With the Republican nomination fight having concluded and the Democratic one all but decided, more and more attention will now increasingly be directed towards general election polling. After months of questionable predictive power, these presidential horse-race polls tend to gain significance around this point in the calendar; it’s particularly important as it marks the usual time both parties confirm their nominees as we enter the summer, and the general electorate begins to calcify their views of these party standard-bearers.
Yet as much as the Democratic fight is not in doubt, Bernie Sanders has chosen to continue his campaign. In doing so, not all Democrats have coalesced around their likely nominee in Hillary Clinton, as both sides are wont to do once we transition from primaries to the general election. On the other side of the partisan coin, despite serious intra-party turmoil, Republicans have settled on their nominee. For them, general election matchup numbers now start to accrue more significance. As a result, this has already begun to create an expected inconsistency when compared to how Democrats fare in these polls.
Throughout the course of the primary season, Clinton held a sizable lead over Trump in general election polls. Even in early 2016, for example, when these types of polls have much less meaning, this advantage was growing towards insurmountability, especially in tandem with Trump’s putrid favorability ratings. Yet in the aftermath of the GOP nomination’s conclusion and the Democratic one’s continuation, three Quinnipiac Polls of key battleground states for the fall that came out this morning have engendered surprise in the political world.
In the context of lopsided national polls, Clinton only led Trump by 43 to 42 percent in Florida, the same 43-42 margin in Pennsylvania, and actually trailed Trump 43-39 in Ohio. Cue the media firestorm of it suddenly being a “dead heat” and “neck and neck,” or even to “be afraid.”
In cases like this, it’s best to focus on the aggregate of surveys for the same population than a single one. But the fact that these polls deviate from means and medians of past ones for each respective state matters less. With Trump winning the GOP nomination, we should expect substantive and lasting variation from the previous polling aggregation. Rather, the problem lies in ignoring the ongoing character of the Democratic primary.
It will feel insufferable at times, but you should expect media speculation to swirl around the closeness that general election polls will show over the next few months. Yet once the Democratic fight concludes, it’s very safe to assume a large part of Democrats unwilling to support Clinton in general election matchups–the ardent Sanders backers–will shift to picking Clinton, who will restore at least a somewhat comfortable lead. To put it in concrete terms, evaluating general election polls right now leads to a very incomplete picture.
Additionally, as political scientist Alan Abramowitz noted, there seems to be some concern with the samples the Quinnipiac polls gathered. I decided to check for a bias for whites and Republicans occupying a larger share of the electorate by comparing their portions to those in the past three elections.
Here’s the comparison for white percentage of each state’s electorate in the 2004, 2008, and 2012 elections to that in the three 2016 Quinnipiac polls for the same states:
And then the same for the portion Republican self-identifiers assume in each state’s electorate:
So it’s very clear that in capturing the percentage Republicans take on in each electorate, the Quinnipiac polls in fact follow the trends–declining ones–of past elections past. The estimation for white share of the electorate is where they slipped up: all three swing states reveal a declining share of white voters over the last three elections, yet the Quinnipiac polls break with this important trend that has manifested at a national level as well. White voters vote Republican more than they do Democrat, and thus contributes to Trump’s likely inflated lead. Even during a juncture in the electoral season where he has the greatest advantage–having sealed his nomination unlike Hillary–Trump still faces a misleadingly small deficit against Clinton.