Pollster.com

Articles and Analysis

 

Whatcha Got "Outliers"

Topics: 2008

Marist pollster Lee Miringoff shares memories of Tim Russert.

And Peter Hart remembers Russert's role in creating the NBC/Wall Street Journal poll.

Mark Mellman sees the underlying fundamentals as "tremendous assets" for Barack Obama.

Kathy Frankovic asks whether voters can look beyond partisanship in judgements about the economy.

Gary Langer and Jon Cohen review the ABC/Post results on the Supreme Court's Guantnamo Bay decision.

Jennifer Agiesta has all the numbers for independents from the ABC/Post poll.

Brian Schaffner ponders what Jim Webb would add to the Democratic ticket.

Chris Bowers compares Obama's trajectory to Kerry's.

John Sides looks back to analysis from the 1980s for a reality check on "Reagan Democrats."

David Hill says campaigners, and pollsters, should do less instant analysis and more thinking.

Frank Newport reprises his public opinion quiz.

Patrick Murray blogs on New Jersey's summer polls.

And if you haven't seen the GraphJam charts yet, take a mental health break and click through (via Sullivan).

 

Comments
Kile Thomson:

any polls coming out tomorrow ?

____________________

hardheadedliberal:

Mark, how carefully did you look at Brian Schaffner's post before you linked to it? I thought his approach to evaluating Webb's performance with various demographic groups bordered on the ridiculous. Following is a comment I left on his site:

Sorry, Dr. Shaffner, ... I think you would fail a graduate student who handed in an analysis based on the reasoning you explained for selecting the universe you used to produce your analysis, which you described this way:

"I included all Democratic senate candidates, no matter how competitive the race. This could produce a problem, but I think it worked out ok since there were races where the Democrat lost by a large margin and also those where the Democrat won by a large margin, which would hopefully cancel each other out. In fact, the exit polls showed Webb winning 54% of the vote, the exact same percentage of the vote that the exit polls showed all other Democratic candidates winning. Thus, the comparison to all other Democratic senate candidates seems to be a fairly reasonable one."

The method you describe looks to me to be wide open to problems akin to the "ecological fallacy." You may have reached a correct conclusion about Webb's drawing power (or lack thereof) with women and working class whites, but the analysis that you presented has zero persuasive value, IMHO.

In the universe you chose, Webb did slightly better than other Democratic Senatorial candidates with seniors, rural voters, suburban voters and voters who regularly practice their religion. He also did slightly worse with women, white voters making less than $50,000 per year and much worse than other Democratic Senatorial candidates with white voters who did not have a college education.

You opined that Webb would be a problem for Obama with women, based on his underperforming in that demographic, and that Obama was too weak in that area to risk Webb on the ticket. But since you posted this analysis, multiple polls are showing that Obama's "unity bounce" upon Clinton's concession speech has already mitigated that problem. If Webb does draw rural voters and seniors more strongly than Democrats in state-wide races in states with comparable demographics, his appeal to those segments of the voting population could vastly improve Obama's performance in Appalachia, including the parts of Appalachia that can be found in WV, VA, PA, OH, and IN.

I would think that for your analysis to be taken seriously you at least need to control for region. Harold Ford's contest in TN was about as close as Webb's vote in VA, although the race of the Democratic candidate might make those two state elections a poor comparison. Clair McKaskill's vote in MO would potentially be a good comparison with Webb's performance. MD has a very different index of partisan identification, but Ben Cardin's race there may be another decent comparison if the vote is adjusted for the partisan index.

Comparison with exit polling on Tim Kaine's support in winning the governorship of VA in 2006 would probably be more useful than comparing Webb's vote to the aggregate of other Senatorial candidates in 32 widely diverse states.

Instead, you included in your universe incumbents like Clinton (NY), Biden (DE), Ted Kennedy, Bingham (NM), Byrd (WV), the two Nelsons (FL & NE) and incumbents in other states, including WI, HI, ND, CA.

Sen. Whitehouse in RI, John Tester in MT, Klobusher in MN, Bernie Sanders in VT, and the sacrificial lamb against Snowe in ME should not be included in the comparison. Those states are almost lily white.

The votes for Sherrod Brown in OH, Bob Casey, Jr., in PA, Bob Menendez in NJ, the sacrificial lambs against Lugar in IN, Kay Bailey Hutchinson in TX, & Trent Lott in MS would also be of questionable relevance. The Democratic votes in the unsuccessful run against Ensign (NV) and also the contests in AZ, UT & WY are from a completely different demographic than the VA demographic.

Any of the comparisons also need to be qualified by the fact that Webb was the only serious Democratic candidate who had never even run for public office before.

I would be very interested in seeing a comparison of the Democratic Senatorial candidates' demographic support by region. Your use of aggregate figures for Democratic Senatorial candidates in 32 states with widely diverse demographic characteristics, states that were about evenly divided between incumbents and first-time candidates for Senate seats is almost bizarre. The analysis you have presented here is the sort of aggregate analysis that a rookie political reporter who took no poli sci or statistics courses might come up with.

____________________



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