respondents claim to have voted than actually
32. Using a statistical test to evaluate correlation
vote. There are questions as to whether this bias
within small samples, we can say the observed
affects Republicans or Democrats unevenly and
relationship is highly significant (Fischer's Exact
Test = 2.51 x 105).
whether the voting behavior of overreporting vot-
ers is significantly different from that of actual
33. Gregory Katz, "A Land at War with Itself,"
voters. ANES has undertaken extensive validity
Financial Times, October 31, 2004.
tests to go back to state, county, and municipal
voter rolls to validate whether respondents who
34. Katherine Q. Seelye, "4 Years Later, Some Voters
said they voted actually voted. But these tests have
Switch Sides," New York Times, October 30, 2004.
proved inconclusive. If anything, ANES overre-
porting is likely to bias estimates towards the elec-
35. MoveOn.org ad, New York Times, August 30,
tion winner, Bush. This would mean that any bias
2004.
would likely underestimate the observed trend of
libertarian voters away from Bush.
36. "Voters in Battleground State of Missouri
Having Very Different Views of Economic Con-
44. Landon Y. Jones, Great Expectations: America
ditions," Morning Edition, National Public Radio,
and the Baby Boom Generation (New York: Coward,
August 26, 2004, accessed through LexisNexis.
McCann, & Geoghegan, 1980), p. 287.
37. Murray and Cummings.
45. Personal communication to David Boaz,
September 1992.
38. "Who's Getting Your Vote? Reason's Revealing
Presidential Poll," Reason, November 2004, http:
46. Maddox and Lilie, figure 20, p. 88.
//www.reason.com/0411/fe.dc.whos.shtml.
47. Richard Nadler, "The Rise of Worker Capitalism,"
39. Daron R. Shaw and Paul Janowitz, "Swing Voting
Cato Institute Policy Analysis no. 359, November 1,
in U.S. Presidential Elections," paper prepared for
1999.
presentation at the annual meeting of the Midwest
Political Science Association, Chicago, IL, April 36,
48. Ryan Sager, The Elephant in the Room (New
2003, http://www.gov.utexas.edu/content/research
York: Wiley, 2006), p. 163.
_papers/midwest_903/janowitzmp sa03.pdf.
49. Gary Larson, ABC News, http://abcnews.go.
40. The observed libertarian U.S. House swing
com/sections/politics/thisweek/nascar_dads_
between 2000 and 2004 is significant at a 99 per-
040215.html.
cent confidence level (t-stat = 3.3077).
50. Jacob Weisberg, "Soccer Mom Nonsense," Slate.
41. Using a two-tailed t-test, a statistical test used
com, October 12, 1996, http://www.slate.com/id/22
to compare changes across two samples, we can say
55/.
the observed libertarian U.S. Senate swing between
2000 and 2004 is significant at a 99 percent confi-
51. In 2000 exit polls measured 14 percent using
dence level (t-score stat = 3.2348). While those data
the question, "Do you consider yourself part of
do not permit a district-by-district analysis, we note
the conservative Christian political movement,
that libertarians are spread across the United
also known as the religious right?" In 2004 the
States, with strong concentrations in the West.
question was changed to "Would you describe
Using our methodology, researchers could further
yourself as a born-again or evangelical Christian?"
refine our analysis for presidential battleground
This question yielded 23 percent of all voters,
states or other contested states and districts.
though this wording likely overestimates the
demographic. See Thomas B. Edsall, "Exit Poll
42. There is some controversy about the proper
Data Inconclusive on Increase in Evangelical
way to calculate voter turnout. The U.S. Census
Voters," Washington Post, November 8, 2004, p.
calculates voter turnout using "voting-age popu-
A07, http://www.washingtonpost.com/ac2/wp-
lation." However, this includes ineligible voters
dyn/ A32794-2004Nov7?language=printer.
such as felons. George Mason University scholar
Michael P. McDonald calculates voter turnout
52. Thomas B. Edsall, Building Red America: The New
using the "voting eligible population," which
Conservative Coalition and the Drive for Permanent
increases the percentage of voters nationally. See
Power (New York: Basic Books, 2006), p. 52.
http://elections.gmu.edu/voter_turnout.htm.
Quotations are from Edsall's interview with Dowd.
See Thomas B. Edsall, "Party Hardy," New Republic,
43. ANES overreporting of voting is a well-known
September 25, 2006.
bias in political science. When ANES researchers
conduct postelection interviews, many more
53. Mark Lilla, "A Tale of Two Reactions," New
25