@fbaezer,
Hey, Fbaezer, thats great! Very cool to hear that they liked the site. And its very nice to get some feedback - and readers! Blogging sure is a bit more lonely a thing than posting on a forum...
Quote:2. The oscillations which obervacionalism.com calls into attention seem to me rather "statistical noise" than major chances, although I don't doubt that after Obama's fast advance the previous weeks, there will be an adjustment on the closing days, leading to a result that could be closer than the 6 points polls are marking today.
Well, thats a reassuring take, about the statistical noise I mean - I admit I'm getting a bit of the jitters! But I do also think that the race will still tighten a bit, interesting to see that he thinks so too.
Quote:3. It's surprising than, even with the divergence between punctual measurements, realclearpolitics.com and even fivethirtyeight.com, besides CNN, show so much stability in the average polling they report, notwithstanding that their "poll of polls" include every day different polling houses.
I suppose the fact that they include so many different polls every day automatically makes the trendlines more stable? (I mean, the average of 15 will be more stable than the average of 5?) Especially because there's seven daily and one near-daily tracking poll now, so those are there every day and usually make up the majority of a day's national polling. Which must make the average more stable than if you'd just have polls from eight pollsters one day, and then polls from eight wholly different pollsters with different house effects the next day.
With fivethirtyeight.com, I think (and I'm sure with pollster.com), they also dont just show a simple running average like realclearpolitics.com does. Pollster.com has this regressive trendline (I posted more info on that some time earlier, anyway they have it on their site), and fivethirtyeight.com, which I dont follow as closely, I think weights in a number of background /context data, at least that's what they seem to do with state polling.
Quote:And surely there have been "house effects" in the US series!
Between pollsters? Yes, definitely. Pollster.com had a nice graph illustrating the house effects of the different daily tracking polls here:
http://www.pollster.com/blogs/tracking_poll_house_effects.php
Quote:5. We must always remember that the election in the US is indirect. The potentially narrow Obama lead in citizen votes becomes an abyss McCain cannot breach: not only the more realist calculations have shown that Obama has tightened enough electoral votes to win, but also that the difference between him and McCain could be over 150 votes.
So, I consider a surprise highly improbable, and more probable a clear Republican defeat, even if the rational bet would be Obama not winning with a two digit difference, but much less.
Wow, thats certainly a reassuring take.
I can see the national vote versus Electoral College argument. The national vote polls are showing a tightening of the race, but there's really no let-down so far in Obama's vote in crucial swing states like Virginia and Colorado and Ohio, let alone New Mexico and Iowa. The only states that seems to have tightened up a bit are Southern states: Missouri, North Carolina and for a bit there, it seemed, Florida, and Obama doesnt need any of those. So I guess that even if the national vote tightens Obama maybe has more of a firewall in the states that count.
Still, over 150 votes! I'm not quite that optimistic, I think it'll be more like 100...