Articles and Analysis


US: National Survey (Harris 4/12-19)

Topics: poll

4/12-19/10; 2,755 adults
Mode: Internet
(Harris release)


Obama Job Rating
41% Excellent/Good, 59% Only Fair/poor (chart)
Reps: 9 / 91 (chart)
Dems: 70 / 30 (chart)
Inds: 36 / 64 (chart)

Congressional Job Rating
16% Excellent/Good, 84% Only Fair/Poor (chart)

State of the Country
39% Right Direction, 61% Wrong Track (chart)



How about using the old grade school system?

A). excellent
B). average
c). poor



Ok. I see that they have broken it down even further. My only question is did everyone responding to this poll know that a job approval of "fair" was considered a negative response. In my book, that's not a negative response.



An internet poll? What?



"Fair" does not equal "dissaprove", thus it's not accurate to report this as a 59% dissaproval rating, as Pollster has done in their National Trends.

Why is Pollster letting a poll like this completely skew the average?




That's the same question I used to ask when a lot of pollsters used these same "excellent, good, fair, poor" choices for Bush. It always makes the candidate look bad because most people don't think they have rated someone bad when they say "fair".

This is the first time I've seen the "excellent, good, fair, poor" options used on Obama since he was elected.

Of course I was slammed around by liberals when I brought up the same point about being rated as "fair" and having the pollster call it disapproval. Go back and look at some of the Bush polls, iVote, and see how many times this was used. You'll be surprised.

If you don't believe me, here is one from 2007 that shows Bush at 46% Excellent/Good and 52% at Fair/Poor. http://blog.al.com/breaking/2007/07/poll_finds_majority_of_alabami.html

I think most of the pollsters were using this method at that time but I'd have to look back to find out for sure. I just remember wondering who on earth decided that saying someone is doing a fair job is a negative rating. It's certainly not something Harris just made up, though.

Research 2000 used this method throughout the Bush presidency but they now use a differnt method ever since Obama was elected. Here is an example of a R2000 poll from 2003 that showed Bush at 51% Excellent/Good and at 49% Fair/Poor. http://www.kcci.com/politics/2597550/detail.html

But now that R2000 runs polls for Daily Kos - they use a method that shows a higher approval for Obama than their old method would have.




It was wrong then and it is wrong now. When there are 4 options, it's extremely misleading to lump them into only two, as Pollster has done here.

However, the two polls you referenced were only for Alabama and Iowa, respectively. And they were done by telephone. And they posted their methodology pertaining to Party ID.

This Harris poll was done in the most unreliable manner possible (on the Internet) and doesn't even report Party ID. For all we know, 60% of the people they polled online could have been Republicans.

And about R2000/DKos. They don't even poll approval for Obama. They poll favorability, which is traditionally better than somebody's approval rating.

This Harris poll is an outlier. End of story.



I'm not sure how you could divide the sentiment up in such a way that WOULD be completely even-handed.

"Excellent" and "Poor" are pretty straight forward, "good" somewhat less so, but "Fair" is a vague term by definition.

If they had five choices including "average" and "fair," would that be better? There are so many ways to interpret them...the attitudes might range from "average at best," to "okay" to "meh." Hard to categorize those kinds of attitudes, which is probably how the plurality of Americans that don't pay close attention to politics feel.



The breakdown in this poll is 9% excellent, 31% good, 26% only fair, and 33% poor, the former two grouped as positive and the latter two negative. I agree that "only fair" =/= "poor."

This poll loses a lot of credibility when it shows that men have a higher approval rate than women and higher among ages 45-64 than ages 18-33.



The point is that the excellent/good/fair/poor was used a lot for Bush. This is the first time I have ever seen it used for Obama.

I agree that it's wrong and rasmussen backs that up with a study on it:


Rasmussen used their standard strong approve, approve, disapprove, and strongly disapprove and they came up with a 47% approval and 52% disapproval in November for Obama.

When they used the excellent/good/fair/poor model like harris did here, they came up with 38% approve and 50% disapprove.

I'm glad that some pollsters are pulling this method back out that the liberals now agree with me that it is nonsense. They didn't agree when it showed Bush with a 38% approval because that's what they wanted to see.




How about just "approve" or "disapprove."


Cyril Washbrook:

@hoosier_gary: it's not "wrong", per se - it's just a different way of measuring public sentiment, and one that Harris has used for a fair while now. The two observations to be made are that (1) it cannot be compared to polls which use the "standard" approve/disapprove measure, and (2) we need to look at how useful or informative Harris' metric is.

On the first observation, Pollster clearly should not be placing these alongside job approval questions from other organisations which use the approve/disapprove system. Obviously, you have to have some leniency with respect to question wording, since different pollsters will use slightly different phrases, but it's another thing to fail to take into account conceptual differences - this isn't just a difference in wording, it's a fundamental difference in the metric being adopted.

On the second count, it is true that Harris' metric is less useful, although again, I'd emphasise that it is not inherently "wrong". This is to a great extent because of the ambiguity in the "(only) fair" term, which - as everyone has established - pulls in people who might respond that they approve or are indifferent to the President in a standard approve/disapprove poll. On balance, we would conclude that an approve/disapprove poll is more useful in terms of allowing reduction of the topline results, because it allows you to create a binary approve/disapprove result. The only way to get a binary result with the Harris poll is to impose a binary interpretation, which is inappropriate.

It does have some utility, as long as you only compare Harris' numbers to previous Harris polls, or other surveys which use the same metric. On those grounds, you could legitimately compare this poll to the previous Harris poll and conclude that nothing's changed since then. So there is some usefulness, but it's limited.



@Cyril, I won't change my mind this this is wrong. It is absolutely wrong to take the word, "fair" - which is a positive word in every other context in which it is used and consider it as a negative response. You might as well call "Excellent" and "Good" in the negative as well.

It was wrong when they did it for bush and it is wrong now that they are starting to do it for obama. Fair is fair. Fair is not disapprove.



At least they are consistent. They should ask the same questions regardless of who is in office.

It may be nit-picky, but the term I see them using consistently is "only fair." I gotta give respondents enough credit when asked if they rate the job done by the Prez excellent, good, only fair, or poor that only fair means less than excellent or good. How do you take it any way except somewhere below average. And we certainly should not be happy with a below average Prez.

I tend to just take it for what it is worth. First it is an iternet poll. Take with a huge grain of salt. Then the respondents are given four choices to rate how the Prez is doing. It is really obvious where each of the four is on the scale. Add it all up and of those who responded a minority rate the job done by the Prez better than average.

One little pixel in the big picture. Not particularly flattering but not the worst news the Prez got that day.


Post a comment

Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.