Making sense of the Gallup 7 percent Romney lead
Gallup used to be the Gold Standard. No more. They need to get busy & solve their problems.
— Larry Sabato (@LarrySabato) October 19, 2012
Top GOP pollster on Gallup’s 7-pt Romneylead nationally among likely voters: “There is nothing I am looking at that resembles that data.”
— John Harwood (@JohnJHarwood) October 18, 2012
Simply put, if Romney were up 7, he’d be advertising in PA, MI, OR, CT and wouldn’t spend a dime in FL.
— Sean Trende (@SeanTrende) October 18, 2012
Just because a pollster appears out of line doesn’t meant that it is wrong
With only nineteen days to go before the election the daily Gallup tracker polls are dominating analysis of the race. Last night a 7% Romney lead was reported which is way out of line with what the other firms are showing in their national surveys.
The state polls are reporting a much more evenly balanced race with Obama having, just, a slight edge in the key swing states.
Gallup is a tracker poll based on findings over seven days. So each new set of numbers is based on the oldest day’s findings being deleted and being replaced with the freshest findings. The result is that the Gallup tracker has this big Romney lead built into its system and we can expect good figures for the GOP challenger for at least the next few days.
The well-respected New York Times analyst, Nate Silver published yet another must read post overnight in which he goes into detail on the Gallup approach.
Nate Silver concludes that it is much more likely that Gallup is wrong &the others right than the other way round.is.gd/3yumAJ
— Mike Smithson (@MSmithsonPB) October 19, 2012
-
My uneasiness about rejecting Gallup out of hand is because of the British experience where there’s been a streak of elections where the poll that appeared the most out of line was the one that was nearest to the final outcome.
Thus Opinium’s 4% lead for Boris in May’s London mayoral race was by far the smallest of all the surveys – yet the firm came out as top pollster. A year earlier, in May 2011, ICM had repeatedly shown the biggest leads for NO in the AV referendum and its final poll was precisely right. It’s been the same at the general elections of 1992, 1997, 2001, 2005 and 2010 – the pollster at the extreme has been the top one.
All the signs are that the US election is going to be one of the closest on record. I find it completely absorbing and have been switching my betting between Romney and Obama.
Mike Smithson
For the latest polling and political betting news from the US and UK