New York Times/CBS NEWS Public Opinion Poll Bias
Methodology is explicitly designed to favor a subset of the poll choices
When conducting a multiple-choice poll, the choices must generally be rotated/randomized before being presented to each respondent. If they always appear in the same order, certain choices will gain a significant advantage due to their position in the list.
Following is an excerpt from the most recent New York Times/CBS News poll, showing question #11 (with highlighting added):
They have decided to rotate some of the names -- so obviously the poll designers have an idea that it's important to do so -- yet only three of the candidates are rotated, and their names always appear first in the list.
When I first read question #11, I speculated to myself, "Maybe all of this is not as nefarious as it seems. Perhaps they rotated only three names due to limitations in the polling system; perhaps they are relying upon a malfunctioning computer prototype from the 1940s, incapable of counting past three for some obscure reason." That desperate hope to save some credibility for the poll operators was cruelly dashed as I saw the discrepancy in #20:
We see here that the Republican question was biased with respect to five selected candidates, confounding any guess about computers that only count to three (but admittedly leaving open the possibility that some pollster's attic was ransacked, between question #11 and #20, to uncover an antique with an unforeseen capacity of five).
I will not attempt to analyze these questions' bias any further, as it is really too blatant to even need analysis. I will, nonetheless, leave you with a few optimistic words.
Opinion polls, like this one, form a privately administered first stage of the official election, deciding who will get the media coverage required to win votes.
These polls determine who gets excluded from debates. These polls are trumpeted by news anchors as respectable, reliable predictors which can separate the wheat from the chaff. These polls are, essentially, a tool to narrow the field of Presidential candidates before an election has even occurred. And they're rigged.
Based on publicly released poll results from January 9-12, 2008, available at:
This also applies to polls from December, 2007 and November, 2007.Please use this form for any comments or corrections.