This is the usual flip out.

update: this is the usual apology for the usual flip out which I always type whenever Drum mentions those polls. I retract the tone. The content may be of interest.
Now we return to our regularly scheduled rant.

Years ago I pointed out Gallup polls showing a solid majority of US adults say the rich pay less than their fair share to Kevin Drum (it was news to him). He noted this in a follow up post, but ever since that post he has been arguing that his guess was sortof right. Now the point is that the solid majority is smaller than it was in 1992 (the first time Gallup asked the question) and even 6% lower (which is statistically significant see below) than in 1994. So the 62% who absolutely utterly reject the central policy proposal of the Republicans are ignored to focus on the estiamted 6% who are no longer diametrically opposed to the central tenet of Republican policy. I totally lost it (as usual when he writes on this topic). Just click the link then come back. I don't want to excerpt. Speaking of the power of repetition, you continue to ignore the main fact reported by all of those polls years after I first pointed them out to you. I hope you recall that you (and Felix Salmon) claimed that people in the US oppose higher taxes on the rich. I told you about those of the polls in your figure which had been conducted before that date years ago. In a follow up post, you admitted that you had been wrong about US public opinion. Since then, you have been arguing against the data. In this post rather spectacularly. I know you suspect polling literalism, but on this topic, I very much think that you are being stubborn. I Now you discuss only the change and not the level. This is a dealing with data 101 error. The fact is that now as in every poll most US adults think the rich pay less than their fair share. Repitition has not worked for Republicans (as it has not worked for me here). Now you note the change 92 to 94 actually makes sense as taxes on the rich were significantly increased by a bill signed into law in 93. The remaining decline of 6% is tiny compared to the current level of 62%. A major party's main policy proposal goes in the opposite direction than the policy desired by Sixty Two percent (62%) of the public. The number is huge compared to the perceptions of commentators who don't bother with data. Comparison with 1992 only makes sense if we can assume that there was nothing unusual about views on tax fairness in 1992. Of course there was. Gallup started asking the question for a reason, which is that rage at the low taxes paid by the rich was extraordinarily intense Compared to anything. Two facts from 1992. First during a debate with focus groups with the dial thingy Bush supporters dialed strong agreement with Clinton at one point. I'm not sure something like this has happened before or since. It was when he said "only rich people have gotten tax cuts" or something (this is an almost 20 year old memory -- I don't remember the exact words). Also an internal Clinton campaign poll found a plurality were in favor of higher taxes on the rich to fund more "waste fraud and abuse" or so Brad DeLong (then deputy assistant secretary of the treasury) told me. Anger at the low taxes paid by the rich was not normal in 1992. It wasn't even especially sane. Yet your whole post is based on the assumption that there was nothing unusual about 1992. Read it. Note all discussion is of change since 1992. The idea that 1992 was a strange year and reversion to the mean could be expected isn't even considered. It is obviously the case. This comment is in the format spittle flecked rage not very rude arrogance, because Gred Sargent misread the post (and ignored Khimm's post). Also www.washingtonpost.com put my comment on Sargent in Khimm's comment thread -- twice. But to me, the practical message is that Democrats do not have to fight the Republicans to convince the US public that the rich pay too little. A very solid majority is convinced. Democrats just have to stand for a more progressive tax code and they will win. I'd say that means cutting taxes for the non rich a little (the full soak the rich and spread (some of it) out thin). This approach has worked very well so far. There is no problem except for sloppy data analysis based on stubborn insistence that your guess was roughly partly in a way semi right. Finally, quick what is the probability that a change as large or larger than the 6% decline from 1994 till 2012 is due to sampling error alone with no change in population public opinion ? You should have the answer immediately, since you should have done the calculation. By pollster usual silly approximation the change is roughly 0.6root(20) standard deviations so the chance is less than 5%. A better estimate is 0.9root(10) actually larger. Both are statisically significant changes. The calculations are quick and should not be optional.