Hey, remember that Castleton Polling Institute survey of the Democratic race for Attorney General? The one that came out a week before primary day, and showed incumbent Bill Sorrell with a big lead?

The one that some experts questioned after it was published? Well…

“That’s a case where – I’ve talked about this here with colleagues – it was probably a case where we never should have released numbers,” said Rich Clark, director of the Castleton Polling Institute. “That’s my fault, probably, for putting out the numbers.”

Ah. Oh. Never mind, then.  

The poll, released on August 22, gave Bill Sorrell a 44-24 lead over TJ Donovan with a whopping 31% undecided. Its major flaw, according to Vermont Pundit Laureate Eric Davis, is that it vastly overestimated voter turnout in a late-August primary with only one significant contested race.

I have some sympathy for Rich Clark. The CPI is a new thing, and it’s risen quickly to prominence in a state that doesn’t get polled very much. I hope it overcomes this self-inflicted wound and becomes a stable, reliable source of polling. But this is not good. The result was wildly inaccurate, and may well have played a part in deciding the outcome of an extremely close contest.

My first reaction (as someone who voted for TJ Donovan) was that the poll result may have dispirited his campaign. Since TJ came within 700 votes of victory, the smallest doubt in campaign workers’ minds could have slowed his momentum enough to cost him the race.

Of course, it’s also possible that the poll could have lulled Sorrell’s troops into a state of complacency; if they thought they had it in the bag, they might have eased off in the homestretch.

For that matter, it might have had little or no effect either way. The poll attracted a lot of attention among insiders and political junkies, but how many voters were even aware of it?  

We’ll never know the truth. But it’s an unfortunate incident in the young life of the Castleton Polling Institute.