# John McCain, risk, and anger

Thomas Levenson has written an interesting post about John McCain and his fascination with high-stakes gambling. While it’s clear that his gambling habit isn’t going to put McCain in any serious financial danger, it does raise questions about his personality. One of the most difficult things voters try to do in an election is to predict the candidate’s future behavior. We know what the candidates have done in the past, but how they’ll respond in future crises is what will matter most if they are elected.

If a new crisis emerges, will the new president take risks, or play it safe? One test psychologists have used to understand risk-taking behavior is the “Asian disease problem.” Imagine that a disease originating in Asia has been transported to the U.S. and is expected to cause 600 deaths. The president has to decide between two possible approaches to the problem. Which should he choose? Let’s use a poll to collect responses (make sure you respond before you read any further!):

Have you answered the poll? Good!

Variants of this test have been used for decades, but there’s a key manipulation I didn’t reveal in the poll. The test has two different forms. In the first variant of the test, both options are expressed in positive terms:

Program A: 200 people will be saved.
Program B: There is a 1/3 probability that 600 people will be saved and a 2/3 probability that no one will be saved.

In the second variant, the same options are available, but they are expressed in negative terms:

Program A: 400 people will die.
Program B: There is a 1/3 probability that nobody will die and a 2/3 probability that 600 people will die.

We’ll do both variants: after we’ve gotten a good response to the first poll, we’ll substitute in a second poll. Then we should be able to compare results. If our results match the pattern of past research, we should find that many more people are willing to “gamble” when the possible results are expressed in negative terms. This is a classic example of the “framing” problem. But there is much more than can be done with this paradigm. After all, aren’t some individuals simply more willing to take risks than others?

Jennifer Lerner and Dacher Keltner gave a personality test to 75 college students, then presented them with both variants of the Asian disease problem, in random order. The personality test was designed to assess two dimensions of personality: anger and fear. Do these emotions affect responses to the problem? Here are the results:

Respondents rated their preference for each program on a scale of 1 to 6, and their responses are shown as averages heare. As you can see, while both angry and fearful people were less willing to take risks when the impact was phrased in positive terms, angry people consistently took more risks than fearful people.

In three additional experiments, Lerner and Keltner confirmed these results. In one crucial variation, rather than studying people with angry and fearful personalities, they induced anger or fear in individuals, and found the same predisposition for angry people: a greater predisposition to make optimistic assessments of risk compared to fearful people.

John McCain has been noted in the media as a person who is quick to anger. This research suggests that angry people are more likely to take risks. As Thomas Levenson points out, McCain also enjoys gambling as personal entertainment (as does Obama, in an occasional low-stakes poker game). Does this mean he would be more willing to take risks as president compared to Obama? Without a similar analysis of Obama’s personality, it’s difficult to know. This interview suggests McCain may have the hotter temper, but that’s probably something voters will have to decide for themselves. They’ll also have to decide whether risk-taking is an asset or a liability in a president.

And what about the responses to our poll? I’ll discuss those in the comments once we have enough answers to assess the results.

Jennifer S. Lerner, Dacher Keltner (2001). Fear, anger, and risk. Journal of Personality and Social Psychology, 81 (1), 146-159 DOI: 10.1037/0022-3514.81.1.146

1. #1 Nelson Muntz
September 29, 2008

Gamblers get a thrill out of taking stupid risks. And they believe in luck, and that they can change their luck (somehow, magically, whatever).

Casinos believe in taking smart risks. Out of the after-tax profits they build bigger casinos. There is no thrill in casino management, it’s a continuous process of managing risks — smartly — all on evidence-based learning.

2. #2 FhnuZoag
September 29, 2008

Hmm, in the case of money, it might be preferable to get the fixed amount that take the lottery. But the specifics of the problem here makes me think again. In the case of saving 200 people, there’s the problem of choosing who gets to live, and who gets to die, and so pressures like that makes me rather gamble and try to save everyone.

3. #3 Dave Munger
September 29, 2008

That’s not really the premise of the problem, FhnuZoag. No one is going to decide who lives or who dies. In Program A, a random 200 people will be saved. In Program B, either a random selection of 600 people will die, or no one will die.

4. #4 Dave Munger
September 29, 2008

Okay, we’ve gotten a pretty good number of responses.

The first variation of the poll gave solutions phrased in positive terms, and here are the results:

So most of our readers preferred the sure solution instead of the risky solution. In an hour or so we’ll be able to compare to the negative version of the poll, which I’ve just posted.

5. #5 Idlethought
September 29, 2008

The fact that the premise is that no-one decides who lives and dies doesn’t mean that answers will not be coloured by feeling otherwise.

One advantage – not I admit a rational one – of the risky solution is that it can *feel* less like you’re deciding who lives and dies.

6. #6 Thomas M
September 29, 2008

The way I understand it in economics, taking a higher risk is usually rewarded by a higher expected payout. In this case the expected payout is the same for both options, but the 400 has no risk attached to it, therefor it is the “better” option? risk in this context can be seen as the variance…

7. #7 Bruce E
September 29, 2008

It wasn’t really said in the post, but statistically speaking, both choices have the same average outcome. That is, assume you are faced with the same decision 1000 times.

If you always select A, then exactly 400,000 people die (I assume there is an unnamed third option of do nothing, with a very bad outcome, say 10,000 people die, such that you would never choose it, since 10,000,000 will have died if you never chose A or B).

If you always choose B, then after 1000 choices you’d expect about 400,000 to have died, with some variation above or below.

The two choices only differ by the variation, over the long term.

Of course, faced this decision only once, and you are choosing to be a gambler or not.

8. #8 Mike Reeves-McMillan
September 29, 2008

As Bruce says, both choices have the same average outcome, which is why I opted for the one that had at least some chance of reducing the deaths below 400 rather than the one in which there would definitely be 400 deaths. One assumes that high-stakes gamblers usually have a good understanding of risk and reward – or are used to losing a lot of money, one of the two.

Poker, when played by experts, is, as I understand it, less a game of chance than a game of being able to read your fellow players while not letting them read you. It’s not like, say, roulette, where it is truly random and favours the house (so the rational thing to do financially is not to play it, though you might do so anyway for the thrill if that’s how you’re wired).

9. #9 Dave Munger
September 30, 2008

Yes, in the long run, all four options would yield the identical result. The point is to determine what conditions lead to a greater tolerance for short-term risk.

Our two polls have now demonstrated that we can reliably lead even our relatively sophisticated reader base in one direction of the other, depending on how we phrase the options: 69 percent chose A when phrased positively, but only 42 percent chose A when phrased negatively. These results are statistically significant because of the very large number of responses.

While A in the “positive” and “negative” phrasing condition represents the identical result (as does B), A and B are different from each other when they are one-time decisions, because B involves more risk.

10. #10 DaveM
September 30, 2008

Is this what we see with the way media presents its data/stories and the way the public reacts to the news? What would happen to the economy or geo-political climate if the news were read with a more “positive” spin? Would we take less chances as a society but be happier about our decisions?

If a company reports a 1 million dollar profit one year then they report an 800,000 dollar profit the next the news will state that they had a 20% loss in revenue. They still made money, increased profits (positive) but it’s the loss that gets reported (negative).

Frankly, I get tired of all the negative reporting.

11. #11 Andrew
September 30, 2008

Since this is in a politcal context it all comes down to spin and damage control: no matter what you chose, the other party will accuse you of A: “Condemning 400 people to die/not trying everything you can to save everyone” -or-B:”Being an arrogant risk taker.”

12. #12 natural cynic
September 30, 2008

I’m with Mike R-M. Obama is a low stakes poker player, where some skill is involved. McCain is a \$100+ a roll craps player.

13. #13 Rebecca
October 1, 2008

which is why I opted for the one that had at least some chance of reducing the deaths below 400 rather than the one in which there would definitely be 400 deaths.

Which is also the one that has an equal chance of increasing the deaths to more than 400.

14. #14 Rebecca
October 1, 2008

And with equal I mean bigger. D’uh.

15. #15 mich
October 1, 2008

I believe the original research by Kahneman & Tversky on framing should have been cited somewhere so readers can understand the underlying manipulation here, which is apparently causing some confusion. The psychological manipulation with personality is interesting, although I’m wondering if fear and anger are more situational. Another possible study could focus on attributes commonly associated with each political party, such as conservative and liberal stances reflected on behaviour and decisionmaking (in which case the assumption would be the opposite as in this article, that liberals would be the ones to take the risks).

16. #16 onsolthree
October 4, 2008

So can I be proud of myself that I picked the non-risky option on the second poll (without having read ahead)? ðŸ™‚

17. #17 Kent Kauffman
October 5, 2008

I would have to say the question only measures risk taking if the person doesn’t go into any analysis of it.

First, there’s the realization, mentioned above, that on average the same amount of people will die in either choice, so in the grand scheme of things your choice doesn’t matter.

Second, you’re the president, and elected official. In option A, guaranteeing the deaths of 400 people is political suicide (no matter the statistics, people do not accept inaction when their family members die). Option B is only screws you 2/3 of the time (and this might actually be higher, since some people will be like, “Well, at least you did something”). And, if you’re president, you didn’t get here by being bad at politics. So the choice is now, 100% career change, versus a 66% career change. The choice is quite easy, then.

And for the record I don’t gamble, whether games of chance (like roulette) or games of skill (poker), because the odds are all on the house in chance games, and I’ve never been a fan of zero-sum games like poker.

18. #18 Kent Kauffman
October 5, 2008

Basically, what I’m saying is that after further analysis of the question, there isn’t any risk involved.

19. #19 Kent Kauffman
October 5, 2008

My first comment disappeared but the second came through…? [the first one got held up in moderation — I’ll approve it in a minute –DM]

Anyways, the question only measures risk if there is no analysis done of the question.

First, as was mentioned above, in the long run the same number of people die no matter what choice is made.

Secondly, you’re asked the question as the president, an elected official. Option A, where 400 are guaranteed to die, is political suicide (people don’t like statistics if their family members die and you did nothing). Option B, only screws you 2/3 of the time (probably less, since some people will say that at least you did something). And, since you’re president, you might make decisions with a political rationale, since that’s how most people get the job. So, Option A, 0% chance of political success, and Option B has a 33% chance of political success. There is no risk, the decision is made for you.

And for the record I don’t gamble, because in high chance games the odds are all on the house, and I’ve never been a fan of zero-sum games like poker.