Media Rant - Presidential Polling

[darvon]

I know most don't like to discuss politics in here, I do, but most don't. This isn't exactly politics, but rather how the media talk about politics.

And the reason I am bringing this up in a Football forum is that NFL people are uniquely experienced to understand the problem I have.

When Big Media presents Presidential Election Polling numbers they present it as : 52% Candidate A and 48% Candidate B.

However, that isn't the question that was asked. The people/readers/viewers want to know "Who is going to win?" or at least "What is your current probability eastimate of winning for each of the candidates?" Just like betting on football games.

And the Media spits out numbers like an adding machine on speed. Yet almost no one gives you the probabilities. They present the polling results (which attempt to predict the popular vote, i.e. the score) without ever turning the score into a probability.

And they also present the score as a percentage, just like probability percentages.


So when the public want to know the answer to the question "What is the probability of Obama winning?", the Media spit out numbers like "Obama 53%" and graphs showing "Obama 53%/ Romney 47%". They don't answer the question about probability, instead they guess the final score.

So the polls actually give you the Media's latest guess on The Score. In a 53%/47% poll, they are essentially saying Candidate A has to give 6 pts. But the presidency is a Winner Take All contest. The fans of Romney only are happy when he wins, not just covers. So probability of winning is THE ONLY number worth finding out, yet the Media never tells it.

You might conclude that the probabilities are unknowable from polling data. Actually that is very wrong. There are zillions of academic studies about turning polls into probabilities. There might be disagreement among the PoliticalSabermetricians, but there isn't a darth of methods. Yet almost ALL Media never give a probability. And the media spends 10 Megawords describing the "horse race" as if it was a sporting event. And they give the spread but they don't want to give the Money Line.

The reason is very clear. There is one exception to this rule. Nate Silver and his 538 blog in the NYT.

Nate's current prediction is:

Popular Obama 50.3% Romney 48.8%
Probability Obama 72.9% Romney 27.1%

And this is with the SAME DATA. Just Moneyline vs Spread.

My beef thus boils down to this.

1) The question interested by the public is "Who's gonna win?" It is reasonable and typical to give probabilities as percentages of 100%.

2) The Media covers the Presidential Elections with numbers and charts for over a year. Almost all Media, almost never (Nate Silver is the only main exception) answer the main question.

3) They answer a different question (i.e. "What is the score going to be?") with numeric data presented in a Percentage Format, just like the real answer to the real question.

4) The Score shows a lot more variability and closeness than the Probability does. The Probability for Obama has been 70% +/- 10% for all of 2012. Yet the Score has been within 6 pts several times during the year. The single time both numbers have been given, the Score had a spread of 1%, while the Probability had a difference of 46%. Thus reporting on the score makes the contest seem MUCH CLOSER and MUCH MORE DYNAMIC than reporting on the probability. Especially if you present the SCORE data in a manner which can be easily misconstrued as the Probability, and NEVER report the Probability. And DOUBLE Especially if most of the public isn't used to dealing with spreads and odds, like we are in the NFL base.



So my rant is that the Media suppress the Probability and report the Score in a percentage format, to justify the 10 Megawords of coverage and numbers and charts to maximize ratings.

And they have been allowed to get away with it so long because most of the Political Press is innumerate. And the pollsters, campaign Sabermaticians and various campaign mercinaries have a vested commercial interest to project a tighter race, thus needed more polls and mercs. And most of the public, unlike us NFL people, don't deal in spreads and moneyline and probabilities.

[wxwax]

I don't think anyone's suppressing anything.

News media are not in the business of making predictions. They report what is happening (television) or what has happened (print.) Trying to predict the future is not their bailiwick.

They leave that to others, and will report on what others are predicting as they see fit. And that, only if they're comfortable with the source and the methodology. Since Mr. Silver's material is doubtless covered by copyright and is also branded by a competing news outlet, I doubt it would be freely lifted and quoted.

If there were a National Weather Service of elections that issued predictions for the outcome, I imagine it would be widely reported.

[packa7x]

They're too busy trying to find polls that fit their slant to worry about to give us solid odds. Just watch the news channel, you'll see the answer hidden in "unbiased" election talk. CBS/NBC/CNN will tell you Obama will win, FOX/FOX News will tell you Romney will win.

[darvon]

News media are not in the business of making predictions. They report what is happening (television) or what has happened (print.) Trying to predict the future is not their bailiwick.

Au contraire, mon ami.

The whole political media is about "the horse race". The entire purpose of polls is to make a forecast of future results in a suspenseful manner. However they never actually give odds, because people would see how much of a mismatch the horse race is and stop watching. They just give poll data dressed up to play odds on TV. And they didn't even stay in a Holiday Inn Expreess last night.

They're too busy trying to find polls that fit their slant to worry about to give us solid odds.

I would love if they gave us ANY odds, not just solid. Then we could actually judge the Nets. There would be some accountability of analysis. Right now, with ONLY poll data, the Media escapes any accountability because they will fall back to the position that poll data is fact not prediction, even though they present it as a probability, which is by definition a prediction.


Bottom line, my ire comes from the fact that if we had an NFL Betting Tout Sheet that wouldn't give odds, used an advanced stat to say it was a close game, changing the favorite a few times, but omitted that their stats actually showed the favorite at 70%/30% for all of the entire year prior.

The NFL betting public would track those guys down and hang them in Effigy (which is a small suburb of Newark).

But the Media gets away with it in spades.

[ScottDCP]

I am ok with their not providing odds. If their prediction causes people to refrain from voting they have affected the outcome abd I might start making torches and sharpening stakes.

[Pruitt]

They're too busy trying to find polls that fit their slant to worry about to give us solid odds. Just watch the news channel, you'll see the answer hidden in "unbiased" election talk. CBS/NBC/CNN will tell you Obama will win, FOX/FOX News will tell you Romney will win.

And what angers me to no end is that in so many cases the media outlet is sponsoring the polls that they are then citing as news.

And consumers and other media outlets than report the poll numbers as if they are news.

Anyone who has been polled (no double entendre intended) knows - or should know - that the way a question is worded will skew the results.

This whole kerfuffle over a statistician presenting a probability of victory instead of merely regurgitating the standard way of presenting polling data is yet another indiction that there are far too many pundits and "experts" in the world.

Read Charles Pierce's thoughts on this nonsense: http://www.esquire.com/blogs/politic...itics-14261107

Here's my favorite part of the column -

Here, for example, is Joe Scarborough, engaged in a remarkable feat of Being Stupid:
"Nate Silver says this is a 73.6 percent chance that the president is going to win? Nobody in that campaign thinks they have a 73 percent chance — they think they have a 50.1 percent chance of winning. And you talk to the Romney people, it's the same thing," Scarborough said. "Both sides understand that it is close, and it could go either way. And anybody that thinks that this race is anything but a tossup right now is such an ideologue, they should be kept away from typewriters, computers, laptops and microphones for the next 10 days, because they're jokes."
Does Squint understand statistics less than he understands the Electoral College? It's a close run thing. And that's not even getting into the fact that we are listening to complaints about Nate Silver from the members of a professional pundit class that, judging by its track record, would have bet against gravity 30 seconds after Galileo dropped the hammer.


Read more: http://www.esquire.com/blogs/politic...#ixzz2Ass3Xuer

[Trumpetbdw]

I am ok with their not providing odds. If their prediction causes people to refrain from voting they have affected the outcome abd I might start making torches and sharpening stakes.

This is the big difference between the elections and the NFL. In the NFL, we're simply making a reasonable guess based on something outside of our control, which is the outcome of a game played by 2 teams that have no direct connection to us in any way whatsoever.

With the elections, all of us are conceivably in control. If the news outlets start making predictions prior to election day, then it would absolutely impact the polling results, as many people would likely not bother to make their voice heard. It's one of the biggest problems with election day, when "projections" are made, often times rendering the west coast voting moot because we already know the outcomes.

The media's potential to directly impact the election by revealing predictions and odds would be akin to a referee's ability to directly impact a sporting event. Not that it would be criminal for the media to involve themselves in this practice, but it would certainly be unethical, and would lend even more credence to those who are already convinced that the media displays pre-conceived bias and manipulates the news when "their guy" loses.

[darvon]

If the news outlets start making predictions prior to election day, then it would absolutely impact the polling results, as many people would likely not bother to make their voice heard. It's one of the biggest problems with election day, when "projections" are made, often times rendering the west coast voting moot because we already know the outcomes.

I have 2 huge issues with this:

1) The Media already ARE making predictions. They presect the polling data in percentage format to mislead the public into thinking that it IS probability data, which is axiomatically a prediction.

2) The Media is using a piece of data which, when presented as probability, implies a much closer and more volatile race (typically) in order to create higher interest in their stories for their commercial success.


The article is interesting, but really talks about whether Nate Silver is right or wrong, not whether the Media portrays polling data as probability.

[Pruitt]

And the media - from networks to local stations - love a close race. Otherwise, would $1 billion be spent on advertising?

Ad reps are going to be getting some nice bonuses in the next quarter.

[FrzzerBwler]

darvon me thinks you're way off base starting w/ they are essentially saying Candidate A has to give 6 pts.
To me, a non gambler, the point spread is a load of crap. After all what is the purpose of the spread? To have a equal amount bet on both teams. when a team actually makes the playoffs based on it's record against the spread I'll actually care what the point spread on a game is.
To try and apply the same logic to a campaign predication is just wizzing into the wind.
I read many polls but always try to discern 2 main things from them.
1) Who did the poll. As others have mentioned, that in itself offers insight to how biased the numbers are.
2) What the numbers are for likely voters. IMO it does not matter what the majority thinks, since the majority of eligible voters never actually vote.