Here is what Sabato, who is very conservative and somewhat authoritative says about Sarvis:
One factor that might have confounded pollsters was Robert Sarvis, the Libertarian candidate who took about 6.5% of the vote. He ended up (unsurprisingly) receiving a smaller percentage of the vote than his polling suggested (he was at about 9% or so in the final polling averages). It’s rare for a third-party candidate to do as well as Sarvis did in a state like Virginia, where Democrats or Republicans generally got nearly all the votes in recent gubernatorial elections. Public polls also seemed to disagree on whether Sarvis was hurting Cuccinelli or McAuliffe more and perhaps his presence in general just confused the polling.
By the way, the exit poll was not definitive on this same subject.
On the one hand, Republicans will note that 3% of Democrats voted for Sarvis; 4% of Republicans; and 15% of independents — a group that, we previously noted, Cuccinelli won.
On the other hand, Democrats will stress that, in a two-way race question asked in the exit poll, McAuliffe would have won 48% to 46%, and 5% of McAuliffe’s voters in that part of the survey said they actually backed Sarvis, while only 2% of Cuccinelli’s support came from the Sarvis column.
We believe the exit poll when it suggests that McAuliffe probably would have won a two-way contest — and maybe the pollsters would have had an easier time surveying that more basic race.
Link:
http://www.centerforpolitics.org/crystalball/articles/the-surprisingly-dramatic-terry-ken-show/