How Accurate Were the Final Polls in the Race for CT Governor?

With the polls closed, how did the final public polls do?

With the race for Governor of Connecticut complete and Ned Lamont confirmed as the winner (685,453) with 49.1% of the vote over Bob Stefanowski (648,086) 46.4%[1] it is a good time to take a look back at the public polling of the race and see where they got it right. Between 10/28 -11/1/18, Quinnipiac University, Hearst/Sacred Heart University, Emerson College and Gravis Marketing all completed polling on citizen preferences in the Connecticut Governor’s race which they then shared with the public.

Tremont Public Advisors concluded two related surveys on 10/31 and 11/3/18 asking Connecticut residents which candidate for governor most “shared their values”.

The polls were conducted in a variety of manners with sample sizes ranging from 500-805 verified respondents with margins of error from 3.7-4.3%. How did they do?

Absolute error on the projected vote margin (or “absolute error”), is computed as the absolute value of the margin in the poll minus the same margin in the certified vote. Quinnipiac and Hearst/SHU polls had the smallest absolute error in the group with 1.3%. Their results most closely tracked the final difference between Lamont and Stefanowski (2.7%) in the final vote tally.

Quinnipiac, Emerson and Gravis correctly identified Ned Lamont as the likely winner of the Governor’s race.

Lamont’s final share of the vote fell within the range (including MOE) identified by Emerson and Gravis. Quinnipiac correctly identified both Lamont and Stefanowki’s share of the final vote within the MOE of its survey.

The final 11/3/18 Tremont survey asking Connecticut residents which candidate most “shared their values” had results that correlated closely with each candidates’ final share of the vote within the MOE with an absolute error of 1.3% (using “shared values” as a proxy for electoral support).

Though Quinnipiac concluded its polling eight days before Election Day, its final survey best reflected the election outcome. The other surveys, in all but one case, correctly identified the winner, yet understated the share of the vote that eventually went to Stefanowski possibly missing the late leakage of Republican voters from independent candidate Oz Griebel to Stefanowski.


[1] Results from CT Secretary of State’s website and the New Haven Independent’s tally of the vote in New Haven.