).
Only if you view religion as superstition (as an atheist myself, I will refrain from offering my opinion) is there any evidence that conservatives are superstitious. Overall, Gallup's data suggests that liberals are consistently more superstitious than conservatives.
This raises problems for users of Wilson's C-scale (which treats superstition as evidence of conservatism). The original form of the Wilson scale, however, included only one item that could be seen as dealing with superstition -- the item: "Horoscopes". Among the students Wilson worked with, the more conservative students DID endorse horoscopes. What Jost et al failed to mention, however, is that the reverse is true if the Wilson scale is applied to a non-student sample. This is shown by the negative correlation with the C-scale total for the "Horoscopes" item reported in Ray (1971). So there has been in the literature for over 30 years evidence that the Wilson scale codes superstition incorrectly. For over 30 years there has been evidence in the academic literature to show that -- contrary to Jost et al., -- superstition is in fact more leftist than rightist.
6. AUTHORITARIAN RELIGIONS
"There is by now substantial archival research suggesting that during times of societal crisis, people are more likely to turn to authoritarian leaders and institutions for security, stability, and structure (e.g., Doty, Peterson, & Winter, 1991; McCann, 1997; Peterson et al., 1993; Rickert, 1998; Sales, 1972, 1973).
Sales (1972), for instance, found that during periods of severe economic threat (the depression years of 1930-1939), people were more likely to join authoritarian churches, such as Southern Baptist and Seventh Day Adventist, and less likely to join nonauthoritarian churches, such as Northern Baptist and Episcopalian, compared with periods of relative prosperity (1920-1930). Similarly, years of heavy unemployment in Seattle, Washington (1961, 1964, 1969, and 1970), were accompanied by higher than usual conversion rates there for an authoritarian church _ Roman Catholic _ and lower than usual conversion rates for a nonauthoritarian church _ United Presbyterian _ whereas relatively good economic years in Seattle (1962, 1965, and 1966) coincided with lower than usual conversion rates for the Roman Catholic Church and higher than usual conversion rates for the United Presbyterian Church.
There is a big problem with this argument, at least as presented. Jost is using church membership information to indicate authoritarian and nonauthoritarian churches and then to generalize that membership information to conservatives. But Jost has made no attempt to determine whether the memberships of the churches he calls authoritarian in the periods studied (1920-39, 1961-70) were indeed politically conservative and whether the nonauthoritarian churches were politically liberal. Since Catholics were a big part of the New Deal coalition and mainstream Protestant churches were thought to be relatively Republican through at least into the 1970s, it may be that people were shifting to liberal authoritarian churches in times of crisis _ a pattern that would undercut not just this argument but the Jost paper's repeated conflation of authoritarianism with conservatism. If Jost's authoritarian churches are at least as liberal than his nonauthoritarian churches, then the equation of conservatism with authoritarianism breaks down.
I have looked at GSS data (1975-89) on members of the churches Jost calls authoritarian -- Catholic and Southern Baptist (no data on 7th Day Adventists) -- and nonauthoritarian _ Episcopalian, United Presbyterian, and Northern Baptist (presumably now the Am Baptist Church in the USA). Most years there are no significant differences on self-described liberal/conservatism, and some years are missing data on some denominations. When there are significant differences (1776, 1982, 1987), Jost's supposed authoritarian churches are more liberal than his supposed nonauthoritarian churches. For example, in 1987 on average Catholics and Southern Baptists are both more liberal than the general US population and more liberal than Episcopelians, United Presbyterians, and Northern Baptists _exactly the opposite of Jost's assumptions. This difference was probably even stronger in the 1961-70 period when most of the data for the studies Jost uses were done, though one would hope for better data than these GSS data starting in 1974.
Once again, Jost's conflation of conservatives with authoritarians breaks down because (where there was a difference) conservatives were more likely to be members of Jost's nonauthoritarian churches than his authoritarian churches.
7. BLIND OBEDIENCE TO AUTHORITY
The idea that conservatives tend to favor blindly obeying authority is at the heart of the literature Jost reviews and of his paper's repeated conflation of conservatives with authoritarians. In General Social Survey, respondents have been asked:
In general, would you say that people should obey the law without exception, or are there exceptional occasions on which people should follow their consciences even if it means breaking the law?
Conservatives are indeed more likely to say that people should obey the law without exception and not follow their consciences _ from 39.5% for slight conservatives to 45.1% for conservatives to 43.5% for extreme conservatives, compared to 46.3% for moderates and 37% for slight liberals to 33.0% for liberals to 33.3% for extreme liberals. The connection between political ideology and full deference to law is a very weak one, not the kind of effect that could give rise to a defining characteristic of conservatism.
Indeed, it can't be a defining characteristic when most conservatives join most liberals in rejecting conformity to law without exception. For example, 63-67% of liberals reject comformity without exception, compared to 54.9-60.5% of conservatives, a significant result only because of a very large sample size (2768). Indeed, political moderates report the highest adherence to following the law without exception (46.3%), just nosing out conservatives.
Let me put it another way, the differences between liberals and conservatives on obeying the law without exception are so small that they would not even be significant with the sample sizes that are used by the scholars in the overwhelming majority of the studies that the Jost group relied on. How could obeying authority without exception be a defining characteristic of conservatism when the effect is so small that it wouldn't be likely to show up even in a sample of 1,000 respondents, if they were drawn as representative of the general public?
I am not arguing that the greater conservative deference to obeying the law is not real or should be ignored, just that the effect is much too small to be a central feature of conservatism, let alone a reason for conflating authoritarianism with conservatism. Like the greater likelihood of liberals reporting having been arrested (9-12% for conservatives, 14-20% for liberals), the difference is real and significant, but being arrested by the police is hardly a defining characteristic of liberals. I would like to see other representative data before reaching a firm conclusion on this.
Interestingly, in exploring the data, I found two groups that expressed adherence to obeying the law without exception, ignoring even conscience. A staggering 70.9% of Asian-Americans said that the law should be obeyed without exception and a majority of African-Americans (55.6%) said that the law should obeyed without exception, compared to just 39.5-45.1% for conservatives. Since both ethnic groups are somewhat more liberal than the US average, this finding of a high deference for obeying authority is a big problem for those who would associate assertions of blind obedience to conservatives.
Another dimension of the Wilson C-Scale is supposed to be a preference for conventional institutions. Yet liberals show as great (or greater) confidence in US institutions as conservatives do. Liberals are more confident of educational institutions, the US Supreme Court, organized labor, the press, the scientific community, and television. Conservatives are more supportive of the military, banks, businesses and corporations, and churches. Interestingly, moderates almost always display higher than average confidence in US institutions (though perhaps usually not significant by itself).
PREVIOUS COMMENTS (revised):
Fear, Anger, and Happiness
July 25, 2003
The Jost article claims that conservatives are angry and fearful and it builds on a literature that claims that conservatives are unhappy. I find this strange, given the decades of superb data showing the opposite. In the NORC General Social Survey (a standard social science database, second only to the U.S. Census in use by U.S. sociologists), the GSS asks the standard survey question about happiness in general. In the 1998-2002 GSS, extreme conservatives are much more likely to report being "very happy" than extreme liberals--47.1% to 31.6%. Earlier years show a similar pattern.
This conservative happiness carries over into most other aspects of life as well. Conservatives usually report being happier in their jobs than liberals. In the 2002 GSS, for example 65.2% of extreme conservatives report being "very satisfied" with their jobs in general, while only 50% of extreme liberals report being very satisfied. When the question is broadened to satisfaction with job or housework, a similar pattern obtains. In the 1998-2002 GSS, 61.0% of extreme conservatives reported being very satisfied, compared to 53.6% of extreme liberals.
As to finances, in the 1998-2002 GSS 34% of extreme conservatives report being satisfied with their finances compared to 26.4% of extreme liberals. More extreme liberals (34.5%) than extreme conservatives (25.8%) report being "not at all satisfied" with their finances.
Conservatives usually tend to report less marital unhappiness than liberals. In the 1998-2002 GSS, 5.1% of those who report being "slightly liberal" say that they are "not too happy" in their marriages, compared to 0.9% of those who are "slightly conservative." Ordinary liberals (3.7%) and extreme liberals (8.9%) also differ from ordinary conservatives (2.4%) and extreme conservatives (4.1%) in the levels of reported marital unhappiness. Indeed, in the 1998 GSS, 18.2% of extreme liberals reported that their marriages were "not too happy," while only 1.6% of extreme conservatives reported marital unhappiness.
Earlier General Social Surveys found that conservatives were more satisfied with their health, their friendships, their family life, and the city or place they live--all in all, a remarkably consistent picture.
Another claim in the Jost paper is that conservativism is driven by anger and fear. Again, their claims conflict with some of the highest quality data available. In the 1996 GSS, questions were asked about anger and fearfulness. Extreme conservatives were much less likely to report being mad at someone or something every day in the last week--7.3% to 24.2% for extreme liberals. Extreme conservatives were also less likely to report being fearful in the last week--32.5% to 56.3% for extreme liberals. In other words, a staggering one-quarter of extreme liberals report being mad at someone or something EVERY DAY and most extreme liberals report being fearful at least once a week.
An ABC News/Washington Post poll in 1995 asked whether you fear dying and 27% of liberals feared dying, while 20% of conservatives feared dying. In David Raden's mostly superb 1994 article on symbolic racism, he extensively analyzes 1988 GSS data. He uses 3 GSS variables "in the authoritarian-conservatism area such as . . . anomia": "It's hardly fair to bring a child into the world," "the lot of the average man is getting worse," and "officials . . . are not really interested in the average man." According to Raden, these feelings of anomia are supposed to be part of the authoritarian syndrome, yet Raden doesn't mention that not one of them is significantly related in the 1988 GSS to conservatism (and the direction of the coefficients is the opposite of that supposed to obtain). If you add in all other years of the GSS where these questions were asked, the last question is still insignificant, but for the other two questions, as respondents get more conservative they display less (not more) anomia of the type hypothesized by Raden.
I wonder whether Jost relied too much on studies that either used unrepresentative samples (such as undergraduates) or used biased questions or indices -- asking about issues on which conservatives tend to be unhappy but not about issues on which liberals tend to be unhappy. In any event, there is a host of data relevant to the questions that the Jost article addresses that appear not to have penetrated the literature of political psychology, including decades of very high quality survey data that undercut their thesis.
Professor of Law, Northwestern University
Director, Demography of Diversity Project
Ph.D. Student, Sociology, University of Chicago
Go to Index page for this site
Go to John Ray's "Tongue Tied" blog (Backup here or here)
Go to John Ray's "Dissecting Leftism" blog (Backup here or here)
Go to John Ray's "Australian Politics" blog (Backup here or here)
Go to John Ray's "Gun Watch" blog (Backup here or here)
Go to John Ray's "Education Watch" blog (Backup here or here)
Go to John Ray's "Socialized Medicine" blog (Backup here or here)
Go to John Ray's "Political Correctness Watch" blog (Backup here or here)
Go to John Ray's "Greenie Watch" blog (Backup here or here)
Go to John Ray's "Food & Health Skeptic" blog (Backup here or here)
Go to John Ray's "Eye on Britain" blog (Backup here or here)
Go to John Ray's "Leftists as Elitists" blog (Not now regularly updated -- Backup here or here)
Go to John Ray's "Marx & Engels in their own words" blog (Not now regularly updated -- Backup here or here)
Go to John Ray's "A scripture blog" (Not now regularly updated -- Backup here or here)
Go to John Ray's recipe blog (Not now regularly updated -- Backup here or here)
Go to John Ray's "Some memoirs" (Occasionally updated -- Backup here)
Go to John Ray's Main academic menu
Go to Menu of recent writings
Go to John Ray's basic home page
Go to John Ray's pictorial Home Page (Backup here)
Go to Selected pictures from John Ray's blogs (Backup here)
Go to Another picture page (Best with broadband)