Are there measurable benefits in using a lottery to select leaders? A scientific experiment

Short answer: Yes, and no!

Longer answer:

Hubris is a tendency of leaders to hold an overly confident view of their own capabilities and to abuse power for their own selfish goals, sometimes with disastrous consequences for organizations. A major reason for hubris is the rigorous selection process leaders typically undergo. This study proposes a governance mechanism used successfully in history to tackle hubris: partly random selections, which combine competitive selections by competence with lotteries. A frequently voiced concern about the use of lotteries is that it takes no account of the competence of the leader chosen. We propose that partly random selections can mitigate the disadvantages of both competitive selections alone and lotteries alone and reduce hubris in leaders. We conduct a test of this governance mechanism by means of a computerized laboratory experiment. Our results show that partly random selections significantly reduce the hubris of group leaders. [my emphasis]

This is the Abstract from the Report. The full citation is: Joël Berger; Margit Osterloh; Katja Rost; Thomas Ehrmann (2020, May 13) ‘How to prevent leadership hubris? Comparing competitive selections, lotteries, and their combination’ The Leadership Quarterly, ISSN: 1048-9843 (paywall)

In order to test their theory, this group of Swiss and German scientists conducted an experiment, using a method instantly recognisable to experimental economists (and others, but they are the ones I’m familiar with). Their hypothesis was that a lottery could play a useful part in limiting hubris when selecting leaders.

We conducted a computerized laboratory experiment   ….  864 students of the University of Zurich and ETH Zurich, were randomly selected from a pool of students who had volunteered to participate in behavioral experiments for monetary compensation. Participants on average gained USD 30 for 45 min……The 864 participants were randomly selected into groups of six and randomly assigned to one of three treatment conditions.

Wow! As you can see, this sort of experiment is not cheap, so well done to the guys in Zurich to obtain the funds from the Swiss government to conduct an experiment on lotteries-for-jobs. Note, too, the use of a randomly selected sample and sub-samples. Ok, so it’s students, it generally is in these scientific tests, but for obvious practical reasons.

Briefly, the experiment proceeded thus: A leader for each group were produced by one of three methods. 1. Using a general knowledge test and appoint the top scorer; or 2. Same test, but select at random from the top three scorers; or 3. A simple lottery where every member of the group has an equal chance.

How ‘hubris’ of the selected leaders was measured was complicated, and if you want know, you’ll have to read the article, but it did involve the well-known ‘Prisoner’s Dilemma Game’. It was from this, and 11 pages of statistical analysis of regression models that the conclusion was reached.

Our study follows a pioneering approach to investigate an unusual selection method for appointing leaders in organizations, partly random selection. This selection method has been extensively used in history but has nearly been forgotten. Today, random decisions are considered by many people to be “irrational”. Our study shows that purposeful random selection, in particular combining competitive selections with a random component, is a rational and promising way of recruiting leaders that tackles hubris in overconfident leaders. Our proposal to “draw your CEO by lot” is provocative but may be promising.

Most of the members of this group engage in philosophical discussion, where the merits of a proposal are a matter of persuasive rhetoric. Elsewhere, exhortations to ‘follow the science’ abound, and mere rhetoric is treated with caution. Even calling in aid ‘common-sense’ can be suspect.

This is, I believe, the first time any hypothesis of us Kleroterians has been subject to what has been described as ‘The gold standard of science’. I have another example from Levitt of Freakonomics fame which almost constitutes Science, which I will post about later.

10 Responses

  1. 1. Using a general knowledge test and appoint the top scorer; or 2. Same test, but select at random from the top three scorers; or 3. A simple lottery where every member of the group has an equal chance.

    Conall, given that the persons selected in 1) and 2) were the same, I’d be interested to hear your speculation on the psychological mechanism that would reduce hubris in the second case. It’s counterintuitive in that they would be the “elect” in both senses of the word (or did they just think they were lucky?). Presumably case 3) involves minimum hubris (but runs the risk of incompetence).

    Liked by 2 people

  2. Keith:> I’d be interested to hear your speculation on the psychological mechanism…

    The answer to could have significant implications for random executive appointment. This seems as mysterious as the placebo effect; we could heal the polity by the mere hint that officers are chosen by lot.

    Liked by 1 person

  3. Unfortunately if it is just a placebo effect, it’s unlikely to last for long.


  4. Keith:> Maybe, maybe not. This test doesn’t suggest what the effect on the general public will be. If the general public views someone chosen partly by lot as less exalted, then the effect might be permanent.

    What I’d like to see experimentally is the effect on the public (in the experiment). How do they view their new leader? One interesting twist would be lie to them: tell them that the leader was elected when they were really chosen at random etc., and see what they effect is.


  5. Corollary: billionaires should not exist ?


  6. Ahmed:> They can exist, but they should be randomly chosen ;-)


  7. Keith & Alex:> It was a bold experiment, but as for psychological effects, which might be longer-lasting, I’m not sure. But at least this was a good start, and some effects can be detected. You can’t say the partial lottery to choose the leader had NO effect.

    As well as taming the conceit of winners, I have also advocated another benefit of jobs-by-lot; that it lessens the false feeling of worthlessness that conventional job appointment systems engender in the losers.

    But overall, eliminating the hubris of winners or the dis-heartening of losers are minor if useful gains from using lotteries for jobs.


  8. The more important evidence this reveals is not that selecting by lot (at least in part) makes people less selfish/corrupt/hubristic … it is the obverse, that selection by ELECTION tends to make people MORE selfish/corrupt/hubristic. We KNOW from dozens of studies that people who have a feeling of righteous “earned” elevation, whether students in a lab setting, elected politicians, movie producers, or corporate CEOs, exhibit greater corrupt tendencies than the general population. This study provides evidence that the ELECTION PROCESS ITSELF can activate this corrupt behavior.

    Liked by 2 people

  9. Terry,

    We should remember the title of Headlam’s book (Election by Lot in Athens) and that the original meaning of hubris was “excessive pride towards or defiance of the gods, leading to nemesis.” Maybe in the ancient world there was a residual sense that sortition involved being chosen by the gods. This led some of the playwrights to have a dim view regarding the arrogance of randomly-selected jurors. I doubt there are many historians who would claim that elected leaders were more hubristic than the assembly (especially as Athenian imperialism was strongly backed by the masses). And what do you mean by “corrupt tendencies”?


  10. Terry:> it is the obverse, that selection by ELECTION tends to make people MORE selfish/corrupt/hubristic

    I agree wholeheartedly. This property can be mitigated by voting for parties instead of individuals, however. Voting for parties also allows cohort voting, so that party strength varies more continuously over time. Both of these things reduce the fever pitch that characterizes democracy in the information age.

    Imagine a sequence like this: we move to proportional voting first, so that everyone votes for parties, not people. Then we shift to cohort voting, perhaps by having a first in/first out legislature in which a few members are added after each cohort, and an equal number of the longest serving members are dropped. At that point, replacing each voting cohort by a large citizen jury is easy.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: