Close to a week ago now Luke Galen posted a statement in the discussion thread to “Reality is just like you think” which appeared to me as an outrageous statement he can’t defend which reveals something significant about his approach to the issue of bias and “religiosity”. This is going to be my last treatment of this issue because there really are other matters I need to get to (not least of which is my overdue treatment of James Sennett’s argument for universal sanction as a criterion of proper basicality; oh yeah, and I need to mow the front lawn).
So for the last time, here is Luke’s statement (henceforth, LS):
Luke’s Statement: “Perhaps you could scrape up a group of christians that are less biased than a group of atheists, but that would be unrelated to their religiosity.”
The moment I read this, I recognized it for the brazen bit of dogmatism that it is. Begin with the fact that the woolly concept of “religiosity” is most certainly not cutting nature at the joints. It is, rather, a concept with very fluid boundaries, the content of which differs depending on the person you’re speaking with.
Let’s consider as an example the simple matter of charting religiosity with respect to whether one checks the box “Believe in God” on a survey. You either check it or you don’t, right? But descend beneath the surface and a dizzying amount of complexity appears. For example, one person may check the box, and a second person may refrain from checking the box, and yet they may hold the same belief about ultimate reality and simply differ as to whether that understanding is properly called “God”. So at most, those who check the box are stating their conviction that there is some entity to which the word “God” refers. But this question itself gives us no clue as to what that thing might be, and there is always a danger that anybody reading the result will end up reading the question in keeping with their assumptions of the sense and reference of “God”.
Then there is the problem of how one checks the box. Did they check it with a Sharpie marker using indelible ink? Or is it checked lightly with a pencil? That is, what sort of conviction is there behind the check of the box? And does the tool chart the conviction or only that the box was checked?
Behind each criterion proffered lie many, many complexities. This isn’t meant as some Luddite attempt to dismiss social scientific surveys and the kind of data they provide. But it is a reminder that this is a field of research that does not support sweeping, categorical statements like LS.
While I have hammered home that point repeatedly in the last week, several readers have repeatedly failed to grasp it. Consider Mike D who commented yesterday as follows:
“The problem is that you’re conflating positive statistical correlations with claims like “always” and “never”. Even the strongest correlation, by definition, includes exceptions to the rule. So the fact that studies show a strong positive correlation between biased thinking and religion isn’t undermined by counterexamples, because the studies by definition include counterexamples.”
The thing that mystifies me is that Mike is making a point that supports my position and yet he seems to think otherwise. You see, LS doesn’t admit exceptions. The whole point of LS is to insist that religiosity never leads to a decrease in bias. And that’s the outrageous claim I’ve been challenging from the beginning.
Mike wasn’t the only one to fail to understand me. Nate immediately followed up Mike’s comment by writing:
“Literally, I copied and pasted this very same point to make this very same point before Mike D. beat me to it. You aren’t going to find exceptionless laws anywhere except outside of physics, Randal. Forget about it in the case of social science.”
I can understand why Mike and Nate would miss the point since Luke himself conveniently equivocates between LS and the statistical generalizations endorsed by Mike and Nate. Yesterday he commented:
“Let’s go back to how this whole thing started: Religious people on average are more biased in their thinking. They are more ingroupy and prejudiced, they are more closed-minded. These statements are solid as the rock of Gibraltar. It doesn’t matter if that is not true of every single person, Just as “my uncle smoked and he lived to be 100” does not overrule “smoking shortens life span”.”
But that isn’t how “this whole thing started”. This whole thing started with Luke’s claim that none of the decreases in bias among those who are religious could be due in any part to their religiosity. This is the dogmatism that Luke needs to defend. My point is not to dispute a statistical generalization by pointing out exceptions to it, but rather to challenge Luke’s categorical assertion regarding any possible exceptions.
So I replied to Nate:
“Luke didn’t offer a statistical generalization according to which a decrease in bias would likely not arise from religiosity. Rather, he asserted without qualification that a decrease in bias would not arise from religiosity. And he has never backed down from that unequivocal assertion. So you’re simply ignoring the claim he actually made.”
Nate then replied:
“You’re right – I was misreading you on that point. I’m left wondering, however, “what’s the big deal?” If the these studies show a causal relationship between cognitive bias and religiousness (according to Luke, bidirectional causation), isn’t the overall effect (which is rather significant, btw) especially damning? What would it prove if you could find one instance of religious activity/belief/whatever that decreased bias, given a strong overall causal trend in the other direction?”
I commend Nate’s concession. As for the “what’s the big deal?” response, it’s a very big deal. Luke has continued to maintain LS even though it goes beyond the data, and even though his own defenders are unwittingly working against his case by agreeing with me that even the strongest correlations admit to exceptions. The more that Luke persists in his unflagging defense of LS, the more ideologically driven he appears and the bigger deal this becomes.
Yesterday in the discussion thread Luke offered two defenses of LS that are worth noting here.
“If liberal religious types are less biased its because they are liberal not because they are religious, axiomatically. because their religious trait is shared with a bunch of authoritarians who aren’t less biased, but more.”
Note how Luke assumes here that there is some discrete “religious trait” which is shared by the “liberal religious type” and the “conservative religious type” (or whatever). That trait is “belief in God” (or that set of traits is a, b, c…). Thus, Luke declares “axiomatically” that the progressive open-mindedness of the liberal is not due to the religious trait. But as I said, these tools are imperfect conventions and they don’t cut nature at the joints.
Let’s say, for example that you do a survey and 80% believe in God and 20% don’t believe in God. And imagine we find that statistically speaking the 80% are more likely to view themselves as better than the average (the “better-than-average effect”) than are the 20%. From this one concludes that people who believe in God are more likely to have a higher self-perception (and consequently be more biased) than those who don’t believe in God.
But hold on a minute. Get a more finely-grained tool and things might look very different. With more precise questions about what it means to believe in God you might find that some sub-groups that “believe in God” score even higher on self-perception while others that “believe in God” score even lower than the 20%. In that case, overlooking these distinctions could really skew the data and yield a very misleading picture.
At this point the LS apologist will interject: “Wait a minute! Where are those studies that show some religious sub-groups are less biased than others?” That misses the point. The point is that simply posing criteria for religiosity like “belief in God” gets us nowhere near the kind of finely-grained detail needed to support dogmatic claims like LS. Consequently, if Luke wants to defend LS, he needs to establish that there are no religious sub-groups with decreases in bias over the general population (or the other religious groups).
Next, let’s look at Luke’s second defense of LS:
“Similarly, if people who engage in secular forms of confession or psychotherapy have positive outcomes just as religious people who engage in what is a form of group therapy, the burden of proof then shifts to the religious advocate to demonstrate that the mechanism of therapeutic effect is religious and is above and beyond what we know to be the secular effect. Just as if you go to a doctor and he injects you with anti biotics then waves his hand and murmurs and says you have to do that part as well. It is up to him to show that the murmuring plus drugs condition is above and beyond just the drug condition.”
As we engage this second defense it will help to have LS quoted once again:
LS: “Perhaps you could scrape up a group of christians that are less biased than a group of atheists, but that would be unrelated to their religiosity.”
Note that in LS Luke claims that any decrease in bias would be unrelated to religiosity. In his second defense he says something quite different, namely that any denier of LS would be obliged to show that a decrease in bias in a religious population was not due to a secular effect. This is, in effect, an attempt to claim that LS is the default position. But of course, it isn’t. This is merely a question-begging attempt to win the argument by default!
In fact, in LS Luke asserts there would be no relation between any aspect of religiosity and a decrease in bias. But then it follows that Luke has a very steep evidential burden to establish that for any decrease in bias within a given population, that this decrease is wholly unrelated to the religiosity of that population. As I first noted close to a week ago, this is an extraordinary claim which Luke can’t defend.
In closing, Luke has been complaining that I haven’t looked at all the studies. Yesterday in response to my request for a study supporting LS, he referred me to Ericksson and Funcke, “Humble self-enhancement: Religiosity and the better-than-average effect.” So I’ve read the study and I’m now back to report.
The focus of the study is the better-than-average effect (henceforth BAE), i.e. the human penchant to think we’re better than the average. The question they asked is “does religion make people humbler when comparing themselves to average others?”
Note first that this study doesn’t deal with bias simpliciter. Rather, it deals with one aspect of bias, the BAE. As a result, it is impossible for this study to justify belief in LS since it does not deal with the claim of LS.
Now let’s turn to the paper itself. This paper combines and synthesizes the results from three studies. To begin with the authors look at a study by Loughnan et al. to measure the BAE across 15 countries. The study measured religiosity in terms of “The country’s percentage of adults who say that religion is an important part of their daily lives…”
Let’s pause here. There are several problems with this criterion. Let me note two. First, the fact that people state that religion is important to them on a survey doesn’t mean religion is important to them. Consider as an illustration the well established fact (see Hadaway, Marler, and Chaves) that people tend to over-report church attendance on surveys due to social expectations. It should be no surprise that if “religion” is viewed positively then people will likewise tend to over-rate their identity as religious. And that in turn presents a problem for any conclusion to the statistical correlation of BAE to religiosity. For example, are people who incorrectly report that they are religious on a survey more likely to have a higher BAE than those who correctly report that they are religious? Who knows? The study doesn’t address that question.
Second, what about people who may under-rate the personal importance of religion due to a high degree of humility? A few years ago I included in a syllabus 10% self-evaluation. In particular, I was asking students to grade themselves on how much they read of the course assigned readings and how well they read those readings. As a result, several of the worst students in class rated themselves at 100% while some of my best students rated themselves at only 50-60%. I abandoned that diagnostic tool after that trial run because I realized how unreliable self-perception was. How many people might check that religion is not important to them out of a more sober self-perception that religion is not as important to them as it ought to be, and yet in point of fact religion is more important to them than the vast majority who check that it is on the survey? Ironically, it could be that people are statistically more likely to downgrade the influence of religion in their life if they are more religious.
Once again, defenders of LS may want to challenge me to show that there are such people. But that isn’t my burden because I’m not making any claims like LS. The burden of proof is instead on the defender of LS who denies that there are such people to show that this is true.
These are complications that really shouldn’t be overlooked. But let’s set them aside and come to the conclusion. Eriksson and Funcke conclude: “Reanalysis of the data from Loughhnan et al. (2011) confirmed that country estimates of the better-than-average effect tended to be higher in more religious countries.” Even if the reservations I noted didn’t exist, we would still have at best a statistical correlation pertaining only to the BAE across religious countries. This provisional result does nothing to support LS.
The second and third studies in the paper were undertaken by the authors. In this case Americans were surveyed to chart their rates of religiosity. In the conclusion they write: “To summarize our answer to the question about the humility of religious people: Yes, when they compare themselves to other people of the same religious persuasion, religious people tend to be humble–but when they compare themselves to the average American, religious people nonetheless seem to self-enhance considerably, and more than non-religious people….”
Once again, we have a statistical trend, and one with all the caveats that attend to this kind of social scientific research. (Keep in mind, for example, the caveat I made that those who rate themselves lower in religiosity may in fact be more religious, assuming that we can come to common agreement on what “religious” is supposed to mean.)
The authors of the study are aware of these limitations. For example, they warn that “It is also worth noting that we used a quite crude measure of religiosity, a single question previously used to measure country levels of religiosity.” They go on to suggest that they believe their study results would likely be confirmed by a more sophisticated tool for measuring religiosity. But of course this remains to be seen. I don’t point this out to dismiss studies like this. Rather, I point this out to challenge indefensible leaps beyond the tentative conclusions of social scientific research to sweeping claims like LS.
So then why has Luke been so insistent on defending LS? Why has he repeatedly defended the claim that for any decrease in bias in a group, that bias would not be related to religiosity? I’ll leave that to the reader to ponder. However, I am curious about the degree of anti-religious bias among those who accept LS versus those who do not.
Perhaps I should do a study.
