I don't get it.
I take it there was a correlation between affluence of the listeners and the stations they listened to. But how does it matter in what order you put that correlation?

It's all a bit confusing to be honest. I understand the mistaking-correlation-for-causality joke, but I don't see how it is a problem of phrasing it as "X tends to be Y".

If you want to sell books, then it's useful to know what the profile of avid readers is. So "people who buy books tend to be female" is useful and appropriate knowledge, regardless of correlation=/=causation jokes it might give rise to. On the other hand, if you want to market something to women, then it isn't necessarily useful information; rather, it depends on whether "women tend to read books". And I suppose that was more in line with what you meant with getting the data analysis backwards, right?

Unfortunately, I still read "because their listening audience ranks above-average in affluence" as meaning the exact same thing as "[because] their listeners tend to be more affluent than the average"; because in both case we look at the profile of the listeners. What I think you meant (and correct me if I'm wrong) would be along the lines of "people of affluence tend to listen to these radio stations (more than to other stations)", which instead tells us something about people of affluence.

And then your friend might still have quipped "Really? How much more money do I need to make before I start listening to their station?"