• User warning: The following module is missing from the file system: toolbar. For information about how to fix this, see the documentation page. in _drupal_trigger_error_with_delayed_logging() (line 1156 of /home/mprc2326/public_html/includes/
  • Deprecated function: Function create_function() is deprecated in eval() (line 1 of /home/mprc2326/public_html/modules/php/php.module(80) : eval()'d code).

MPRC Most Accurate of 2011 Polls

Submitted by admin on Mon, 11/14/2011 - 10:40

The Maine People's Resource Center (MPRC) released by far the most accurate polls of the November, 2011 election. The organization also released more information about its results and was more transparent in its methodology than any other local or national pollster.

Three polling organizations provided information on Question 1 on the ballot. Critical insights (CI) projected an 8-point margin (with 6% undecided). Public Policy Polling (PPP) projected a 4-point margin, (with 7% undecided). MPRC projected a 16.5-point margin (with 7.1% undecided). All three projected the "Yes" side ahead.

The actual result was a margin of 20.74 percentage points. [1] By allocating the undecided voters proportionally and comparing the projected margins to the actual results, [2] we find that MPRC had an error of 2.98, CI had an error of 12.23 and PPP had an error of 16.44.

The chart below shows error for both unallocated (Error) and allocated undecided voters (Error2).

Q1 accuracy chart

The Maine Heritage Policy Center also released a poll on Question 1, not included in this evaluation due to biased instrument wording and questionable methodology. That poll's error on Question 1 would have been 26.74.

Only CI and MPRC released results for Question 4, which, due to a higher number of undecided voters, can be considered much more difficult to project. CI's results showed a one-point advantage for the "No" side, while MPRC showed a 4.1 point margin for "Yes." On Election Day, Maine voted "Yes" with a margin of 5.74 percentage points.

With undecided voters allocated proportionally, CI had an error of 7.28 on Question 4, while MPRC had an error of .03.

Q4 accuracy chart

MPRC was also the only organization to release results on Questions 2 and 3 and successfully projected the direction and scale of the margins on both issues, although not as precisely as on Questions 1 and 4.

Additionally, MPRC was the only organization to release a poll of the Portland mayoral race. This was the inaugural race for this position as well as the first use of instant-runoff voting in Maine.

MPRC projected the vote percentages of the top two candidates with an average error of .7 percentage points. For all nine candidates measured, the average error was only 1.06. This is surprisingly precise and is less than would be expected due to simple statistical variation.

Portland mayor accuracy chart

MPRC also accurately projected the number of rounds of instant-runoff voting and the eventual winner.

"I believe these results will contribute to MPRC's reputation for accuracy," said MPRC communications director Mike Tipping. "I'm proud of what we achieved this year and I hope MPRC's example of methodological transparency will contribute to greater openness and honesty in public opinion research."

The National Council on Public Polls has also found MPRC to have produced the most accurate results of evaluated polls for the 2010 Maine Gubernatorial Election.[3]

1. All results from Bangor Daily News unofficial totals
2. A Mosteller 5 measure of accuracy. For a more complete description of measures of accuracy, see