TAMPA, FLORIDA—It’s been a lousy year for North American pollsters and all those who have been guilty of paying too much attention to the horse race.
North of the border, polling in Alberta last spring was flat-out wrong, and in Quebec this autumn underestimated the Liberal vote. Federal Liberals, we are expected to believe, are in majority territory based on Justin Trudeau’s name.
So what to make of the blizzard of numbers in the U.S. presidential race?
Alarm bells are already ringing.
Republicans and Democrats claim to have diametrically opposite numbers in crucial states and are given free air time and website space to trumpet those numbers, unalloyed by any analysis.
But polls are heroin for cable news and social media here and they keep going back to their dealer for constant fixes.
Wednesday alone, there were 27 different polls released in battleground states.
In Ohio, eight polls released in the past week show Barack Obama winning the state by one to five points, but Rasmussen has Mitt Romney winning by two points.
Similarly in Florida, eight polls in the past week have yielded a Romney victory five times, a narrow Obama win twice, and one tie. Wednesday, four Florida polls had Romney leading in two, Obama in one and one tie.
The most closely-watched math belongs to the FiveThirtyEight blog by Nate Silver of The New York Times, who now ranks Obama’s chances of capturing the winning 270 Electoral College votes at 79 per cent, likening it to the odds of an NFL team nursing a field goal lead with three minutes left in the fourth quarter.
How often does that team win? Well, 79 per cent of the time, Silver says.
It is difficult to argue that this deluge of numbers is furthering democracy here.
The campaign in its final days more closely resembles a nationwide calculus class, with pollsters standing at a giant blackboard, chalking out logarithms to a numbed electorate.
It is confusing even the campaigns. As Reid Wilson, the editor-in-chief of The National Journal Hotline wrote Thursday, pollsters for both parties have spent weeks studying the same data and coming to different conclusions because they are using different assumptions.
U.S. polling is muddled on a number of levels. It is not a new insight to report that fewer people have land lines and it is more difficult to engage mobile users in polls, but that problem is accelerating. There is a mishmash of early and absentee voting in different states and voter affiliation is recorded, leaving each campaign to claim their voters are more likely to cast their ballot on election day if their man is behind.
But there is another factor that may be depressing Obama’s actual support, particularly in the battleground states of Florida, Colorado and Nevada.
As the Centre for American Progress pointed out Thursday, pre-election polling is underestimating minority voters, especially Latinos.
“Every year the electorate becomes less white,” said Ruy Teixeira, a senior fellow at the think-tank. “It is starting to penetrate into people’s minds that polls are not doing a 100 per cent accurate job of capturing this change.”
Two national polls had minority participation at 20-23 per cent this year, actually down from the 26 per cent participation in 2008—at a time when the minority population is growing. “It all doesn’t compute very well,” he said.
He and Matt Barreto of Latino Decisions say there is too little polling of Latinos in Spanish and that Latino opinion is disproportionately gleaned from those who are more highly educated, speak English and live adjacent to Caucasian neighbourhoods.
This showed up most starkly in Nevada in 2010 when Democratic Senate Majority Leader Harry Reid was shown trailing Republican Sharron Angle by an average of three points in 16 pre-election polls—but won by five points.
If, as they contend, Obama has 70 per cent of the country’s Latino vote, there could be a systematic under-representation of Democratic votes across the country, particularly in the 29 states where Latinos are the largest minority.
Somebody—perhaps many people—will be wrong Tuesday, but in the meantime we can amuse ourselves with international polling showing Obama favoured in 31 of 32 countries, with Israel alone backing Romney.
Some 88 per cent of Canadians and the same number of Israelis, according to Gallup International, believe Tuesday’s outcome will have an impact on them.
And, according to Upworthy.com, Clint Eastwood would overcome the empty chair fiasco and beat Oprah Winfrey for president 42 per cent to 38 per cent.
As Ohio’s Republican governor John Kasich told the Fox News Channel this week: “I am convinced that God created pollsters to make astrologers look accurate.”
Of course, Kasich doesn’t like the numbers in his state.
Original Article
Source: hill times
Author: TIM HARPER
North of the border, polling in Alberta last spring was flat-out wrong, and in Quebec this autumn underestimated the Liberal vote. Federal Liberals, we are expected to believe, are in majority territory based on Justin Trudeau’s name.
So what to make of the blizzard of numbers in the U.S. presidential race?
Alarm bells are already ringing.
Republicans and Democrats claim to have diametrically opposite numbers in crucial states and are given free air time and website space to trumpet those numbers, unalloyed by any analysis.
But polls are heroin for cable news and social media here and they keep going back to their dealer for constant fixes.
Wednesday alone, there were 27 different polls released in battleground states.
In Ohio, eight polls released in the past week show Barack Obama winning the state by one to five points, but Rasmussen has Mitt Romney winning by two points.
Similarly in Florida, eight polls in the past week have yielded a Romney victory five times, a narrow Obama win twice, and one tie. Wednesday, four Florida polls had Romney leading in two, Obama in one and one tie.
The most closely-watched math belongs to the FiveThirtyEight blog by Nate Silver of The New York Times, who now ranks Obama’s chances of capturing the winning 270 Electoral College votes at 79 per cent, likening it to the odds of an NFL team nursing a field goal lead with three minutes left in the fourth quarter.
How often does that team win? Well, 79 per cent of the time, Silver says.
It is difficult to argue that this deluge of numbers is furthering democracy here.
The campaign in its final days more closely resembles a nationwide calculus class, with pollsters standing at a giant blackboard, chalking out logarithms to a numbed electorate.
It is confusing even the campaigns. As Reid Wilson, the editor-in-chief of The National Journal Hotline wrote Thursday, pollsters for both parties have spent weeks studying the same data and coming to different conclusions because they are using different assumptions.
U.S. polling is muddled on a number of levels. It is not a new insight to report that fewer people have land lines and it is more difficult to engage mobile users in polls, but that problem is accelerating. There is a mishmash of early and absentee voting in different states and voter affiliation is recorded, leaving each campaign to claim their voters are more likely to cast their ballot on election day if their man is behind.
But there is another factor that may be depressing Obama’s actual support, particularly in the battleground states of Florida, Colorado and Nevada.
As the Centre for American Progress pointed out Thursday, pre-election polling is underestimating minority voters, especially Latinos.
“Every year the electorate becomes less white,” said Ruy Teixeira, a senior fellow at the think-tank. “It is starting to penetrate into people’s minds that polls are not doing a 100 per cent accurate job of capturing this change.”
Two national polls had minority participation at 20-23 per cent this year, actually down from the 26 per cent participation in 2008—at a time when the minority population is growing. “It all doesn’t compute very well,” he said.
He and Matt Barreto of Latino Decisions say there is too little polling of Latinos in Spanish and that Latino opinion is disproportionately gleaned from those who are more highly educated, speak English and live adjacent to Caucasian neighbourhoods.
This showed up most starkly in Nevada in 2010 when Democratic Senate Majority Leader Harry Reid was shown trailing Republican Sharron Angle by an average of three points in 16 pre-election polls—but won by five points.
If, as they contend, Obama has 70 per cent of the country’s Latino vote, there could be a systematic under-representation of Democratic votes across the country, particularly in the 29 states where Latinos are the largest minority.
Somebody—perhaps many people—will be wrong Tuesday, but in the meantime we can amuse ourselves with international polling showing Obama favoured in 31 of 32 countries, with Israel alone backing Romney.
Some 88 per cent of Canadians and the same number of Israelis, according to Gallup International, believe Tuesday’s outcome will have an impact on them.
And, according to Upworthy.com, Clint Eastwood would overcome the empty chair fiasco and beat Oprah Winfrey for president 42 per cent to 38 per cent.
As Ohio’s Republican governor John Kasich told the Fox News Channel this week: “I am convinced that God created pollsters to make astrologers look accurate.”
Of course, Kasich doesn’t like the numbers in his state.
Original Article
Source: hill times
Author: TIM HARPER
No comments:
Post a Comment