How not to Analyze Political Results

Just when I think Ben Shapiro has hit bottom with ridiculous arguments, he comes up with one even worse. Speaking of Barak Obama, he says:

"What do you call a candidate who wins 90 percent of the African-American vote, between 30 percent and 50 percent of the Hispanic vote and 40 percent of the white vote in a tight Democratic primary race? A general election loser. Apply those percentages to the general election, and the candidate will bomb."

Every sports fan knows you can't take a team's stats against one opponent and apply them to the next opponent. Different teams have different strengths and weaknesses. It you think the Patriots are going to score 42 against the Giants because they scored 42 against the Raiders, you are going to be very disappointed. Likewise, one cannot apply primary percentages to general elections. Yet that is exactly what Shapiro is doing here. To make comparisons that make sense, they need to be apples to apples, or in this case Republican vs Republican, and Democrat vs Democrat. One cannot do this:

"In 2004, President Bush won 43 percent of the Hispanic vote, 58 percent of the white vote and 11 percent of the African-American vote. That means that John Kerry did better among Hispanics than Barack Obama has done in the Democratic primaries; better among whites than Obama has done in the Democratic primaries; and almost as well among African-Americans. Obama's coalition is Kerry's, but weaker."

This analysis is so bad it is comical. Of course Kerry did better among Hispanics against Bush than Obama did against Hillary Clinton. Practically any Democrat would. As for the whites, one cannot compare % of Democratic whites to % of total whites, especially with the proportion of whites, and voters in general, who consider themselves Democrats vs Republicans, growing rapidly. One get more votes with a smaller percentage of the pie if the pie grows. And it is not the least bit clear that Obama's percentage of the pie will be smaller against McCain.