10.27.2004
The Nonscientific Reality Behind US Election Polls (Scoop.co.nz)
By Paul G. Buchanan
25 October 2004
The worst mistake a political analyst can make is to offer predictions about future events. The study of politics is nowhere close to being an exact science, and in fact makes meteorology look downright precise. This has not stopped those who study politics for a living from calling themselves "scientists," and from attempting to emulate the real sciences when describing political phenomena in terms of absolutes, trends, and indexes of predictability. At best, this is informed opinion masquerading as fact, and at worst it is nothing more than playing at being economists. Unlike the laws of nature and physics that govern scientific inquiry, or the stylized models upon which most macroeconomic theory is premised, the study of real human political behaviour is as contingent and precarious as the subject is fickle and unpredictable. Thus political analysis is no more than an exercise in focused speculation.
Nowhere is this more evident than in the work of those who conduct political opinion polls. Consider, for example, the polls that showed Mark Latham and John Howard in a near dead heat in the run-up to the recently concluded Australian national election. Needless to say, the pollsters proved wrong, much to Mr. Latham�s chagrin. But to really understand the deception potentially inherent in political opinion surveys, we must look to those conducted in the United States, where election polling is considered to be at its highest stage of evolution. Case in point: the current race for president.
With less than a week to go before the US presidential election, most opinion polls show that the contest is neck and neck, with Republican George W. Bush slightly ahead of Democrat John Kerry by 2-4 points either side of the 50 percent mark. These figures may be misleading. Leave aside for the moment all the voting irregularities being discovered, the attempts (mostly Republican) at voter suppression now being given wide exposure, and the massive (mostly Democratic) voter turn-out drives currently underway (including unprecedented efforts to register Americans living abroad), all of which lies outside the purview of polling surveys. Ignore the fact that people lie and misrepresent their views in sample groups for a variety of reasons. Forget that polls commissioned by partisan groups tend to reflect the wishes of those paying for them.
(More ... Scoop: The Nonscientific Reality Behind US Election Polls)
¶ 21:43
25 October 2004
The worst mistake a political analyst can make is to offer predictions about future events. The study of politics is nowhere close to being an exact science, and in fact makes meteorology look downright precise. This has not stopped those who study politics for a living from calling themselves "scientists," and from attempting to emulate the real sciences when describing political phenomena in terms of absolutes, trends, and indexes of predictability. At best, this is informed opinion masquerading as fact, and at worst it is nothing more than playing at being economists. Unlike the laws of nature and physics that govern scientific inquiry, or the stylized models upon which most macroeconomic theory is premised, the study of real human political behaviour is as contingent and precarious as the subject is fickle and unpredictable. Thus political analysis is no more than an exercise in focused speculation.
Nowhere is this more evident than in the work of those who conduct political opinion polls. Consider, for example, the polls that showed Mark Latham and John Howard in a near dead heat in the run-up to the recently concluded Australian national election. Needless to say, the pollsters proved wrong, much to Mr. Latham�s chagrin. But to really understand the deception potentially inherent in political opinion surveys, we must look to those conducted in the United States, where election polling is considered to be at its highest stage of evolution. Case in point: the current race for president.
With less than a week to go before the US presidential election, most opinion polls show that the contest is neck and neck, with Republican George W. Bush slightly ahead of Democrat John Kerry by 2-4 points either side of the 50 percent mark. These figures may be misleading. Leave aside for the moment all the voting irregularities being discovered, the attempts (mostly Republican) at voter suppression now being given wide exposure, and the massive (mostly Democratic) voter turn-out drives currently underway (including unprecedented efforts to register Americans living abroad), all of which lies outside the purview of polling surveys. Ignore the fact that people lie and misrepresent their views in sample groups for a variety of reasons. Forget that polls commissioned by partisan groups tend to reflect the wishes of those paying for them.
(More ... Scoop: The Nonscientific Reality Behind US Election Polls)
¶ 21:43
