I really appreciated your comments on the 538 podcast about this topic — salience to the election at hand is key. I’m a volunteer grassroots activist in the NY19 congressional district and have been leading canvassing teams contacting ‘drop off’ Dems and ‘high scoring’ Dem friendly unaffiliated voters all spring and summer.
In that context — people know that I’m canvassing for Dems & giving them election info — when asked the open-ended question “what issues concern you,” I’m very consistently seeing these issues as the most frequently mentioned by all ages of rural/exurban/town voters: Medicare/Social Security (I think that this is a bit of a stand-in for all govt programs of this type incl the ACA, Women’s Reproductive Freedom/Abortion, Climate, Housing, and to a lesser degree, Inflation/Economy. A good sign for Dems in my opinion is that these voters don’t trust Republicans on M/SS & ACA, Abortion and Climate. They proactively tell us this — and that they don’t like extremist Republicans. Many were willing to sign petitions to direct Congress to not cut M/SS and in support of Biden’s plan for M/SS solvency.
In contrast to 2022, here Dems won many local elections this year due to (still looking at this) what appears to be increased Dem turnout and somewhat decreased R turnout. In some cases by one or two votes at the local level! I do think — no data, just my gut feeling based on door knocking — the press coverage of how right wing religious the new GOP House Speaker hitting right around our early voting period DID impact Dem voters, encouraging them to turn out to be able to make some sort of stand.
I do wonder a bit if the nationwide Dem turnout dynamics flip is related to who’s now in the Dem coalition — folks more likely to vote — but not sure that applies in my area, as college grad is a pretty low percentage of the population. Will be taking a closer look at town by town turnout in Dec.
I know that what I’m sharing are just a slice of life, but maybe it’s helpful as you analysts think about what questions to ask…
Compared to the attention given to sampling, survey research is blindfolded on what it is doing as a social process built upon the social processes going on in society. Hence, when the underlying modes of discourse and signaling change in the society, survey researchers are clueless about what is happening to their attempt to elicit meaningful responses. Nowhere truer than for issue polling.
It seems that the issue of social influence is not being addressed. When I am thought of by my cyberspace and meatspace communities as knowledgeable, admirable, a person to follow, then more people will want to do as I do. If I express concerns about Climate publicly and daily, then my gang will too. That's different from watching Fox, because I have actual friends, colleagues, followers, associates, I interact, etc. So I'm an influencer. Now substitute You, We, They for I.
One more point about issue polling that I discussed in the Mystery at the Midterm RCP piece. There was a great deal of confusion about the impact of the abortion issue in the 2022 midterm. It appears that much of this centered around different surveys offering different ranges of possible responses to their “the most important problem/issue” questions. Some offered a very constricted range and others a big smorgasbord. I think some discussion among pollsters should give guidance about best practices. A choice among 4 is too few (I recall that’s what the exits provided), but subdividing issues like the economy into many components is not illuminating either (though users could always combine for themselves if they were on the problem).
I explained the discrepancy between the 2022 polls and the blocked red wave results in my post-election analysis, “Mystery at the Midterm: What Happened to the Red Wave?” in RealClear Politics, May 16, 2023. The archived version includes the full analysis. One point of this is that public opinion is not the only influence on election outcomes. Another is that all actual voters are not picked up in survey samples and, even if many are, they may be filtered out as unlikely voters, especially when voting systems change (expanded absentee voting) and intense mobilization campaigns are waged.
I was just looking at the RCP final presidential preference polls for the last four presidential contests. The mean absolute error for the two-party national vote was only about one percentage point. Nothing wrong with recent polling. The error is in its analysis.
I really appreciated your comments on the 538 podcast about this topic — salience to the election at hand is key. I’m a volunteer grassroots activist in the NY19 congressional district and have been leading canvassing teams contacting ‘drop off’ Dems and ‘high scoring’ Dem friendly unaffiliated voters all spring and summer.
In that context — people know that I’m canvassing for Dems & giving them election info — when asked the open-ended question “what issues concern you,” I’m very consistently seeing these issues as the most frequently mentioned by all ages of rural/exurban/town voters: Medicare/Social Security (I think that this is a bit of a stand-in for all govt programs of this type incl the ACA, Women’s Reproductive Freedom/Abortion, Climate, Housing, and to a lesser degree, Inflation/Economy. A good sign for Dems in my opinion is that these voters don’t trust Republicans on M/SS & ACA, Abortion and Climate. They proactively tell us this — and that they don’t like extremist Republicans. Many were willing to sign petitions to direct Congress to not cut M/SS and in support of Biden’s plan for M/SS solvency.
In contrast to 2022, here Dems won many local elections this year due to (still looking at this) what appears to be increased Dem turnout and somewhat decreased R turnout. In some cases by one or two votes at the local level! I do think — no data, just my gut feeling based on door knocking — the press coverage of how right wing religious the new GOP House Speaker hitting right around our early voting period DID impact Dem voters, encouraging them to turn out to be able to make some sort of stand.
I do wonder a bit if the nationwide Dem turnout dynamics flip is related to who’s now in the Dem coalition — folks more likely to vote — but not sure that applies in my area, as college grad is a pretty low percentage of the population. Will be taking a closer look at town by town turnout in Dec.
I know that what I’m sharing are just a slice of life, but maybe it’s helpful as you analysts think about what questions to ask…
Compared to the attention given to sampling, survey research is blindfolded on what it is doing as a social process built upon the social processes going on in society. Hence, when the underlying modes of discourse and signaling change in the society, survey researchers are clueless about what is happening to their attempt to elicit meaningful responses. Nowhere truer than for issue polling.
Thanks for this rundown, though I just can’t get over the typo in the title.
It seems that the issue of social influence is not being addressed. When I am thought of by my cyberspace and meatspace communities as knowledgeable, admirable, a person to follow, then more people will want to do as I do. If I express concerns about Climate publicly and daily, then my gang will too. That's different from watching Fox, because I have actual friends, colleagues, followers, associates, I interact, etc. So I'm an influencer. Now substitute You, We, They for I.
One more point about issue polling that I discussed in the Mystery at the Midterm RCP piece. There was a great deal of confusion about the impact of the abortion issue in the 2022 midterm. It appears that much of this centered around different surveys offering different ranges of possible responses to their “the most important problem/issue” questions. Some offered a very constricted range and others a big smorgasbord. I think some discussion among pollsters should give guidance about best practices. A choice among 4 is too few (I recall that’s what the exits provided), but subdividing issues like the economy into many components is not illuminating either (though users could always combine for themselves if they were on the problem).
I explained the discrepancy between the 2022 polls and the blocked red wave results in my post-election analysis, “Mystery at the Midterm: What Happened to the Red Wave?” in RealClear Politics, May 16, 2023. The archived version includes the full analysis. One point of this is that public opinion is not the only influence on election outcomes. Another is that all actual voters are not picked up in survey samples and, even if many are, they may be filtered out as unlikely voters, especially when voting systems change (expanded absentee voting) and intense mobilization campaigns are waged.
I was just looking at the RCP final presidential preference polls for the last four presidential contests. The mean absolute error for the two-party national vote was only about one percentage point. Nothing wrong with recent polling. The error is in its analysis.