These Monmouth approval crosstabs match YouGov's almost exactly, but M's topline number is about 4-5 percentage points higher than YG's. This fits a broader pattern of party/vote-unweighted data being better for Biden that makes me think partisan nonresponse is still pretty high.
To restate the tweet: I am worried that polling aggregates are overestimating Biden's approval rating just like they overestimated his vote share last November
I thought it could be a mode effect, but I think this has to do with weighting, given that the party-weighted Civiqs data is also less favorable to Biden but other online polls aren't. I'm not saying 100% but this fits the narrative
It is possible that giving people the explicit "Don't Know" option on their screen is bumping up the unsure pop and deflating Biden approval, but we can't say for sure. Hence why I said I'm "worried about" this and not that I'm 100% confident in it.
To illustrate the point more: The unweighted YouGov data have 2020 recall at Biden +9 & 2016 at Clinton +4. Weighting by party/past vote would fix that issue, but most pollsters aren't doing that (and there's no way to check the accuracy of demo weights cause they're not panels.)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
If I take the Economist/YouGov poll, sale the J&J safe/unsafe question as a scale from -2 to 2, and then plot the trend over the course of the poll, there seems to be a large dip on Monday too. Any ideas?
Historically speaking, presidents that bucked the polls and ignored the majority — or, worse, used propaganda to manipulate opinion so they could cite polls later — have done more damage than those who just listened to the people most of the time & led when they thought necessary
The criticism of polls that some politicians just blindly follow them to win elections missed the mark. It’s a straw man set up by contrarian opinion columnists, minority lawmakers, and other elites so they can argue against giving the public opinion its proper due in Washington.
The other big argument in the lit — that “leadership” stands in tension with looking at polls — is both a false dichotomy and predicated on the idea that the people make bad judgments and are too dumb for self-government, which isn’t borne out in the scholarship.
One thing to note, across Pew’s postmortems and others’, is that the “solution” to 2016/2020 seems to be an increased reliance on weighting, more investment in sophisticated sampling techniques, or both — neither of which are readily available to firms without a ton of resources.
High-quality public opinion research is still possible, both online and off, but this means that we should expect more variance in good polls and more bias in bad polls. Not a great situation to be in, and the bandaids being proposed don’t really fix the underlying issues.
Lots of right-leaning commentary on democracy recently has advocated for restricting the franchise to people who are “better” at making decisions, with very little — if any — attention paid to the vast social science literature on this topic. Some things worth thinking about:
Obviously, these arguments are situated in a context of historical racism — whites used the exact same justification to disenfranchise black voters throughout the Jim Crow south. Oh, if they can’t pass literacy tests, why should they get to vote? *wink*
In addition to that, tho:
The fundamental problem with this is that, in a democracy, “majority rules” really is the only legitimate decision rule for government action. You can talk about the dangers of crowds, etc, but those fears are relatively unfounded in representative govs. gelliottmorris.substack.com/p/democracy-is…
It will be impossible to enact federal laws/rules preventing state election subversion — or at least substantially lowering the risk of it from the current (relatively) high level — so long as Republicans are driving partisan radicalization against democracy and free outcomes.
I view the point from @Nate_Cohn and others that Dems have missed the mark on HR1 bc of an overestimation of harms to turnout as a valid , tho maybe a bit beside the point that the gov cannot pass reasonable remedies so long as (a) our institutions are biased toward a party...
...that (b) views their opponent’s victories as illegitimate regardless of the conditions of their victory. HR1 probably won’t save democracy, but the solution is probably not attainable right now anyway. It is better to go ahead and reduce harms to voters in light of that.