The Efficacy of Coercion

The study of the effect of anything on behavior is perennially impeded by a simple fact:

To be sure that X influences Y you need at least, at a minimum, to measure the relevant behavior before X begins. To know if a new medicine causes people to eat less, you have to measure their eating before you give them the drug.

But the act of measuring often causes those whose behavior is being studied to change the relevant behavior, precisely: A change in eating habits following the administration of the new drug risks being a mixed effect of the drug and of people’s awareness of their eating habits and of a desire to do well, both of the latter triggered by the original measurement. Sometimes, the effects of the measurement are greater than those of the X factor, the drug in this case. One can imagine a case where the effect of X seems very large while it is in reality zero because the effects of the measurement would explain all the change in behavior. The problem is that the experimenters usually don’t know one thing from the other. In my admittedly extreme example, this would lead to declaring effective a drug that has zero effect on eating habits.

Once in a while, reality, Destiny, if you will, gives us a natural experiment that bypasses this methodological problem. Here is the story of one such.

My area of central California has been in a drought for more than two years. I think everyone more or less agrees. If there are dissenting voice, they are whispering, not shouting. About two years ago, (don’t hold me to this), the City of Santa Cruz where I live mandated involuntary rationing. The arsenal the city displayed included encouragements to spy on neighbors, stealth inspection by city personnel (to catch the odd old lady rinsing down her car outside of permitted hours) and, especially, powerful fines for overusing. The later came about in the form of stepwise progressive water rates: Go over you quota of 100 by one unit, you rate increase by 50%, by 30 units, your rate shoots up by 70%, etc. (These are only illustrations, not real numbers.)

The adjacent town of Soquel and its water district which includes several other small towns simply recommended that customers voluntarily reduce their water consumption.

Practically everyone in the general area of the central coast must have been well aware of the wisdom of collective decrease in water usage because it was trumpeted for weeks by the local media including radio and television stations. And, by the way, there was no detectable push back. Of course, in this case, the measurement act did not interfere with the involuntary study implicit in these divergent courses of action. They did not because water consumption measurements ( “reading the meter”) are both frequent and routine. They are also mostly unobtrusive.

A week ago, one of the local newspaper had the good idea of performing a bit of simple bookkeeping. The results:

In the city of Santa Cruz, water consumption went down during the period of observation by an impressive 26%.

In the Soquel water district, water consumption went down by …. 22%.

Apparent superior efficacy of coercion over voluntary restraint: 4 percentage points!

I don’t have the relevant figures and I won’t search for them unless someone makes it worth my while but I am fairly sure that the superior performance of coercion – measured at 4 percentage points – could have been achieved much more cheaply by bribing a few recalcitrant consumers rather than coercing everyone else.

I know it’s a little more complicated than this, maybe. I am just saying we should know more about the comparative efficacy of the use of force, of coercion vs every other solution.