« A Six Week Comparison of TPM Tracker Poll and Major Campaign Events: More Than Noise | tonnyb's Blog | Clinton as Secretary of State »
The Effects of the Infomercial and Other Campaign Events on the TPM Tracker: More Than Noise
I decided to plot the national poll numbers against the major campaign events on a graph. The graph contains the TPM Tracker over the past 6 weeks. Please note the black line running through the data points are trend lines. Using those lines, Obama's support increased from 48.4% to about 51.5%. During the same time, McCain's support dropped from 43.8% to about 43.2%.

A larger version of the image can be seen here. The first major events (reflected in the first blue panel) in this time frame that moved the numbers are the trio of 1) McCain's 'campaign suspension, 2) the first debate and 3) the slow release of the Couric-Palin interview. Obama's numbers jump from 47.5% to 49.9% a 1.4% increase. At the same time, McCain dropped from 45.5% to 43.5%.
The second major event is the VP debate. Obama's numbers rose from 49.5% to 50.5%. McCain suffered a fall of about 1% point.
The third event is the second debate. Obama's numbers improved about 0.5% point. McCain's remained unchanged.
There doesn't seem to be any effect for Obama from the third debate. McCain seems to have gained briefly after the debate.
The event that has created greatest single effect, so far, came after the Powell endorsement. Obama gained about 1.5% points. The effect lasted about 6 days.
Obama's infomercial has increased Obama's numbers by 1.5%. Obama began a 1.5% point drop 4 days prior to the infomercial, as the glow of the Powell endorsement faded. Simultaneously, McCain began a steep upward climb that was abruptly stopped by the infomercial.
The acute effects of all these events lasted approximately 5-6 days and in many cases begin a day before the event (probably a result of the pre-event hype). The infomercial was well timed as it reversed Obama negative trend and, if past is prologue, will allow the effect to peak on election day.
Let me know what you think.

A larger version of the image can be seen here. The first major events (reflected in the first blue panel) in this time frame that moved the numbers are the trio of 1) McCain's 'campaign suspension, 2) the first debate and 3) the slow release of the Couric-Palin interview. Obama's numbers jump from 47.5% to 49.9% a 1.4% increase. At the same time, McCain dropped from 45.5% to 43.5%.
The second major event is the VP debate. Obama's numbers rose from 49.5% to 50.5%. McCain suffered a fall of about 1% point.
The third event is the second debate. Obama's numbers improved about 0.5% point. McCain's remained unchanged.
There doesn't seem to be any effect for Obama from the third debate. McCain seems to have gained briefly after the debate.
The event that has created greatest single effect, so far, came after the Powell endorsement. Obama gained about 1.5% points. The effect lasted about 6 days.
Obama's infomercial has increased Obama's numbers by 1.5%. Obama began a 1.5% point drop 4 days prior to the infomercial, as the glow of the Powell endorsement faded. Simultaneously, McCain began a steep upward climb that was abruptly stopped by the infomercial.
The acute effects of all these events lasted approximately 5-6 days and in many cases begin a day before the event (probably a result of the pre-event hype). The infomercial was well timed as it reversed Obama negative trend and, if past is prologue, will allow the effect to peak on election day.
Let me know what you think.
Advertisement





Maybe it's also the fact so many magazines have obama on the cover compared with mccain. Maybe some media bias?
November 2, 2008 8:48 PM | Reply | Permalink
Did you run an r value and test for significance? An eyeball guess is that the McCain trend does not differ from noise.
And while the overall upward trend for Obama appears to be real, the ups and downs you've tried to tie to specific events may not pass statistical muster (correlation does not equal cause).
I think you'd need to pull out some serious multi-regression tools to make any connections that would demostrate anything more than matrixing.
November 2, 2008 9:18 PM | Reply | Permalink
I was expecting more analysis from pollsters on the effects of the infomercial. Thanks for your work. I won't get all technical about it. There has got to be a bounce from that since there were SO many viewers!!!
November 2, 2008 11:07 PM | Reply | Permalink