All Hallows Eve


View 748 Wednesday, October 31, 2012


I’ve been a bit under the weather the last couple of days. Much of my family is on the east coast, and all came through Sandy unharmed. I have friends in Manhattan who still have no electric power, and I understand that the looting has begun – I saw pictures of people carrying huge flat screen TV monitors down the street, and other such matters. Our best wishes to all.


A long time correspondent sends this:

Screwy Polls


A lot of people seem to be stressing over the difference between the national polls that show a modest Romney lead, and the state-by-state polls that still point toward an Obama win in the Electoral College.

I’ve been following polls closely for months, particularly the electoral-vote map with state-by-state poll averages at RealClearPolitics, and one thing I’ve noticed is that some polling outfits have been playing pro-Dem games with their state polls – in at least one blatantly obvious case, temporarily dropping their ~5 point Dem skew earlier this month, then after a week or so readopting it. (In aid of some sort of "comeback kid" narrative? Because word came down "damn salvaging your credibility"? Who knows.)

It’s gotten bad enough that the people at RealClearPolitics have written a story about it,

They’re very diplomatic about it; they don’t ever actually state that some of the state polls in their averages are rigged (or to be charitable, merely a mix of incompetence and wishful thinking). But they do calculate that the nationwide average of the state-by-state polls is biased 2 to 2.5 points toward the Dems compared to the national polls. They go on to state that the correct answer is very unlikely to be an average of the two, and leave it at that.

I’ll go out on a limb and say that it’s the national polls that are correct. I’ve come to apply about two points of Kentucky windage to the RCP averages in key swing states (or in some cases I just ignore the more blatant pro-Dem outliers) and the RCP piece reinforces my belief I have it about right.

Nobody should draw too much comfort from this. Even applying RCP’s Kentucky windage to the current swing state poll averages yields only the narrowest of Romney wins – and several key states, notably Ohio, Nevada, and Wisconsin, will be well within the margin of cheating that their regional Dem machines have demonstrated in the past. (Iowa is also razor-close, but less historically prone to ballot-box stuffing.)

My current take: Romney wins if he takes any two out of those four by greater than the margin of cheating. This assumes he also holds onto current narrow leads in Colorado, Virgina, and Florida. And in at least one case it assumes he also knocks off one electoral vote from Maine (as seems likely).

Bottom line: Absent a couple more points Romney pickup nationally (possible but nothing to count on) it’ll be a damned close-run thing.

sign me


Polls are subject to manipulation, and they are also vulnerable to mistaken assumptions. Few people are competent in both the mathematical complexities of statistical inference, and the subtleties of political thinking. In particular, there are powerful social forces at work that make it very difficult to factor in dishonesty in answering pollsters. When I was Managing Director of the Pepperdine research Institute we won contracts to study polling and results from both the FDA and the Department of Justice, and it was very clear that training pollsters is both difficult and expensive, and few polling companies can afford to do it right.

In the present situation we have a President who was elected not only to the Presidency, but to something greater than that. He was the bringer of Hope and Change, The One you’ve been waiting for. Not surprisingly he did not meet all those expectations. Those who believed he would can either denounce him, or retreat into silence, or react in other ways. There are numerous studies of what happens to Believers when their expectations are not met. None of them look to be too useful for pollsters.

My own view, not based on much evidence, is that there are a number of people disappointed in Mr. Obama who do not care to admit that, and that many of them will react to this by abandoning electoral politics – and by not voting. That’s not much more than a guess, but I would not be astonished if a week from today we find that the election wasn’t even close.


The technology march continues, and there’s a lot to write about. We have an embarrassment of riches in cell phones, iPads , Tablets, communications devices other than phones, desktops, laptops, and all of them at consumer prices. I’m trying to keep up with all of it. I note that I am not alone in being not quite overwhelmed with the latest development – not quite overwhelmed, but faced with a very great deal to write about.

But first it’s getting late and it will be dark in an hour or so. It should be a good Halloween. I rather enjoy seeing the spectacle.








Bookmark the permalink.

Comments are closed.