There are significant differences between IVR, live operator, and internet polls of the race for Massachusetts governor

Which polling methodology is seeing the right set of voters won't be known until after Tuesday's election
by BRENT BENSON

As the polls of the race for Massachusetts Governor have steadily moved from a Martha Coakley lead in early September to a Charlie Baker lead in late October, a complicating factor has arisen: each of the three polling methodologies used to survey the race are showing different results. Interactive Voice Response (IVR) or robopolling firms show a reasonably large 5 point lead for Baker. Traditional live operator telephone polling shows Baker with a 2 point lead. And internet panel surveys show Coakley with a 4 point lead in the race.

Which of the methodologies are seeing the appropriate set of voters? IVR-only firms like Emerson are not able to reach large swaths of the population without landlines, meaning fewer younger, urban, and non-white voters. Live operator telephone polling suffers from severe non-response effects as more and more people screen their phone calls. Internet surveys are the new kids on the block and have their own (probably smaller) set people who cannot be reached.

Pollsters of each persuasion use best-practices and weighting to compensate for these effects, but it seems that each of the methodologies is getting a different view on the Coakley/Baker race. If the IVR firms and/or the live operator pollsters are having trouble seeing all of the Democratic voters, Baker could be in for a Tuesday shocker. If the internet pollsters are getting it wrong, Martha Coakley will suffer another tough loss and Charlie Baker will be the next in a long line of Republican governors of Massachusetts.