It is one thing to say that something is broken; it is quite another thing to fix it. In such a case, if it ain’t broke, don’t fix it doesn’t cut it. Any pathological fear of change must give way or the brokenness must be endured. During the last half of 2011, over a year before the U.S. presidential election, the election season was already in full swing. Without any primaries or caucuses, the media and “debate” (i.e., talking points) organizers divided the Republican candidates into two tiers. Besides being an artificial dichotomy given the spectrum of support revealed in polls, that they were being used to prioritize among the candidates in the “debates” and more generally in terms of electability is problematic.
Most importantly, a poll is not an election. To relegate some candidates while privileging others (even crowning a “front runner”) ahead of any primary or caucus is without democratic legitimacy because polls have no decision-making authority. In other words, for the media (and “debate” organizers) to rely on polls to discriminate between candidates both in terms of electability and in allotting attention deprives the electorate at the polls or caucuses to break up the pack. Indeed, even aside from the fact that a poll does not constitute an election, relying on polls is problematic.
For one thing, relying on a poll can give the false impression of permanent preferences. Bill McInturff, a Republican pollster who directed the WSJ/NBC poll with Peter Hart in October 2011, remarked, “How quickly candidates have risen and then, like a soufflé, how quickly they’ve fallen back down.” A poll is at best a snapshot, and is thus obsolete even on the next day. Furthermore, the proliferation of cell phones in lieu of LAN lines and of do-not-call lists makes the “science” of polling less exact than even it portends. Indeed, to rely on the telephone in conducting surveys introduces a certain bias. For example, how many homeless people have phones? Less obviously, how many survey callers get through to corporate executives who have gate-keepers poised to divert such calls?
Even if polling were capable of giving an accurate snapshot, the polls would have to be interpreted correctly. That obvious error can be perpetuated even by a media heavy-weight like The Wall Street Journal should give us all pause in whether we should rely on the reporting of polls. For example, the October 13, 2011 headline reads, “Can Vaults to Lead in Poll.” The WSJ/NBC News telephone poll of Republican primary voters that month put Herman Cain at 27% and Mitt Romney at 23 percent. Unfortunately for the poll and for the headline, the margin of error of the poll is 5.35 percentage points. Now, math is not my strong suit by any means, but the last I checked, 4 (27-23) is less than 5.35. The poll itself does not justify the headline.
Given the margin of error, Romney
could claim to be in the lead! This
conclusion too is within the poll’s result. In fact, Romney in the lead is just
as valid a conclusion from the poll. So it is indeterminate as to which of the
two candidates is in the lead. This why the newspaper's headline is dogmatic.
Within a margin of error, no particular result is privileged over any other. To
pick out one “result” within the margin of error is dogmatic in the sense of
being arbitrary, and yet it is done all the time. Given the graphic above,
readers who do not understand statistics and surveying as a social-scientific method are apt to fixate on
the 27% and the 23% as if these figures were engraved in stone.
Most importantly, a poll is not an election. To relegate some candidates while privileging others (even crowning a “front runner”) ahead of any primary or caucus is without democratic legitimacy because polls have no decision-making authority. In other words, for the media (and “debate” organizers) to rely on polls to discriminate between candidates both in terms of electability and in allotting attention deprives the electorate at the polls or caucuses to break up the pack. Indeed, even aside from the fact that a poll does not constitute an election, relying on polls is problematic.
For one thing, relying on a poll can give the false impression of permanent preferences. Bill McInturff, a Republican pollster who directed the WSJ/NBC poll with Peter Hart in October 2011, remarked, “How quickly candidates have risen and then, like a soufflé, how quickly they’ve fallen back down.” A poll is at best a snapshot, and is thus obsolete even on the next day. Furthermore, the proliferation of cell phones in lieu of LAN lines and of do-not-call lists makes the “science” of polling less exact than even it portends. Indeed, to rely on the telephone in conducting surveys introduces a certain bias. For example, how many homeless people have phones? Less obviously, how many survey callers get through to corporate executives who have gate-keepers poised to divert such calls?
Even if polling were capable of giving an accurate snapshot, the polls would have to be interpreted correctly. That obvious error can be perpetuated even by a media heavy-weight like The Wall Street Journal should give us all pause in whether we should rely on the reporting of polls. For example, the October 13, 2011 headline reads, “Can Vaults to Lead in Poll.” The WSJ/NBC News telephone poll of Republican primary voters that month put Herman Cain at 27% and Mitt Romney at 23 percent. Unfortunately for the poll and for the headline, the margin of error of the poll is 5.35 percentage points. Now, math is not my strong suit by any means, but the last I checked, 4 (27-23) is less than 5.35. The poll itself does not justify the headline.
Given the margin of error, Romney
could claim to be in the lead! This
conclusion too is within the poll’s result. In fact, Romney in the lead is just
as valid a conclusion from the poll. So it is indeterminate as to which of the
two candidates is in the lead. This why the newspaper's headline is dogmatic.
Within a margin of error, no particular result is privileged over any other. To
pick out one “result” within the margin of error is dogmatic in the sense of
being arbitrary, and yet it is done all the time. Given the graphic above,
readers who do not understand statistics and surveying as a social-scientific method are apt to fixate on
the 27% and the 23% as if these figures were engraved in stone.
Moreover, typically people apply a false sense of exactitude when numbers are reported. It is like driving from Pittsburgh to Indianapolis and remarking while passing Youngstown, we will arrive in Indy at 5:45pm. Airline pilots in particular enjoy exercising the error of exaggerated exactitude by announcing the arrival time for the destination even from the departure airport before takeoff. At New York’s Kennedy, a pilot might announce, “we will touchdown in Paris at 6:47a.m. tomorrow morning.” Not 6:48a.m. Not 6:46a.m. Rather, 6:47a.m. on the dot. On your next flight, you might point this error of excessive exactitude out to your pilot (after you have landed—you wouldn’t want to provoke a nervous breakdown before you are back safely on the ground at the appointed time).
Besides the problems in interpreting and reporting particular polls, relying on them before the first primary or caucus is problematic from the standpoint of democracy itself. To the extent that Cain’s 27% “front-runner’s status,” anointed by the October poll more than a year before the presidential election, impacts or detracts from the electoral choices of the primary or caucus voters, the democratic process itself suffers. In other words, introducing error into an electoral process does not bode well for a democratic system or a specific electoral outcome. Moreover, willowing down a field of candidates by polls circumvents a democratic system wherein elections are the means by which decisions are made with respect to candidates for public office. In effect, pollsters, commentators and news editors (and executives) have supplanted, or at least jumped the gun on, the primaries and caucuses. The elongated electoral “season” (measured in years) even relative to the primaries and caucuses has provided the vacuum in which the “early decisions” can be made by (unelected) pollsters, commentators, and journalists. The U.S. presidential election process is broken. Whether anything will be done to fix the process remains to be seen. I’m not holding my breath, given the magnitude of the problem and the headlines. Indeed, saying that the process in the fall of 2011 is democratically illegitimate would be like saying that the emperor is wearing no clothes. Resisting the urge to make a joke at Gov. Christie’s expense (he has said the jokes have to be funny at least—he is indeed a class act), I will rest my case here—hoping that I have adequately reserved my readers’ respect.
Source:
Neil King, Jr. and Jonathan Weisman, “Cain Vaults to Lead in Poll,” The Wall Street Journal, October 13, 2011.
