Is It Sometimes Rational to Select Leaders Randomly? A Cool Old Study

This term at Stanford, I am teaching a doctoral seminar on leadership.  Of course, this one of the broadest and most confusing topics on earth.  I am not qualified to teach a seminar on love or religion; so, for me, this is the most vexing topic I can teach.  The topic for the first meeting was "cynicism."  I started out by assigning academic papers that brought evidence and perspectives that undermined conventional assumptions about leadership and that even questioned why scholars bothered to study the topic at all (my friend and co-author Jeff Pfeffer raised this question in a 1977 paper called "The Ambiguity of Leadership").

The most entertaining paper we read was by S. Alexander Haslam and a long list of coauthors, called  "Inspecting the emperor's clothes: evidence that random selection of leaders can enhance group performance" (Group Dynamics: Theory, Research, and Practice, Vol. 2, No. 3, 1998, pp. 168-184).  The two key studies in the paper entailed assigning student groups to play various versions of the "survival exercise" (see some of the variations here), where the group imagines that they have experienced some kind of disaster and are stranded (a plane crash, a broken car in the desert, and a nuclear war were used in these studies).  The group's task is to rank order the importance of a dozen or so items that might help them survive the ordeal (e.g., a compass, map,  loaded pistol, newspapers, cigarette lighter).  The performance of the group is determined by comparing their rank-ordering to those produced by experts.  This is, of course, just a simulation of reality.  But I've participated and led these exercises and they are quite engaging -- I suspect many of you have had similar experiences. 

Overall, the researchers compared the performance of these student groups under four conditions:

1.  A leader selected via a formal selection process (self-ratings by group members)

2. A leader selected by an informal process (group members had a discussion and picked a leader)

3. A leader who was randomly selected.

4. No leader selected. 

The consistent finding was that groups with RANDOMLY selected members performed significantly better than groups in all other conditions, and there weren't significant differences found between the other conditions.  The researchers also did some follow-up surveys, and revealed some mildly interesting findings; notably, groups with randomly selected leaders rated their leaders as LESS effective even though their performance was BETTER.

The authors assert that this rather surprising finding -- which was fairly strong and replicated across two (albeit modest) studies -- occurs because performance on this task requires cooperation, input, and effort from all group members.  They suggest that the very act of selecting one individual, of singling him or her out as better than the rest or simply focusing attention on that person, undermines the group's sense of unity and shared identity. They suggest that doing so may lead to social loafing.   As they put it, in describing the impact of a contest for the "best" leader:

  'In effect, their thoughts about the leader may have been of the form "if you're so wonderful, you can get on with it.' 

I am still not entirely sure that these arguments are right, but I guess they make some sense (although they do not quite explain why groups that did not select leaders at all did equally badly -- the researchers suggest this is because the leadership role is necessary).  Yet the study, imperfections aside, is provocative.  I like it because it challenges so many deeply held assumptions about groups and organizational life.  I especially like how it implies that just THE PROCESS of selecting the leader can provoke group dynamics that undermine the performance of the group as a whole.  That is worth considerable attention as this is something that selection committees and such often forget -- and consistent with findings from many corners of the behavioral sciences that show "what you do is as important as how you do it."  Also, while the survival games probably do not generalize well to most tasks in organizational life, another possible implication is that, if you are doing a task where no one has any special expertise or experience, you might try randomly selecting your leader.

What do you think? Does this have any implication in real life, or is it just one of those crazy studies that is irrelevant to real people and organizations?

P.S. As veteran readers of this blog may remember, I have written about the virtues of randomness before; check out this post about Karl Weick's cool ideas about randomness and wisdom.

P.P.S. Do not miss the link to the study from Arie below.  More evidence that randomly promoting people might work! Thanks Arie, fantastically weird.

 

Comments

[freelance writer]

I like what the academic work indicates and as you've rightly expressed, it not just indicates haphazardly chose guides generating preferable consequences

[Seattle IT Services]

This is an concept that we have lately confirmed in longitudinal research of scholars on its way at school and authorities on its way at a army training higher education.

[Larry Ford]

I am reminded of a very successful name-brand consulting firm that specialized in executive assessments. My own company spent hundreds of thousand of dollars on these assessments (2 and 1/2 days of paper and pencil tests, work simulations, presentations and the like, exhausting). The consultants were all well trained PhD psychologists. I asked them how they validated the testing process. They said that they collected these data over many years and got ratings of leadership effectiveness from supervisors, colleagues, and subordinates. I observed that such ratings themselves were notoriously invalid and lack of a validly measured criterion made validation with respect to that criterion problematic. They indicated that I was a jerk and should just shut up. Since they were probably right, I did. It then occurred to me that our standard approach to defining, measuring and predicting leadership confines us to only what is current in pop culture. For example, I predict that the most significant, impactful and enduring work in leadership today is the reality TV show "Undercover Boss." My personal favorite leadership icon is Cincinnatus, who fits nicely with the no-leader paradigm..

[Web design melbourne, web design sydney]

I like what the study points out and as you've rightly said, it not only points to randomly selected leaders producing better results (maybe?) but also that the leader selection process impacts the groups dynamics.

[Bob Sutton]

Alex,

Thanks so much for writing. Your comments are really interesting and remind me of a new study that shows that groups with narcissistic leaders rated their leaders as most effective, but exhibited the worst performance, apparently because they inhibited the free and creative exchange of information. They sound a lot like the ones, as you described from your study, who had "tickets on themselves."

Thanks so much again for your comment

See this http://www.psychologicalscience.org/index.php/news/releases/narcissists-look-like-good-leadersbut-they-arent.html

[Alex Haslam]

Thanks very much for this article, and the great discussion that follows it -- there are some really interesting points here. One point to note about the method of the original studies is that the formal selection process required would-be leaders to rate themselves on a range of dimensions said to be predictive of leadership. So, in effect, those who emerged as leaders through this process were those who were willing to say "I'm great at communication", "I'm great at listening", "I'm great at goal-setting", and so on. In other words, to coin an Australian expression, they had tickets on themselves.

My sense (along the lines of our argument in the paper) and some of the comments here was that this really disrupted the group dynamic because those who emerged as leaders thought that leadership was "all about me" (and how wonderful I am) rather than "all about us".

This is an idea that we have recently validated in longitudinal studies of students arriving at university and officers arriving at a military training college. In these it turns out that those who start out with the view that "I am a great leader" are far less likely ultimately to emerge as leaders (as judged by their peers) than those who start out with the view that "I want to be a good follower". What this suggests is that a key to leadership is finding out about the group you want to lead (rather than simply thrusting yourself upon it), and this is where the formally-selected leaders in the above studies went wrong.

Anyway, in case you're interested in reading more about this, this is an analysis we elaborate upon in our recent book "The New Psychology of Leadership: Identity, Influence and Power" (Psychology Press, 2011). And thanks again for rekindling the debate here in such an illuminating way.

[Jonathan Bollag]

The randomly chosen leadership study outcome is very interesting and somewhat eye opening. The fact that the members of the groups rated leaders as less effective even though they performed better leads me to the topic of management vs leadership. Usually with leaders it is more important to have a team or group respect the leader while with management it is less important. The Fundamental Interpersonal Orientations Behavior (FIRO-B) Assessment measures leadership qualities and I wonder how The Randomly chosen leaders would score? At Career Assessment Site we administer leadership assessments and it is great to see such an interesting study.

[Preben Ormen]

Fun piece, enjoyed it. The study is provocative, but I am missing a definition of leader to go with it.

What is the function or role of the leader we are appointing? Sometimes the 'leader' is a figurehead, somebody has to be in the position and do the 'leader tasks'.

Partnerships often manage themselves this way. People take turns at the helm.

We may be overrating the role of the leader, in our culture we have leader-as-hero. Other cultures do not, and they get on fine.

We really do need a definition of leadership and what a leader is supposed to do that's different from the rest of the group/team to make sense of the study's findings.

I have done the survival games several times and do not think they relate to leadership at all. They just prove that people get better results in a group (but not necessarily the best overall result - blogpost here: http://prebenormen.com/commentary/bias-and-experts)

[Kevin Rutkowski]

I wonder if one of the keys to this working is that everyone is aware that the leader was selected randomly. If people know that the leader was selected randomly, they may be less likely to expect the leader to have all of the answers than if they knew that the leader was selected because they were the most qualified.

Also, the leader may be more likely to ask for input from the team than if they were presented as being the most qualified.

It may be that a leader who everyone knows was selected randomly can best take into account the input from the entire team.

It also sounds like these groups were made up of people who don't know each other and who work together on a very short task. A randomly selected leader may have a very different effect over the course of a year or with a group of people who know each other very well.

[C. E. Goepfert MD]

In medicine, especially in the 'Old World' (and where else steep hierarchical structures exist), assholes in the leadership as well as getting rid of assholes in general, are a big problem. From experiencing and observing different structures in departments/hospitals, a variant of what you describe wonderfully seems to work well (no leadership at all is not going to work here):

1. People with similar qualifications and education get the same title, but individual responsibilities (so that each narcissistic self is served well), and payment is adjusted according to individual performance. Fairness.

2. Important decisions are done in agreement of the group of senior and (in an ideal world) capable people. Equality.

3. Leadership either rotates through this senior group for 2-4 years each or longer, on a more random basis, if the department is smaller (highly unlikely it would work in a large group); or there is one leader who is stepping back and dividing responsibilities according to individual competencies (see point 1) of his senior staff, if the department is large. This is what I would call 'wisdom'.

To my knowledge there is no evidence-based medicine here (would be interesting though). This is all expert opinion.

[Joe Marchese]

I'm less concerned with the leader's selection process than in the leader's approach to leading. Is there something missing in the predictive leadership attributes tht drives us to make suboptimal choices?

[web design London]

It not only points to randomly selected leaders producing better results but also that the leader selection process impacts the groups dynamics..

[Gonzalo Valdes]

Hi Bob,

I think there is certain situations in which a leader just diminish group performance... especially when you have small groups attaining small tasks.

For instance, the spaghetti game
http://agilethinking.net/games/game_spaghetti.html

I've played it myself... if you put a leader (boss or whatever you want to call him/her) to guide the group on untangling the spaghetti it takes much more time than if the group solves it without one... (is something like everyone tries to solve their local situation... collaborating with a neighbor... and then the sum of those collaborations results in solving the problem)

Cheers,

--
Gonzalo (one of you students in the seminar!)

[Glen Davis]

How fascinating. Thank you for sharing.

You can see an ancient illustration of this principle in Acts 1:23-26 - the apostles replaced Judas by a random process.

"So they nominated two men: Joseph called Barsabbas (also known as Justus) and Matthias. Then they prayed, 'Lord, you know everyone’s heart. Show us which of these two you have chosen to take over this apostolic ministry, which Judas left to go where he belongs.' Then they cast lots, and the lot fell to Matthias; so he was added to the eleven apostles."

[Dria]

Would be interesting to see this study redone and included a scenario in which whomever least wanted the job was appointed as leader.

[Larry Ford]

1. If random assignment is the baseline, then the other conditions are treatments. It's not that random assignment is better, but it's not as damaging as other methods of assignment.

2. Is there really any such thing as leadership? We can't agree on it's definition, or what's good or bad. Maybe there is no such thing. Prove that it exists first.

3. Maybe the real finding is that leadership is irrelevant. Or, at best, all it can do is harm. Wouldn't that be fun?

[Darbsnave]

A link to the Pfeffer article: http://bit.ly/pekX1H.

[Account Deleted]

Hey Bob,

It's been a while since your last post... but that's a separate topic.

I like what the study points out and as you've rightly said, it not only points to randomly selected leaders producing better results (maybe?) but also that the leader selection process impacts the groups dynamics (absolutely!!).

Do you think it would negatively impact the performance of the group if ALL the group members were experts?

[Dominic]

this is very interesting,

It certainly seems possible, there would be less "ego competition" with random selection. But really this theme needs to be researched further...

but maybe since research starts within an organisation who's leaders are not randomly selected, who is gonna fund research that may prove that even their organisation would be better off if someone random was running it, rather than themself, who has striven to attain leadership?

there is the old saying... those who wish to persue a career in politics should be automatically barred from doing so.

[Ariegoldshlager]

Bob,
Please see also this study on Random Promotions as a best strategy:
http://www.nytimes.com/projects/magazine/ideas/2009/#business-2
http://arxiv.org/abs/0907.0455

Is is possible that random promotions are an effective antidote to the Peter Principle?
Arie.