With only a month left until the New Hampshire state primaries, the gubernatorial race is heating up on both sides. Major news outlets in the state have released several polls showing the Republican races to be narrowing to just two candidates: Chris Sununu and Ted Gatsas. On the Democratic side, the race is reportedly between Colin Van Ostern and Mark Connolly.
However, just how accurate are these polls? Are they really representative of the likely primary voters in the state? To answer that question, we covertly ran a poll that was selectively and anonymously released to groups of people, both Democrats and Republicans, who were identified as high probability primary voters. We found a significant difference between the major polls and our own. This poll had a sample size of 201 to represent the ~1,327,000 residents of the state. We omitted responses that did not provide an area code to substantiate residence. The following data was collected between 8/3/16 and 8/12/16 at a confidence level of 95% with a margin of error of (+/- 6.9%):
Edelblut (R-Wilton): 30.3% overall, 57.3% among Republicans
Sununu (R-Newfields): 14.6% overall, 23.2% among Republicans
Gatsas (R-Manchester): 10.8% overall, 12.2% among Republicans
Forrester (R-Meredith): 5.4% overall, 4.9% among Republicans
Marchand (D-Portsmouth): 23.0% overall, 24.7% among Democrats
Van Ostern (D-Concord): 21.9% overall, 42.0% among Democrats
Connolly (D-New Castle): 10.7% overall, 17.3% among Democrats
Freeman (D-Keene): 3.7% overall, 4.9% among Democrats
We targeted likely primary voters by tracking social media groups and the accounts of the people who the poll was ultimately released to. We looked for accounts that were actively posting either about New Hampshire politics or specifically the Governor’s race. The poll was withheld from any person who could be proven to be working for a campaign. In all cases the poll was sent either via private message or email.
Edelblut blows his competition out of the water.
Perhaps the most notable difference is that both Sununu and Gatsas are dwarfed by Frank Edelblut among likely Republican primary voters. Among all likely primary voters considering Republican candidates, Edelblut had almost double the support of Sununu and triple the support of Gatsas. Among only self-reported Republican voters Edelblut had over twice as much support as Sununu. Edelblut was also listed as an option by 22.2% of the undeclared voters. Gatsas was close behind with undeclared voters at 18.5% and Sununu was right behind him at 14.8%.
Just last week, Frank Edelblut released a list of over 50 current State Representatives who support his bid for the corner office. He also touts endorsements from the New Hampshire Liberty Alliance and the Coalition of New Hampshire Taxpayers; both libertarian leaning policy advocacy groups. Meanwhile, Chris Sununu flaunts the endorsements of former governors Judd Gregg and Steve Merrill. While Edelblut seems to be trying to appeal to the libertarian-leaning base it appears that Sununu is looking to gain favor from the ‘traditional’ conservatives. Republicans showed a tendency to be somewhat more partisan than Democrats; with 75.6% reporting that they would not vote for a Democratic candidate. By contrast, 67.9% of Democrats reported that they would not vote for a Republican candidate.
Marchand could flip the script with Undeclared voters.
Among the Democrats, Marchand and Van Ostern appear to be neck and neck. However, upon closer inspection, it is revealed that Marchand’s support is largely from swing voters. Marchand was listed as a potential candidate by a whopping 48.1% of the undeclared voters. Additionally, several Republican respondents selected Marchand in addition to their Republican candidate selection. This may mean that if Marchand was given his party’s nomination he would have more crossover voter support than his opponents. However, he only has a little more than half the support that Van Ostern has among self-reported Democrats.
Connolly, who was reported in other polls to be second to Van Ostern, did not even match Marchand in partisan support; trailing by a little over 7 points. Connolly, a former deputy Secretary of State and Director of the State Bureau of Securities Regulation, is a party insider favorite who has endorsements from several prominent elected Democrats including Mayor Paul Grenier of Berlin and State Senator Andrew Hosmer.
How accurate are the major news polls?
The responses to this poll call into question how scientific major state news networks’ polls really are. Why is there such a drastic disparity between our responses and those of the large networks? After all, the major networks have more resources and are (at least theoretically) better able to target likely primary voters. The answer is that the data collected by most newspapers and TV/radio stations is not drawn from likely primary voters. Often times the polls run are sent out to people who are mostly politically inactive. The polls are usually administered over the phone and are not in any way targeted other than the fact that the person has at some point in the past voted in a primary. Many times the polling methodology is left open so that the only criteria for a ‘likely primary voter’ is that the person in question is registered to vote. There is no further investigation as to whether or not the voter is actually a high propensity voter.
Why is this targeted social media method not a standard practice of pollsters looking to have a concentrated, representative sample size? It is presently considered a form of bias to use social media to target poll subjects. But in a time where social media is the most popular method of communication, would it not make sense to conduct polls, if not using social media, with information gathered from social media, rather than phone-based polling? One thing is for sure: the election on September 13th will prove which method is best.
All polling and data compilation conducted by Caleb Dyer