Political pimps - dishonest science and cooperative news media (my research)

Good morning, boys. I am laughed out of class when I suggest that science does not represent reality but here is the proof along with some of my writing on the matter. These people are political pimps and science is there little whore sometimes. Now on the topic:

With a non-ideal survey design, and a margin of error that challenges our tolerance for accuracy, the biggest problem with the survey design is that respondents were not actually New Hampshire Democratic primary voters, despite the survey’s topline title, “Survey of New Hampshire Democratic Primary Voters”. A closer look at the methodology reveals that the sampling frame accounted for “known characteristics of Democratic Party primary voters from the New Hampshire voter file” (Dyck and Cluverius 2020). According to the stated methodology, “Respondents were selected from YouGov and three other proprietary online panels”, with a total of 453 eligible candidates being interviewed before “409 were then matched down to a sample of 400 to produce the final dataset” (Dyck and Cluverius 2020).

A sample was drawn from one population to measure attitudes in another. Scientific and complex? Yes. Dishonest and misleading? Absolutely. This is an example of scientists (studying politics at the highest level, no less) spreading misinformation with a sensational survey item that one might consider as circumstantial evidence that UMass did not actually fund this research or design this study in-house, as only a yellow tabloid writer like Rolling Stone's Peter Wade could concoct such a question to be included in a public opinion study.

As described by New York Times journalist Nate Cohn, “This [characteristic-matching] step is intended to mimic probability sampling” in creating their set of respondents (DeSilver 2014). Nonetheless, UMass' Center for Public Opinion has done a harm to their analytical reputation. Sampling the attitudes of the public is often compared to drawing a blood sample; you don't need to vamp all the blood from a body to tell if a person has diabetes - you just need a little sample which is sufficiently representative. Yet one cannot make the claim that I myself am a diabetic by sampling another person's blood simply because we are both males of the same age, race or partisanship, and this sort of technique is what public opinion research has evolved to employ in 2021.

 

So how do you get from sampling to science doesn't represent reality?

 

Well they are making a claim about the attitudes of NH Democratic primary voters based on their public opinion research. In order to draw conclusions about the attitudes within a population you must actually sample that particular population. That's not what UMass did here. These are political scientists in a professional sense, not pundits, journalists, commentators, staffers, or historians. They have to play by the rules of science.

 

@Kranes56 It's sort of depressing because YouGov isn't the only firm using this technique anymore. Pew is starting to do it. Gallup does it. 538 does it now too. And researchers are relying on these pollsters to produce research based on bad survey design imo. My analogy above kind of puts it into perspective if you've never heard of these new techniques: