Lab ko La Salle

by Bombit Largoza. Dr. Bombit Largoza was Associate Editor in 1991 and Editor in Chief in 1992. He specializes in Experimental Economics.

How much noise should we tolerate in a public space like Medrano Study Hall, and how do we keep it at that level? A couple of students, Andrew Nico Dalisay and Dwight Keifer Dee, put their heads to it in an Experimental Economics class of mine last year. We quickly realised the term “study hall” doesn’t adequately describe all the legitimate ways students can use it, from individual reading to one-on-one conversations to full-blown group discussions. As proper economics majors, they reckoned both extremes to be socially inefficient. A room with too much noise is obviously unacceptable, but a room kept totally quiet would likely entail discouraging, or worse suppressing, otherwise beneficial activities. The optimal noise level – one that allows sufficient conversation to take place without distracting those who need quiet – clearly lay somewhere between the two extremes.

But it wasn’t just a question of figuring out how much noise should be tolerated, then laying down the rules to enforce it. They worked out that the optimal level of noise had to be flexible as well, depending on the composition of room users. With more readers, less noise should be tolerated, but with more groups meeting, more noise should be allowed. Think of the wastefulness in insisting that a room be quiet for the sake of a couple of people reading when everyone else is there to hold group discussions. We took care to avoid the mistake of many rule-makers, who care more about the costs of being too lax with a policy, ignoring the costs of being too strict. We knew we had to be just as sensitive to the equally real costs that too much silence imposes upon groups meeting: losses from forgone opportunities to meet and discuss in a room with fewer readers to bother.

They soon realised, as many of us have, that rules are pretty poor at dealing with “moving targets” and situations that require flexibility. But the alternatives – norms (“students are mature enough, let them work it out among themselves”) aren’t always practical either, say when there is a psychological cost to approaching someone and asking them to tone it down a bit. Incentives on the other hand can be costly, not to mention objectionable to people who believe you shouldn’t reward students for something they ought to be doing anyway. So is there anything else, they asked, that can help us arrive at an optimal noise level that is responsive enough to the composition and needs of public space users?

After a bit of nudging, they found Thaler and Sunstein’s bestseller on, well, “nudges”: ways of carefully presenting choices so that it’s more natural for people to do the good thing and avoid the bad. In this case, they needed a “nudge” that would make Medrano Hall users more empathic to each other’s needs, allowing a more well-developed relational sense to prevail over impractical incentives and unenlightened penalties. A bit more googling brought them to thefuntheory.com, the folks that brought Stockholm the subway “piano stairs” (to nudge commuters away from convenient but environmentally costly escalators), Munich the “speeding ticket lottery” (in which drivers staying below the speed limit are given tickets to a lottery whose prize comes from fines paid by violators), and other clever, effective, but relatively inexpensive ways of getting people to do right by their communities.

So for several days close to finals week last December, Medrano Hall users would find themselves randomly experiencing one of three conditions. They’d either be monitored or admonished by a Discipline Officer (“authority treatment”), left unsupervised (“norms treatment”/control condition), or invited to write the purpose of their stay on a large sheet of cartolina by the door as they entered, and then to indicate the quality of their stay with a smiley or a frownie as they left. This last “nudge treatment” served as a signaling device meant to request empathy from other users in a clear but non-confrontational way.

It was a small field experiment involving 230 students and best viewed as a pilot, but even the preliminary results were interesting. Did it work? Depends what you mean. Evaluations obtained from users showed that noise levels were lowest in the presence of Discipline Officers, with the nudge treatment results statistically no different from the control. But evaluations of user satisfaction were nevertheless highest in the nudge treatment – despite the noise levels – and lowest in the control condition. (Results also confirmed our fears about inefficient outcomes: many of those reprimanded by Discipline Officers didn’t just pipe down – they left the study hall altogether).

I get a kick out of designing and running experiments with students because of results like these. Sure they’ll be messy and unpredictable, and will require larger-scale replication. But they also have a way of opening windows to reasoning that’s often grown stale from relying for too long on pet stereotypes, ideology, and “experience”. Who would have thought that noise levels weren’t all that really mattered to users of Medrano Hall? And maybe it doesn’t always have to be about increasing penalties, increasing surveillance, waving carrots and sticks, or creating a cult of “compliance”. With the raft of research coming out of social psychology, evolutionary biology, behavioural economics, and the neurosciences, we can design much more behaviourally-sensitive ways of getting people to do good by themselves and their communities.

But results, no matter how interesting, aren’t as important as the methods we use for testing our intuitions. In the past, we’ve resolved issues on campus (“do certain shirts worn during campaign periods really exert undue influence on voters?”) and beyond (“do CCTs just make poor people more mendicant?”) by “debates” that are, without fail, long on rhetoric and cherry-picked examples, and short on robust evidence. These days, we’re finally able to bring the best traditions of laboratory science and experimentation into the realm of social policy. By subjecting policy questions to randomised controlled trials, the way we would if we were testing medical claims in a lab, we can augment observational and correlational evidence (from the usual case studies, surveys, etc.) with evidence of the proper counter-factual kind. The kind of course that, though stronger, invites yet more hard thinking, more testing.