New Poll Has Tom Perriello Up 54%-46%; Corey Stewart (!) Up 42%-41%

First, I’m going to summarize some key findings from this new poll, then I’m going to lay out a few reasons why you might – or might not – believe it.

• “On the Democratic side, Tom Perriello has a clear, 8-point lead, 54–46, over Ralph Northam. Former congressman Perriello does well among a wide array of Democrats: young voters, former Bernie Sanders supporters, black voters, and voters unenthusiastic about the economy.”
• “Tom Perriello’s lead among young voters is huge: he leads by nearly 30 points among voters under 45. Voters under 45 made up 44% of the Virginia electorate last November, and our models predict that they will make up 39% of the electorate on Tuesday.”
• “Northam’s margin among the oldest voters — those 65 and up — is indeed above 20 points. But Northam’s lead is far smaller among those between 55 and 64, and voters between 45 and 54 are evenly split between the two Democratic candidates.”
• “Perriello also fares quite well among minority voters, leading by 20% or more among blacks, Hispanics, and Asian Americans. Collectively, these three groups make up nearly half of the Democratic electorate in Virginia. Whites planning to vote in the Democratic primary are evenly split between Perriello and Northam.”
• “On the Republican side, we see a very tight race (in contrast to other pollssuggesting Ed Gillespie will rout his opponents), and we give a very slight edge to Corey Stewart. Stewart’s advantage comes from his enthusiastic base: his supporters are both more likely to vote and less likely to change their minds. We project Stewart ahead of former RNC Chair Ed Gillespie, 42–41, with Frank Wagner in third place at 16%.”

And now some reasons you might – or might not – put credibility in this poll.

• It’s an almost brand-new polling outfit, so no track record.
• On the other hand, the first poll they did – of the Montana At-Large race recently – was very accurate (they forecast a 5-point win for Republican Greg Gianforte over Democrat Rob Quist, and the final results were a 6-point win for Gianforte).
• Also seemingly lending the poll credibility: “Change Research was founded by two former LinkedIn data scientists, to make public opinion polling accessible to candidates at every level. Change Research finds voters online (cheap!) and uses data science to ensure a representative sample (accurate!).”
• Looking at the “cross tabs” of this poll, nothing in particular’s jumping out at me as glaringly wrong on the Democratic side. Also, a lot seems right when you look at age distribution of support (e.g., younger voters heavily for Tom; older voters strongly for Ralph), the issues distribution (e.g., those who list environment/Dominion as top concerns go HEAVILY for Tom; those who rank gun safety as most important go for Ralph heavily); ideological (e.g., those who voted for Bernie Sanders go heavily for Tom; those who voted for Hillary Clinton in the primary lean slightly to Ralph); geographic (e.g., Tom wins Charlottesville big time and does very well in rural Virginia; Ralph wins Norfolk by a significant margin), etc.
• On other other hand – and this is by far my biggest concern with this poll – what on EARTH is going on with the Republican race for governor???  Corey Stewart ahead of Ed Gillespie, when every other poll has found Gillespie crushing Stewart?
• Maybe this helps explain those seemingly outlier-ish results? “Overall enthusiasm on the Republican side is light — and Corey Stewart may fill the void…Support for Trump ties closely with support for Stewart. 14% of Virginia’s electorate rates Trump’s performance a 9 or 10 out of 10. These Trump fans overwhelmingly plan to vote Republican on Tuesday, and they strongly support Corey Stewart, giving him a 15-point lead.”
• This gives me more confidence; huge number of voters surveyed! “3623 adults were surveyed. Unregistered respondents were excluded from the ultimate sample, leaving a total of 3232 voters.”

Bottom line: I see no particular reason to discount this poll, but I’m not sure I’d totally accept it either, given the lack of a track record and its cutting-edge online polling methodology (as they describe it: “Founded by two of LinkedIn’s first data scientists, Change Research brings an innovative, Silicon Valley style to the traditional field of polling. Big changes are coming!”).