Analyzing the latest survey from St. Pete Poll showing a to-the-decimal-point tie

in Uncategorized by

In analyzing the St. Pete Polls’ last take on the governor’s race (shock, it’s tied at 45.5%) a question dawns on me, and this question should dawn on every analyst and every politico in the state.

Why is it, no matter how we poll this race, every one comes up even?

Recall the ending of the 1983 sci-fi thriller “WarGames.” No matter how the game (in that case, thermonuclear war) was played, it always ended with the same ending; no winner could be projected. The WarGames machine was trying to play out every scenario to see who the winner would be, and in each case, it could not find one.

USA first strike? Winner: None.

USSR first strike? Winner: None.

Far East strategy? Winner: None.

Hong Kong variant? Winner: None.

NATO territorial? Winner: None.

It’s kind of that way with all of these polls. No matter the methodology, the party balance, how the questions are worded, or even who the pollster is, it seems that nearly every poll taken over the past two weeks basically says this race is a statistical dead heat and none of them can project a winner.

Survey has too many Democrats? Winner: None.

Survey sample has too many younger voters? Winner: None.

Pollster uses live operators? Winner: None.

Pollster uses Internet respondents? Winner: None.

Pollster uses a blended sample? Winner: None.

Pollster asked party preference or didn’t link to voter file? Winner: None.

Pollster threw darts while spinning dervishly and hopping around on one foot? Winner: None.

Quinnipiac, Public Policy Polling, Survey USA, 0ptimus, St. Pete Polls, SEA, YouGov, Gravis, and others all show the race deadlocked. And, of course, the polling aggregator sites are in on the act.  Huffington Post has Crist up by 0.3%, RCP has Crist up by 1.0%, and Nate Silver’s 538 has Crist leading with 0.6%.

It’s a question I can’t answer, but it seems that no matter how skewed, biased, sloppy or exceedlingly well done a poll is … no matter whether the pollster used live operators, Internet Avatars, or robo-polls … the results all have this thing coming in for a razor-thin margin of victory.

So back to St. Pete Polls’ final survey headed into the election showing the race not just “statistically tied” but actually tied-tied at exactly (and befittingly) 45.5% each.  t’s well balanced by party, by age and by geography.  It’s a robo-poll, but at this point does that even matter?

Put away the saltshaker because this poll is saying the same thing every other poll has been saying for the past two weeks …Winner: None.