In this post, I will analyze Jonathan Haidt’s study (in partnership with Brian A. Nosek and Jesse Graham) about liberal and conservative perceptions of and stereotypes about moral foundations:
Haidt did this research on self-identified conservatives and self-identified liberals which invalidates it from the start. Self-report data is notoriously unreliable.
Here is a good summary of the study and in summarizing the author unintentionally pointed out the problem of self-reports:
One of the applications of those pairings is a study that Haidt describes in Reason this way:
“In a study I conducted with colleagues Jesse Graham and Brian Nosek, we tested how well liberals and conservatives could understand each other. We asked more than 2,000 American visitors to fill out the Moral Foundations Questionnaire. One-third of the time they were asked to fill it out normally, answering as themselves. One-third of the time they were asked to fill it out as they think a ‘typical liberal’ would respond. One-third of the time they were asked to fill it out as a ‘typical conservative’ would respond. This design allowed us to examine the stereotypes that each side held about the other. More important, it allowed us to assess how accurate they were by comparing people’s expectations about ‘typical’ partisans to the actual responses from partisans on the left and the right. Who was best able to pretend to be the other?
“The results were clear and consistent. Moderates and conservatives were most accurate in their predictions, whether they were pretending to be liberals or conservatives. Liberals were the least accurate, especially those who described themselves as ‘very liberal.’ The biggest errors in the whole study came when liberals answered the care and fairness questions while pretending to be conservatives. When faced with statements such as ‘one of the worst things a person could do is hurt a defenseless animal’ or ‘justice is the most important requirement for a society,’ liberals assumed that conservatives would disagree.”
In other words, conservatives understand liberals better than liberals understand conservatives. More precisely, conservatives’ version of liberals matches liberals’ version of themselves better than liberals’ version of conservatives matches conservatives’ vision of themselves.
That last sentence hits the nail on the head, without the author realizing it.
Haidt was studying perception. He oversimplified his conclusions by stating the liberal perceptions of conservatives were wrong for the reason they didn’t match conservatives’ perceptions of themselves. This is oversimplified because, as Haidt should know, self-perceptions are often inccorect (while apparent stereotypes aren’t always incorrect). Haidt would need to also measure the accuracy of self-perceptions among conservatives and liberals.
Haidt is measuring the symbolic ideology rather than the substance on specific issues. This is a failing not only of this particular study by Haidt but also a failing of his other studies as well. As far as I can tell, he is only using self-report data in developing his Moral Foundations Theory. He is asking people what they identify as and asking people which values they identify with (i.e., symbolic ideology). Neither of these gets at people’s pragmatic ideology or gets at whether people’s stated beliefs conform to the less-conscious values they act according to. It certainly doesn’t get beyond the superficialities and biases of self-perceptions and self-reports.
Maybe the difference Haidt is measuring is being incorrectly analyzed. Maybe liberals are correct in their views of conservatives while conservatives themselves have less accurate self-awareness about their own conservative values. Maybe liberals are looking past the rhetoric and talking points to the actual behavior of conservatives. Actual behavior says a lot more about someone’s actual values than their own claims about what they theoretically or idealistically value.
One of the conclusions that Haidt comes to in the study’s paper is that liberals and conservatives are closer together than either side realizes. This is probably true in one sense and untrue in another sense. This is true when speaking of the average American in terms of the average liberal and the average conservative. However, the problem is that the conservative movement includes a significant number of people who are fairly liberal in their political positions.
If these politically liberal self-identified conservatives were removed from the measure of conservatism, then the average conservative would be much further to the right. There is no similar percentage of politically conservative self-identified liberals and so the average liberal would remain about the same. So, Haidt would come to different conclusions if he did a study that categorized people according to their political positions rather than their political labels.
There is an even further point of possible confusion. Haidt does at least distinguish a third group of ‘moderates’. Polls show that most Americans will identify as conservatives if ‘moderate’ isn’t given as a choice. But if ‘moderate’ is given as a choice, most Americans identify as ‘moderate’.
So, there is a certain amount of overlap between moderates and conservatives. First, this would exacerbate the other overlap of political liberals self-identifying as conservatives. Second, considering there is a large percentage of Americans who will switch between the labels of ‘moderate’ and ‘conservative’, it makes issues of ideological conflation even more fuzzy. Furthermore, considering most Americans are politically liberal, these self-identified moderates are probably in ideological alignment with the politically liberal self-identified conservatives.
All of this leads one to wonder what ideological labels even mean. What is Haidt measuring? And what does Haidt think he is measuring?
Let me continue my analysis with some other types of questions and criticisms.
Were the subjects of the study a representative sample? If the sample was, for example, all or mostly college students who are more liberal, then it would mean that the conservatives were around more liberals and the liberals were around fewer conservatives. In the paper, this is what they say about the participants:
“The participants were 2,212 visitors (62% female; median age 28; only U.S. residents or citizens) to ProjectImplicit.org”
This basically fits in with my doubts. Like college students, females and the younger tend to be more liberal. In general, more liberals are probably found online than in other environments. So, Haidt’s sample would include more people who are liberal, liberal-minded, or otherwise familiar with liberalism and liberal-mindedness. I’d argue that this doesn’t offer a fair and accurate representation of the general population.
This brings me to other confounding factors.
Most older people are conservatives. Simply being older can potentially give one more perspective and experience. Younger generations are more liberal than the older generations were at the same age. However, when these younger liberal generations grow older, they probably will maintain their higher rates of liberalism (as did the older generation maintain their lower rates of liberalism) and will also gain more perspective and experience. Furthermore, younger conservatives may have no better understanding of liberals than younger liberals have of conservatives.
So, are the results of the study merely pointing to a demographic fluke at this point in history? If the study controlled for age, would different results be found?
Ignoring all of that, maybe there is something going on that Haidt isn’t even considering.
First, authoritarianism and social conservatism have been shown to have strong correlation:
“For several decades Bob Altmeyer, an American scholar at the University of Manitoba, has been a tireless and dedicated researcher. According to the Oxford Handbook of Political Psychology, Altmeyer’s work “powerfully predicts a wide rang of political, social, ideological and intergroup phenomena.” Altmeyer’s work is largely directed at other psychologists and social scientists. He has undertaken hundreds of experiments and his work is reliable and valid according to Paul Nesbitt-Larking reporting in
“Political Psychology in 2004. His work goes the distance in understanding conservatism.
“In an article titled “What Happens When Authoritarians Inherit the Earth? A simulation,” Altmeyer explains that, “When I started out, and ever since, I was not looking for political conservatives. I was looking for people who overtly submit to the established authorities in their lives, who could be of any political/economic/religious stripe.” His work identified “right-wing authoritarians” but he was not using the term “right-wing” in the political sense. Rather he used the designation in a psychological sense.
“But as he continued his work he reports that “it turns out that in North America persons who score highly on my measure of authoritarianism test tend to favor right-wing political parties and have ‘conservative’ economic philosophies and religious sentiments. He goes on to say that this empirical finding has been repeatedly duplicated in his continuing studies and has been replicated in studies by others.
“The extensive research on the behavior and personality characteristics of right-wing authoritarians and conservatives concludes that they are people who do not see themselves as they actually are and have little facility for self-analysis.
“The research demonstrates that conservatives delight in hurling invectives against their enemies and often prove to have the thinnest of skins if the same is done to them. Many conservatives are unaware of their illogical, contradictory and hypocritical thinking. And if they are forced to address it, either rationalize it away, fail to care, or go on the attack against those who reveal their human weaknesses.”
And authoritarianism has been correlated with higher rates of hypocrisy:
“Research reveals that authoritarian followers drive through life under the influence of impaired thinking a lot more than most people do, exhibiting sloppy reasoning, highly compartmentalized beliefs, double standards, hypocrisy, self-blindness, a profound ethnocentrism, and — to top it all off — a ferocious dogmatism that makes it unlikely anyone could ever change their minds with evidence or logic.”
It’s certainly not hard to find inconsistencies among conservatives, especially social conservatives. For example, conservatives have the highest rate of porn consumption and conservative states have the highest rates of teen pregnancy. I could list many more examples, but that isn’t as helpful as research on authoritarianism and hypocrisy. By the way, liberals test high on ‘openness’ which is something authoritarians, of course, test extremely low on (‘openness’ probably disinclines someone towards the worst forms of hypocrisy as found in authoritarian groupthink).
So, it is possible that many individual social conservatives even if only moderately authoritarian, enough to skew the conservative sample in that direction, are less consistent in their beliefs than the average liberal. This might skew the entire study for conservatives are familiar with their own inconsistencies and it is easier to be aware of your own inconsistencies (or the inconsistencies of those in your own group) than to be aware of the inconsistencies of someone who is entirely different from you, although being intuitively aware of inconsistencies doesn’t necessarily imply a broader self-awareness. On the opposite side: If liberals are more consistent in their beliefs, then it probably would be easier for a conservative to understand liberal beliefs. It is harder to have accurate views of an inconsistent group, especially one that hypocritically betrays its own stated values on a regular basis.
I really don’t know how this factor might play out in Haidt’s study. I have no data about ideological self-awareness as it might relate to authoritarianism and hypocrisy. My main point is simply that it is a confounding factor not being controlled for.
Secondly, it seems obvious that self-identified conservatives as a group are less ideologically homogenous than self-identfied liberals as a group. This would contribute to the conclusion that conservatives are collectively more inconsistent which would make it harder for an outsider to assess their beliefs on average. To determine this, testing for the average beliefs of liberals and conservatives wouldn’t be adequate, and so the rate of diversity of beliefs would also need to be tested. This is particularly problematic for Haidt’s entire theory as he is relying on self-defined labels which are notoriously unreliable because people’s defintions of such labels aren’t consistent. I’ve analyzed the complexity of this problem before.
The problem of the study is that self-described liberals are a smaller and more narrowly defined demographic group whereas self-described conservatives are a larger and more broadly defined demographic group, diversity making the conservative movement itself less consistent. As Pew data shows, almost 1 in 10 Americans holding strong liberal beliefs self-identify as conservative, but you don’t find a large number of Americans holding conservative beliefs self-identifying as liberal.
The conservative label, besides including 9% of Pew’s “Solid Liberals” (liberal across the board, both fiscally and socially), includes neocon progressives and war-hawks, free market neoliberals, fundamentalists as well as theocrats, some libertarians and socially liberal fiscal conservatives verging on libertarianism, patriotic statists wanting a militaristic empire, anti-statists wanting a weak government, openly gay Republicans, WASP culture warriors, white supremancists, gun-toting militants and survivalists, constitutionalists, small town rural types, elderly people remembering a conservatism that no longer exists, former Cubans who hate communism, small business owners fighting free trade globalism, big business defenders promoting free trade globalism, corporatists verging on fascism, anarcho-capitalists, Randian Objectivists, traditional Catholics who have high rates of membership in unions, union-bashing think tank employees, ordinary people wanting to conserve progressive reforms such as social security, politicians promoting the ending of the progressive reforms such as privatizing social security, etc.
You don’t find such massive diversity among self-described liberals.
A further problem is that self-described liberals don’t represent all liberals. As I pointed out, many liberals self-identify as conservatives. Other Pew data also has shown almost half of liberals self-identifying as Independents. I would suspect that most moderates hold a majority of liberal beliefs, values and policy positions. The data shows that most Americans in general, despite large numbers self-identifying as conservative, are actually very liberal on many key issues.
One other factor to consider is the mainstream media. Maybe some of the most popular pundits (or, most popular or not, pundits with the most ability to make themselves heard) such as Limbaugh and Beck aren’t representative of the average conservative. So, a liberal might mistakenly base their views of the typical conservative solely or largely on these few far right pundits. However, maybe the opposite isn’t true. As liberals are more narrowly defined as a group and because liberal activists are less radicalized, popular pundits of liberalism might be more representative of the average liberal and so the media ends up, intentionally or not, giving conservatives more accurate information about liberal beliefs.
Even if it is true that liberals are inaccurate in their assessment of the beliefs of the average conservative, liberals may be accurate in their assessment of the beliefs of conservatives in the media and other powerful conservatives who control the political narrative of the conservative movement. This is an important difference since pundits and other powerful people have more influence and control over party politics than does the average citizen. Political movements are defined more by their activists and leaders than by the average person identifying with the movement. It might be possible, though, that liberal activists and leaders are closer to the average liberal than is found with conservatives in the conservative movement.
How is the liberal supposed to know what the average conservative thinks when the spokespersons for the conservative movement don’t represent the average conservative? If this is the case, this would be more of a criticism of the conservative movement that causes such confusion than a criticism of the liberal who is confused by it. Going by this interpretation, I would posit that this possible liberal misperception of conservatives would be based on the mischaracterization of the average conservative by the conservative media itself and based on how the rest of the MSM mostly accepts this right-wing framing of the conservative movement.
All of the mainstream media and all of mainstream politics is similarly confused. You’ll never see acknowledged in the MSM that, although the average American would rather self-identify as a conservative than a liberal, the average American holds liberal views on many if not most major social, economic and political issues. The average liberal is simply repeating what they’ve learned from the MSM which is problematic in itself. It is sad that we must judge liberals for believing what they’ve been told by supposedly trusted news institutions. The MSM has misinformed the American people about the general public being more conservative than it actually is and misinformed the American people about the average conservative being more right-wing than they actually are.
I admit that it is a sad state of affairs. I wish liberals better understood that the average conservative is closer in opinion to the average liberal than the average conservative is to the radicalized activists, leaders and pundits of the conservative movement. This is hard for most liberals to wrap their minds around for these non-radicalized average conservatives keep being manipulated by the radicals in their movement and hence voting the radicals into power, at least in recent decades.
Still, being manipulated by radical rhetoric isn’t the same thing as being radical oneself. Even liberals can be manipulated by the radical rhetoric of the right-wing which is what happened after 9/11. Fear works, sad but true.
Another factor to keep in mind relates back to hypocrisy. Haidt is testing for self-described labels and self-described values. This might in some cases have little to do with actual behavior. If on average conservatives show more inconsistency between their stated values and their actual behaviors, why would Haidt judge liberals as being inaccurate for basing their assessment on the actual behaviors of conservatives rather than on their stated values? Liberals shouldn’t be blamed for assuming that conservatives are more like liberals in thinking that their actual behaviors match up with their stated values. This speaks maybe to the naivette of liberals in not appreciating conservative hypocrisy, but such naivette certainly isn’t a moral failure. I think that ‘moral consistency’ (i.e., lack of hypocrisy) should be added to Haidt’s moral values.
There is also an irony in Haidt doing this kind of research. As I pointed out in my reviews of his theory, it seems obvious to me that Haidt lacks an accurate and unbiased assessment of liberal moral values. The fact that the theory itself is problematic makes any research based on it problematic. Maybe Haidt merely ends up testing for which group ends up agreeing the most with his own personal bias.
Another irony in a scientist like Haidt promoting conservative values is that research shows that conservatives mistrust science more and that this mistrust has been increasing:
But a similar strong and persistant bias can’t be found among liberals:
“So now the big question: Are liberals also “smart idiots”?
“There’s no doubt that more knowledge—or more political engagement—can produce more bias on either side of the aisle. That’s because it forges a stronger bond between our emotions and identities on the one hand, and a particular body of facts on the other.
“But there are also reason to think that, with liberals, there is something else going on. Liberals, to quote George Lakoff, subscribe to a view that might be dubbed “Old Enlightenment reason.” They really do seem to like facts; it seems to be part of who they are. And fascinatingly, in Kahan’s study liberals did not act like smart idiots when the question posed was about the safety of nuclear power.
“Nuclear power is a classic test case for liberal biases—kind of the flipside of the global warming issue–for the following reason. It’s well known that liberals tend to start out distrustful of nuclear energy: There’s a long history of this on the left. But this impulse puts them at odds with the views of the scientific community on the matter (scientists tend to think nuclear power risks are overblown, especially in light of the dangers of other energy sources, like coal).
“So are liberals “smart idiots” on nukes? Not in Kahan’s study. As members of the “egalitarian communitarian” group in the study—people with more liberal values–knew more science and math, they did not become more worried, overall, about the risks of nuclear power. Rather, they moved in the opposite direction from where these initial impulses would have taken them. They become less worried—and, I might add, closer to the opinion of the scientific community on the matter.
“You may or may not support nuclear power personally, but let’s face it: This is not the “smart idiot” effect. It looks a lot more like open-mindedness.”
So, the very act of scientifically studying biases, including liberal biases, is typically going to get strong support from liberals and weak support (if not outright antagonism) from conservatives. Even if liberals were more biased about certain issues, that may be less relevant in that liberals also show a stronger desire to correct their own mistaken views.
To me, this relates back to the issue of consistency and hypocrisy. If liberals are more aware of inconsistencies when they occur, they will put more effort into becoming more consistent. I’d love to see Haidt not only study moral values but also how those values relate or don’t relate to moral behavior, especially the specific moral behaviors that the moral values imply.
I should clarify that I’m not arguing that liberals are morally better in all or even most ways. As I see it, there are strengths and weaknesses to both conservative and liberal predispositions. What I am suggesting, though, is that it might be possible that liberals are more self-aware of their own moral failings, at least in terms of being less prone to confirmation bias and the smart idiot effect when it comes to their own cherished beliefs and opinions. The hypothetical part would be whether being more self-aware of moral failings actually leads to lessening those moral failings and hence seeking to morally improve oneself, beyond merely being willing to change one’s mind according to new info.
As a liberal-minded person, what I care about are the facts even when or especially when they contradict or put doubt to my beliefs. However, conservatives don’t equally share my concern and this bothers me, almost causes me to lose hope.
Because of my liberal respect for science, I feel compelled to take Haidt’s theory seriously and to carefully look at his data. What I’ve come to is doubts about how Haidt is going about his research. My doubts are only increased as his conclusion doesn’t seem to fit the broader range of research about biases in terms of conservatives and liberals. I’d like to see Haidt’s response to all this other research and why it seems to point away from his preferred conclusion.
So, I honestly don’t know what to make of it. If someone can fix some of the problems of Haidt’s research model and yet come to similar results, I would be more convinced of his conclusion. Until then, it’s just data, just as likely to turn out to be meaningless as meaningful.
Here are a few of my previous posts about Jonathan Haidt, Moral Foundations theory, and the conservative/liberal distinction:
And here are some relevant commentary on Haidt’s theory and research: