Why the polls missed Hillary’s big win in California and the dilemma she faces
Hillary Clinton won California big, by 12 percentage points, in what figures to have ended whatever slim chance Bernie Sanders had in the Democratic race. This race was assumed by many to be much closer based upon late polls that had her leading Sanders by a shrinking 2 percent. How wrong they were. Why? In all fairness, an outlier poll, by the LA Times/USC, had Clinton winning by 10 percent. And the reason they called it more accurately is that they ultimately relied on their model of the “likely voter” rather than their sample of eligible voters. By the way, when the “eligible voters” sample was used in their poll, it yielded results similar to the other polls showing the race a virtual dead-heat. So the more refined “likely voter” sample generated much more accurate results.
A little background will help in understanding why different samples resulted in such wildly varying results, and why polling techniques that worked in several prior primaries weren’t appropriate for the California race.
You will recall that the very early primaries saw most polls underestimating Sanders vote. This was mainly because they had used samples that relied on voting patterns in those states from prior elections. That generally assumed a much lower turnout by the independents and first time voters. The same assumption by the way underestimated Trump in the Republican race.
Pollsters learned from their early mistakes and adjusted their later polls to account for the increased participation by independents and new registrations. Well, in many of the earlier primaries, this increased their accuracy. In California, however, un-registered voters couldn’t simply show up at the polls and choose, on the spot, which party’s primary they would vote in as they did, say, in New Hampshire. In California, an independent or first-time voter had to first request to vote in one of the partisan primaries. This extra step reduced Sanders vote totals.
Going into the real heavy campaigning period in the California Primary, one would have expected Hillary to run strong there. After all, she had a much larger and experienced “boots-on-the-ground” apparatus, many of whom campaigned for her back in the 2008 race. Others joined the Hillary band-wagon before Bernie announced his candidacy. Even more important, California was the model of population diversification with 38 percent Hispanics and 14.5 percent African Americans. Hillary had swept nearly every state with large numbers of minorities. Sander did best in mostly white, smaller states. So California figured to be Hillary country. And that’s exactly what happened. We were only misled by polls that assumed voting turnout based upon the earlier states where it was much easier for young first-time voters and independents to vote in the Democratic Primary. The importance of the assumptions contained in sample construction cannot be overstated. Making simple changes, such as what percentage of the sample to be polled were landlines and what percentage were cell phones, and how many young first-time voters would actually make the effort to register and then to vote could affect the accuracy of the polls. Historically, the young first-time voter turnout is very low. In this year’s election, they turned out in larger numbers, and they went by a large majority for Sanders in the Democratic Primary and Trump in the Republican one. This lead many pundits to conclude that the anti-establishment attitudes that had risen in recent elections reached a crescendo this year. But even with that motivation, their turnout was higher in the states where it was easier for non-party regulars to choose which party primary to vote in. In California it wasn’t that difficult to choose to vote in one party primary or the other, but it did involve one extra step, and that apparently deterred some of Sanders supporters, many of whom were first-time voters.
Sanders has asked to meet with President Obama, who will almost certainly stress the need for party unification and for Sanders help in keeping his voters in the Democratic column in order to defeat Trump. I believe Sanders shares that same goal, but he still wants to continue his fight up to the convention in order to move the platform closer to his more left agenda. Hillary could probably give him assurances right now that the platform would contain many of his leftist agenda items and gain his immediate support. But here’s the conundrum: Pushing the party platform to the left will make it almost impossible for her to gain support from the Republicans who, at this time, can’t stand the idea of voting for Trump. Party platforms are generally worth not much more than the paper they are printed on. But it could saddle Hillary with positions that can cost her the election. Is it better to win a platform fight and lose the election or win the election and then actually affect the direction the country goes in? Sanders and his supporters will have to weigh that question. To accomplish this is quite a challenge. It will be interesting to see what effect President Obama will have on Sanders.