This is the first post-debate edition of this post.
Background: As in 2004, 2008, and 2012, I’m indulging in a favorite pastime, namely, playing with numbers to project the outcome of the general election in November. Specifically, I’m trying to estimate the popular and electoral vote for president, and the balance of power in Congress. I begin with the popular vote, then use statistical relationships that I’ve derived from past elections to translate the popular vote split into electoral votes and changes in the numbers of House and Senate seats held by Republicans.
This year, my main method of projecting the popular vote for president is based on a Reuters poll that’s been consistently pro-Clinton. This method doesn’t paint a worst-case scenario for Trump, but it paints bad-case scenario. It’s my way of compensating for my loathing of Clinton, who would perpetuate and extend the damage done to America by Barack Obama.
The latest version of the Reuters poll is now three days old, and doesn’t reflect Clinton’s post-debate bounce. The poll nevertheless favors Clinton. Here’ my interpretation of it:
- Clinton will win 50.4 percent of the two-party popular vote to Trump’s 49.6 percent.
- The outcome of the electoral vote is too close to call.
- The GOP will lose 5 House seats, retaining a majority of 242-193.
- The GOP will lose 1 seat in Senate, leaving that chamber with 53 Republicans and 47 Democrats (counting the so-called independents as Democrats).
My projection of the two-party popular vote is close to the one that’s implicit in the latest election forecast at Nate Silver’s FiveThirtyEight. There, Silver projects a 46.6-44.2 edge for Clinton, with most of the balance (9.2 percent) going to Gary Johnson (7.8 percent). Clinton’s 46.6-44.2 edge translates to 51.3 percent of the two-party popular vote (46.6 divided by 46.6 plus 44.2). Rasmussen’s most recent poll points to a 50.6-49.4 split in the two-party popular vote — in favor of Clinton. In other words, unless the post-debate bounce continues and Clinton pulls away, the race is essentially tied, given the margin of error in the polls.
However, barring a game-changing event (e.g., a major terrorist attack on American soil), the “internals” of the Reuters poll suggest that Trump is more likely than Clinton to gain support as election day approaches:
- The “don’t know” vote is mainly a hiding place for shy Trump supporters; about three-fourths of the “don’t know” contingent prefer Trump to Clinton. “Don’t know” will, by definition, dwindle from its current level of 9.1 percent to 0 percent by election day. I’ve already credited the gain to Trump in my estimate of the two-party popular vote. But there’s more…
- Johnson is polling at 7.9 percent, which is almost an order of magnitude higher than his 1 percent of 4 years ago and the all-time Libertarian Party high of 1.1 percent recorded by Ed Clark in 1980. As voters choose not to waste votes on Johnson, Trump will pick up 3 votes for every 2 that go to Clinton. That’s consistent with what’s happened in Silver’s projections: Trump gains relative to Clinton when Johnson’s share drops.
- “Other” is now polling at 3 percent, which is about triple the “other” vote in recent elections. The “other” vote is trending toward 2 percent, which will help Trump. He will gain 1 vote from every “other” supporter who abandons “other.” Those voters will be replaced by current Clinton supporters who will abandon her, thus adding to Trump’s popular-vote margin.
- Clinton holds a slight edge among those who say that they won’t vote, now 7.6 percent. But it looks as if the “won’t vote” contingent will stay about where it is, so there’s some more bad news for Clinton.
- Jill Stein is currently polling 3.7 percent. If voters abandon her, as some will, Clinton will pick up a lot more votes than Trump. But…
- The bottom line is that Trump’s prospective gains are greater than Clinton’s.
There’s also the all-important electoral vote, where GOP candidates have a slight built-in advantage. Electoral votes are weighted in the direction of the less-populous States, which — more often than not — go Republican. Based on the results of presidential elections from 1948 to 2012, a GOP candidate can expect to win a bare majority (270 electoral votes) with 49 percent of the two-party popular vote. (G.W. Bush, for example, won 271 electoral votes with 49.7 percent of the two-party popular vote in 2000.)
Now, for a look at some indicators of the popular-vote outcome:
The indicators measure Clinton’s net advantage (or disadvantage) in the Iowa Electronic Markets (IEM) Winner Take All (WTA) market; the RealClearPolitics “poll of polls,” adjusted to eliminate pro-Clinton bias (see this post); Rasmussen’s poll, which (currently) is updated weekly; two national polls that are updated daily, Reuters and USC/LA Times; and Rasmussen’s net approval rating for Obama (percentage of respondents strongly approving of his performance minus the percentage strongly disapproving). I include the last item because perceptions of Obama’s performance are likely to rub off on Clinton.
Four indicators have recently become more favorable for Clinton or have leveled off after trending down: IEM WTA, the Reuters poll, the USC/LA Times poll (though it’s still unfavorable to Clinton), and Rasmussen’s net approval rating of Obama. Two indicators still seem to be trending against Clinton (despite post-debate upticks): the adjusted RCP “poll of polls” and Rasmussen’s presidential-race poll.
I’ll continue to update this post as the polls are updated. Stay tuned.