Inflated Clinton National Poll Theory Update
by Chris Bowers, Mon Apr 16, 2007 at 09:33:43 AM EDT
Here is the latest national poll from Rasmussen Reports: (4/9-4/12, 774 LVs, MoE 4, 4/2-4/5 numbers in parenthesis)
Clinton: 32 (34)
Obama: 30 (29)
Edwards: 16 (16)
Now, I hardly ever post individual national polls on the front page, but I do so in this case because it is relevant to my ongoing discussion on whether or not national primary polls inflate Clinton's advantage. While every other national Democratic primary preference poll that I know of, except possibly the ever-questionable Zogby, includes the entire self-identified and leaning Democratic registered voter population in their polls, Rasmussen only includes "likely Democratic primary voters." To look at this from a more brass tacks perspective, while other national polls on the Democratic nomination include roughly 45% of the national registered voter universe, Rasmussen includes just under 39% of the national likely voter universe (774 of 2000 likely voters surveyed across four days). Now, while I would prefer a presidential primary poll that sampled only around 20% of the likely voter universe, as the universe of likely Democratic primary voters is usually only around that size or smaller, for now, in order to examine the validity of my thesis, I will have to take what I can get. While not as small as I would like, 39% of likely voters is substantially smaller than 45% of registered voters.
With a more narrowly targeted universe of Democrats, Rasmussen has consistently put up numbers that are less favorable to Clinton than virtually every other national poll taken this year. Over the last three months, six of the fourteen Rasmussen polls have shown Clinton ahead by five points or less, while only one of the other thirty-three national polls have done the same. Further, nine of fourteen Rasmussen national polls taken this year show Obama within eight points or less, while only six of the other thirty-three polls have done the same (including three from Time that only included Clinton, Edwards and Obama in the question). Clinton holds a median lead of 7% in the fourteen Rasmussen polls, and a median lead of 15% in the other thirty-three polls. Similarly, Clinton's mean advantage of 7.7% in Rasmussen polls is more than doubled in the other thirty-three polls.
Simply put, Rasmussen shows a significantly closer race than any other polling outfit, with the possible exceptions of ARG and Zogby (who both poll "likely" Democratic voters, whatever that means to those firms) and Time, which usually only includes Clinton, Obama and Edwards in their questions. Given my thesis, it isn't hard for me to wonder if the main reason for this is that Rasmussen polls a more representative sample of the Democratic primary electorate than other national polls. As I already stated, they sample roughly 39% of likely voters, while most other polls sample 45% of registered voters. While it is possible that other factors are at play, this certainly seems to be a decent amount of evidence to support my thesis. Clinton's lead is not as large as most national polls make it look, because most polls include too many irregular, low-information voters--among whom Clinton holds a disproportionate advantage--to be an accurate sample of the Democratic primary electorate.
Also, since I know we live in an era where most people's first reaction to an argument they don't like is to accuse the arguer of bias and a hidden agenda, you might enjoy these three posts where my poll analysis went quite decidedly against the hopes and conventional wisdom of the blogosphere:
Also, in other polling news, a new CNN poll (PDF) shows Clinton 36%, Obama 28%, and Edwards 15%. Obama seems to be getting some sort of post-money announcement bump, which strikes me as a little strange. Then again, considering the proliferation of shows akin to Lifestyles of the Rich and Famous over the past decade, maybe it isn't surprising that announcing you have tons of cash makes you more popular among the voters.
Clinton: 32 (34)
Obama: 30 (29)
Edwards: 16 (16)
Now, I hardly ever post individual national polls on the front page, but I do so in this case because it is relevant to my ongoing discussion on whether or not national primary polls inflate Clinton's advantage. While every other national Democratic primary preference poll that I know of, except possibly the ever-questionable Zogby, includes the entire self-identified and leaning Democratic registered voter population in their polls, Rasmussen only includes "likely Democratic primary voters." To look at this from a more brass tacks perspective, while other national polls on the Democratic nomination include roughly 45% of the national registered voter universe, Rasmussen includes just under 39% of the national likely voter universe (774 of 2000 likely voters surveyed across four days). Now, while I would prefer a presidential primary poll that sampled only around 20% of the likely voter universe, as the universe of likely Democratic primary voters is usually only around that size or smaller, for now, in order to examine the validity of my thesis, I will have to take what I can get. While not as small as I would like, 39% of likely voters is substantially smaller than 45% of registered voters.
With a more narrowly targeted universe of Democrats, Rasmussen has consistently put up numbers that are less favorable to Clinton than virtually every other national poll taken this year. Over the last three months, six of the fourteen Rasmussen polls have shown Clinton ahead by five points or less, while only one of the other thirty-three national polls have done the same. Further, nine of fourteen Rasmussen national polls taken this year show Obama within eight points or less, while only six of the other thirty-three polls have done the same (including three from Time that only included Clinton, Edwards and Obama in the question). Clinton holds a median lead of 7% in the fourteen Rasmussen polls, and a median lead of 15% in the other thirty-three polls. Similarly, Clinton's mean advantage of 7.7% in Rasmussen polls is more than doubled in the other thirty-three polls.
Simply put, Rasmussen shows a significantly closer race than any other polling outfit, with the possible exceptions of ARG and Zogby (who both poll "likely" Democratic voters, whatever that means to those firms) and Time, which usually only includes Clinton, Obama and Edwards in their questions. Given my thesis, it isn't hard for me to wonder if the main reason for this is that Rasmussen polls a more representative sample of the Democratic primary electorate than other national polls. As I already stated, they sample roughly 39% of likely voters, while most other polls sample 45% of registered voters. While it is possible that other factors are at play, this certainly seems to be a decent amount of evidence to support my thesis. Clinton's lead is not as large as most national polls make it look, because most polls include too many irregular, low-information voters--among whom Clinton holds a disproportionate advantage--to be an accurate sample of the Democratic primary electorate.
Also, since I know we live in an era where most people's first reaction to an argument they don't like is to accuse the arguer of bias and a hidden agenda, you might enjoy these three posts where my poll analysis went quite decidedly against the hopes and conventional wisdom of the blogosphere:
- CT-Sen: Expect High Turnout And A Very Close Election, August 5th, 2006
- Fingernail Biting Time, November 5th, 2006
- Calling Bullshit On The Blogosphere's National Trial Heat Narrative, January 30th, 2007
Also, in other polling news, a new CNN poll (PDF) shows Clinton 36%, Obama 28%, and Edwards 15%. Obama seems to be getting some sort of post-money announcement bump, which strikes me as a little strange. Then again, considering the proliferation of shows akin to Lifestyles of the Rich and Famous over the past decade, maybe it isn't surprising that announcing you have tons of cash makes you more popular among the voters.
Tags: inflated Clinton poll theory, polls, President 2008 (all tags)
-----------
myDD - skin
-----------
57 Comments