Rasmussen makes me sick

[ps1074]

Okay, today's tracking poll shows Bush at 47,2% and Kerry at 46,5%. On his home page it is shown as Bush 47 and Kerry at 46.

I have no problem with that really but let's take a look at the congressional daily tracking poll.

Dems are at 44,8 and Republicans are at 41,5. On his home page that is presented as Dem at 45 and Repubs at 42.

Why the fuck this guy keeps Kerry at 46 (46,5) and gives repubs and extra point 42 (41,5)?

I guess this is because if he rounded Kerry's numbers at 47 it would have been the first time in a month that Bush is not longer leading and it will probably make it into the headlines of some of the newspapers...

I know it is just a small detail but it is all about perception. People look at the home page and see Bush leading instead of race tied.

[NMDemDist2]

1. link ? n/t

[ps1074]

2. sorry, here are the links

http://www.rasmussenreports.com/index.htm

http://www.rasmussenreports.com/Presidential_Tracking_Poll.htm

http://www.rasmussenreports.com/Congressional_Ballot_Survey.htm

[goclark]

3. I agree, George Step. said

that a Time Poll had Bush "ahead by six points."

Real interesting how he picked just that poll when lots of others have Bush/Kerry tied.

Huuum

And George was once a Democrat - LOL

[ps1074]

5. did he also mention

that they oversampled republicans in that Time poll?

I forgot what the numbers are but they had 6 or so percent more republicans. If weighted it will show the race tied of Kerry ahead.

[goclark]

9. Did not hear those words from his tiny lips

[Goldmund]

6. It's "round to even" rounding.

The rule is that when you have a .5 number you round it to the nearest even number. That's the way I was taught to round decimal numbers in school.

Same goes for rounding, for ex, 3.75 (3.8) and 3.85 (also 3.8).

[LisaL]

7. Well, I was taught differently.

The number 3.75 would be rounded to 3.8, while 3.85 would be rounded to 3.9. You do not round to an even number, you round to a next number if its .5 or above and to the previous number if it's below 0.5

[Goldmund]

15. That will accumulate your rounding error

If you're doing a calculation with lots of numbers and always round up, the error will add up; if you round up half of the time and round down half of the time, then the errors will tend to cancel themselves out.

[unfrigginreal]

17. That's very interesting

I don't remember being taught that but it makes sense. Thanks for posting.

[faithnotgreed]

19. yeah, he also said "iraq is stabilized" (about a month ago)

what is going on with him? hes just as disinterested in truth as all the rest. perhaps thats how he got his job, taking over for sam donaldson and cokie

[M_Demo_M]

4. Call them on it.

Send them an email and ask WTF is going on with their rounding.

[e-mail]

[featherman]

8. Rasmussen has it right

What he wants to display is the differential.
In Bush v. Kerry, the differential is .7 which should be rounded to 1
In Congress, the differential is 3.3 which should be rounded to 3
So the rounded differentials do indicate more accurate margins

Both of these numbers are good news for DEMS so what the heck.

[Quixote1818]

13. I agree, as stated in my post below.

[Nana]

10. Yes, write them!

I just did..let them know this is crap!!!

[HFishbine]

11. I had a math teacher

who instructed us when rounding a .5 to a whole number to round to the even number. It was a matter of the rounding balancing out over time. This may be what Rassmusen is doing.

[Quixote1818]

12. Sorry, I don't have a problem with this. Here is why:

The difference between 47,2% and Kerry at 46,5% is .7 points which if you round that off is closer to 1 point difference between the two. The way you want them to round it off would make it a tie when it's actually closer to a 1 point lead for Bush.

The difference between 44,8 and Republicans are at 41,5 is 3.3 points.
So wouldn't you round that off to make the difference three points? If you rounded it the way you want them to round it off it would become a four point difference when in fact it's closer to a 3 point difference.

Sorry but you would be fudging the numbers too much with your math.

[Nana]

14. I am a math teacher..

and, yes, .5 and up is rounded up by all standards. However, he is not rounding the percentage, but rather the spread..which is a choice he is making. It does conveniently, in this case, work "against" the Democrats. However, I suppose it could work "for" them at a later date. So..I don't agree with it..but understand his reasoning behind it now.

[progressivebydesign]

16. I'd say the math teacher has the last word.

Your account of it is exactly the way I was taught in school.. so I'm going with your answer.

[Goldmund]

18. Here is the rounding to even rule

http://www.uop.edu/cop/psychology/Statistics/Rounding.html

Rule 3- To prevent rounding bias, if the remainder is exactly 5, then round the last digit to the closest even number.Thus the number 3.55 (rounded to 1 digit) would be 3.6 (rounding up) and the number 6.450 would round to 6.4 (rounding down)if rounding to 1 decimal.