Thank you for your interest in the matter.

1) I think even in your example model, the answer chosen by the method would still be C, the correct conclusion, for, as the author says, "The percentage of smart and dull groups choosing each answer is compared and the largest ratio of the smart to dull percentages is the Smart Vote."(emphasis added) As you see, it's not the difference (a subtraction) that matters, but the ratio: A = 50/70 ~ 0.71 B = 40/27 ~ 1.6 C = 10/3 ~ 3.3 Thus, C > B > A.

2) and 3) I don't grok totally Regression Analysis yet (dropped out College, akrasia+depression won), but he emphasises in many comments that he controls for many variables such as income, education, to avoid non-cognitive motivations (I'm not sure this is the right term, but I hope you understand me) and only get the 'smart' decisions.

Regarding biases, he once gave an answer to User:gwern who once commented on his blog, that even though it is true that even bright people are prone to some biases as everyone else, still in these cases they're a little bit less prone, so what counts is the 'trend' from the dumb to the smart. In his words:

"In cases like the sunk cost fallacy high IQ people usually make mistakes like everyone else. The point however is that on average they make them less often. Typically you see stuff like 70% of a smart group get it wrong but 90% of the dull group do. That trend, and not the % correct, is what points to the better answer.

I have a big collection of common logical mistakes people make (gleaned from Tversky & Kahneman mostly, but also many others). On all those I've tried so far - including the sunk cost fallacy - the brighter group does quite a bit better, even though most still get them wrong. Contrary to what you are saying here the Smart Vote does particularly well on such fallacies."

Regarding the question posed in the EDIT, he seems to use the General Social Survey (GSS). According to Wikipedia ,

"The General Social Survey (GSS) is a sociological survey used to collect data on demographic characteristics and attitudes of residents of the United States. The survey is conducted face-to-face with an in-person interview by the National Opinion Research Center at the University of Chicago, of a randomly-selected sample of adults (18+) who are not institutionalized. The survey was conducted every year from 1972 to 1994 (except in 1979, 1981, and 1992). Since 1994, it has been conducted every other year. The survey takes about 90 minutes to administer. As of 2010 28 national samples with 55,087 respondents and 5,417 variables had been collected. The data collected about this survey includes both demographic information and respondent's opinions on matters ranging from government spending to the state of race relations to the existence and nature of God. Because of the wide range of topics covered, and the comprehensive gathering of demographic information, survey results allow social scientists to correlate demographic factors like age, race, gender, and urban/rural upbringing with beliefs, and thereby determine whether, for example, an average middle-aged black male respondent would be more or less likely to move to a different U.S. state for economic reasons than a similarly situated white female respondent; or whether a highly educated person with a rural upbringing is more likely to believe in a transcendent God than a person with an urban upbringing and only a high-school education."

So, yes, I think it's a fairly comprehensive and diverse sample.