The statement that “Reilly is slipping” is incorrect. In fact, the only candidate to lose percentage points from the previous CBS4/Survey USA pollis Deval Patrick.
<
p>
Tom Reilly gained a percentage point in the current poll and closed the gap between Deval Patrick and himself from 11% to 8%. Considering the poll’s margin of error (+/- 4.9%), all three candidates are in a statistical tie.
<
p>
This poll confirms what we all intuitively knew: the Democratic Primary is a dead heat and the campaign that performs best between now and September 19 will prevail.
leftisrightsays
memory I thought TR had 28 in the last poll. I competely agree this is a dead heat.
tom-msays
This poll says only 8% of likely voters are undecided. If anyone believes that’s even close, then I have a bridge you might be interested in.
<
p>
Maverick’s right: this is essentially a dead heat at this point. However, since this appears to be a fairly soft, name-recognition poll I’d much rather be the guy who has yet to do any significant advertising than the two guys whose faces are all over the TV.
Look at the question. It asks “if you were in the voting booth right now” who would you vote for? How many people are going to vote for Undecided in the voting booth?
<
p>
This phrasing of the question is deliberately designed to push soft supporters one direction or another. This does not, however, mean that the poll is invalid.
tom-msays
You’re exactly right- the phrasing is designed to PUSH people one way or the other. And, more often than not, that direction is going to be based on name recognition more than any real support.
<
p>
We eat, drink and breath this stuff, but the vast majority of voters haven’t even begun to pay attention, so a poll where 92% of the voters have made up their mind means very little, in my opinion.
<
p>
What’s more telling is that the only candidate who is not on TV still has a comfortable statistical lead. THAT’S the story.
This is not a “dead heat” — comeon, guys, this is supposed to be “reality based” discussion, not parroting the lazy media garBAGE.
<
p>
The MOE concept is a statistical tool used to gauge how likely it is that the sample drawn (if truly randomly, which is a whole nother issue) is representative of the population at large.
<
p>
You can’t willy-nilly add and subtract numbers and call the results “a statistical tie” — that’s just nonsense.
<
p>
Here’s how to interpret the results in a more mathematically sound way (and I don’t pretend to be the ultimate authority, but at least I do have a little training in statistics…):
<
p>
The margin of 35% (Patrick) over 30% (Gabrieli) suggests that there is a 91% chance that Patrick actually leads Gabrieli. Similarly, Patrick’s lead in the poll over Reilly is 98% likely to represent the fact that he really is in the lead.
<
p>
In like vein, there is an 80% chance that Gabrieli’s lead over Reilly is a true phenomenon in the population at large.
<
p>
Now, having said all that, I am concerned, as a Patrick supporter, that his lead has narrowed. Still, given the factors mentioned by other posters, I’m by no means discouraged.
alexwillsays
The margin of 35% (Patrick) over 30% (Gabrieli) suggests that there is a 91% chance that Patrick actually leads Gabrieli. Similarly, Patrick’s lead in the poll over Reilly is 98% likely to represent the fact that he really is in the lead.
In like vein, there is an 80% chance that Gabrieli’s lead over Reilly is a true phenomenon in the population at large.
<
p>
Doesn’t the ~4.9% lead that Patrick has over Gabrieli mean a 95% chance of a lead? SUSA uses 1/sqrt(N) for the MOE, which is two standard deviations, so 95% confidence. Maybe I’m still not quite getting it (I’m a physicist not a statician) but that was the way I was interpreting it…
jonfromwistasays
time since Ive done any statistical work but according to SUSA’s methodology ” In theory, with the stated sample size, one can say with 95% certainty that the results would not vary by more than the stated margin of sampling error, in one direction or the other, had the entire universe of respondents been interviewed with complete accuracy.”
p>
As I’ve noted there, it’s not the final word, but I think it helps to clear up a lot of the confusion.
<
p>
The part that is not addressed is the three-dimensional aspect of the race. Partial-equilibrium analysis is dangerous in such a situation. Still, we’re gettin’ there!
tom-msays
Look, you can plot the standard deviations any way you like, but unless the election were held today and all the undecideds break exactly like they did on this phone poll, it is essentially (“essentially” being the key word) a dead heat.
<
p>
Accusing us of “parroting the lazy media garbage” is lazy in its own right and, quite frankly, obnoxious. This race is tight and there’s a long way to go and no on here can dispute that. If that’s “parroting,” then so be it.
… but a 91% chance that Patrick is leading is not exactly a “dead heat” — unless I don’t know what you mean by that term. I would say a “dead heat” would be characterized by a number somewhere around 50%.
<
p>
Now, if you want to argue about the results from a more subjective point of view, that’s fine, but let’s not play fast and loose with the statistics.
The statement that “Reilly is slipping” is incorrect. In fact, the only candidate to lose percentage points from the previous CBS4/Survey USA pollis Deval Patrick.
<
p>
Tom Reilly gained a percentage point in the current poll and closed the gap between Deval Patrick and himself from 11% to 8%. Considering the poll’s margin of error (+/- 4.9%), all three candidates are in a statistical tie.
<
p>
This poll confirms what we all intuitively knew: the Democratic Primary is a dead heat and the campaign that performs best between now and September 19 will prevail.
memory I thought TR had 28 in the last poll. I competely agree this is a dead heat.
This poll says only 8% of likely voters are undecided. If anyone believes that’s even close, then I have a bridge you might be interested in.
<
p>
Maverick’s right: this is essentially a dead heat at this point. However, since this appears to be a fairly soft, name-recognition poll I’d much rather be the guy who has yet to do any significant advertising than the two guys whose faces are all over the TV.
Look at the question. It asks “if you were in the voting booth right now” who would you vote for? How many people are going to vote for Undecided in the voting booth?
<
p>
This phrasing of the question is deliberately designed to push soft supporters one direction or another. This does not, however, mean that the poll is invalid.
You’re exactly right- the phrasing is designed to PUSH people one way or the other. And, more often than not, that direction is going to be based on name recognition more than any real support.
<
p>
We eat, drink and breath this stuff, but the vast majority of voters haven’t even begun to pay attention, so a poll where 92% of the voters have made up their mind means very little, in my opinion.
<
p>
What’s more telling is that the only candidate who is not on TV still has a comfortable statistical lead. THAT’S the story.
This is not a “dead heat” — comeon, guys, this is supposed to be “reality based” discussion, not parroting the lazy media garBAGE.
<
p>
The MOE concept is a statistical tool used to gauge how likely it is that the sample drawn (if truly randomly, which is a whole nother issue) is representative of the population at large.
<
p>
You can’t willy-nilly add and subtract numbers and call the results “a statistical tie” — that’s just nonsense.
<
p>
Here’s how to interpret the results in a more mathematically sound way (and I don’t pretend to be the ultimate authority, but at least I do have a little training in statistics…):
<
p>
The margin of 35% (Patrick) over 30% (Gabrieli) suggests that there is a 91% chance that Patrick actually leads Gabrieli. Similarly, Patrick’s lead in the poll over Reilly is 98% likely to represent the fact that he really is in the lead.
<
p>
In like vein, there is an 80% chance that Gabrieli’s lead over Reilly is a true phenomenon in the population at large.
<
p>
Now, having said all that, I am concerned, as a Patrick supporter, that his lead has narrowed. Still, given the factors mentioned by other posters, I’m by no means discouraged.
<
p>
Doesn’t the ~4.9% lead that Patrick has over Gabrieli mean a 95% chance of a lead? SUSA uses 1/sqrt(N) for the MOE, which is two standard deviations, so 95% confidence. Maybe I’m still not quite getting it (I’m a physicist not a statician) but that was the way I was interpreting it…
time since Ive done any statistical work but according to SUSA’s methodology ” In theory, with the stated sample size, one can say with 95% certainty that the results would not vary by more than the stated margin of sampling error, in one direction or the other, had the entire universe of respondents been interviewed with complete accuracy.”
…I still remember that course…100 polisci undergrads forced to learn statistics. Still makes my head hurt…thanks to the number crunchers.
David has done a great service with a thoroughly-researched post on this topic.
<
p>
As I’ve noted there, it’s not the final word, but I think it helps to clear up a lot of the confusion.
<
p>
The part that is not addressed is the three-dimensional aspect of the race. Partial-equilibrium analysis is dangerous in such a situation. Still, we’re gettin’ there!
Look, you can plot the standard deviations any way you like, but unless the election were held today and all the undecideds break exactly like they did on this phone poll, it is essentially (“essentially” being the key word) a dead heat.
<
p>
Accusing us of “parroting the lazy media garbage” is lazy in its own right and, quite frankly, obnoxious. This race is tight and there’s a long way to go and no on here can dispute that. If that’s “parroting,” then so be it.
… but a 91% chance that Patrick is leading is not exactly a “dead heat” — unless I don’t know what you mean by that term. I would say a “dead heat” would be characterized by a number somewhere around 50%.
<
p>
Now, if you want to argue about the results from a more subjective point of view, that’s fine, but let’s not play fast and loose with the statistics.