Lies, Damn Lies, and Polling Statistics

We’re looking at the latest Boston Globe poll on the Senate race in Massachusetts and something doesn’t add up.

On the one hand, this poll purports to show that Martha Coakley has a commanding lead, with 43% for Coakley, 22% Capuano, 15% Pagliuca, 6% Khazei, and 13% Don’t Know. On the other hand, the poll states that 75% haven’t made up their minds. That all adds up to way more than 100%. So we took a closer look at the numbers. Once the huge undecided vote is properly factored in, it is clear the race is still wide open:

Percent	Candidate
79%	Undecided (plus Don’t Know)
11%	Martha Coakley
5%	Mike Capuano
4%	Steve Pagliuca
1%	Alan Khazei

Now, here is where it really doesn’t add up. If you apply the 1% for Khazei to the 450,000 voters that might be expected for the special election primary on December 8, that translates to 4,500 votes. However, Citizens for Alan Khazei has 4,348 friends on Facebook. And over 3,400 citizens have contributed to his campaign. It’s simply not credible that 97% of the people who intend to vote for Alan Khazei have friended him on Facebook and 75% have given him money. This poll must have a huge Khazei undercount.

So what gives? I have two theories. The first is that it is impossible to poll most Khazei supporters. If they are not working at their day jobs, Khazei supporters are out knocking on doors to canvass for votes, or making phone calls, or not answering their phones to avoid calls from the Khazei campaign to go out and canvass or make calls.

The second is that the Khazei campaign may be playing a great game of Build, Hide, and Overwhelm (BHO). This strategy is based on the difference in dynamics between a head-to-head race with two candidates and a multi-candidate race such as we have with four candidates in this race.

In a two-person race, it makes sense to compete head-to-head, competing for strength in every poll up to election day, with a strong dose of negative advertising to undercut your opponent’s support, who will have nowhere else to go but to you.

In a four-person race, any head-to-head competition or negative advertising may drive voters to one of the other candidates. In this type of race, if you are not the frontrunner, it makes sense to hang back and build your strength quietly, hide your strength while the other candidates battle it out, and then bring out your strength in an overwhelming final push.

Mike Capuano may be playing BHO too. He got his current job as 8th District Congressman by winning a 10-person Democratic primary in September 1998. His strategy was to let his opponents battle it out until the final week or two of the election, then he came on strong with a series of TV ads positioning himself as a man of accomplishment in his years as Mayor of Somerville. He may be playing to let Martha Coakley and Steve Pagliuca slug it out with their TV advertising buys, then come on strong at the end. It worked for him in 1998, but so far Coakley and Pagliuca are not slugging.

All of this raises the question, just how many votes will be needed to win this Senate primary? In a two-way race, a candidate needs 51% to win. But in a three-way race a candidate can win with 40% and in a four-way race with around 33%. Theoretically a candidate in a four-way race needs just 25.1% to win, but that assumes the other three candidates are tied at 24.9% (unlikely).

How does that translate into votes? The primary winner will need to get 150,000 to 300,000 votes on December 8. Shannon O’Brien won a four-way primary for Governor in 2002 with 33% and 243,039 votes. John Kerry won a four-way primary for U.S. Senator with 41% and 322,000 votes in 1984. However, those were regular election primaries held in September with candidates running for a number of offices. This is a special election primary for a single office to be held in December where turnout could be as low as 450,000. That means 150,000 votes could win this Senate primary.

None of the candidates are there yet. Based on the Boston Globe poll, I count just less than 50,000 solid votes for the putative frontrunner Martha Coakley, well short of what is needed to win. But the real frontrunner is Undecided, and how the undecided voters of Massachusetts make up their minds in the next 2 weeks will determine this Senate primary election.