A big ‘What If?’ emerges on the Warner-Gillespie race

[UPDATE: Travis Fain at the Daily Press reports that a Hampton University spokesman has said the poll was not released because it showed the opposite of what we wrongly suspected.[read_more]  Instead of showing Gillespie gaining ground, it showed Warner up by double digits, apparently causing the organization to doubt its accuracy.  We regret our error in believing otherwise.  However, in the interests of transparency, we do hope that Hampton University releases the full data on the poll to help analysts better assess the reliability of future polls they conduct.  Kudos to Travis Fain for the scoop, and the good journalism.]

The dominant news in the closing days of the recent race between Ed Gillespie and President Obama’s Virginia handmaiden, Mark Warner, was that Gillespie was closing the gap and rapidly gaining ground on Mark Warner. As it turns out, there likely was public proof of that fact that appears to have been quashed to avoid having an impact on the race.[read_more]

As we wrote the day before the election,

If the word I’m hearing about internal polls are true, Warner’s decision to finally take Gillespie seriously may be too little, too late.
…
The latest public poll from Christopher Newport University has Warner at +7. That same shop had McAuliffe at +7 in their final pre-election poll last year. Ken Cuccinelli outperformed that poll and came within 2.5 points of winning. Taken together with a likely oversampling of Democrats (seriously, how can a pollster in this environment assume more Democrats will vote than Republicans?) and some internal numbers showing Gillespie is actually ahead, I think the Gillespie momentum has reached its peak and that Ed pulls off a close win tomorrow.

As it turns out, the CNU poll wasn’t the last public poll prior to the election. There was another one, by Hampton University, that was ready for release at least two days before the election. As reported by The Daily Caller, “The Hampton University poll was scheduled to be released before the elections, however, it was held for further review,” said Yuri R. Milligan, the school’s director of University Relations. “It was then too late in the day — the day before the election — to release the poll.”

So why was the poll not released? We believe it was for one of two reasons, both likely related to the poll confirming what the Gillespie campaign maintained all along: that Gillespie had pulled even or overtaken Mark Warner. They either decided the poll must have been flawed, or they decided the poll would have too positive an impact for Ed Gillespie. The “objective, non-partisan” center at HU that handled the poll had no such compunction when, earlier in the race, they published a poll (with a straight face) showing a 25 point lead for Warner and suggesting that signaled his likely re-election. Presumably they used the same methodology in the earlier poll, where they appear to have had no doubts about accuracy. That point weighs pretty heavily in favor of a conclusion that the decision to not publish a poll favorable to the Republican—at precisely the point when it could have given him the momentum he needed to finish first across the finish line—was, at bottom, politically motivated (or at the very least, done in a way that was quite conscious of the potential electoral importance of the decision).

We can never know if the publication of that poll would have made the 0.5% difference. But what we do know was that Gillespie was finishing strong, and Warner was finishing weak. Gillespie told The Bull Elephant on the Saturday before the election that everything he had been seeing suggested that his campaign was overtaking Warner’s on that last weekend, just in time to pull out a close win. If a favorable poll had been published in time to make all the newspapers and television reports (and in time for the campaign and friendly online venues to make hay of it), it is quite possible that it could have encouraged enough Republicans (or discouraged enough Democrats) to erase and overtake the narrow statewide 16,000 vote gap on election day.

In the end, that appears to have been the point.