3 ms·

Trump still has approval ratings in the fourties, which is slightly less than half. Surely those opinion polls are biased in some way or another, but that's st

Trump still has approval ratings in the fourties, which is slightly less than half. Surely those opinion polls are biased in some way or another, but that's still the best numbers we have, in my opinion. > As an empirical matter, this perhaps demonstrates that a little more than 18% supported him at that time. No it doesn't. It demonstrates that at least 18% supported him for president, at the time. It tells you nothing about the people that couldn't or didn't vote, but may have supported him in spirit. For instance, it's pretty pointless to actually go out and vote for Trump in California. > Having traveled the country by land for years and discussed this topic... Of all the numbers that are available, your anecdata is the least relevant.

jMyles 7y ago
> Trump still has approval ratings in the fourties, which is slightly less than half. I do not have confidence in any of the current methodologies for polling for presidential approval ratings, and of course I'm not alone in this view; I'm joined by a substantial cohort within academic political science (in which I have a degree, if that's important) and, occasionally, Pew itself. [0] At a minimum, even if the methodology is spot-on, all of the current polls (at least of which I'm aware) exclude people under the age of 18, incarcerated people, undocumented immigrants, homeless people, and, in many cases, people who do not have a landline telephone. This relegates them to joke status if the conclusion being discussed is support of the entire population. > No it doesn't. It demonstrates that at least 18% supported him for president, at the time. It tells you nothing about the people that couldn't or didn't vote, but may have supported him in spirit. For instance, it's pretty pointless to actually go out and vote for Trump in California. It sounds like we're on the same page here: elections are not a good basis for drawing conclusions about the support a particular leader has within a population. I particularly object to your contention that this demonstrates that "at least 18% supported him" - depending on your view of "support", I suggest that this excludes the enormous contingent of "anybody but Hillary" voters. These people did not support Trump, do not support his administration, and used their vote as a tactic to show opposition to a candidate rather than support for a candidate. > Of all the numbers that are available, your anecdata is the least relevant. Feel free to disregard it then. It's meaningful to me. I have every confidence in the conclusions that I'm able to draw from my extensive travels and discussions on this topic, particularly in the deep south, whose politics I had never understood until traveling there and discussing the 2016 election with people on a daily basis for several months. 0: https://www.politico.com/story/2019/02/27/phone-polling-crisis-1191637 https://www.politico.com/story/2019/02/27/phone-polling-cris...
- gridlockd 7y ago
  You can have all the doubts you want about bias in phone polling, but unless you can explain to me why it should be biased towards Trump supporters in a significant way, I would still go by these numbers as a "best effort" estimate. I could certainly come up with reasons why those numbers would be biased against Trump: Coming out in favor of him is "undeniable support for white supremacy" or "caging children" or "pussy grabbing" or whatever scandal Trump is currently involved in. That's also the likely reason why the election polls were so utterly wrong. > people under the age of 18, incarcerated people, undocumented immigrants, and homeless people Okay, let's say we count all of these as "non-supporters", you'd still have approval by roughly a third of the country. Is "roughly a third" really substantially different from "fourty-ish"? ________________________________________________________ This is an edit as reply, since I'm "posting too fast": > I'm making the case that these polls are not a good basis for drawing any conclusions Given the task of estimating the number of Trump supporters, "I can't do it" is not an acceptable answer. Again, best effort matters, even though we know it's not the true number. > I'm not following the math here. Are you suggesting that, taken together, these groups represent only 7% of the population? Obviously that's not correct. The math is as follows: 330 million people in the US (ballpark figure) 80 million undocumented and/or underage (ballpark figure) therefore, (330 - 80 = 250) million polling candidates 42% of 250 million supporting Donald Trump = 107.5 million 107.5 of 330 million supporting Donald Trump = roughly a third
  - jMyles 7y ago
    > You can have all the doubts you want about bias in phone polling, but unless you can explain to me why it should be biased towards Trump supporters in a significant way, I'm not making this case. I'm making the case that these polls are not a good basis for drawing any conclusions about support for a particular candidate or elected official throughout the entire population. > I could certainly come up with reasons why those numbers would be biased against Trump: Coming out in favor of him is "undeniable support for white supremacy" or "caging children" or "pussy grabbing" or whatever scandal Trump is currently involved in. That's also the likely reason why the election polls were so utterly wrong. Agreed - the polls are not useful for this particular set of conclusions. > Okay, let's say we count all of these [people under the age of 18, incarcerated people, undocumented immigrants, and homeless people] as "non-supporters", you'd still have approval by roughly a third of the country. Is "roughly a third" really substantially different from "fourty-ish"? I'm not following the math here. Are you suggesting that, taken together, these groups represent only 7% of the population? Obviously that's not correct.

jMyles 7y ago

> Trump still has approval ratings in the fourties, which is slightly less than half. I do not have confidence in any of the current methodologies for polling for presidential approval ratings, and of course I'm not alone in this view; I'm joined by a substantial cohort within academic political science (in which I have a degree, if that's important) and, occasionally, Pew itself. [0] At a minimum, even if the methodology is spot-on, all of the current polls (at least of which I'm aware) exclude people under the age of 18, incarcerated people, undocumented immigrants, homeless people, and, in many cases, people who do not have a landline telephone. This relegates them to joke status if the conclusion being discussed is support of the entire population. > No it doesn't. It demonstrates that at least 18% supported him for president, at the time. It tells you nothing about the people that couldn't or didn't vote, but may have supported him in spirit. For instance, it's pretty pointless to actually go out and vote for Trump in California. It sounds like we're on the same page here: elections are not a good basis for drawing conclusions about the support a particular leader has within a population. I particularly object to your contention that this demonstrates that "at least 18% supported him" - depending on your view of "support", I suggest that this excludes the enormous contingent of "anybody but Hillary" voters. These people did not support Trump, do not support his administration, and used their vote as a tactic to show opposition to a candidate rather than support for a candidate. > Of all the numbers that are available, your anecdata is the least relevant. Feel free to disregard it then. It's meaningful to me. I have every confidence in the conclusions that I'm able to draw from my extensive travels and discussions on this topic, particularly in the deep south, whose politics I had never understood until traveling there and discussing the 2016 election with people on a daily basis for several months. 0: https://www.politico.com/story/2019/02/27/phone-polling-crisis-1191637 https://www.politico.com/story/2019/02/27/phone-polling-cris...

gridlockd 7y ago
You can have all the doubts you want about bias in phone polling, but unless you can explain to me why it should be biased towards Trump supporters in a significant way, I would still go by these numbers as a "best effort" estimate. I could certainly come up with reasons why those numbers would be biased against Trump: Coming out in favor of him is "undeniable support for white supremacy" or "caging children" or "pussy grabbing" or whatever scandal Trump is currently involved in. That's also the likely reason why the election polls were so utterly wrong. > people under the age of 18, incarcerated people, undocumented immigrants, and homeless people Okay, let's say we count all of these as "non-supporters", you'd still have approval by roughly a third of the country. Is "roughly a third" really substantially different from "fourty-ish"? ________________________________________________________ This is an edit as reply, since I'm "posting too fast": > I'm making the case that these polls are not a good basis for drawing any conclusions Given the task of estimating the number of Trump supporters, "I can't do it" is not an acceptable answer. Again, best effort matters, even though we know it's not the true number. > I'm not following the math here. Are you suggesting that, taken together, these groups represent only 7% of the population? Obviously that's not correct. The math is as follows: 330 million people in the US (ballpark figure) 80 million undocumented and/or underage (ballpark figure) therefore, (330 - 80 = 250) million polling candidates 42% of 250 million supporting Donald Trump = 107.5 million 107.5 of 330 million supporting Donald Trump = roughly a third
- jMyles 7y ago
  > You can have all the doubts you want about bias in phone polling, but unless you can explain to me why it should be biased towards Trump supporters in a significant way, I'm not making this case. I'm making the case that these polls are not a good basis for drawing any conclusions about support for a particular candidate or elected official throughout the entire population. > I could certainly come up with reasons why those numbers would be biased against Trump: Coming out in favor of him is "undeniable support for white supremacy" or "caging children" or "pussy grabbing" or whatever scandal Trump is currently involved in. That's also the likely reason why the election polls were so utterly wrong. Agreed - the polls are not useful for this particular set of conclusions. > Okay, let's say we count all of these [people under the age of 18, incarcerated people, undocumented immigrants, and homeless people] as "non-supporters", you'd still have approval by roughly a third of the country. Is "roughly a third" really substantially different from "fourty-ish"? I'm not following the math here. Are you suggesting that, taken together, these groups represent only 7% of the population? Obviously that's not correct.