What Went Down In The April 26 Primary Elections

So do tonight’s results essentially make Indiana make-or-break for #NeverTrump?

The farther south you go in Delaware, the better Trump is performing. Trump is taking just 51 percent in New Castle County (which is a majority of the state’s population), but farther south, he is winning 64 percent in Kent County and 72 percent in Sussex County (which are where most GOP primary voters live). This shouldn’t be a big surprise. In 2010, Kent and Sussex County’s GOP voters were responsible for Christine O’Donnell, essentially handing a Senate seat to Democrats.

Exit polls from tonight are telling us some things about the divided hearts and minds of the GOP electorate, and unlike most years, those hearts and minds have very definite feelings about the Republican nomination process itself. According to exit polls in Pennsylvania, Maryland and Connecticut, pretty much all of Trump’s supporters said the person with the most votes, not the most delegates, should be given the nomination (the opposite of the party’s official rules); six in 10 supporters of Kasich and Cruz said that those with the most delegates, not votes, should get the nomination. Another interesting trend: this is a spite-motivated GOP electorate. A quarter of Cruz supporters and a third of Kasich’s voters said they were voting not for their candidate per se, but “against his opponents.”

It does not look like Clinton will sweep at this point. Sanders is currently running 20 percentage points ahead in Rhode Island, although none of Providence is reporting. Even if Sanders wins in Rhode Island, only 24 delegates were at stake.

Fox News has declared Clinton the winner in Pennsylvania.

https://twitter.com/GenderWatch2016/status/725122513483735041

As we watch Connecticut’s Democratic primary, keep in mind there should be some major regional differences in the vote. Clinton should do well in the southwest part of the state and around Hartford. Sanders holds a very small lead right now, but a lot of the southwest vote is out.

So far in Connecticut and Rhode Island, Clinton is winning upscale white areas (Portsmouth, Rhode Island and Salisbury, Connecticut) and heavily minority areas (Hartford/Bridgeport, Connecticut and Central Falls, Rhode Island), but losing just about everywhere else to Sanders. Sanders is well ahead in Rhode Island, but none of Providence is reporting yet. Like Massachusetts almost two months ago, both states look extremely competitive early in the night.

All five states have now been called for Trump, and the margins are large enough that he could sweep nearly every delegate, except in Rhode Island where allocations are highly proportional. Suppose that Trump finishes with 100 of 109 delegates tonight, for instance, not counting any uncommitted delegates in Pennsylvania. That would put him on pace for 1,209 delegates, based on the state-by-state projections that our expert panel issued last month, close enough that he could probably get over the top to 1,237 with uncommitted delegates from Pennsylvania or elsewhere. True, that panel had him winning Indiana, which is a long way from assured. But the point is that Trump has made up the ground he lost in states like Wisconsin and Colorado and put himself in the strongest position he’s been in since March 15.

As more results are reported, we’ll be posting election returns for today’s GOP primaries from each state and (where relevant) congressional district, along with what the delegate allocation would be based on those results. The delegate counts will change throughout the night as more results come in; they should be considered preliminary, as an illustration of how delegates would be awarded if the current results didn’t change. The results come from each state’s election agency. (Connecticut and Rhode Island aren’t reporting results by congressional district, so we’re estimating them based on past voting patterns.) While Pennsylvania’s GOP voters do choose delegates in each congressional district, they’re not bound to a presidential candidate, so we’ll only be showing the statewide totals in these tables (we may post the results of those delegate elections separately as they become available).

OK, Sanders is fired up about process. He just criticized the New York primaries because independents were unable to vote. Sanders does tend to do better with independents than with Democrats, as the political scientist Hans Noel points out. And his speech – like, frankly, a lot of American political discourse – makes far too much of the role of independents, who are not so much represented among voters as among non-voters. What’s more interesting about Sanders, though, is why independents who do vote – or try to – in primaries are drawn to him. Most political scientists agree that politically engaged independents tend to be closet partisans who mostly vote for one party or another. But clearly there’s a difference between them and Democrats when it comes to this year’s choice. This could be about process – about Sanders as an outsider to the party and a challenger to establishment figures like Clinton. But it could also be that independents are more likely to dislike Clinton’s foreign policy record, especially on Iraq. Last year I discovered that in poll after poll, independents were more hands-off than Democrats on a variety of questions. In this post, I outlined the range of global threats where independents thought the best U.S. option was to stay out.

NBC News declares Clinton and Trump the winners in Delaware.

Trump has won Maryland, but the question is whether he will sweep all 38 delegates by winning all eight congressional districts. If there’s a district he’s likeliest to lose, it’s Maryland’s 8th District, which takes in the well-educated Montgomery County suburbs and should be favorable territory for Kasich. More than 53 percent of non-Hispanic whites 25 and older hold at least a bachelor’s degree; that’s usually a poor sign for Trump. If there’s another district where Kasich could do well, it’s probably the 3rd District, which was one of the ugliest gerrymanders of the 2012 redistricting cycle. It snakes from the D.C. suburbs to Annapolis, ultimately forming a ring around Baltimore, and it’s the second best-educated district in Maryland. A Trump loss here is unlikely, but not out of the question. If there’s any place Cruz could possibly surprise, most Marylanders think it’s the western 6th District. The 6th stretches all the way from the rural, very conservative western tip of the state, where Cruz could do reasonably well, to some of D.C.’s most rapidly diversifying outer suburbs, such as Germantown (now home to an outpost of my favorite area Korean BBQ joint, Honey Pig), where Kasich could hold down Trump’s vote share. It’s still a long-shot, though.

NBC News declares Trump the winner in Rhode Island.

Keep an eye on Rhode Island. Right now Trump is barely under 67 percent of the vote, and Cruz is just at the 10 percent threshold to win delegates. If Trump gets at least two-thirds of the vote in each congressional district, he’ll get two versus the one delegate from each district he would otherwise. If Cruz gets below 10 percent statewide, Trump will also get more delegates.

Yes, I know that Trump was greater than a 99-percent favorite in Maryland, Pennsylvania and Connecticut tonight, but his margins are still extremely impressive. Based on exit polls, it looks as though Trump will get into the mid-50s in each state, something he’s previously done only in his home state of New York. A few weeks ago, Trump was only at about 40 percent in polls of Pennsylvania and Maryland, so I don’t think you can ascribe all of tonight’s performance to demographics. It seems likely to me that something — perhaps Trump’s argument about the system being “rigged” against him — shifted undecided voters toward Trump or at least encouraged his voters to turn out.

A lot has been said about the racial divide in the Democratic primary; less so about the big economic divide among white voters. That’s clearly the case in the fairly white state of Connecticut tonight. Clinton is losing by 5 percentage points among voters making less than $30,000, but she is winning by 37 percentage points among those who make greater than $200,000.

At times throughout the past couple of weeks, we at FiveThirtyEight, along with a lot of other news organizations, have referred to today’s voting as the “Acela Primary.” That’s produced some backlash: former Obama advisor Dan Pfeiffer, for example, tweeted: “calling it the Acela Primary is a great way to highlight the great chasm between political/media class and the rest of America.” I take issue with that backlash for a couple of reasons. First, the people who live in the states that are voting today probably know what Acela is, so it’s not media elitism we’re talking about here so much as geographic parochialism. Yes, the Acela is the high speed train, aka, more expensive, but “Acela” is a bit more sonorous than “Northeast Regional” (the slow train), right? Second and more importantly, nicknaming any primary days is dumb, so why single out “Acela” in particular? “Super Tuesday,” “Super-duper Tuesday” — the internet feels a need to come up with a name for every single day deemed “crucial” for the election. The cute-sification and Twitter noise surrounding the naming is a bit grating from a sheer aesthetic point of view, but also frames election days in the same way we do sports contests. They’re not the same thing at all. Of course, my writing this is inherently unhumorous and hypocritical — I’ve dropped the terms and my bread is partially buttered by participating in, ahem, blogs of these events and podcasts dissecting them after the fact — but a girl can bite the hand that feeds her once in a while. Naming primaries, Acela or otherwise, is both irritatingly in-crowdy, and trivializes the votes of citizens not in states deemed “important.” It’s a bit moony to say so but there you have it. Voting in America matters from a philosophical point of view, not just the hard campaign math side of things.

https://twitter.com/TheStalwart/status/725116281150705664

And just to prove I’m an idiot, exit polls in the Maryland Senate primary have Van Hollen ahead of Edwards by 8 percentage points, quite close to pre-election polls. https://twitter.com/Nate_Cohn/status/725114436197965824

https://twitter.com/GenderWatch2016/status/725114844597370881

And here’s where the early exit polls have it on the Democratic side:

• Connecticut: Clinton 50, Sanders 47
• Pennsylvania: Clinton 54, Sanders 45
• Maryland: Clinton 64, Sanders 33

If these results hold, Clinton will win Maryland by a larger margin than polls anticipated but will win Pennsylvania by a smaller margin, consistent with what we might expect based on the states’ demographics.

We know that Trump has won all 17 statewide delegates in Pennsylvania, which doesn’t include the district delegates allotted in the loophole primary. He has also won at least 14 delegates in Maryland. Trump will win a lot more than that, but we don’t know how many yet.

Right now, Trump seems to be in a fairly strong position to sweep all the delegates in Connecticut. He’s above 50 percent in the exit polls and leads in the New York suburbs, which is where Connecticut’s 4th district is located.

Living in Connecticut for five years did not yield nearly as many politically interesting moments for me as living in Wisconsin has. But there were a few good moments, like when I happened to pick up a copy of the Economist while traveling, only to discover that my governor was headed to prison. Or when businessman Ned Lamont ran against three-term incumbent Senator Joe Lieberman, a former vice presidential candidate, in the Democratic primary and won. Lamont’s 2006 bid was essentially a challenge to Lieberman’s support for the Iraq War, in a year when the electoral tide turned on the president and the war. It wasn’t enough to bring Lieberman down in the general, and he won his seat again as an independent. The Lieberman-Lamont story in some ways marks the beginning of two related arcs that have since come to animate party politics. One was the ideological sorting of the two parties. The 2006 midterms turned moderate New England Republicans like Nancy Johnson out of their seats. And Lieberman, who continued to caucus with the Democrats, eventually broke with the party. He became a public supporter of John McCain in 2008 and is now trying to lead the “No Labels” campaign. The 2006 midterms were a major turning point in the Democrats’ effort to become a more uniformly liberal party. Perhaps more consequentially for this year’s contest, the Lamont-Lieberman showdown foreshadowed some of the importance of the Iraq War as a wedge between Democrats. It was the big policy issue difference in 2008 between Clinton and Obama. These themes have come back this year. One question I’d love answered is whether this is a real rift in the Democratic Party or if it’s somehow specifically stuck to Clinton. I’ve said elsewhere (in a totally speculative way, to be clear) that her Iraq vote is one thing that Clinton likely did with the idea of running for president in mind – and that has backfired. Few such Democrats are still in the Senate, although some, like John Kerry and Joe Biden, have managed to stay afloat politically. Many others lost their seats in 2010 and 2014, like Blanche Lincoln and Mary Landrieu, or left the Senate around then, like Evan Bayh. So as we watch this all go down this evening, in some sense we’re back where it all started. But Clinton’s standing in the polls suggests that this division may no longer be motivating Democratic primary voters in the Northeast.

What Jody called the “second-northernmost-quarter-of-the-Appalachian-Trail-primary” is taking place in states with significant black and Latino populations, both key in recent years to Democratic victories. Black voters, particularly in the South, have also been instrumental to Clinton’s margin of victory in the primaries. Maryland, at 29 percent, and Delaware at 22 percent have the largest proportion of black citizens. Rhode Island and Pennsylvania have the largest proportion of Latino residents, each with 13 percent. Not surprisingly, as is the case in the rest of the country, the Latino population is growing far more rapidly than the black population. In fact, in Lancaster County, Pennsylvania — known as the horse-and-buggy-filled Amish Country — Latino residents surpassed the number of Amish by 2010.

https://twitter.com/Taniel/status/725112775584964611

ABC News has declared Trump the winner in Connecticut, Maryland and Pennsylvania. Clinton has been declared the winner in Maryland.

It’s appropriate that Trump weighed in on the threat by the actress Lena Dunham, a Clinton supporter, to move to Canada if he were to become president. According to an analysis by TheRedPin, a Toronto-based brokerage that has a business interest in knowing about migration to Canada, Trump is by far the main subject of tweets about moving to Canada. Keep in mind that there is some circularity in that ranking: Some of the counted tweets are by Americans urging Dunham and others to move to Canada.

I’m more interested in which delegates win in the loophole primary in Pennsylvania. Remember, all 54 of the district delegates in Pennsylvania will be selected separately from the statewide winner, and none of the district delegates are bound to support a specific presidential candidate. The presidential winner of each congressional district only matters if the unbound delegates on lists like this one say they’ll support the presidential candidate who wins in the district. The number who says they will support the candidate that wins their district has been shrinking over the last few weeks.

In Maryland, the most recent polls show Chris Van Hollen ahead of Donna Edwards in the Democratic primary for U.S. Senate, a de facto general election given that Democrats are highly likely to win Maryland in November. But there are a couple of things that make me think the race might be more of a toss-up. One is that locally-based pollsters like the Washington Post and the Baltimore Sun showed Edwards narrowly ahead in March, which I’d give some weight to even though the polls are less recent. The other is that Edwards is both black and very progressive, making her a good match for the turnout we’re expecting tonight.

Donald Trump does better with voters who are men than he does with women, thanks in part to a series of misogynistic remarks. Trump also usually gets a higher share of women than his two Republican rivals. Both of these things are true because Trump has won so many states by large margins. In the 22 states that voted before today and for which we have exit polls, Trump did, on average, 7 percentage points better among men than women, and nowhere did he poll higher among women. Cruz did about the same, and Kasich was a point stronger among women than men. (Marco Rubio explains some of the gap: He tended to do better among women than among men.) Nonetheless, Trump beat Cruz on average by 10 points among women, and Kasich by 22 points. For Cruz and Kasich to stop Trump, they’ll have to start beating him more regularly among women.

I know it’s sort of amusing for those of us in the Northeast to see people in other parts of the country mispronounce the term Acela, which refers to the “high-speed” Amtrak trains that run from Washington to Boston and pass through all five states holding primaries today. I’ll also admit to getting a kick when out-of-towners mispronounce New York’s Houston Street like the city in Texas instead of as “how-stun,” as it’s pronounced locally. But we should be forgiving; Acela is an extremely obscure term. According to Google Trends, it’s searched for only one-thirtieth as often as the term Amtrak, and almost all of those searches are in the Northeast. It’s also roughly as obscure on Google as the city of Pierre, South Dakota, which I only learned a few weeks ago is pronounced more like “peer” instead of like Pierre as in Trudeau. How many people riding the Acela would get that one right?

Twitter follower counts are consistent with Trump’s expected edge in the states voting today: He has a greater share of his U.S. followers there than do his Republican rivals. These are estimates based on Twitter users’ profile texts from TweepsMap, a Toronto Twitter analysis company.

Trump has a big Twitter following in the states voting today

	PERCENTAGE OF U.S.-BASED TWITTER FOLLOWERS WHO ARE IN
CANDIDATE	PENNSYLVANIA	MARYLAND	CONNECTICUT	RHODE ISLAND	DELAWARE
Trump	3.7%	1.4%	1.0%	0.4%	0.2%
Cruz	2.8	1.1	0.7	0.2	0.2
Kasich	3.1	1.3	0.9	0.3	0.2
Clinton	3.3	1.6	0.9	0.4	0.2
Sanders	3.6	1.4	0.9	0.4	0.2

Keep in mind two things with these numbers: 1) Not every follower is a supporter; and 2) Trump has way more followers than his Republican rivals. For instance, the number of Trump followers in Pennsylvania is about 63 percent the number of Kasich followers throughout the country.

One thing to watch out for tonight is whether Trump exceeds his polling averages. Before New York, Trump rarely exceeded his polling average by more than a couple of points. In New York, he got over 60 percent of the vote after polling at just 52 percent in the FiveThirtyEight polling average. If Trump does far better than his polling average, it could mean that he is winning over undecided voters in a way he hadn’t done previously. That would be very dangerous for Cruz in a state like Indiana, where Trump is leading but only with 38 percent of the vote.

And here are the GOP numbers:

Our weighted polling average for today’s GOP contests

	POLLING AVERAGE AS OF APRIL 26
CANDIDATE	CONNECTICUT	MARYLAND	PENNSYLVANIA	RHODE ISLAND
Trump	53.2%	49.5%	47.0%	55.7%
Cruz	11.2	21.7	25.3	11.2
Kasich	25.5	21.0	20.9	21.6

Our polls-plus and polls-only forecasts for today’s GOP contests

	CONN.		MARYLAND		PENN.		R.I.
CANDIDATE	POLLS+	POLLS	POLLS+	POLLS	POLLS+	POLLS	POLLS+	POLLS
Trump	>99%	>99%	>99%	>99%	>99%	>99%	>99%	>99%
Cruz	<1	<1	<1	<1	<1	<1	<1	<1
Kasich	<1	<1	<1	<1	<1	<1	<1	<1

We’re about 30 minutes from the polls closing in all five states, so we’ll have the real votes to look at soon. But here’s one last look at our weighted polling averages and forecasts for today’s Democratic contests:

Our weighted polling average for today’s Democratic contests

	POLLING AVERAGE AS OF APRIL 26
CANDIDATE	CONNECTICUT	MARYLAND	PENNSYLVANIA	RHODE ISLAND
Clinton	48.6%	52.9%	54.6%	44.7%
Sanders	44.4	37.7	37.9	45.8

Our polls-plus and polls-only forecasts for today’s Democratic contests

	CONN.		MARYLAND		PENN.		R.I.
CANDIDATE	POLLS+	POLLS	POLLS+	POLLS	POLLS+	POLLS	POLLS+	POLLS
Clinton	76%	75%	97%	96%	>99%	>99%	41%	42%
Sanders	24	25	3	4	<1	<1	59	58

Last Friday, Gov. Terry McAuliffe of Virginia signed an executive order restoring voting rights to 200,000 felons who have served their time. It’s an issue worth keeping an eye on as the political season unfolds. Laws vary widely by state, with some allowing all felons to vote, including those currently incarcerated; others allowing those who’ve completed their sentences to vote; and some barring convicted felons from voting for life. Unlike some fights that initially took place or are taking place on a state level, like same-sex marriage or marijuana legalization, it’s not as clear whether the re-enfranchisement of felons will become a state-by-state trend. Just this February, Maryland passed a law similar in effect to Virginia’s executive action, giving 40,000 convicted felons the right to vote. But in Kentucky, Gov. Matt Bevin initially struck down a wide-ranging felony re-enfranchisement executive order by the previous governor, and then just this month signed a more restrictive law granting some felons the chance to vote. It applies only to non-violent offenders who’ve not been convicted of a crime in five years, and they must pay $100 to reinstate their voting rights. Many legal advocates have noted that felony disenfranchisement laws are the most stringent in areas with high black populations in the South, and date from the post-Reconstruction era when white politicians sought to suppress new black voters. The effect of laws like Virginia’s could be small but significant in the presidential race. Most convicted felons are probably Democratic voters, but — like many other Americans, disproportionately those lower-income and lower-educated — they do not exercise that newfound right to vote.

One of the overlooked stories of the 2016 primaries might be how decisions made in the congressional redistricting process four or five years ago are having unforeseen consequences in this year’s GOP presidential fight. Democrats didn’t get to redraw a lot of states’ congressional boundaries after the 2010 Census (in fact, Republicans drew about four times as many districts), but there were two important states where Democrats resorted to some creative tactics to draw themselves additional seats: Illinois and Maryland. It just so happens those gerrymanders are helping Trump. In Illinois, Democrats diluted a lot of Republican votes in the Chicago suburbs by adding them to minority-majority districts based in the city. Under Republican National Committee rules, every district gets the same number of delegates, so these voters had a lot of power. And GOP voters living in close proximity to minorities have tended to be among Trump’s best supporters. In Maryland, the situation is a little different tonight. In 2012, Democratic mapmakers decided to dilute Republican votes in western Maryland by splitting them into two different districts: The 6th and 8th congressional districts, both anchored by the Montgomery County suburbs of D.C. Kasich should do well in wealthy, well-educated Montgomery County, but as others have pointed out, Trump could do well enough in Western Maryland precincts to overcome Kasich in both districts.

Speaking of Google Trends, there were signs earlier in the Republican primary that Election Day searches were decent indicators of how candidates would do at the polls. Those numbers suggest what the polls do: Trump leads by big margins in Google searches today in Connecticut, Delaware, Maryland, Pennsylvania and Rhode Island.

A quick point about “momentum,” which I should probably take up at more length at some point in the future. There’s a tendency among data-y types (like us here at FiveThirtyEight!) to dismiss momentum every time the term is invoked on television or in the newspaper. That skepticism is mostly for the good:

• Demographic differences between the states are often mistaken for momentum — the Democratic primaries of 2008, which were highly predictable from demographic patterns, are a great example of this.
• The way most political analysts use the term “momentum,” it’s more often a lagging indicator (in which direction were things moving last week?) than a leading one (how will things develop from here?).
• The news cycle progresses so quickly that momentum, even it exists for some period of time, may not last long enough to persist until Election Day. Sometimes, perceived momentum may even backfire on a candidate because it raises the media’s expectations for his future performance. Rubio’s purported momentum after Iowa probably contributed to the schadenfreude over his poor debate in New Hampshire or his mediocre showing on Super Tuesday, for example.
• Relatedly, whether a candidate is perceived to have momentum in the first place is also largely a function of expectations. A candidate who wins a state overwhelmingly may be perceived to have lost momentum if the media was expecting an even more overwhelming victory. Since the media’s expectations fairly quickly adjust to new polls and other types of information, this can make momentum fairly random.

But like most things, the anti-momentum arguments can sometimes be taken too far. True, momentum usually isn’t all that helpful in a predictive sense. (If candidate X wins state A, that might or might not increase her likelihood of winning state B; it depends on a lot of things.) That sometimes leads to the dubious conclusion, however, that primary campaigns are static and that the timing doesn’t matter at all. To take a more specific example, if Trump wins Maryland with more than 50 percent of the vote tonight after only having gotten 35 percent in demographically similar Virginia on Super Tuesday, that would suggest that something about the Republican race had changed. It might not be “momentum” per se and it might not tell us all that much about what will happen in Indiana next week (then again, it might). But it would suggest that demographics alone weren’t enough to predict Trump’s share of the vote.

Question “Is it too early to ask who Clinton will choose for her VP?” — commenter Bilal Muhammad Answer Well, it’s too early to ask a normal person. But I am mildly obsessed with VP picks. For the sake of the argument we’ll start by assuming Clinton becomes the nominee. There have, of course, been calls for her to select Sanders. This would clearly be an important gesture toward party unity. But it would also be two older, white candidates on the ticket for a young and diverse party. Let’s think a bit more about one of the main factors that go into VP selection: balancing. The research says the evidence for this is mixed. Figuring out what constitutes a “balanced” ticket can be a bit of an empirical challenge. How much of an age difference is big enough to balance a ticket’s age? Should we look at the candidates’ years of experience, or the type – legislative, executive, etc.? We’ve certainly seen some presidential tickets with big age and experience differences recently. When President Obama picked Joe Biden in 2008, Biden had been in the Senate for most of Obama’s life. In 2012, Mitt Romney chose a running mate who was both decades younger and also had a lot of legislative experience that Romney, a former businessman and governor, lacked. For Clinton, this is an even more interesting conundrum. She’s got lots of varied experience, but she’s also an unconventional candidate by virtue of being a woman and the former first lady. Will she pick someone – as Obama did – who brings conventional credentials and demographics to the ticket? Or will she consider women and people of color? The latter seems likely given the make-up of the Democratic coalition. Perhaps paradoxically, I think that Clinton may be more likely to make a bold pick if things look electorally safe. If she’s likely to win, then perhaps she’ll make a point about the many, many all-male presidential tickets we’ve had in the past and select another woman – Elizabeth Warren, Kamala Harris or Kirsten Gillibrand – perhaps? There are some other important considerations. William Adler and I have been working on some research about whether Democratic nominees have different running-mate priorities than Republican nominees; one of our hypotheses is that Democratic nominees use the VP selection to reach out to specific constituency groups while Republicans use the pick to affirm their ideological commitments. It’s possible that Clinton won’t have to pick between these. But in the event that she needs to make offer an olive branch to the Sanders-Warren wing of the party, her choices for a nominee who also represents the black or Latino (to name a few) constituencies within the party will be pretty slim.

“Super Tuesday” interest on Google has gotten supersized. According to Google Trends, relative searches for the term in the U.S. in 2004 peaked at 4 percent of this spring’s high, reached 47 percent of this year’s peak in 2008 and fell to just 11 percent in March 2012.

The American people have spoken, and there was only one Super Tuesday this election: the one on March 1. Search activity for Super Tuesday the day after the March 15 primaries was just 8 percent of its level on March 2, and is basically imperceptible today.

Parts of the media can keep trying to make Super Tuesday happen again, but Americans’ Google searches suggest it already happened almost two months ago.

Look at the preliminary exit polls, and you can’t help but think this is going to be a very strong night for Clinton and Trump. More than half of Democratic voters in Connecticut, Maryland and Pennsylvania say they want the next president to continue President Obama’s policies. Clinton has cleaned up among those who feel that way in previous contests. The same exit polls show that more than half of Republican voters would “definitely vote” for Trump in the general election, while the numbers are far worse for Kasich and especially Cruz.

Here are the results from the most recent competitive Republican (2012) and Democratic (2008) primaries in Pennsylvania:

Those tempted to believe the Trump phenomenon will usher waves of new voters into this fall’s electorate may be right — except, those may be waves of anti-Trump voters. Ever since Trump began dominating the 2016 news cycle, Democrats have been salivating over the prospect of record engagement among young voters and Latinos. Now, there’s some evidence that surge may be arriving, with potentially disastrous consequences for the GOP. California political data guru Paul Mitchell sounded a siren late last week, observing in the Sacramento-based Capitol Weekly that new voter registrations doubled among Latinos, rose 150 percent among young voters and tripled among Democrats compared to this point in 2012. Consider that Trump’s favorability/unfavorability has recently been pegged at 17/74 among millennials and a mind-boggling 9/87 among Latinos. Sure, Republicans never had a chance to win California to begin with. But if we begin to see similar registration numbers in many other states, Republicans could be headed for the electoral apocalypse many Democrats have predicted.

After tonight, Sanders and his advisors are going to “reassess” his campaign, according to The New York Times, though the paper said Sanders “is adamant that he will remain in the race until the Democratic convention this summer.” What Sanders is doing here is basically the political equivalent of the freshman year of college “Thanksgiving talk,” that five-day period during which high school sweethearts would initiate a slow, painful breakup that would be completed by the end of Christmas break so as to be free-and-clear to date, say, that cute baseball player in their hallway come second semester. If polling is to be believed — an existential question that’s too weighty to be taken on in the space of this em dash — Sanders will not do well in the five states voting tonight, with his chances to win the nomination devolving more and more into the realm of magical thinking. He’s probably not going to be the Democratic nominee, and his campaign advisors know it, but it’s a difficult thing to leave an enterprise one is so passionate about. While the campaign won’t end tomorrow, this reassessment might be the first step in strategizing how Sanders will make his exit from the race. “If we have a really good day, we are going to continue to talk about winning most of the pledged delegates because we will be on a path toward it,” Tad Devine, a Sanders advisor, told The Times. “If we don’t get enough today to make it clear that we can do it by the end, it’s going to be hard to talk about it.” Confused? This might speak to a certain internal divide in the Sanders campaign. After last week’s New York primary, Sanders’s campaign manager, Jeff Weaver, said on MSNBC that the campaign would try to “flip” superdelegates between the last primary contest and the convention, while Devine gave the more tempered response of reassessing. In the House of Bern, there is dissent, perhaps.

The Political TV Ad Archive tracks select media markets around the country, including Philadelphia, the fourth-largest local television market in the nation. Pennsylvania’s quirky party primary rules mean that there are only 17 bound GOP delegates (out of 71), while Democrats have 189 pledged delegates (out of 234 total). The campaign ad spending in the media market tracks with the stakes at hand, with the Democratic contenders making more ad buys, and making them earlier in the season.

CAMPAIGN		TOTAL CAMPAIGN ADS AIRED 4/1-4/25
Sanders	1064	–
Clinton	712	–
Trump	184	–
Cruz	132	–

Clinton and Sanders began airing ads within the first seven days of the month, while Cruz and Trump only began airing ads on April 20th and 21st respectively. The campaigns also differed in how many different types of advertisements they aired.

CAMPAIGN		UNIQUE ADS IN PA 4/1-4/25
Clinton	5	–
Sanders	11	–
Trump	1	–
Cruz	1	–

The Clinton campaign aired an ad boasting of her foreign relations credentials 217 times between the start of April and Monday. “The world a president has to grapple with,” a narrator says in voiceover. “Sometimes you can’t even imagine.” Sanders’s most-played ad, with 187 airings during that time span, focused on him speaking directly to the camera, asserting that there are “two Democratic visions for regulating Wall Street,” and on his plan to “break up the big banks.” Trump and Cruz only ran one ad each. Cruz’s ad, aired 132 times, uses a voiceover to assert that “Obama was a failure, Hillary could be worse,” before turning to Cruz’s plans to end the Affordable Care Act. Trump’s ad, aired 187 times, has the candidate speaking to the camera, declaring that “Washington is broken” and going through a list of his plans, from blocking lobbyists to building a border wall. Super PACs were absent from the ad buys during this period in the Philadelphia market. One strategy they abandoned: anti-Trump attack ads. Those had a brief run… and didn’t work.

https://twitter.com/Taniel/status/725089844964265984

One big disadvantage for Sanders tonight is that all states but Rhode Island are holding a closed primary, as New York did last week, which means that only registered Democrats can vote. (Rhode Island is holding a modified primary instead; both Democrats and unaffiliated voters may participate, although not Republicans.) As we’ve described before, Sanders does better with voters who identify as independent than he does with only Democrats. In fact, the differences are quite dramatic. Among states where exit polls have been conducted, Sanders has won independents in all states but Alabama, Georgia and Mississippi. However, Clinton has won Democrats in all states but Vermont and New Hampshire (she also tied Sanders among Democrats in Wisconsin). Overall, in the average state where exit polls have been conducted, Sanders has won independents by 27 percentage points while Clinton has won Democrats by 29 percentage points.

Clinton dominates among Democrats, Sanders among independents

STATE	ELIGIBILITY	MARGIN AMONG DEMOCRATS	MARGIN AMONG INDEPENDENTS	INDEPENDENT SHARE OF ELECTORATE
Alabama	Open	Clinton +73	Clinton +8	20%
Arkansas	Open	Clinton +58	Sanders +19	24
Georgia	Open	Clinton +55	Clinton +3	20
Illinois	Open	Clinton +15	Sanders +39	21
Michigan	Open	Clinton +18	Sanders +43	27
Mississippi	Open	Clinton +75	Clinton +32	13
Missouri	Open	Clinton +11	Sanders +34	24
South Carolina	Open	Clinton +60	Sanders +7	16
Tennessee	Open	Clinton +48	Sanders +9	23
Texas	Open	Clinton +50	Sanders +6	26
Vermont	Open	Sanders +64	Sanders +84	41
Virginia	Open	Clinton +42	Sanders +16	22
Wisconsin	Open	Tie	Sanders +44	27
Iowa	Modified	Clinton +17	Sanders +43	20
Massachusetts	Modified	Clinton +20	Sanders +33	33
New Hampshire	Modified	Sanders +4	Sanders +48	40
North Carolina	Modified	Clinton +32	Sanders +25	27
Ohio	Modified	Clinton +29	Sanders +33	24
Oklahoma	Modified	Clinton +9	Sanders +48	27
Florida	Closed	Clinton +42	Sanders +13	17
Nevada	Closed	Clinton +18	Sanders +48	18
New York	Closed	Clinton +24	Sanders +44	14
Average		Clinton +29	Sanders +27	24

How much, exactly, does a closed primary hurt Sanders? There are a couple of problems that make this trickier to analyze than you might think. For one thing, the exit polls ask about voter self-identification — whether they usually think of themselves as Democrats, Republicans or independents — rather than which party they’re registered with. As you can see from the table above, for instance, some voters who self-identify as independents were nevertheless registered as Democrats and voted in the New York Democratic primary last week. Another complication is that the states with open primaries tend to be different than the ones with closed primaries. In particular, open primaries are more common in the South, where Clinton performs strongly. Also, there are different degrees of openness and closedness; New York has a very early deadline for changing party registration, for instance, which may have especially hurt Sanders. Still, a regression analysis suggests that closed primaries hurt Sanders by at least a couple of percentage points and possibly a bit more than that. Tonight’s results will provide some useful data on the magnitude of the effect.