GAYETI SINGH | 6 NOVEMBER, 2016
NEW DELHI: In the months, and now days, leading up to the November 8 United States presidential elections, the polls have constantly thrown up varied numbers. At the time of writing, Democrat candidate Hillary Clinton is in the lead, but within a span of a week saw this lead go from an unassailable 12 point margin in later October, to shrinking as dramatically as low as 1 point, with rival Republican candidate Donald Trump then taking a 1 point lead in the ABC/WashingtonPost poll for the first time a few days ago. At the time of writing, Clinton is back up again now to 1.9 points ahead of Trump.
The story has been the same throughout, with polls seeing Clinton and Trump going from being virtually neck and neck before the presidential debate, to Clinton’s lead climbing up and down. The data therefore suggests that a large number of American voters are wavering in their decision to back either candidate.
Pollsters explain the wide swings through parallel political developments: Trump’s sexual allegations scandal saw Clinton climb to a 7 point lead, Clinton’s email drama, in turn, has seen her lead shrink dramatically. The reality, however, may just be that the swings are driven not by changes in voter preference, but by who responds to the survey. As such, the 2016 US presidential race may just be a lot closer than one expects.
Benjamin Lauderdale of the London School of Economics and survey research expert Doug Rivers point to this possible explanation when they say, “We believe that most of the bounces seen in surveys this year represent sampling noise that can be reduced or eliminated by adopting by better statistical methodology. We risk a repetition of 2012 where polling swings were largely statistical mirages. The convention and first debate bounces in 2012 were mostly the consequence of transitory variations in response rates. Fewer voters were changing their minds than were changing their inclination to respond to surveys.”
Most polling methodology uses independent samples, meaning that the respondents change from week to week. If the surveys were tracking the same sample of respondents and we saw the swings that are in play, a change in voter preference would be the likely explanation. However, current methodology makes it impossible to distinguish change in individual vote intentions from changes in sample composition from week to week. As an article by Lauderdale and Rivers notes, “ It is possible that five percent of the electorate switched from Clinton to Trump over the past week (decreasing Clinton’s lead by 10 points). But it’s also possible that nobody switched and apparent swings are due to differences in sample composition.”
A finding from Lauderdale and Rivers’ work that lends support to the above is that Clinton and Trump supporters’ willingness to participate in the polls varied by a significant amount depending upon what was happening at the time of the poll. “When things are going badly for a candidate, their supporters tend to stop participating in polls. For example, after the release of the Access Hollywood video, Trump supporters were four percent less likely than Clinton supporters to participate in our poll. The same phenomenon occurred this weekend for Clinton supporters after the announcement of the FBI investigation: Clinton supporters responded at a three percent lower rate than Trump supporters (who could finally take a survey about a subject they liked).”
Similarly, data collected by the Institute for the Study of Citizens and Politics panel, corroborates the above by showing that candidate preferences have remained relatively stable over a period of months. As an article by one of the researchers, Dan Hopkins, notes, “The table below shows the responses for the 1,227 respondents who completed our October poll, with the rows showing their candidate support in January and the columns indicating candidate support this October. When asked about a matchup among Clinton, Trump and other candidates, a significant majority of the respondents gave the same answer both in both months: 37.2 percent backed Clinton both times, 30.9 percent backed Trump both times, and 12.7 percent opted for another candidate or said they wouldn’t vote both times. In all, that means that around 80.7 percent of respondents gave the same answer in both surveys. (Of course, that doesn’t rule out the possibility their support shifted in between — just that they settled back where they were.)”