• comment on a post Where We Go from Here over 7 years ago

    With all due respect to Jerome, to me Stoller and Bowers were mydd, so the loss is huge.

    What I like best about mydd is detailed, data-based voting analysis.  That's where I think the strength is and where it should continue.  That's what differentiates mydd from dkos, FDL, or TPM.  Trying to make mydd more like any of those sites seems like a quick way to obscurity.

  • on a comment on Senator Cheney? over 7 years ago

    "If she wants it, it's hers."

    The governor will choose one out of the 3 names submitted, and you can be sure he will choose the weakest one.

  • on a comment on New Iowa, South Carolina Polls over 7 years ago

    When the chi-square is inconsistent that all the polls are measuring the same thing, then probably the polls are not all measuring the same thing.

    And when they are not all measuring the same thing, then averaging is no longer a statistical procedure, but an ad hoc one.  It may very well be a historically accurate ad hoc procedure.

    Incidentally, for those who care, RCP does a straight equal-weighting averaging, not an uncertainty-weighted average.

  • on a comment on New Iowa, South Carolina Polls over 7 years ago

    Actually, I went to the realclearpolitics page and did the calculations myself.  Formed the uncertainty-weighted-average of the Iowa poll results (using binomial uncertainties) and looked for the chi-squared of the fit to three averages for clinton, obama, and edwards.  For this last I used the data-measured correlations.  The result is a chi-squared of 45 for 12 dof, which is essentially a statement that: these polls are not statistically consistent.  The ARG is clearly an outlier of the group;  if you remove it, the chi-squared is 26 for 9 dof, which is merely "nearly impossible" rather than flat-out impossible.

    Incidentally, the measured correlations are
    Obama-Edwards: +0.70
    Obama-Clinton: -0.76
    Edwards-Clinton: -0.91

    The uncertainties on the correlations are large (probably 50% or so), so don't take them too seriously.  They are actually quite surprising if taken seriously.

    Caveat averagor: when the chi-square is telling you the polls are not statistically consistent, tread carefully in averaging.

  • comment on a post New Iowa, South Carolina Polls over 7 years ago

    Overall, the fluctuation we see between many early state polls is probably just as connected to a highly volatile early state electorate as it is to different methodologies.

    There's also the simple fact that all the polls come with substantial statistical uncertainties.  Is there any evidence the fluctuations are any larger than one would expect, given the uncertainties of the individual polls?

  • ...I read blogs like mydd and not people like Barone!

    It drives me nuts to no end when someone like Barone mindlessly speculates on a question that can be easily, reliably, and quantitatively answered.

    Thank you for looking at actual data!

  • comment on a post NY 29: Bowling-Gate? over 7 years ago

    Let's imagine that the $2.5M federal loan is about 3 percentage points below what it would have been otherwise commerically--say a loan at 6% instead of 9%.  The interest saved over the course of 5 years (though I imagine it's more likely a 10 year loan) is about $375,000.

    That's a great return on the initial $1000 investment.  

    Where do I sign up?

  • comment on a post Opening the Congressional Research Service over 7 years ago

    Here's a link to the Washington Post's story on last week's memo from the director of CRS restricting circulation of the reports; and a dKos diary, with additional links, on the new limitations.

  • comment on a post List of Blue Dog Saboteurs over 7 years ago

    Sestak was on NPR yesterday afternoon saying the date certain was the most important thing an Democratic Congress could do to make progress on Iraq.

    Unfortunately my IE has suddenly taken leave of its senses and I can't seem to load npr.org to provide a link.

  • on a comment on MyDD Madness over 7 years ago

    I can hardly complain about my alma mater getting in, but let's be honest--Drexel was robbed, and we probably got the last at-large spot.  I'm sticking with the story that it was Purdue's fault that Syracuse was left out...

    And what's with Po' Boys?  Is this some new-fangled place, I grew up in C-U (yes, I was a townie) but left about 15 years ago...I've been told there are now some actual nice restaurants in town--is this one of them?

  • I hope TX-17 and TX-22 are considered places to bleed Republicans of cash rather than places to seriously hold onto.

    Given that incumbents traditionally pick up about 8 points just for being the incumbent, all the others look very holdable.

  • If you use IE you can simply highlight the table, then paste it directly into Excel, then Data->Sort.  It can be re-exported (File->Save As) then as HTML.  I just did this in under 2 minutes--the hardest part was that R+15 comes between R+1 and R+2 by the sort (which was pretty easy to fix.)

  • comment on a post Questions for Edwards, Obama and Reid over 7 years ago

    For Reid: Does the presidential signing statement for HR5631 indicate that the President will ignore any Boland-type amendment, esp. vis-a-vis Iran?

    For Edwards: can you explan what you see as the risks of American military action against Iran?  How do they compare, in probability and magnitude, to the probability and magnitude of risks of an Iranian nuclear program?

  • This whole US attorney massacre, with Lam among them, reminds me of the high school kid who goes to the drugstore to buy toothpaste, deodorant, chocolate, aspirin, socks, cough syrup, band-aids, and Playboy.  

    Which one did he really come for?

    And who does he think he's fooling?

  • comment on a post The Californication of the West over 7 years ago

    Actually, I think a similar effect is why New Hampshire has been changing into the Blue column the past few years.  Home prices have been driving many Bostonians into New Hampshire, where prices are lower, but it is still close enough to commute.


Advertise Blogads