Friday, November 09, 2012

The Media and Mathematical Ignorance



 Going into the Presidential election, many of my pro-Obama friends and co-workers were nervous.  They had been hearing ad infinitum that Romney had caught up to Obama, and that this race would be a cliffhanger.  In dogged defiance, I insisted on this blog and elsewhere that Obama would win by about 100 electoral votes (Yes, I am permitting myself some gloating).

The disconnect between the reality of Obama’s impending decisive victory, and the supposed neck-and-neck nail-biter being reported by the mainstream media (especially in the final week or two of the campaign), can only be attributed to one or two factors:

1) The mainstream media needed to lie about the election to retain audience viewership, and thus command the highest-possible amount of advertising dollars; or

2) The mainstream media are truly incompetent when it comes to understanding and analyzing statistics.

Liars, or Stupid.  Or both. Take your pick.

The maps displayed by CNN, NBC, and other major media outlets all contained a collection of so-called ‘swing states,’ where the races, according to an effervescent John King and a hyperactive Wolf Blitzer, were ‘too close to call’ because the polls were all ‘within the margin of error.” Interspersed among the Reliably Blue and the Reliably Red states were a chain of unknown “yellow” states, that simply contained all the uncertainty of a well-matched, fever-pitch sporting event.

Yes, they were within the statistical margin of error.  No, that did not mean they were toss-ups.  And herein lies the media’s thorough misreporting of facts.

Consider the actual results and the polls from four of the so-called “too-close” swing states in the week leading up to the election:

Nevada
Nov 1  Survey USA,  Obama 50%, Romney 46%
Nov 4 You Gov,  Obama 49%, Romney 45%
Nov 5 Public Policy, Obama 51%, Obama 47%
Actual: Obama 52%, Romney 46%

Wisconsin
Nov 1 NBC-Marist, Obama 49%, Romney 46%
Nov 2 We Ask America, Obama 52%, Romney 45%
Nov 3 Public Policy, Obama 51%, Romney 48%
Nov 4 You Gov, Obama 50%, Romney 46%
Actual: Obama 53%, Romney 46%

New Hampshire
Nov 5 Rasmussen, Obama 50%, Romney 48%
Nov 5 New England College, Obama 50%, Romney 46%
Nov 5 Granite State/UNH, Obama 50%, Romney 46%
Actual: Obama 52%, Romney 47%

Colorado
Nov 4 You Gov, Obama 48%, Romney 47%
Nov 5 Reuters, Obama 48%, Romney 47%
Public Policy, Obama 52%, Romney 46%
Actual: Obama 51%, Romney 47%

In each case, the media insisted these were too close to call.  In each case, the media insisted that all the polls were within the margin of error, and implied – or stated outright – that they could not be reliable indications as to what was happening in those states.

And that’s where they were either ignorant or lying.

Fact 1:  Take a look at those polls again.  In not one case did Obama fall behind Romney in those polls.  If anything, the media should have reported that Obama was consistently ahead of Romney in those states.

Fact 2: Statisticians allow themselves only a 5% possibility of error.  That means that poll results that pollsters consider a ‘safe bet’ is more than 95% likely to accurately reflect voter sentiment. 

A result that lies within a poll’s margin of error does NOT mean that the difference between the candidates is negligible, or that the numbers could even be reversed; it means that the certainty of the poll accurately reflecting the population is something just under 95%....like 90% or 92%.

Imagine if John King had announced the following:

“Polls in the key states of Colorado, New Hampshire, Nevada and Wisconsin show a 90% certainty that Obama will win each of those states.”

That would have been an accurate reporting of what it means to be within the margin of error.

It also would have made for poor drama, and low advertising revenues.  But the fact is that “within the margin of error” means nothing more than that. 

And the reality is that Obama consistently outpolled Romney in those states, and won in those states by margins at least as big as the polls.

The polls were indeed accurate.  But the media was crippled by a combination of stupidity and desire for drama.

No comments: