Pollster.com

Articles and Analysis

 

Morning Status Update for 9/30

Topics: Status Update

We logged ten new statewide polls yesterday, including seven -- five from Fox/Rasmussen and one each from SurveyUSA and PPP -- that tracked results from last week or mid-September. All seven showed improvements of between two and five percentage points on the margin for Barack Obama. Although our state classifications remain unchanged since yesterday, our estimates have shifted slightly in Obama's direction in 4 of the 6 battleground states for which we have new surveys (Florida, Pennsylvania, North Carolina and Virginia).

2008-09-30 trends.jpg

We did see a shift in the North Carolina Senate race, where a new PPP survey shows Democratic challenger Kay Hagan leading by eight points (46% to 38%) over Republican Senator Elizabeth Dole. That result confirms an existing upward trend for Hagan, and increases her lead to 4 points on our overall estimate (45.1% to 41.1%), enough to shift the race from toss-up to lean Hagan.

Back to the presidential race. I have noted previously in these updates that our trend estimator tends to be "small-c" conservative by design: It smooths out the line, which may sometimes understate a trend, especially if we see a significant shift nationally but only one or two new surveys in each state. We saw that lag just after the Republican convention, when it took about a week for the McCain-Palin bump to show up clearly on our map, and we have seen something similar in the Obama-Biden rebound over the last two weeks.

The "more sensitive" setting on the "smoothing" tool in our new Flash charts provides a way to check for potential short term trends. Consider Pennsylvania. Our standard trend estimate shows Barack Obama with a 3.3% lead (48.0% to 44.7% at the time of this posting**):

However, the more sensitive estimate, which gives greater weight to the more recent surveys in Pennsylvania that show Barack Obama running "above trend," gives the Democrat a lead of just over seven points (49.9% to 42.8%). That margin would be more than enough to shift Pennsylvania back to the "lean Obama" category:

The "more sensitive" trend estimate setting comes with an important warning: It is more apt to be "fooled" by outliers or other random noise in the data, so some of the short terms shifts it plots may turn out to be illusory.

To try to check for that possibility, I ran the more sensitive estimates -- again, as of this writing -- for the 12 of the 14 battleground states I have been monitoring for these updates (we lack sufficient polls to calculate a more sensitive trend in West Virginia and Indiana). As the table below shows, Obama does better on the more sensitive trend estimate in 9 of 12 states.


2008-09-30 sensative trends


The bottom line: The national trackers are showing additional gains for Obama over the last week. These gains are probably not yet fully registered on the standard state trend estimates that drive our map. If these gains hold -- a big "if" -- we will likely see continuing movement to Obama on the national map over the next week.

PS: Charles Franklin is traveling today and has not had a chance to run and send the "national forces" charts. I will update when he does.

**The numbers in the text match the chart as of 7:00 a.m. Eastern Time. Any polls added after that will change the estimates.

 

Comments
m913:

Thanks for the explanation. That standard/sensitive variation had me flummoxed. Especially when the less sensitive setting also showed a significant variation. Now I think I understand.

____________________

RalphW:

Thanks. I keep wondering as I look over the map each day, why Washington state with a 5.9 margin for Obama is still a lean. Is this a low data state?

____________________

markc:

What happened to posting all the polls on the main page?

____________________

StoryPronto:

I've been wondering about WA also. In low data states such as WV & IN & perhaps WA, do your contributing polling sources inform you in advance of polls they are conducting - where and when the results will be available?

successstory@lnf.com

____________________

bythesea:

Mark, have you written anything about the potential effect of greater young person and African-American participation in voting? Their improved performance is the key to Democratic success here in North Carolina. Are pollsters considering this likelihood and including it in their calculations, or are they assuming traditional performance levels? If pollsters aren't including their effect then the election is over its President Obama. If they are including it then a slight slip from expectations and its President McCain.

____________________

nkoren:

I'll echo Markc's question. We're only seeing two or three polls per day on the main page, when there are plainly more polls happening than that. What's going on?

____________________

Stuart:

Why aren't you updating your polls?

Thanks

____________________

thoughtful:

@Mark

Its very subtle, when there is still quite a lot of churn, i think how you are doing it is very insightful.

Thank you, is this going to be your new method of presenting latest polls on the site?

____________________

m913:

@Bythesea

Not to steal from Pollster.com (love ya'), but there is a nice essay by Nate Silver at fivethrityeight.com criticizing the GWU poll for failing to account for the age shift.

____________________

enivel:

one piece of the first (conservative) chart that was not changed on the second (sensitive) one is the projected EV counts based on even slight leanings.

using the conservative estimates, obama had a 269-252 lead with new hampshire and virginia as legitimate toss-ups -- no lead for either candidate.

those numbers were not adjusted to show the sensitive estimates. if they had been, both new hampshire and virginia, as well as north carolina, would have fallen on obama's side, pushing the count to 301-237.

____________________

PHGrl:

Muhlenberg is releasing a PA poll daily.
Today is 49/41 yesterday was 49/42

http://www.muhlenberg.edu/studorgs/polling/documents/Release_sept30.pdf

is the most current one going to be included?

And tomorrow's release i believe has NO overlap with 9/26's release.. both of those should go in.

____________________



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