Here's what the turnout has been in the highest-vote-total primaries in Virginia since 2006.
That top bar, obviously, is Tuesday's result in Cantor's district. The second-highest total is the same district in 2012, with nearly 20,000 fewer people coming out to vote. (The Virginia primaries in 2012 didn't coincide with the presidential primary.) Turnout in Virginia's 7th on Tuesday was at 13.7 percent. No other congressional primary in 2012, 2010, 2008, or 2006 topped even 10 percent.
In an interview with the National Journal magazine Wednesday, Cantor's pollster, John McLaughlin, explained that his turnout estimate was that about 45,000 people would vote, not the 65,000 who actually did. That incorrect turnout estimate is almost certainly why McLaughlin's polling, which showed Cantor with a wide lead at the end of May, was so far off. If you misunderstand who's going to come out to vote, your estimate of how much support you'll see on Election Day will be wrong, too.
But it wasn't a crazy estimate. As you can see in the graph, 47,719 people voted in the 7th District in 2012 — 9.7 percent of the registered voter base. The average turnout in Republican primaries in 2010 was about 31,000, or 6.9 percent turnout, and lower in both regards in 2012. There may well have been signals that McLaughlin missed showing that turnout would surge, but it's not clear what those might have been. If anything, 2012 could have been a high-water mark.
Which brings us back to Klein's argument. In 2012, Cantor got 37,369 primary votes against a gentleman named Floyd C. Bayne, winning 79.4 percent to 20.6. Klein's argument is that those 37,369 voters didn't come out to vote for Cantor in 2014, because Cantor got only 28,898 votes on Tuesday. (Actually Klein's argument is that the people who gave Cantor 223,000 votes in the 2012 general election didn't turn up to vote for him, but that's a particularly weird argument.)
The implication is that voters never change their minds, that Cantor voters are Cantor voters forever and will always be more than 50 percent of the district. Yes, as Klein says, it is "possible that the majority of Republicans in Virginia's 7th District wanted Cantor to win and just never considered the idea that he could lose." It is also possible that a majority of the Republicans in the district changed their opinions of Cantor between 2012 and 2014. There were 223,000 Cantor voters in 2012 — but there were zero for David Brat, the man who defeated Cantor. And those votes had to come from somewhere. Between 2012 and 2014, Cantor lost 8,471 primary votes. It's not crazy to think that many of those votes ended up in Brat's 36,000.
Looking at a turnout of 45,000 people, Cantor's pollster estimated that he'd win handily. Who knows what he would have seen in the numbers if he'd predicted the turnout correctly.
But more voters clearly worked against Eric Cantor in Tuesday's election. If, as Klein theorizes, Cantor's campaign had gotten even more people to come to the polls, the only available evidence suggests that the result for Cantor would have been even worse.