I think Nate Silver is brilliant when it comes to statistics so I was very alarmed at his article today when he, for lack of a better word, hap-haphazardly wrote that Romney got a 6-point swing in the Rasmussen Poll.
From Nate Silver's FiveThirtyEight NYT Blog:
On the favorable side for Mr. Romney is the Rasmussen Reports tracking poll, which now shows him four percentage points ahead of President Obama. That represents a 6-point swing toward Mr. Romney compared with the poll Rasmussen Reports released immediately before the convention ...
~Sept 3, 2012
Statistically speaking, Nate Silver is wrong, Rasmussen data does
not show Romney got a 6 point. At best, Romney got a 3 point swing from Scott Rasmussen and here is why:
Pre-Convention Rasmussen
Rasmussen's Margin of Error =3
Romney 47
Obama 45
In Statistics, you must use the Margin of Error. Therefore, the Rasmussen polling data Pre-Convention are:
Romney (45 + 3 = 48) and (45 - 3 = 42)
Obama (47 + 3 = 50) and (47 - 3 = 44)
Thus, Pre-Convention data, with Rasmussen's Margin of Error of 3,
- Romney's numbers were: 42 and 48
- Obama's numbers were: 44and50
Post-Convention Rasmussen (Sunday)
Rasmussen's Margin of Error =3
Romney 48 (not a 6point swing as 48 is below Pre-Convention 50)
Obama 44
Again, in Statistics, you must use the Margin of Error. Therefore, the Rasmussen polling data Post-Convention are:
Romney (48 + 3 = 51) and (48 - 3 = 45)
Obama (44 + 3 = 47) and (44 - 3 = 41)
Thus, Post-Convention data, with Rasmussen's Margin of Error of 3,
- Romney's numbers were: 45 and 51
- Obama's numbers were: 41and47
Which is a 3 point swing for Romney, not a 6 point swing.
Unfortunately, many people reading this Diary do not seem to understand what the Margin of Error is.
1) Margin of Error is the "Confidence Level" of collected data within a sample size.
2) Margin of Error cannot simply be "canceled out" because it is a "confidence interval" where the
Margin of error = Critical value x Standard deviation of the statistic
Anyone who ignores or who 'cancels out' the Margin of Error does not know what they are doing and should take a refresher course in Statistics.
I must underscore:
In statistics you MUST include the values of the Margin of Error thus, Romney's Post Convention Rasmussen:
3 point swing, not 6 point swing
Because the Margin of Error must be used in Statistics, it is statistically inaccurate to say Romney got a 6 point swing in the Rasmussen poll at best Romney got a 3 point swing period.
Having someone with Nate Silver's statistical genius ignore the Margin of Error and incorrectly proclaim Romney got a 6 point swing is disturbing because that type of mistake would be expected from a student taking their first test in Stats 101, not an accomplished person like Nate Silver.
Additionally, the only short-lived solace Romney can get from Rasmussen's post-convention poll is that Scott Rasmussen placed Romney above the Margin of Error, by 1 point. I say "short-lived" because today's Rasmussen's polling data is identical to their Sunday poll which is a disaster for Romney because one would expect polling data to be better on Monday Post-Convention than the numbers on Sunday Post-Convention.
For the record, I think Rasmussen polls are BS. Not only does Scott Rasmussen keep most of the data used secret, the worst part is Rasmussen uses "online survey tool" to conduct their poll. Meaning, much of the data in Rasmussen's polls come from online surveys ... hmmm ....
Rasmussen Website:
To reach those who have abandoned traditional landline telephones, Rasmussen Reports uses an online survey tool to interview randomly selected participants from a demographically diverse panel.
Naturally, Scott Rasmussen does not say which "online survey tool" they use and for all we know their online survey tool is FoxNation -- which I would not be surprised to learn.
Due to all the secret information Scott Rasmussen intentional keeps from those looking at his polls, determining the veracity, reliability and confidence of Rasmussen data points is like trying to get bubble-gum off the bottom of a shoe with tread on its sole.
Some comments here are implying that one can simply cancel out the Margin of Error and that is false.
A 3 percent margin of error means that if you asked a question from this poll 100 times, 95 of those times the percentage of people giving a particular answer would be within 3 points of the percentage who gave that same answer in this poll.
(WARNING: Math Geek Stuff!)
Why 95 times out of 100? In reality, the margin of error is what statisticians call a confidence interval. The math behind it is much like the math behind the standard deviation. So you can think of the margin of error at the 95% confidence interval as being equal to two standard deviations in your polling sample. Occasionally you will see surveys with a 99% confidence interval, which would correspond to 3 standard deviations and a much larger margin of error.
(End of Math Geek Stuff!)
Don't overlook that fact that the Margin of Error is a 95% confidence interval, either. That means that for every 20 times you repeat this poll, statistics say that one time you'll get an answer that is completely off the wall.
It is that reason, that I say, at best Romney got a 3 point swing, not a 6 point swing.