Political pundits took to the airwaves and their ink-wells after Hillary Clinton surprised the press by posting a first-place finish in New Hampshire this week. The talking heads have credited Senator Clinton’s atypical show of emotion for helping her to “pull-out a win.” Senator Obama was expected to win over Clinton in a “double-digit landslide.” But was there anything that Hillary did or said that changed the minds of New Hampshire’s astute electorate?
Allow me to suggest that the candidates did nothing to change any voter’s mind. The pollsters, instead, sampled the wrong people. The famous photograph of Harry S Truman—with glowing face—holding up the election-day front page of the Chicago Daily Tribune with its banner headline reading “Dewey Defeats Truman” is an instructive metaphor for us today.
How did the pollsters of Truman’s day get it wrong? Simple answer: Using the telephone directory to select interview participants skewed their results. Most wealthy (Republican) households had telephones in the mid-twentieth century while those of more modest means (Democrats) had fewer and were not as well represented in the survey. The November 1948 pollsters got it wrong because they did not understand that the media environment of their day created a political distinction between groups of voters.
The January 2007 pollsters of the New Hampshire primary, likewise, do not understand today’s media environment. Cell-phone-only users are not only more likely to be younger voters (under age 30), but are more likely to respond to a pollster’s questioning.
Consider the Pew Research Center for the People and the Press study, released in May 2006 and titled “The Cell Phone Challenge to Survey Research.” The full report is available online at http://people-press.org/reports/display.php3?ReportID=276. The Pew researchers discovered that “characteristics associated with age are also distinctive in the cell-only population.” Here is the error made by all political posters conducting their surveys by telephone in 2008. The Pew report reported, “Nearly three-in-ten (29 percent) cell-only respondents are married, compared with 57 percent in the landline sample.” The Pew Research Center’s report highlights other distinctions of cell-only people: “[O]nly 24 percent say they own their own home; in the landline sample, 71 percent do so. The cell-only population also includes a higher proportion of minorities, especially Hispanics (14 percent vs. 6 percent among landline users).” In addition, “The landline sample includes a higher proportion of college graduates than does the cell phone-only group (36 percent vs. 28 percent).”
The Pew Research Center for the People and the Press study includes an additional significant finding: “[H]eavy users of cell phones are more easily reached and interviewed on their cell phones than are lighter users, resulting in a potential bias on some types of measures.” Today’s media environment means that younger, unmarried, rent-paying minorities who have not completed college at the same rate as their landline-using fellow-citizens are better represented in phone-based surveys.
Nearly 60 years after Truman defeated Dewey, we are still creating invalid poll results because we rely on unsuitable samples from the targeted population. Young Americans, ages 18-29, are much more likely to use only a cell phone (48 percent according to the Pew study) than are those over the age of 64 (4 percent). Of course those in the younger demographic are more likely to be “heavy user[s] of cell phones.” Maybe they are also more likely to vote for Senator Obama?
The truth is this: Pollsters still do not recognize that voters’ use-patterns of specific media break along political lines. And that was their double-digit mistake this week.