Making complicated decisions by vote: It's complicated

Blogging on Peer-Reviewed ResearchNothing brings out conflict an academic department like a hiring decision. Adding a new faculty member is complicated by dozens of factors: What field should the new hire be in? Is it more important to hire someone in precisely the right field, or with a better publication record? Will the new hire impinge on the turf of a current faculty member?

And consider this: even if the requirements for the job are agreed upon by all, what if the member of the hiring committee disagree on the qualifications of an applicant? For example, a psychology department might have decided that two requirements are essential: Three years teaching experience and a focus on adolescent development. The applicant in question has been teaching for three years, but only one course per semester. Her research focuses primarily on elementary school-age children but she's coauthored two papers on teens. Four of the seven hiring committee members believe she meets the teaching requirement, four out of seven feel she meets the research focus requirement, but only two out of seven believe she meets both requirements. Should she be offered the job?

Similar conflicts can be found in a number of different arenas, from juries to social networking software. A jury might acquit or convict someone while knowing their decision doesn't conform to the letter of the law. A computer search might return results that don't strictly match the search criteria. These are all examples of the doctrinal paradox. But according to Jean-François Bonnefon, although it's well-known to researchers, the doctrinal paradox has never been studied empirically.

Bonnefon surveyed 1,092 people about how they'd resolve the doctrinal paradox in four different contexts. Here's one of the contexts:

The seven administrators of a company are considering whether an employee will move to a new position; the employee will have to comply with the decision. The position is much coveted. Having the profile for the position amounts to being both young and trilingual.

As in my example above, four of the administrators judged the employee to be "young," and four said the employee was "trilingual," but just two administrators agreed the employee was both young and trilingual. The survey respondents were asked to rate their agreement with two possible procedures for deciding whether to moving the employee to the new position.

  • In the first strategy, a straight vote, the administrators would vote on whether the candidate met both job requirements at the same time -- thus, rejecting the candidate.
  • In the second strategy, a "conjunct procedure," there would be two separate votes, one on each job requirement, and thus the candidate would be approved for the move to a new position.

The key to this study was the four different contexts. Half the participants were told that the new position was "much coveted," but half were told it was "one nobody wants to fill." For each of these cases, half the time the two qualifications were youth and trilingual ability, but half the time the second qualification was "having strong experience in team management" -- a qualification rather incompatible with the "youth" requirement.

Here are the results:


Respondents rated their preference for voting procedures on a scale of 1 (agree with straight vote) to 5 (agree with conjunct procedure). When the decision had negative consequences -- in this case, when an employee was being transferred to a job no one wants, then respondents favored the straight vote. But when the job was desirable, they preferred the more lenient conjunct procedure that would land the employee the job.

They also favored the lenient procedure when the job requirements were incompatible (young and experienced) and thus suggested that the position would be difficult to fill.

Clearly the ideal pattern is the straight vote, and indeed this was rated as "simpler" by the survey respondents. But when the decision becomes more complicated, most people are willing to bend the requirements of the decision making process in order to come to a decision.

Bonnefon, J.F. (2007). How do individuals solve the doctrinal paradox in collective decisions? An empirical investigation. Psychological Science, 18(9), 753-755. DOI: doi:10.1111/j.1467-9280.2007.01972.x

More like this

Imagine that, over the course of a conversation with a friend from work, she makes the following two statements: It's possible that my brother will be coming into town tomorrow It's possible that our boss knows about the affair you had with the intern (You might also have to imagine a more…
California often takes the lead in responding to public health issues, so itâs not surprising that their state legislature was the first to take up a bill to ban the artificial butter flavoring chemical diacetyl from California workplaces by 2010. California is home to 29 flavoring plants, and…
APPLYING for a job? The weight of the clipboard to which your CV is attached may influence your chances of getting it. Negotiating a deal? Sitting in a hard chair may lead you to drive a harder bargain. Those are two of the surprising conclusions of a study published in today's issue of Science,…
Cross-posted by Revere at Effect Measure You wonder when they will ever learn -- or IF they will ever learn. In the wake of yesterday's announcement that the Director of the National Institute of Environmental Health Sciences, Dr. David Schwartz, will step aside while NIH does an inquiry into…

...a good argument for an enlightened monarch.

By Joel McKellar (not verified) on 20 Nov 2007 #permalink

If you indicate to the committee what outcome you want (i.e.difficult to fill) they will tend to find ways to rationalise the wishes.

By peter evans (not verified) on 20 Nov 2007 #permalink

Worse yet, if you're trying to select using 3 or more criteria, you can lose transitivity: for example, Candidate A is better than B, B is better than C, and C is better than A.

I think Peter Evans has got it... hacking the rules is a standard part of human politicking.

By David Harmon (not verified) on 26 Nov 2007 #permalink

There's a fairly recent New Jersey Supreme Court death penalty case, I believe, that presents a similar question. I don't remember the details, but it was something like this: there were two criteria, both of which had to be satisfied for the death penalty to be applicable. Out of the seven justices, four thought one criterion had been satisfied; four thought the other criterion had been satisfied. Only three thought both had been met. The Court decided that the death penalty should not apply, i.e., applied the straight vote, rather than the conjunct procedure.