Just some thoughts off the top of my head:
I would imagine that clients with a higher median age in general prefer slower music.
I would also imagine that clients with a higher median age in general are more affluent (or more likely to be shopping for more than one person).
As a result, I would expect spend to be higher in areas with an older customer base.
Managers with an older customer base would likely select music that appeals to their customer base, likely slower music. Shoppers who do not like the manager's music taste may opt to instead shop somewhere where the music is more in line with their own tastes, so I'm not sure that the independent variable here can be properly controlled.
Therefore, the study may actually be finding a correlation which depends rather on the affluence of the community, rather than what they purport to have discovered.