Several people have brought this article, "I analyzed the chords of 1300 popular songs for patterns. This is what I found." to my attention.
While the methodology is well thought out, I don't think simple frequency statistics with regard to key and chord occurrence will tell anything particularly meaningful about composing popular songs with effective hooks.
Keeping in mind that this is only the first post in a promised series, there are several further avenues of inquiry that the author may intend. So, perhaps he will do more than simply scratch the surface.
With regard to proceeding, an additional level of frequency statistics could be applied to the present data by simply labeling the chords already ID'd with their Roman Numeral degree symbols: I, ii, iii, IV, V, vi, vii(d).
That metric would allow the research to proceed to identifying harmonic succession patterns, and their frequency, which would begin to hone in on what an effective pop song hook looks like in terms of harmonic structure. Here's where details begin to become important: The root motions have to be labelled first in order to get a handle on recognizing the harmonic succession patterns. I have no time to get into that in detail, but what I call Progressive (+4th), Regressive (-4th), Half-Progressive (-3rd), Half-Regressive (+3rd), Super-Progressive (+2nd), Super-Regressive (-2nd), and Tri-Tone motions have to be labelled and added to the database.
The next step would be identifying all of the groups of two types of root motion, then three types, &c. THEN, and only then, could the composer begin to draw some concrete conclusions about what kinds of root progression patterns create harmonic continuities that are catchy.
I'm not going to pursue that myself, as I'm simply not personally interested, but it could be a genuinely scholarly inquiry if it is followed as I suggest.
