Thinking about God makes you behave virtuously -- depending on what you believe

Blogging on Peer-Reviewed Research

Does religion increase "virtuous" behavior? So far the research on the topic has been inconclusive. A big problem with most studies is that they tend to be correlational. For example, religious people are more likely to say they are willing to help others. But are they willing to help others because they are religious, or is it the other way around? Or could some other factor cause both the religiosity and the helpfulness?

Few experimental studies have been conducted on the effect of religion, but Azim Shariff and Ara Norenzayan believe they have come up with a paradigm that does demonstrate some effects of religion on behavior. Their study combines two well-researched phenomena: the dictator game and the scrambled-sentence paradigm.

In the dictator game, a volunteer research participant is told she can split a pool of money ($10 in this case) with the next research participant. The first participant -- the dictator -- gets to take as much money as she wants, and the second person will get whatever she leaves. The experimenter promises the dictator that her identity will never be revealed. Of course there is no second volunteer: the researchers are interested in how the dictator behaves.

Shariff and Norenzaya wanted to know if volunteers who were first subtly directed ("primed") toward "God" concepts would behave differently in the dictator game.

The priming was done using a scrambled-sentence paradigm. Before playing the dictator game, half of the 50 volunteers were asked to unscramble 10 short sentences, while dropping an extra word. For example:

felt she eradicate spirit the

would be unscrambled as:

she felt the spirit

For those who were primed toward "God" concepts, five of the ten sentences included the words spirit, divine, God, sacred, or prophet. The other participants weren't primed at all, and everyone went on to play the dictator game. Here are the results:

i-bc99867bf317ed3cc917bd391d7d4db5-Shariff1.gif

As you can see, the people primed with God concepts were much more benevolent dictators. They left an average of $4.22 for the next participant, compared to just $1.84 for the unprimed group. After completing the study, the participants (all college students) answered questions about their religious status. Twenty-six people reported a religious affiliation and 19 were categorized as atheists, but the researchers found no difference in the results based on religious status: Atheists were affect by the God-prime just as much as religious people.

But this experiment has a couple of problems. The people in the no-prime group had a different task from the God-prime group: they never unscrambled sentences. Perhaps the act of unscrambling on its own is what causes dictatorial benevolence. It's unlikely, but it's possible. More importantly, perhaps priming with nonreligious words like "justice" can also cause this behavior. And maybe college students aren't representative of all adults.

A second experiment addressed these concerns. Now 75 participants were recruited from the community using newspaper ads. Their ages ranged from 17 to 82. This time, there were three types of primes -- a neutral unscrambling task with benign words, the God prime from the first experiment, and a new secular prime which included five sentences with the target words civic, jury, court, police, and contract. Here are the results:

i-f1a7921c5f1bdd038a0fbdd86cd2da6b-Shariff2.gif

Once again, the God prime led to more generous dictators compared to the neutral prime. But so did the secular prime. Indeed, there was no significant difference between the results for the God prime and the secular prime. But in this experiment, there was a difference between atheists and theists: the atheists didn't respond to the God prime; atheist results in the God prime group were indistinguishable from the neutral prime group. In the neutral prime group, there was no difference between atheists and theists. I'm a little surprised that Shariff and Norenzayan didn't make more of this in their discussion.

Instead, the authors offer two possible explanations for the results: Either the God prime activated concepts of generous individuals and the participants acted in line with those perceptual activations, or people in the God prime condition imagined that some supernatural force was watching over them and judging them as they decided how much money to take.

Neither of these explanations take the results of the secular prime, or the fact that atheists didn't respond to the god prime, into account. I can see how the first explanation could be modified to accommodate both the responses of atheists and responses to the secular prime, though. The secular prime could call to mind images of generosity and fairness for both atheists and religious people. Meanwhile, for atheists, the God prime wouldn't bring that imagery to mind, while it would for theists.

One question remains: why did college student atheists respond differently to the God prime? Perhaps older atheists are more "die-hard" in their beliefs. This seems plausible to me -- after all, many young atheists have been raised with an expectation of belief in God, and so God primes could still invoke imagery of fairness and justice.

Shariff, A.F. & Norenzayan, A. (2007). God is watching you: Priming God concepts increases prosocial behavior in an anonymous economic game. Psychological Science, 18(9), 803-809.

More like this

One question remains: why did college student atheists respond differently to the God prime? Perhaps older atheists are more "die-hard" in their beliefs. This seems plausible to me -- after all, many young atheists have been raised with an expectation of belief in God, and so God primes could still invoke imagery of fairness and justice.

What were the numbers involved in the second case? The 19 self-identified atheists in the college sample is already pretty small, and if there were a total of 75 subjects in the second trial, I'd be surprised if you had more than ten. That's such a small sample, I'm not sure I'd trust any generalization from it.

I think both the religous and social primes motivate moral thoughts which include the idea of fairness and monetary need which could explain the similarities in the results of the two primes.

I also think that college students would respond very differently than an average adult because of the constant awareness most college students have of a lack of money or an abundance (compared to peers). Many college students think about such things when confronted with a money issue no matter what the value is whereas many adults place less emphasis on a small amount of money.

By Julie Halsema (not verified) on 16 Oct 2007 #permalink

Chad: 22 out of the 75 participants in experiment two were atheist. That gives you about seven in each condition -- enough to get significant results given this effect size. But the point is well-taken; before we make any generalizations about the overall population, a larger study would need to be conducted.

What would the "real life" equivalent of the game be? They get to keep the money in the game, and they believe that whatever they leave goes to the next participant.

I'm not sure you could duplicate that in real life. If you find $10 on the street, you're either going to keep it for yourself or try to figure out whose money it is so you can return it. You don't know where the money came from, so it doesn't make sense to leave some of it behind for the next person.

The real life equivalent? Maybe something like:

Your inventor friend creates super widget 2000 and asks you to help market it. You go to MegaCorp and sell the rights for $10,000. Your best friend knows you made the sale, but not how much you actually got.

So, what's his cut?

http://www.arcade-history.com/index.php?page=detail&id=3397

By Marc Connor (not verified) on 16 Oct 2007 #permalink

See, that's not exactly the same game, since your friend knows who you are. It's highly unlikely that in that scenario you'd give him zero dollars. But lots of people do that in the dictator game.

Good point, it's not exactly the same. Many similar dealings occur with large groups or corporations or board members, though (studio accounting, anyone?), so there's a bit of blurring there. You could say to your friend that "the board decided to only give you $0" and not that you cast the deciding vote.

Of course, with groups and collective entities, the whole 'dictator' thing goes out the window, so I'll just stop.

Not exactly the same, you're right, but close enough to make an interesting study, I think.

By Marc Connor (not verified) on 16 Oct 2007 #permalink

The experiment is interesting, but does it measure 'virtuousness'? 'Altruism' would be closer, but the participants surely suspect that they are are under scrutiny and tend to act cautiously and within accepted boundaries. How about letting them take the 10 dollars home and ask them to mail whatever they feel like, whenever they feel like to the "other participant", anonymously (but with a key to their scores attached).

thorK:

You're right -- technically it doesn't measure virtue. The researchers call it "prosocial behavior," but that's not as catchy in a headline.

hmm, that is an interesting experiment, i think it would also be interesting to see the differences between gender, education etc.

I wonder if the atheist college students' susceptibility to God priming had to do with a desire to prove to oneself that they are virtuous without God. It seems that morality often used as a point of argument for religion, and in an attempt to fend off feelings of moral inferiority, atheists may be generous themselves.

Either that or the college atheists may have bought into the above mentioned arguments, or perhaps there is a combination effect.

Atheists do worse than those for whom organized religion is important, across a range of civic and altruistic measures. It's not that they do poorly absolutely; they do quite well. It's just that those who are more religious do even better. I reviewed some of this stuff here:
http://akinokure.blogspot.com/2007/10/sin-hypocrisy-and-blank-slate-soc…

Starting in the paragraph with "But there is evidence..." I'm technically an atheist.

I question the assumption that giving money to some anonymous partner in a dictator game is virtuous. Would it not be more virtuous to keep the entire sum and later spend it on some truly worthy cause? The anonymous partner is no more deserving of the money than any other random person. (In fact, in this setup, there isn't even such a person; the dictator is merely throwing money away!) Rather than virtue, I think we're measuring guilt here. The dictator has done nothing to earn this money; does the guilt of knowing he is unworthy drive him to diffuse that guilt by passing money on to other equally undeserving parties?

It might be interesting to directly address the question by asking dictators about how guilty they feel after the game. You could create a control group of people who are asked how much to give but then told that the researcher decided not to carry out the dictator's request. Do guiltier-feeling people give more money? Does successfully giving away money reduce guilty feelings? Therefore, does mentioning God to religious people make them feel bad about themselves?

One might also test this by removing the feeling of guilt; let the dictator perform some simple task to earn the money. Will the dictator give up "hard-earned" money? Do they feel guilty for refusing? Even if religiously primed?

"Atheists do worse than those for whom organized religion is important, across a range of civic and altruistic measures"

Most religious sects have a loosely or tightly defined notion of what level of altruism is needed to be virtuous. Forr example, some religions tithe a certain amount. That "primes" people with a metric, maybe.

"I question the assumption that giving money to some anonymous partner in a dictator game is virtuous." Agreed. Not to sound, say, Republican, but maybe I believe can use money more efficiently than an arbitrary unknown person.

Two things.
1: I find it both interesting and/or irritating that the experimenters decided to go with punitive words for the secular side of the experiment. For example--unless i misunderstood and those five words weren't the only primes--they didn't include positive ethics words like "justice" or "honor"? I guess they either think that god has a mostly cowing effect, or they think that we need fear in the absence of god. Or some combination. I would be very curious about the results of that study.

and 2. It seems like you guys don't have trackbacks? I went off-topic on my blog and would have liked to tell you about it without sticking a comment with a link to it down here.

Another interesting study. Couple of comments:

(1) I find it interesting that the study DIDN'T show that God priming significantly reduced dictatorial benevolence relative to neutral priming, particularly as it concerns atheists. Such a finding would be consistent with the underlying assumption that the notion of God should trigger negative emotional reactions (and possibly antisocial responses) in atheists.

In relation to commenter 16's first comment, it would be interesting to modify the primes. For example, it would seem that in their totality, all the primes connected to God would trigger spiritual rather religious thoughts; with religion having more of a negative connotation. As an example, whether the primes God, church, commandment, Bible, and faith would reduce dicatorial benevolence relative to neutral primes and/or God primes used in this study for atheists.

(2) Concerning the discussion of a real-life study related to the current, there was an interesting one conducted by Darley and Bateson (1973) where the participants were religious students at Princeton Theological Seminary. That experiment attempted a recreation of the Good Samaritan study from the Bible, whereby a Samaritan helped a person who was robbed and beaten while travelling on a road from Jerusalem to Jericho.

In the study, the students were divided into two groups: first group read an article about future job opportunities as seminarians; the second group read the Good Samaritan story. They were then requested to go to another building and give a short talk on what they had read. However, among both groups, 1/3 were told they were late and needed to hurry; 1/3 were told they were not late but had little time to make the appointment; 1/3 were told they had plenty of time to make the appointment. On their way their, they accounted a "victim" asking for help. Among the three conditions: 63% helped if they were told they had time; 45% helped if they were told they had little time; 10% helped helped if they were told they were late. What story the participants had read prior to this encounter made no difference. This was explained as a conflict between the desire to help and cost of helping.

A really cool update to this experiment and the one reported here, would be to see what happens when you place participants under the different priming conditions indicated above, and see whether this impacts helping behavior in the same situation described.

By Tony Jeremiah (not verified) on 17 Oct 2007 #permalink

The Decline of Psychological Research
By
Gil Gaudia, Ph.D.

The confounding of science and religion has sunk to new lows with the publication of an article "God Is Watching You" by Azim Shariff and Ara Norenzayan, in the September issue of the journal Psychological Science.
Here, "evidence" is presented, the authors claim, that "religion increases prosocial behavior." The interpretation of this "psychologese" is that if you casually mention words that have a religious connotation to people, they will go out into society and behave altruistically. Specifically, the researchers say "implicit priming of God concepts did increase prosocial behavior," where "priming" refers to the surreptitious presentation of words that make the subject more likely to manifest certain behaviors, and "prosocial behavior" means giving a stranger $2.38, on average, in a cutesy game originally designed for economists, called "The Dictator."
This significant breakthrough or "groundbreaking discovery," as the editor describes the criteria for publication, has about the same amount of scientific validity as the children's game of "Ouija Board."
Hearing casually the so-called God concepts, "spirit, divine, God, sacred, and prophet," and then sharing a trivial amount of money with someone in a game, no more makes that individual "prosocial" than does a murderer sharing his cellmate's TV guide at one moment prevent the inmate from stabbing him to death a few minutes later.
The authors then announce that their findings have "implications for theories positing religion as a facilitator of the emergence of early large-scale societies of cooperators," a somewhat grandiose expectation, given the data. What these experts are expecting the intelligent reader to swallow is that when people hear words like those mentioned earlier (they call them "God concepts") they become more prosocial, or to use their words, "religion increases prosocial behavior."

Which would be the more convincing evidence of the "link between . . . religious beliefs and cooperative behavior among unrelated strangers?":

1. Sharing several coins with someone in a few minutes of a contrived situation set up by a graduate student, or;

2. Climbing aboard a bus with dynamite strapped around your waist so that you can blow to pieces thirty people, including children, who are "unrelated strangers" but have a different religious belief than you do.

Which is more convincing evidence of the "link between . . . religious
beliefs and cooperative behavior among unrelated strangers?"

1. Playing a childish game designed for economists who are perennially searching for a way to beat the stock market, or;

2. Flying four commercial aircraft loaded with people and fuel into buildings because the "unrelated strangers" you will thereupon incinerate have a different religious belief than you do.

Going further, the paper shows a lack of understanding of statistical inference. The claim that "Those who were primed with God concepts left" more money for the recipient at a "sigificance level of .001" is embarrassing for its inaccuracy, implausibility, and its complete misuse of significance testing. They say that there is only one chance in a thousand that this finding is incorrect or is a result of chance. Even a physicist predicting the speed of a falling billiard ball under windy conditions would be reluctant to make that claim, using centuries-old and well-tested laws of gravity.

Claiming a .001 level of significance is saying that if one drew a sample like this, a thousand times, only one of the samples would erroneously give the results that were obtained here by chance, all the other 999 being "real" differences. . . . and this sample was far from random, having been solicited "through posters displayed at the University of British Columbia, Canada."

If you begin with two groups that are essentially equivalent on some measure and then remeasure them after a period of time, (where no intervening treatment was involved), there would still be differences in the mean scores--some very small and others quite large--based only on chance fluctuations. The trick for the researcher, after a treatment, would be to be able to say, "As a result of our treatment, our score was not one of these chance fluctuations, but rather one that is much more likely to occur when something had interfered with chance." This is called rejecting the null hypothesis. In the example, the researcher's claim that they would be correct more than 99.9% of the time! And is stated as; "the test results were significant at the .001 level;" quite unusual in psychological research.

It is not clear how the subjects were grouped, and to what extent the assumption of "no difference" in the independent variable was met. Since many of the subjects were "religious" at the outset, what measures were taken to see that the religious subjects were not all in one group? What justification was there for the 19 subjects being "categorized as Atheists?"

Which variable was the independent variable, the "priming" of the subjects or the category of Atheist/Theist? It appears as if there is a serious confounding of at least two variables: Comparing "Religious Primed" subjects with "Neutral Primed" subjects, and comparing Atheists with Theists. What are the authors seeking? Is it the effect of the priming, or the already-present condition of of Atheism/Theism? Atheists amount to less than five percent of the North American population--perhaps even less--and yet the authors managed to select thirty-eight percent in their sample. This alone would be enough to cast doubt upon the validity of this study, since having Atheists represented eight times over their expectation in the population, is a suspicious event.

How certain may we be that in the "Dictator Game" leaving some money for the receiver demonstrates altruism or prosocial behavior on the part of the giver? It is entirely plausible that sophisticated participants are looking for "the catch" and may be trying to leave money thinking it was to their advantage. Hardly anyone who responds to a solicitation to engage in an experiment is unaware that something is "going on." Most are guessing at what it is from the outset, and if they haven't immediately guessed right, they will be much closer by the time the game is over.

One wonders why a study like this is attempted. What ultimately is to be gained? On the microlevel, this seems like an innocuous doctoral thesis, but when one examines the list of references, it is difficult to see it as other than an effort to inject religion into science; into politics; and into education and secular culture. There seems to be a hidden agenda akin to the studies carried out by proponents of Intelligent Design.

This "experiment" represents an extremely low level of scientific investigation, a complete misunderstanding of statistical techniques, and a transparent attempt to marry religion and science that exceeds Stephen Jay Gould's Non Overlapping Magisteria (NOMA). It defies credulity that a professional magazine would publish it.

By Gil Gaudia, Ph.D. (not verified) on 21 Oct 2007 #permalink

Gil Gaudia:

I'll ignore the misconceptions about statistics and experimental design in your comment and respond only to this:

One wonders why a study like this is attempted. What ultimately is to be gained? On the microlevel, this seems like an innocuous doctoral thesis, but when one examines the list of references, it is difficult to see it as other than an effort to inject religion into science; into politics; and into education and secular culture. There seems to be a hidden agenda akin to the studies carried out by proponents of Intelligent Design.

As I mention in my post, I find it a bit odd that the researchers focus on the God-prime when it's clear that secular primes have just as strong an effect. That said, there's little evidence of your assertion that this study is akin to "injecting religion into science." On the contrary, the study is attempting to do the reverse -- apply science to understand religious beliefs.

Dr. Gaudia,

Here's some info that addresses two of your comments:

Comment #1:

It is not clear how the subjects were grouped,and to what extent the assumption of "no difference" in the independent variable was met. Since many of the subjects were "religious" at the outset, what measures were taken to see that the religious subjects were not all in one group? What justification was there for the 19 subjects being "categorized as Atheists?"

****************
From Study 1:

"...Twenty-six indicated identification with a religion, and 24 did not. Of the religious subjects, 19 identified themselves as Christians, 4 as Buddhists, 2 as Jews, and 1 as a Muslim. Of the remaining 24 subjects, 19 were categorized as atheists and 5 as theists without an organized religion. Subjects were defined as atheists if they both indicated "none" for religion and scored below the midpoint of the scale on a question assessing belief in God. Subjects who did not indicate a religious identification but nonetheless scored higher than the midpoint on the belief-in-God question were categorized as theists, along with those who did state specific religious identifications..."

***********************
From Study 2:

"...Categorization of subjects in this study was based in part on responses to a new question, not used in Study 1, asking them specifically to mark whether they considered themselves religious, spiritual, agnostic, or atheistic. Of the subjects who indicated their religion, 25 identified themselves as Christians, and 3 as Jews. Of the remaining 47 subjects, 21 reported being "spiritual," 22 reported being agnostics or atheists, and 4 declined to answer. Subjects were categorized as atheists only if they both identified themselves as atheist or agnostic and scored below the midpoint of the scale on a question assessing belief in God--a more stringent criterion than in the first study. All other participants were classified as theists..."

*********************

Comment #2:

Which variable was the independent variable, the "priming" of the subjects or the category of Atheist/Theist? It appears as if there is a serious confounding of at least two variables: Comparing "Religious Primed" subjects with "Neutral Primed" subjects, and comparing Atheists with Theists

********************************************
I'm assuming you're referring to study #2. In which case, this info (also from the article) addresses the questions presented here:

"...Unlike in the first study, there was a weak religiosity-by-prime interaction in this sample, F(1, 46) = 2.22, p = .14, prep = .78, indicating that the effect of the religious prime appeared to be stronger among theists than among atheists..."

The F notation indicates an ANOVA analysis was conducted. In fact, a 2 Factor ANOVA with the 2 factors (independent variables) being Religiosity (levels of the independent: atheist vs. theist) and Priming condition (levels of the independent: God vs. neutral vs. secular).

By Tony Jeremiah (not verified) on 21 Oct 2007 #permalink

I know this isn't a scientific test or anything, but I think this has some reasonable relevance to the subject:

When I'm driving and something happens on the road that gets me upset, if I'm listening to 90.5 (A Christian station here) I am far less likely to get ticked off or do something rash because while images of tossing grenades at the guy on the cellphone that just cut me off start to enter my mind, the lyrics I am hearing on the radio quickly remind me that really isn't anything I would want to think or do. Aside from the hypocrisy of listening to Christian music while screaming obscenities at the car next to me, the music alone injects a more positive state of mind than if I was cruising around listening to Slayer or KMFDM. Older KMFDM - the new stuff isn't that good.

By John Albano (not verified) on 20 Feb 2008 #permalink