I just wanted to paste a post from 538’s comments section about this Ziegler poll thing

From "Personal":

A truly bizarre thing happens when survey makers try to enter the field of experimental research. They fail, miserably.
First of all, the results actually make sense from two different perspectives. The vast majority of American voters are uninformed. I ran a set of voter knowledge studies for my lab for the 2004 elections and found that 90% of people did not know the NAMES of the two legislative branches, much less who was leading them. Thus, the results actually show a slightly higher amount of knowledge then would be expected from voters (with something like 88% of people not recollecting Harry Reid).
Moreover, neuroscience has repeatedly shown that voter’s are much more likely to remember certain things based on their innate set of beliefs and biases. I.e. an Obama supporter is, of course, more likely to remember bad things about McCain, and vice versa. This occurs because of psychological constructs like the fundamental attribution error, personality biases, and even things as basic as race, socioeconomic status, etc.
Because of these two factors, the results obtained in the study would be expected…
However, what I find most frustrating about all the coverage of this survey and Nate’s subsequent interview is the lack of comments about the true, fundamental issue that is in this study: the inference of causation from what tries, and fails, to be correlation.
As a psychological researcher, it is impossible for a study which simply asks people questions about what they know to infer that media coverage was the cause (which is the principle purpose of this study as I understand it.) You must have such amazing things as "control groups" and "causation questions." Moreover, even with those things it is difficult to ascertain cause. It remains, just as likely that, even assuming the data is correct – which may be a stretch, the discrepancy in knowledge arises out of the inherent biases or low voter knowledge that I just spoke of. Or, possibly, people are just more likely to attribute negative things to the opposition (notice most negative things were attributed to Palin, rather then McCain, following data plots that indicate more people have an unfavorable impression of Palin then McCain). It is possible that people just think badly of Palin, and not knowing what is true, are more likely to attribute bad things to her. Whereas with positive things about people, they just don’t remember (this follows with psychological research that shows people are more likely to respond to negative then positive stimuli under certain conditions).
Any of these explanations are in line with the data collected. It is…incredible to me that Zogby would stand by these fundamentally flawed data and the inferences that Ziegler tried to draw from them. It is also incredible that people are assuming that "media coverage" is the cause.
http://www.fivethirtyeight.com/2008/11/final-words-for-now-on-zieglergate.html#comments

Anyway, that’s enough about that.

Advertisements