Those who follow the news have no doubt heard that, according to the latest Fox News Poll, if the 2020 election were held today, Donald Trump would be defeated by Joe Biden 49% to 39% and Bernie Sanders 46% to 40%. Additionally, more Americans apparently disapprove of the way Trump is doing his job than approve, 51% to 46%, with 3% responding "I Don't Know". All the "talking heads", particularly those representing the Democratic Party, have been quick to seize on these poll results as proof positive that Trump is in serious trouble seeking reelection.
One of the graduate school seminars I took for my Masters in Government was an in-depth study of public opinion polling. The most fundamental point made was that, unless one obtains a copy of the poll, the poll results should not be taken seriously. The way in which poll questions are written, as well as the order in which they are asked, will have a direct influence on the outcome. The most important questions to be considered are whether the poll results reflect "informed" opinion, whether the poll was truly random, and whether the results have been controlled for possible bias.
When a poll taker calls a potential respondent, there is no way of knowing what that person is doing. Perhaps they just sat down to dinner. Perhaps he/she has just been watching a news cast or has been perusing their favorite "news" site on the Internet. Their answer to the first few questions will reflect whatever has been most recently on their mind. If the poll is intended to gauge political attitudes, the results will also be skewed in favor of their political affiliation.
In the case of this latest Fox News Poll, Question #1 asks "Do you approve or disapprove of the job Donald Trump is doing as president? [If Approve/Disapprove: is that strongly (approve/disapprove), or only somewhat?]
Asking this as the very first question is problematic for a number of reasons. It does not necessarily reflect an informed opinion. The respondent may be someone who simply doesn't follow the news and has little or no knowledge about any of the national issues and what Trump is, or is NOT, doing regarding those issues. The individual may simply provide a response based on what he/she heard their neighbor say, (i.e. Donald Trump is a racist/white supremacist, hates immigrants, etc.) The respondent's political affiliation will definitely affect their answer. If they've just watched a "news cast" on CNN or MSNBC, they will be more likely to answer that they disapprove. Conversely, if they watch Fox News, their answer will likely be the opposite.
In this particular poll, the respondent's political affiliation is not ascertained until Question #17: "In the presidential [primary election/caucus] in [state] next year, are you more likely to vote in the Democratic or Republican [primary election/caucus] for president, or are you unlikely to vote in either?" Although the polling methodology is claimed to be random, in actuality the people who answered the phone and participated are skewed toward the Democratic Party by 8 percentage points.
This skews the entire poll, particularly regarding Questions #20 and 21. "If the 2020 presidential election were held today, how would you vote if the candidates were: [If Don't Know] Well, which way would you lean?" For Question #20, Joe Biden is inserted. Biden is preferred 49% to 39%. Question 21 inserts Bernie Sanders - result Bernie 46%, Trump 40%.
These are the numbers being thrown around this week. However, if one considers the 8% Democratic bias revealed by Question #17, Biden would only lead Trump by 2% and Trump would be ahead of Bernie.
In order for poll results to more accurately reflect informed opinion, respondents should be asked questions dealing specifically with various issues BEFORE being asked whether they favor one candidate over another. This forces them to think not only about each issue, but also to consider where the different candidates stand on those issues. Question #38 asks a very relevant question:
38. Do you think it would be a good thing or a bad thing for the United States to move away from capitalism and more toward socialism?
39.-43. I am going to mention several policy proposals. For
each one, please tell me if you favor or oppose the idea. [IF FAVOR / OPPOSE: Is that strongly (favor/oppose), or only
somewhat? [RANDOMIZE]
The above question tells us nothing because it is does not ask the respondent to even consider what the "buy in" would cost them.
Here is the next question in the sequence:
------------------Favor-------------- ------------Oppose------------------- (Don’t know)
Allowing refugees from TOTAL Strongly Somewhat TOTAL Strongly Somewhat
One of the graduate school seminars I took for my Masters in Government was an in-depth study of public opinion polling. The most fundamental point made was that, unless one obtains a copy of the poll, the poll results should not be taken seriously. The way in which poll questions are written, as well as the order in which they are asked, will have a direct influence on the outcome. The most important questions to be considered are whether the poll results reflect "informed" opinion, whether the poll was truly random, and whether the results have been controlled for possible bias.
When a poll taker calls a potential respondent, there is no way of knowing what that person is doing. Perhaps they just sat down to dinner. Perhaps he/she has just been watching a news cast or has been perusing their favorite "news" site on the Internet. Their answer to the first few questions will reflect whatever has been most recently on their mind. If the poll is intended to gauge political attitudes, the results will also be skewed in favor of their political affiliation.
In the case of this latest Fox News Poll, Question #1 asks "Do you approve or disapprove of the job Donald Trump is doing as president? [If Approve/Disapprove: is that strongly (approve/disapprove), or only somewhat?]
Asking this as the very first question is problematic for a number of reasons. It does not necessarily reflect an informed opinion. The respondent may be someone who simply doesn't follow the news and has little or no knowledge about any of the national issues and what Trump is, or is NOT, doing regarding those issues. The individual may simply provide a response based on what he/she heard their neighbor say, (i.e. Donald Trump is a racist/white supremacist, hates immigrants, etc.) The respondent's political affiliation will definitely affect their answer. If they've just watched a "news cast" on CNN or MSNBC, they will be more likely to answer that they disapprove. Conversely, if they watch Fox News, their answer will likely be the opposite.
In this particular poll, the respondent's political affiliation is not ascertained until Question #17: "In the presidential [primary election/caucus] in [state] next year, are you more likely to vote in the Democratic or Republican [primary election/caucus] for president, or are you unlikely to vote in either?" Although the polling methodology is claimed to be random, in actuality the people who answered the phone and participated are skewed toward the Democratic Party by 8 percentage points.
This skews the entire poll, particularly regarding Questions #20 and 21. "If the 2020 presidential election were held today, how would you vote if the candidates were: [If Don't Know] Well, which way would you lean?" For Question #20, Joe Biden is inserted. Biden is preferred 49% to 39%. Question 21 inserts Bernie Sanders - result Bernie 46%, Trump 40%.
These are the numbers being thrown around this week. However, if one considers the 8% Democratic bias revealed by Question #17, Biden would only lead Trump by 2% and Trump would be ahead of Bernie.
In order for poll results to more accurately reflect informed opinion, respondents should be asked questions dealing specifically with various issues BEFORE being asked whether they favor one candidate over another. This forces them to think not only about each issue, but also to consider where the different candidates stand on those issues. Question #38 asks a very relevant question:
38. Do you think it would be a good thing or a bad thing for the United States to move away from capitalism and more toward socialism?
Good
thing Bad
thing (Don’t know)
34% 54 22
Questions 39-43 represent examples of how questions can elicit information that answer the direct question posed, but do not reflect any serious knowledge of the complete issue:
----------------Favor-------------- ------------Oppose------------------- (Don’t know) TOTAL Strongly Somewhat TOTAL Strongly Somewhat
Changing the health care 65% 34 31 23% 9 14 12
system so that every
American can buy into
Medicare if they want to?
The above question tells us nothing because it is does not ask the respondent to even consider what the "buy in" would cost them.
Here is the next question in the sequence:
------------------Favor-------------- ------------Oppose------------------- (Don’t know)
Allowing refugees from TOTAL Strongly Somewhat TOTAL Strongly Somewhat
Central America to seek
political asylum in the 50% 24 27 41% 18 24 9
U.S.?
Here we have a number of problems. In the preceding question, the TOTAL percentages were the sum of the Strongly and Somewhat categories (i.e. 65% = 34+31, 23% = 9+14). Here, we have 24% as Strongly plus 27% as Somewhat. Yet, the TOTAL is shown as 50% instead of the correct sum of 51%. Likewise, 18 Strongly plus 24 Somewhat under the Oppose column should add up to 42% instead of 41%. One is compelled to ask what's going on here? Do the people at Fox News not know how to add? Why would the TOTAL percentages be displayed differently?
The main problem with that question, as posed, is that it does not require the respondents to reflect on exactly what it is they understand by the word "asylum". An individual truly seeking escape from a threat of violence for their political beliefs is one thing. But, if they are simply seeking to enter the U.S. to escape poor economic conditions in their home country, it is something entirely different. Are the respondents making such a distinction? Of those coming to the southern border, what percentage fall into what group? The distinction matters because those simply seeking to improve their economic condition do NOT qualify for asylum.
The next 3 questions, if asked BEFORE the question of which candidate the respondent would choose if the election were held today, would elicit completely different results.
-----------------------Favor-------------- ------------Oppose------------------- (Don’t know)
Getting rid of private TOTAL Strongly Somewhat TOTAL Strongly Somewhat
health insurance and
moving to a government 43% 22 22 48% 17 31 9
run health care system
for everyone?
Here again, we have a math problem. 22% Strongly plus 22% Somewhat = 44%, not 43%. Yet, under the Oppose column, they've added 17% and 31% correctly. Why should anyone give this poll any credence when there are such inconsistencies in simple math?
-----------------------Favor-------------- ------------Oppose------------------- (Don’t know)
Decriminalizing entering TOTAL Strongly Somewhat TOTAL Strongly Somewhat
the U.S. without proper
documentation? 34% 18 17 57% 18 39 8
-----------------------Favor-------------- ------------Oppose--------------- (Don’t know)
Providing health insurance TOTAL Strongly Somewhat TOTAL Strongly Somewhat
coverage to undocumented
immigrants? 32% 13 19 60% 16 44 8
The last two questions show clear opposition to the positions being put forth by not only Biden and Sanders, but most of the other Democratic candidates, as compared with Trump's position.
Finally, there is Question #45, which is posed in this way: "Do you approve or disapprove of the job the U.S. Immigration and Customs Enforcement agency or ICE is doing? The answers are as follows:
Approve Disapprove (Don’t know)
45% 49 5
This is an extremely loaded question because it does not ask the respondent whether he/she is aware that ICE is following the laws passed by Congress and is conducting business in exactly the same manner as it did during the previous administration, when Obama was characterized as the "Deporter-in-Chief".
As stated previously, for the results of a public opinion poll to be useful in any way, they must seek to reflect "informed" opinion, not simply "gut" reaction to a phone call. If Questions #1, #20, and #21 were asked at the END of this poll, I have no doubt the results would be entirely different and would more properly reflect a more thoughtful response.
The most salient point is NOT to put any stock in the poll results touted in the media. They called the 2016 election wrong. It's likely they're equally wrong in stating current public opinion.
