What was the margin of error on the Truman poll, and the margin of error on the Biden poll, that you claim as evidence that Biden rated higher than Truman somewhere around this time in their presidencies?
I have no idea what the margin of error was. Do you?
No.
So I'm curious. Since you don't know, why would you be confident that Biden rated higher than Truman, and repeatedly defend that notion?
Why don’t you ask Fivethrityeight whether it knows? After all, they are the ones putting up the data. It‘s not unreasonable to assume they think the data is meaningful.
But notice how you have shifted the goalposts. Initially you were bitching that I claimed Biden’s approval relating was equivalent to, or better than, seven of 13 presidents for whom such data is available. Now you are focusing on Truman. Why?
I've already explained why, multiple times. But I'll do it again.
I will take this as your concession that there is no meaningful distinction between Reagan’s 43.0 and Biden’s 41.5 on day 661 of both of their presidencies, because, well, there isn’t.
No. That is not a concession. Biden's rating
might be lower (a lower point estimate where the 95% confidence intervals do not overlap), or there might be
no evidence of a difference (the confidence intervals overlap).
For all the comparisons where Biden
might be lower or there's
no evidence of a difference, I have left these comparisons 'on hold', because we do not know the margins of error and your argument appears to be based on
no evidence of a difference.
You touted the Truman comparison as the one where Biden is higher (the 'or higher' part of your statement), presumably because whatever two figures you compared had a difference of at least 6 percentage points (using your 'three point margin' shortcut).
I don't know what those two figures were, but even a difference of 6 percentage points might have overlapping confidence intervals, because the confidence interval on the Truman poll
might be very much larger than the short cut 'three point margin' established in more recent decades.
That is what that three-point margin of error thingie means, you know.
fivethirtyeight lists dozens of polls with all sorts of margins of error, and pools those polls for their headline figure. When they are pooled, the margin of error on the point estimate would go down. I don't know what it goes down to, but it is smaller than the margins of error in the individual polls.
And so I think some of your comparisons where you claim there is no difference are probably wrong, and if we knew the real margins of error, it could show Biden to be lower (given the point estimates are lower). Not for the Trump-Biden comparison, though, as the ratings are very, very close and the margin of error on each poll would need to be tiny to show a difference if there really was one.
I don’t know this for certain, but I believe modern polling is not relevantly different now from what it was dating all the way to the Eisenhower administration, which means that if modern popularity polling means anything for Biden, then it means pretty much the same thing for all the presidents surveyed dating at least to Eisenhower. And my point stands.
Truman MAY be a bit of an outlier in terms of polling because I’m not sure polling was fully developed then, and polling is based on science. The Truman polls in question were taken in 1947. The next year, 1948, the polls, including Gallup, pretty much unanimously called the presidential race for Dewey over Truman, but Truman won handily. HOWEVER, I don’t think this happened because the polls were conducted in some manner relevantly different from the way they are now. The difference was that all the pollsters, thinking the election was settled, stopped polling about two weeks before election day. They did not realize then, as is well understood now, that public opinion can shift dramatically in the final two weeks of a political race.
I hope that answers your questions. It is tiresome to keep wiping the egg of your face. You should try doing it for yourself some time.
It of course does not answer my questions, except that perhaps it is relevant to say your Biden-Truman comparison is not a solid comparison for all sorts of social and technical reasons, and your statement should be 'using a three point margin of error as a stand-in, there is
no evidence of a difference between Biden's approval ratings and five of the last 12 presidents'.
That statement is far more honest than where you say 'no difference or higher', which is ambiguous as to how many have no difference and how many are higher.