Pushing back the frontiers of science

A paper lands in my inbox: "On the Efficiency of AC/DC: Bon Scott versus Brian Johnson," by Robert Oxoby (U. Calgary). Oxoby's contribution:

Among musicologists, researchers of popular culture, and rock and roll lovers of all ages there exists a common debate. That is, with respect to the rock band AC/DC, who is the better vocalist: Bon Scott or Brian Johnson? The band’s original vocalist, Scott, performed on seven of the band’s albums (excluding live albums and compilations), passing away in 1980. Brian Johnson joined the band in 1980, serving as vocalist on nine albums (excluding live albums and compilations). Since 1980, there has been near constant contention regarding who was the better singer.

In this paper, we explore this issue. Since it is difficult to ascertain which vocalist was better given the heterogeneity of musical tastes, our analysis does not focus on the aural or sonic quality of the vocalists’ performances. Rather, using tools from the field of experimental economics, and we consider which vocalist results in individuals arriving at more efficient outcomes in a simple bargaining game.

[Oxoby has students play the "ultimatum game": one student gets a sum of money (say $10) which is to be shared with another student. The student who gets the money makes a take-it-or-leave-it offer to the other student. If the offer is accepted, the students split the money as stipulated; if it is rejected, neither gets any money. The Nash equilibrium is for the first student to offer a penny to the other student, who accepts it because a penny is better than nothing. But experimentalists have long observed that the offer of a penny is routinely rejected as insulting (despite the fact that the rejecting student is made worse off as a result of his decision), and so the average offer is somewhere in the neighborhood of $3-$4.]

A total of 36 participants from a large Canadian university took part in the experiment (two sessions of 18 individuals each). In one session, “It’s a Long Way to the Top” was played while participants made their decisions; in the other session “Shoot to Thrill” was played while participants made their decisions. To maintain anonymity among subjects and bargaining pairs, participant pairings were made by computer and decisions were entered via computers located in separated experimental stations.

To analyze the results, we compared the offers extended by participants across each treatment. For each participant we also calculated their minimum acceptable offer (MAO; representing the lowest offer a participant would accept) and compare these across treatments. Note that in any participant pair, an efficient outcome (i.e. an offer that was not rejected) was more likely in the presence of a higher offer and
a lower MAO.

When the music of Bon Scott was played, participants extended offers (albeit in Canadian dollars) with mean (standard deviation) of 3.28 (1.18) whereas participants in the Brian Johnson treatment extended offers with mean (standard deviation) of 4 (0.97). The distributions of offers in each treatment are presented in Figure 4.
Using nonparametric Wilcoxon rank-sum tests, we can reject the hypothesis that the distribution of offers across treatments are the same (p = 0.064). In terms of MAO, participants of Bon Scott treatment had minimum acceptable offers with mean (standard deviation) of 3.94 (0.87) while participants in the Brian Johnson treatment had minimum acceptable offers with mean (standard deviation) of 3.17 (1.25). The
distributions of MAO are presented in Figure 4. Again, using nonparametricWilcoxon rank-sum tests, we can reject the hypothesis that the distribution of MAO in each treatment are the same (p = 0.050). Thus, offers were lower and MAO were higher when participants heard the music of Bon Scott. This suggests that more offers would be rejected when listening to Bon Scott than when listening to Brian Johnson.

Our results suggest that having participants listen to songs by AC/DC in which Brian Johnson served as vocalist results in participants realizing more efficient outcomes. Thus, in terms of a singer’s ability to implement efficient behavioral outcomes among listeners, our results suggest that Brian Johnson was a better vocalist than Bon Scott....

Well, you can't argue with the Wilcoxon rank-sum test. But the interpretation of results seems to me to be exactly backward. Oxoby should ask himself, what was the message the singers were trying to get across, and how successful was each in getting that message across? Look at the lyrics for "Shoot to Kill" (Brian Johnson, vocalist):

All you women who want a man of the street
But you don't know which way you wanna turn
Just keep a coming and put your hand out to me
'Cause I'm the one who's gonna make you burn
I'm gonna take you down - down, down, down
So don't you fool around
I'm gonna pull it, pull it, pull the trigger
Shoot to thrill, play to kill
Too many women with too many pills
Shoot to thrill, play to kill I got my gun at the ready, gonna fire at will
I'm like evil, I get under your skin
Just like a bomb that's ready to blow
'Cause I'm illegal, I got everything
That all you women might need to know
I'm gonna take you down - down, down, down
So don't you fool around
I'm gonna pull it, pull it, pull the trigger
Shoot to thrill, play to kill
Too many women with too many pills
Shoot to thrill, play to kill
I got my gun at the ready, gonna fire at will
Shoot to thrill, ready to kill
I can't get enough, I can't get the thrill
I shoot to thrill, play to kill
Pull the trigger

A lovely little ditty, whose purpose is clearly anti-social. The fact that players arrived at more efficient outcomes (indicating more generous behavior on the part of the offerer and less churlish behavior on the part of the recipient of the offer) when listening to this song should be interpreted as failure of the vocalist to effectively transmit his message. Hence I would argue that Bon Scott, not Brian Johnson, is the better vocalist. Clearly, this is a subject worthy of more research.