When you receive an unexpected call from a private number or 1-800 number, do you answer the phone? Most people don’t, and those who do are hardly a representative sample of the American population. Yet the results of all major political polls are based on the assumption that the 9 percent of us who answer the phone are perfectly representative of the 91 percent who don’t.
According to a report from the Pew Research Center, only about 9 percent of Americans answer the phone and respond to opinion polls. When response rates fall this low, polls tell us less about public opinion than about who answers the phone.
It gets worse. Do Democratic and Republican voters have, on average, the same numbers of phones? If Republicans have more phones than Democrats, they’re more likely to be represented in telephone polls (and, of course, vice versa).
To read more articles by Salvatore Babones and other authors in the Public Intellectual Project, click here.
Pew is a first-class nonprofit research organization that puts a premium on getting the numbers right. Most major political polls are quick-response jobs for impatient commercial organizations. Pew’s 9 percent response rate is likely better than that of any major political poll. No major political poll reports its non-response rate.
The Politico/George Washington University poll conducted by Lake Research Partners and the Tarrance Group publishes 329 pages of detailed poll results without a single mention of non-response. The Washington Post/ABC News poll conducted by Abt SRBI provides somewhat less detail and still no mention of non-response. There’s also no mention of response rates in the detailed results released by the Public Policy Polling poll.
Other polls do no better. The Reuters poll conducted by polling firm IPSOS makes no mention of non-response, nor does the UPI poll conducted by CVoter. To its credit, at least Gallup acknowledges in its (literally) small print that “practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls.” It gives no further explanation of the potential for non-response biases.
Many of these polls, including the Rasmussen poll, use – no kidding – computerized robo-callers to poll potential voters. That’s right. Many of the poll numbers you see quoted in the news are based on data collected by robo-callers.
The truth is that we have no idea who is ahead in the presidential race. Opinion polling has entered uncharted territory. In just the last 15 years, Pew’s response rates have dropped from 36 percent to 9 percent. Rumors are that commercial political polls sometimes have response rates that are less than 1 percent. The fact that no major political poll reports its response rate is not reassuring.
We’ve been here before. In 1936, the Literary Digest poll tipped Republican Alf Landon to win a landslide victory over Democrat Franklin Roosevelt. With 2.4 million respondents (compared to 1,200 for a typical poll today), the poll had essentially zero margin of error. Nonetheless, the poll was dead wrong, because its 2.4 million respondents were not a truly random sample of the US population.
A decade later, polling was much better developed, and the 1948 elections were covered by professional polling companies. National polls confidently predicted that Republican Thomas Dewey would handily defeat Democrat Harry Truman. In one of the most famous photos of the 20th century, Truman is seen holding aloft a newspaper bearing the headline “Dewey Defeats Truman.” The problem: bad sampling, again.
In 2012, we’re right back where we were in 1936. In 1936, the problem with telephone polls was that Republicans were much more likely to have a phone than Democrats.
In 2012, it may be that Republicans have more phones than Democrats, or are more likely to answer the phone than Democrats. It beggars belief to suggest, as the polls do, that half of all Americans in 2012 are going to vote for a multimillionaire private equity executive for president. Either way, we’ll know next week.
Or maybe we won’t.
The only polls that have consistently high response rates are exit polls. These are the polls conducted outside voting stations immediately after people have voted. Exit polls are the gold standard of political polling.
The Washington Post has reported that the major news networks plan to cut back dramatically on exit polling in 2012. This is a problem because with many states moving to paper-free electronic voting, exit polls are the only way to detect errors and fraud. Without a paper trail and with no solid polling, there will be no way to know that the results reported by private voting machine operators are in fact correct.
Although not designed to catch errors and fraud, exit polls are well-suited to this purpose, as Jonathan Simon points out on BuzzFlash. Victoria Collier writes in Truthout that a lack of exit polls opens the door to vote-rigging, especially in an environment in which most voting technology companies are owned and operated by partisan (Republican) firms.
In a scenario eerily reminiscent of Jeb Bush’s role in delivering Florida to his brother in the 2000 election, it turns out that the Romney family are major investors in one of the nation’s leading voting-machine companies. Ownership is not itself evidence of an intention to commit fraud, but Republican domination of the voting-machine industry combined with Republican advocacy of electronic voting certainly gives solid grounds for suspicion.
Exit polling provided strong statistical evidence of electronic vote fraud in Ohio that delivered the 2004 election to George Bush. Without strong exit polling, fraud (or even honest errors) will be impossible to detect in the future. Telephone polls just aren’t robust enough to provide a check on the integrity of our voting systems.
The lack of effective political polling is more than just a problem for political junkies in need of their daily news fixes. It skews our knowledge of the political process and ultimately undermines the confidence we can have in American democracy. Whether or not anyone has ever committed large-scale electronic vote fraud, if we don’t have the tools to detect it, eventually someone will.