[terry]

My only complaint is there seems to be no thought as whether Facebook is a good data source for the project.

I agree. An analysis based on Facebook "likes" clearly is not a valid statistical analysis. People who join Facebook, then choose to "like" a team, are self-selected, not randomly chosen. Then there's a mystery algorithm that was used to "smooth the data and fill in the gaps." Is there some reason to believe that this method provides a good reflection of the overall population? The authors don't tell us.

Bringing this topic back around to Stanford, it's interesting to note the preferences in Stanford's zip code (94305):

62%  Giants
8%    A's
7%    Red Sox

Those numbers are typical of much of the Peninsula and the South Bay. The A's barely register on the radar, even after 45+ years in Oakland, 16 division titles, 6 World Series appearances, and an active campaign to move the A's to San Jose. Even in the East Bay, most areas prefer the Giants by a wide margin. The A's really look like a poor step-child under this methodology.

[terry]

My only complaint is there seems to be no thought as whether Facebook is a good data source for the project.

I agree. An analysis based on Facebook "likes" clearly is not a valid statistical analysis. People who join Facebook, then choose to "like" a team, are self-selected, not randomly chosen. Then there's a mystery algorithm that was used to "smooth the data and fill in the gaps." Is there some reason to believe that this method provides a good reflection of the overall population? The authors don't tell us.

Bringing this topic back around to Stanford, it's interesting to note the preferences in Stanford's zip code (94305):

62%  Giants
8%    A's
7%    Red Sox

Those numbers are typical of much of the Peninsula and the South Bay. The A's barely register on the radar, even after 45+ years in Oakland, 16 division titles, 6 World Series appearances, and an active campaign to move the A's to San Jose. Even in the East Bay, most areas prefer the Giants by a wide margin. The A's really look like a poor step-child under this methodology.