George,

Alan Abramowitz, in his Crystal Ball article of August 20, is a clear example of a political scientist who can take perfectly good National Election Studies (NES) data and grossly distort it. His conclusion that the Independent voter is a myth is simply not true. I have worked with NES data for well over 30 years and am currently an advisor to their study directors. I know these data well.

The question Abramowitz is referring to asks respondents the following questions: “Generally speaking, do you usually think of yourself as a Republican, a Democrat, an Independent, or what?”

If the respondent answers “Republican” or “Democrat,” he/she is then asked: “Would you call yourself a strong [Republican][Democrat] or a not very strong [Republican][Democrat]?” In this way the Republicans and Democrats can be divided into “Strong” and “Weak.”

If the respondent answers “Independent”, he/she is then asked: “Do you think of yourself as closer to the Republican Party or to the Democratic Party?” If the respondent said “Democratic Party”, Abramowitz categorized those respondents as “Ind Democrat”; if the respondent said “Republican Party, he/she was categorized as “Ind Republican”. If the respondent said “Neither,” he/she was labeled “Pure Independent”.

A serious problem with this series of questions has been known about for many years. When respondents who initially say they are “Independents” are pushed into saying which party they are closer to, they often report their current voting intention. This follow up (push) question asked of Independents should never be asked. It does not measure party identification. Party identification is supposed to be a long-term predisposition, not a measure of current voting preference for a particular candidate running that year. Note that the initial question asks “Generally speaking, do you USUALLY think of yourself as …”. If a respondent says he/she is an Independent, leave it at that. The initial response is the most accurate response.

Since the follow up question to Independents is really largely a measure of current voting intention, it is not at all surprising that Abramowitz finds in his Table 2 that very high proportions of Independent Democrats and Independent Republicans voted for the presidential candidate that they were “leaning” toward.

I have also found in my studies of this question over a great many elections that Weak party identifiers are very susceptible to being wooed by a presidential candidate of the opposite party. They, along with Independents, often move in the direction of the most popular candidate. I have estimated that, as of 2004, almost 55 percent of the electorate are “floaters” – Independents and Weak party identifiers with no ideological leanings. That is a rather sizable chunk of the electorate that can be moved in one direction or the other.

Last July, I published a major study showing exactly how many liberals and conservatives there were in the electorate in 2004 as well as how many party identifiers (and how many were both ideologues and party identifiers). Here's one of the major graphics from that study.

If you want to read the whole article, it was published in a refereed online journal called The Forum. Here is the link: http://www.bepress.com/forum/vol6/iss2/art5

David RePass (Electronic mail, August 21, 2009)