27.11.12

De Radikales ideologi?

Er Det Radikale Venstre fortsat et "social-liberalt" parti? Det har jeg dd. en ikke helt videnskabelig men dog veldokumenteret kommentar om.

19.11.12

De Konservatives meningsmålinger

Det ser ikke godt ud for De Konservative i meningsmålingerne.  (Update: Jeg har en klumme om det i Berlingske: "Konservativ dødsspiral".*)  Her er to næsten identiske figurer baseret på knap 200 meningsmålinger fra juli 2011 til nu.  Den første viser et glidende gennemsnit af de seneste to målinger samt en ikke-lineær tendenslinje.


Den anden er mere detaljeret og viser

7.11.12

Comparison of US 2012 election results, polls and forecasts

There have never been more polls and more "predictions" in a US election than in 2012.  So much so of both that we may have reached a point of saturation.

Nonetheless, here is a quick comparison of the actual election result (in terms of national popular vote shares for the two major party candidates) and the last polls from various pollsters as well as the last "predictions" made by various individuals and websites.  The comparison is based in "gross errors" vis-a-vis the actual election result.*

As is evident the poll aggregates do quite well, with the exception of Huffington Post's Pollster, although all get the "outcome" (winner) right.

Among the pollsters the variation is greater, ranging from the very precise (PPP, IBD/TIPP) to the far off (Reuters, Zogby, Pew et al.).  Most pollsters are within the statistical margin of error.  Rasmussen, which did very well in the last two presidential elections, was mediocre this time, while Pew,

6.11.12

Amerikansk præsidentvalg 2012: Vægtet, dagligt gennemsnit af "tracking polls" af "sandsynlige vælgere"

Dagens§ tal:4.-5.11.: Obama: 49,21%Romney: 48,73% (-0,5%) [N = 7.257. Kilder: ABC News/Washington Post; Gallup; Rasmussen; TIPP]
Tidligere datoer§:
  • 3.11.: Obama: 49,00%Romney: 48,42%% (-0,6%) [N = 3.569.  Kilder: ABC News/Washington Post; Rasmussen]
  • 2.11.: Obama: 48,50%Romney: 48,50%% (+0,0%) [N = 2.985.  Kilder: ABC News/Washington Post; Rasmussen]
  • 1.11.: Obama: 48,00%Romney: 48,51% (+0,5%) [N = 3.035.  Kilder: ABC News/Washington Post; Rasmussen]
  • 31.10.: Obama: 47,93%Romney: 48,54% (+0,6%) [N = 2.793.  Kilder: ABC News/Washington Post; Rasmussen]
  • 30.10.: Obama: 47,92%Romney: 49,00% (+1,1%) [N = 2.788.  Kilder: ABC News/Washington Post; Rasmussen]
  • 29.10.: Obama: 47,46%Romney: 49,00% (+1,5%) [N = 2.771.  Kilder: ABC News/Washington Post; Rasmussen]
  • 28.10.: Obama: 46,97%Romney: 49,99% (+3,0%) [N = 5.459.  Kilder: ABC News/Washington Post; Gallup; Rasmussen]

5.11.12

Too close to call: Growth and the cost of ruling in US presidential elections, with an application to the 2012 election

I have finally written-up a small research note trying to collect together the little model for the explanation of outcomes of US presidential elections that I have played around with for years.  What could be more relevant than do so on the eve of the 2012 election?  At this risk of being proven totally wrong, here is the title and abstract and link to files for download:


Too close to call: Growth and the cost of ruling in US presidential elections, with an application to the 2012 election
The note briefly outlines a new model for the explanation of US presidential elections, founded on (a) recent economic growth and (b) a measure of what may be called “’the cost of ruling”.  The former is based in changes in real disposable income for the period following a mid-term election, while the latter combines factors of incumbency and terms-in-office.  The model is applied to data from the US presidential elections 1932-2008 and has considerable explanatory power for the variation in the incumbent party’s candidate’s share of the two-party vote (R2=0.74).  The model is controlled against a number of other frequent explanations and is found to be quite robust.  When augmented with approval ratings for incumbent presidents, the explanatory power increases to 83 pct. and only incorrectly calls one of the last 15 US presidential elections.  Applied to the 2012 election as a forecasting model the prediction is that President Obama will win 49,6 pct. of