In his National Journal article, “Forget the 2016 Polls: Nobody Knows Anything Yet,” S.V. Dáte writes, “At the same point in the 2012 race, just over two months before the Iowa caucuses, pizza-chain executive Herman Cain had a clear lead in Iowa, while eventual winner Rick Santorum was at 4 percent.”
Citing “large percentages of respondents who say they still have not settled on a candidate,” Dáte notes some interesting technical reasons why early polling is less influential:
Layered onto this fundamental lack of deep voter interest are the logistical difficulties in modern political polling. More and more Americans do not have home landlines anymore, only cell phones. And those numbers, by law, must be manually dialed, driving up costs. The majority of Americans, regardless of what type of phone they have, do not answer incoming numbers they don’t recognize. These factors produce a response rate in surveys of 8 percent, compared to 80 percent or so a few decades ago.
And then there are the sample sizes, often so small that the margins of error are larger than the spreads among a host of candidates. An ABC News/Washington Post poll released this weekend had Trump leading nationally with 32 percent, Carson in second at 22 percent, and then 10 candidates ranging from Sen. Marco Rubio at 11 percent down to Sen. Lindsey Graham and former Sen. Rick Santorum at 1 percent.
But because the sample size of 423 Republican respondents produces a 5.5-point margin of error, those 10 candidates from Rubio to Santorum were statistically tied.
John Dick, founder of the polling and research firm Civic Science, said such dependence on obviously imprecise surveys is actually doing voters a disservice. “It is categorically irresponsible, in my opinion,” Dick said.
As with church attendance and charitable contributions, notes Dáte, there is also the tendency of too many poll respondents to say they will vote, but don’t show up at the polls on election day. “A Fox News poll released on Sunday similarly had 79 percent of respondents saying they are likely to vote…If 77 or 79 percent of registered voters truly wind up voting in their primaries, it would shatter turnout records across the country.”
The early polls are consequential in other ways. Dáte acknowledges the power of early polls in attracting contributions and in selecting those who get to participate in televised debates, which has played a significant role in the subsequent allocation of media attention.
The polls are of interest to the candidates themselves as a way to pinpoint weaknesses with different demographic groups, define the popularity of policy positions and geographic vulnerabilities. But for those following political campaigns, using the polls to determine who is actually leading the horserace is pretty much a waste of time.