Rather than simply guessing what happened to Rep. Eric Cantor, a political firm called Silver Bullet LLC has gone into the field to find out what went wrong. They surveyed people who voted in the Republican primary election in Virginia’s 7th district, and found a few things of interest, including the fact that:
Voters who said they saw very few advertisements from David Brat were actually more likely to support Brat. Unaffiliated voters were less likely to have seen advertisements from Brat than Republicans, so a possible explanation for the odd, inverse relationship between advertisements seen from Brat and support for Brat could be that the campaigns were not targeting unaffiliated voters. Those voters had a propensity against supporting Cantor but immigration reform is not a likely culprit, only 7% of unaffiliated voters cited that as the main reason for not supporting Cantor. These unaffiliated voters made up 24% of the electorate and broke for Brat 2.3:1.
Also worth noting:
26% of voters who supported Brat changed their mind in the last two weeks, so there is evidence of a significant last-minute shift in the electorate. Combine this with a lack of accounting for the large number of unaffiliated voters who pushed turnout higher and the polling misses can be almost fully explained, if not fully justified. [Emphasis mine.]
This was an automated survey, for what it’s worth. You can see the full methodology report here.