Conducted: June 11-12, 2013
Margin of Error: +/-4.52%
Harper Polling conducted an experiment intended to shed light on a topic of much debate among political pollsters: Who do you call? Is your call file composed of registered voters or voters with a predetermined vote history? In both scenarios, the pollster will ask each respondent whether they intend to vote in the upcoming election. Those who say 'no' get dumped out of the survey.
Despite screening out self-identified non-voters, does a poll yield different results based on who you call in the first place?
Our Massachusetts Senate Special Election poll for Conservative Intel on June 10th & 11th showed Congressman Ed Markey in control of the race by 12% with two weeks to go. On that survey, we called registered voters and screened out self-identified non-voters in the special election.
On the second night of that survey, we began a simultaneous second survey that ran from June 11-12th using the exact same polling instrument and a nearly identical weighting model. Once again, we screened out self-identified non-voters.
But this time, our call file was composed of those who voted in at least two of the last four general elections including newly registered voters. No respondent participated in both surveys.
That poll produced very different results with Markey leading by just 1%. This test was not intended to predict the outcome of the special election. It simply reveals that who pollsters call does, in fact, matter.
A 12% lead (49-37%) for Markey shrinks to 1% (44-43%).