" Adjusted" Rasmussen Poll: Bush "approval/disapproval" at 37/42

[kansasblue]

Our analysis of the latest Rasmussen Report concludes Bush support is weaker than shown. Here our not so scientific analysis base on the latest Rasmussen Report polling.

If someone reports they STRONGLY approve or disapprove then these votes are given a value of '2'. SOMEWHAT approve or disapprove are give a value of "1". We've added additional value to anyone expressing strong support (either for or against the President).



President Bush Job Approval
Updated Daily by Noon Eastern from Rasmussen.
http://www.rasmussenreports.com/Bush_Job_Approval.htm

Bush Job Approval
Strongly Approve 25% x 2 = 50
Somewhat Approve 23% x 1 = 23
Somewhat Disapprove 16% x 1 = 16
Strongly Disapprove 34% x 2 = 68

the adjust value give Bush approval rating of 73 / 2 = 37
and the adjusted disapproval at 16 + 68 = 84, 84 / 2 = 42


For example:

What if 1 percent said Strongly approve
49 percent said Somewhat approve
1 percent disapprove
49 percent Strongly disapprove


the poll would say 50% / 50%


so would this poll:

49 percent said Strongly approve
1 percent said Somewhat approve
49 percent disapprove
1 percent Strongly disapprove

This also is a 50 / 50 poll 50% approval

but they are very very different in support.

Study Bush's current number with this is mind. He's less popular than the number indicate.

[geek tragedy]

1. Whose analysis is that?

Seems dubious to me.

[mtnester]

2. So, he is now below 40%

Maybe he and Bob Taft can git together...compare NOTES

[Jackpine Radical]

3. This is a totally arbitrary weighting system

with no psychometric or empirical validity, and no particular meaning.

That said, the poll results of 48-52 do not bode well for a sitting President.

[DCal]

11. No, it isn't arbitrary

It is similar to calculating an average on a 4-point scale. Each of the points on the scale have meaning.

Here is how the means can be calculated:

25% Strongly Approve (4 on 4-point scale)
23% Somewhat Approve (3 on 4-point scale)
16% Somewhat Disapprove (2 on 4-point scale)
34% Strongly Disapprove (1 on 4-point scale)

Average = 2.40

If the percentages were as follows:

35% Strongly Approve (4 on 4-point scale)
13% Somewhat Approve (3 on 4-point scale)
25% Somewhat Disapprove (2 on 4-point scale)
25% Strongly Disapprove (1 on 4-point scale)

The average would be 2.59.

The same % approve and disapprove in the two examples, but Bush does worse in the first example (which is the actual result of the Rasmussen survey).

So, it DOES make a difference how the percentages are distributed within the approval and disapproval categories.

[Jackpine Radical]

13. The 1-2-3-4 scale is at best a rank-order scale.

You are inappropriately applying interval-level, normal statistical procedures to data that are too crude to merit that treatment.

Yes it does presumably make a difference whether people choose "somewhat" or "strongly," but there is too little information present to say anything about how much of a difference.

[DCal]

14. Taking into consideration the strength of approval or disapproval

is infinitely better than ignoring it, as you would have us do.

It does not "presumably" make a difference whether people choose "someewhat" or "strongly", it DEFINITELY makes a difference. If you look at any analysis of voting behavior, you can see dramatic differences between partisan voting intention between those who strongly approve of a candidate vs. those who somewhat approve.

At the very least, tracking changes in each of the four categories over time is preferable to lumping all approves and disapproves together. And, if there was a 4-point scale with only labels for points 1 and 4 (i.e., interval level data), you would not have a problem with using a mean, right?

[kansasblue]

4. I'm just playing around. Don't take it so serious.

But it makes sense.

What if 1 percent said Strongly approve
49 percent said Somewhat approve
1 percent disapprove
49 percent Strongly disapprove


the poll would say 50 / 50


so would this poll

49 percent said Strongly approve
1 percent said Somewhat approve
49 percent disapprove
1 percent Strongly disapprove

This also is a 50 / 50 poll 50% approval

but they are very very different in support.

Study Bush's current number with this is mind. He's less popular than the number indicate.

[rogerashton]

7. Um, right --

Equal weights at one are just as arbitrary as any other scale.

But -- if this were a pre-election poll -- wieghts according to empirical probability of voting makes sense. For a mid-off-year, its hard to justify any specific weights.

Your point is, I gather, that Bush loses more support at the extremes and does a bit better with the wishy-washy middle. Which is what I would expect for a sitting president.

[jobycom]

5. I think extra weight should be given to those who eat Ben & Jerry's

Cherry Garcia with a wooden spoon on Sundays. See how that would affect the poll.

[patdem]

6. I kind of agree..why not just ask approve/disapprove..it doubt many

who somewhat approve would say they simply approve..and by the same token i do not think anyone who says they somewhat disapprove would say they approve....why the four questions? if there is not some diffference?

[EST]

8. I like it.

Fits in better with my own opinion.

[wli]

9. i.e. the polls are falsified; old news

Would you really expect anything less?

[ChiciB1]

10. Polls Don't Make A Damn Bit Of Difference...

They get their way anyway!!!

[Nicholas D Wolfwood]

12. sorry - i'm not biting on that.