Pollster.com

Articles and Analysis

 

Sunday's Leftover "Outliers"

Topics: Outliers Feature

(Late because I fell asleep early last night...)

The Wall Street Journal looks at likely voters, party weighting and how they affect poll results.

PPP shows how their party ID results have fluctated since the conventions.

The Washington Post and ABC News share their poll's "final verdict" on the presidential debates.

Gary Langer summarizes the numbers on Colin Powell.

Michael McDonald creates a web page to track early voting turnout statistics.

Sam Wang expounds on why so few news organizations aggregate polls.

Andrew Gelman tests whether America is a "center right" nation.

Jon Martin sees McCain aping polling spin from Drudge.

The San Francisco Chronicle has love for polling geeks.

 

Comments
Mark Lindeman:

Another data point on Powell: in the 2004 National Election Study, Powell's average "feeling thermometer" score was 70. It ranged from 58 for strong Democrats to 81 for strong Republicans. So the favorables are pretty favorable (although of course these numbers might be lower now).

I don't think this endorsement will sway many votes; I do think it will make it even harder for McCain to make inroads.

____________________

kglore in PA:

So that apples can be compared to apples, why can't pollsters agree to weight their samples by the same Party ID % for Dems, Reps, and Independents? This could serve as a constant whether it is real or not. Then we could see if independent polling nation wide showed a trend or not. Otherwise, the numbers can be quite misleading!

____________________

kglore,

As long as pollsters report the horse-race numbers separately among self-identified Democrats, Republicans, and Independents, you'll get the information needed to compare trends irrespective of how many D, R, and I voters are in the sample. As a simple default model, just assume each subgroup comprises 33.3% of the full sample.

Readers are invited to visit my sample-weighting website at:

http://courses.ttu.edu/hdfs3390-reifman/weighting.htm

____________________

Mark Lindeman:

kglore, problem with that is that if the pollsters measure party ID differently (even asking it a different context, perhaps), the comparables may not be comparable.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR