The weather in November

The Hartford Courant is at it again. Front page banner headline above the fold:

McCain-Clinton a Tossup.

Two things wrong with this article.

After editorially excoriating media who used polls as news in New Hampshire, and getting it fiercely wrong, The Hartford Courant once again presents polls as news, polls as fact. This headline will sit on coffee tables, and in honor boxes across the state all day long, creating an impression that is wrong and speculative. They may as well be trying to predict the weather on election day in November.

Another thing, the headline reveals a secondary facet of the story. The real story is that if Barack Obama is the candidate, which evermore so, it appears he will be, then, according to the polls, he is predicted to beat McCain by a whopping 17 points.

Shame on the Courant for once again misusing polls as news. If I never hear "Quinnipiac" for the next seven months, it will be too soon.

NOTE: Had the gall bladder removed today, so liklihood of additional posts is slim.

Smells kinda like teen spirit

The bad news: Harp Magazine has declared bankruptcy. With Harp and No Depression gone, the niche of music publications for adults narrows considerably.

The good news: Dave Grohl is running for president. Be on guard for the Nader-like backlash. Foo on yet another third-party candidate.

The problems with reporting polls as news

Is that lately, they have frequently been wrong.

Karen Hunter, reader representative at the Hartford Courant, examine that fact today in her column.

Interestingly, she concentrated on a UCONN poll managed by Monika McDermott, which turned out to be spectacularly wrong in it's projections. Apparently, defensively, McDermott explained how the poll could have gotten it so wrong, but she blamed the media for misrepresenting the facts of the poll by concentrating on two question, which, she says "sell newspapers"

I'm afraid no one today knows what sells newspapers, or more of them would be sold. McDermott is partially right. Newspapers and other media like to concentrate on the simple facts of a horse race, and abandon the complexities of a campaign. The media likes to sell poll results as facts, but unfortunately, polling companies market the most sensational elements of the polls they conduct.

In fact, McDermott herself was on the Colin McEnroe show making projections, based on her poll, that were more in the realm of opinion then they were anchored in any kind of fact.

The two big problems are that pollsters are trying to make money by predicting how an entire population feels, or will act, based on a small sample (and it's now clear that they can be wrong well beyond their predicted percentages). And the pollsters themselves get a little starry-eyed when being interviewed as if they were some sort of seers. For their part, the news organizations have treated poll results as if they were actual facts, instead of what they really are, guesses.

In a perfect world, pollsters would retire to the back rooms of phone banks, computers and statistics, and news organizations would print poll results on page 29, for what they are, pretty good hunches about what might happen.

Short memory at the Courant

On January 10th, the Courant published an editorial which criticized the role and effect of polls in the electoral process, saying:

Poll-watching and premature analysis tend to reduce this quadrennial exercise in self-determination to the level of a sports contest.

Today, at the top of the front page, in bold headlines, the Courant declares the favorites in the upcoming presidential primary, declaring:

Connecticut Favorites - Choices are Clear, For Now.

I guess that "for now" absolves them from using polls as predictive news, but they have shamefully ignored the advice their editors offered a week ago. Keep a record of the poll numbers and we'll see just how wrong the Courant is a few weeks from now, and whether they'll offer an apology for misleading readers. Of course, if they end up being correct, they can be accused of affecting the outcome.

So, just as editors advised, it's safer not to emphasize polls, which have proven so unreliable in the past.