Behavioral economists Devin Pope and Maurice Schweitzer have a new paper showing that professional golfers perform differently on putts that are identical in all respects except that one is for par and one is for birdie.  What does “identical in all respects mean?”  From a summary in the New York Times:

The professors, Devin Pope and Maurice Schweitzer, seemingly anticipated every “But what about?” reflex from golf experts. The tendency to miss birdie putts more often existed regardless of the player’s general putting or overall skill; round or hole number; putt length; position with respect to the lead or cut; and more.

They find that par putts are made more often than birdie putts.  One natural response might be the following.  If the putt is for par, then the golfer is, on average, farther behind than if the putt is for birdie.  And when you are farther behind you have an incentive to take greater risk.  In putting, you can increase the chance of making a putt at the expense of a more difficult next putt if you miss (by say using a firmer stroke.)  You would have more incentive to do this when you are farther behind.  (Then you care less about the consequences in the event of missing since in that event you are even further behind.)

But they apparently control for this by matching par putts with birdie putts that are identical in terms of the total score that would result from sinking them.  They find the bias is still there. (See Column 8 of Table 3 in their paper.)

However, you might say that this means that the bias is not due to loss aversion.  Because in these two matched settings the golfer is at the same point relative to any reference point.  And if you appeal to “narrow framing” by saying that the players are using a hole-by-hole reference point of par, then the same narrow framing makes it rational to take risks when the putt is for par and play it safe when the putt is for birdie.

A real smoking gun would be the following.  Take the birdie and par putts matched in terms of score conditional on sinking the putt.  Now ask what is the expected final score, or tournament rank, or prize money, or other measure of success, conditional on whether the putt was for birdie or bogey.  The null hypothesis is that these would be the same.  Loss aversion would imply that they are not the same, although it is not obvious which direction it would go.  (The authors do a back of the envelope calculation to address a related question in their concluding section.  They find that the apparent loss aversion matters for final scores but they don’t seem to include any of the controls from the earlier parts of the paper in this calculation.)

(Gatsby greeting: kottke.org)