A top GOP pollster who had House Majority Leader Eric Cantor winning in a walk last week acknowledged in a post-primary memo that his team’s methodology was “woefully inadequate” and said the poll should have taken into account the possibly of non-Republican voters participating.
Pollster John McLaughlin had put Mr. Cantor up 34 points, 62 percent to 28 percent, two weeks before his stunning double-digit loss to economics professor David Brat on June 10. While Mr. McLaughlin still retains a list of top-tier clients, at least one report says Republicans are encouraging candidates to rethink their using his services in the wake of Mr. Cantor’s defeat and some other polls that missed the mark in 2012.
In a memo released this week, Mr. McLaughlin writes that the sample used in last month’s poll was selected from a pool of voters who participated in either the 2008 or 2012 GOP presidential primaries or the June 2012 congressional primary in Virginia.
“Without a parallel Democrat primary, this election was very similar to a wide-open jungle-style primary,” the memo reads. “It created an organic turnout of new voters not included in our previous poll of past primary voters.”
The poll says Mr. Cantor won among voters who had participated in at least one of the three previous primaries, while Mr. Brat won among people who had voted in a Democratic primary and those who had never voted in a primary, with independents breaking for Mr. Brat by a 62 percent to 38 percent margin.
“Since they were not included in any past Republican primary voting samples, that is why our poll was wrong,” Mr. McLaughlin wrote in a memo accompanying a survey of 400 voters who participated in the June 10 primary. “In retrospect we needed to see this possibility and poll this race as if it were a special election and not just a typical Republican primary. The polling that we and others did from samples of past Republican primary voters was woefully inadequate in this case of an open primary for a national figure with no simultaneous primary on the other side.”
Indeed, voters do not register by party in Virginia, but political parties and groups frequently obtain voting records that indicate which elections in which they have recently participated in order to predict future behavior.
The survey showed that in poll conducted last month, 64 percent self-identified as Republican, three percent as Democrats and 31 percent independent. In the post-primary poll of people who participated, 54 percent said they were Republicans, 13 percent said they were Democrats and 32 percent said they were independents.
“Clearly Eric Cantor’s support was limited to within the Republican party, which was split, but organically, first time Republican primary voting Independents and Democrats made the difference,” the memo reads.
Some analysts have been skeptical that Democrats ultimately swung the outcome of the contest, as Mr. Cantor’s unofficial vote total of 28,902 actually lags his 2012 total of 37,369.
Nevertheless, Mr. McLaughlin says the sample should have been larger.
“While the Republican primaries that we have polled for successfully this year so far have been served well by polling past Republican primary voters, this race needed to be polled from a general election voter sample that could account for our influx of new voters into the open primary,” he wrote.