Monday, September 29, 2014

The Real Battle in #MAgov: WBUR vs. the Boston Globe

Look at a graph of Massachusetts governor polls this year and you'd think it's been an extremely volatile race: the last five polls have gone Baker +2, Coakley +10, a tie, Coakley +3, and Coakley +9. Problem is, you'd be wrong. The horse race has actually been remarkably stable; it's the campaign's two premier pollers that have diverged.

Almost all of the polls that have shown a statistically tied race have come from SocialSphere and been sponsored by the Boston Globe. SocialSphere conducts a weekly tracking poll of the race, and its last three results have been Baker +2, Coakley +3, and Baker +1. On the other hand, the MassINC Polling Group runs a weekly tracking poll of its own, sponsored by WBUR, Boston's public-radio station. MassINC's results have consistently given Democratic Attorney General Martha Coakley a solid lead: Coakley +10, Coakley +9, Coakley +5, and Coakley +9 in the same time span as the Globe polls.

So even though the polling average suggests a "Leans Democratic" race, this contest is probably either "Likely Democratic" or an outright "Tossup." But which pollster is right? Let's poke around into each poll to see where their discrepancies might originate. The most recent Globe/SocialSphere poll is here with crosstabs here. The latest WBUR/MassINC polling memo is here, and crosstabs are here.

The differences aren't immediately obvious to find. Both polls use registration-based sampling and called both landlines and cell phones. Both firms conduct live telephone interviews, and the sample sizes were not significantly different (502 likely voters for MassINC, 400 likely voters for SocialSphere). The similar methodologies produced similar samples, too. The MassINC poll breaks down as 36% Democrats, 12% Republicans, and 52% unenrolled. The SocialSphere poll's partisanship is 35% Democratic, 13% Republican, and 52% unenrolled. MassINC's sample was 84% white, and SocialSphere's was 85%.

There were some slight differences. The SocialSphere sample was 55% female, while MassINC's was 52%. And although the age categories did not line up perfectly between the two polls, it looks like SocialSphere surveyed a much older electorate. For MassINC, 38% of respondents were age 44 or younger vs. 29% for SocialSphere. Half of SocialSphere's sample was 55+, while just 29% of MassINC's was 60+. But the biggest difference of all was in level of education. The electorate projected by SocialSphere was 30% postgraduate degrees, while only 16% had some college, but no degree. For MassINC, only 21% had an advanced degree, and a full 29% said the highest education they had reached was some college but no degree.

However, none of these explains the 12-point discrepancy. SocialSphere's bigger pool of female voters should have helped Coakley do better in the Globe poll than the WBUR poll, not worse. (More to blame is the fact that Coakley led women by a whopping 25 points according to MassINC and "just" 15 points according to SocialSphere.) And MassINC found that the candidates were more or less tied among "some college, no degree" voters but that Coakley was way ahead with holders of postgraduate degrees (the SocialSphere poll was too small to provide crosstabs for each specific education category), which, again, should have underestimated her support. Finally, although both polls agreed that Martha Coakley did best with young voters (52% to 28% among 18- to 29-year-olds per MassINC; 57% to 19% among 18- to 34-year-olds per SocialSphere), Coakley won every age group in the MassINC poll, so her lead wasn't just due to millennials. It was more that the polls wildly disagreed on how older voters felt. SocialSphere gave Baker a whopping 51%-to-29% lead among voters ages 55–64.

As you can probably tell by now, the real disagreement between the polls has to do not with their sampling, but with how each candidate is performing with different segments of the electorate—the most confounding and unresolvable way to disagree. Is Baker winning independents 45% to 23%, as SocialSphere says? Or is it basically a tie game, 43% to 38%, as MassINC says? How do women voters genuinely feel—massively pro-Coakley, or does Baker still have a fighting chance with them? And are voters making $150,000 or more really spurning the millionaire Baker to vote for Coakley, 46%–38% (MassINC), or is Baker the one up by 47%–38% among those making $100,000 or more a year (SocialSphere)? We won't truly know until Election Day.

Normally I would defer to the better pollster, but neither firm in this case has really given us a reason to doubt it. MassINC is an extremely reputable firm up in the Bay State, with a long record of excellence and a solid B rating from FiveThirtyEight's newly released Pollster Ratings. SocialSphere, meanwhile, is a newish firm, so it doesn't have a pre-2014 track record to base any judgments off (indeed, FiveThirtyEight doesn't even give it a rating). And if you're looking to outside pollsters to break the tie, you're out of luck; since the beginning of September, two national firms polled the state, and one agreed with MassINC and one agreed with SocialSphere. When it comes to how Massachusetts's key constituencies are feeling in this surprisingly close race, the best any of us non-professionals can do right now is guess.

UPDATE: Maybe we don't have to wait until Election Day after all. A spate of polls Monday all agreed with the Globe results, and then WBUR's tracking poll released on October 1 also moved in line with the SocialSphere numbers.

Friday, September 26, 2014

Introducing a Different Breed of 2014 Race Ratings: Lieutenant Governor

You might've heard there's an election coming up. You probably know it will decide control of the US Senate. If you're lucky, you've heard that a handful of governors are on the ballot too. The US House of Representatives will also be elected, you're told, but those elections don't really matter.

If you listen to professional political handicappers, that's the end of it. In fact, it's not. It may not have the sex appeal of a presidential election, but in this even-numbered midterm, pretty much everything but president is on the ballot: 30 lieutenant governors, 26 secretaries of state, 31 attorneys general, 87 state-legislative chambers, 173 mayors, 144 statewide ballot measures, and even a heck of a lot more. These are the elections that people should truly care about, because these are the elections that truly make the policy that we feel, on the ground, in our daily lives.

Yet these races suffer from a lack of analysis and handicapping. State-based political observers may have a good idea of where the wind is blowing downballot, but they rarely formalize their thoughts into race ratings; meanwhile, big names in forecasting, like Charlie Cook, stick to, well, the big races. It's that gap I want to try to bridge.

I thought about releasing Senate and gubernatorial race ratings this year, as I did in 2012, but I realized anything I put together would just be a reheated take on the conventional wisdom. Leaving the top of the ticket to the true experts, I'm going to try something that I don't think has been done yet for 2014: evaluating this year's state elections below governor and handicap them on the familiar Solid-Likely-Leans scale.

My goal between now and Election Day will be to Cook up (see what I did there?) race ratings for all the state constitutional offices—adjutant general, agriculture commissioner, attorney general, auditor, commissioner of labor, commissioner of state/public lands, comp/controller, insurance commissioner, lieutenant governor, mine inspector, railroad commissioner, secretary of state, superintendent, tax commissioner, and treasurer—that don't get the treatment they deserve and then to classify those races in a language politicos are accustomed to. The ratings will be published to the new "2014 Ratings" tab in the menu above, and each new set will be accompanied by an explanatory blog post. This post announces the debut ratings: those for lieutenant governor.

There are only 11 states to rank for lieutenant governor—a relatively modest lift—so if you'll indulge me for just a bit longer, I'd like to spend more time outlining my methodology for these and future ratings.

First, although I consider quantitative factors, the ratings are qualitative—this is not a model. Unfortunately, with these downballot races, there is scant polling to base a prediction upon, so the ratings are openly subjective. I make the best guess I can based on a combination of factors, including candidate quality (including incumbency), fundraising totals, the overall political climate in the state, and the state's partisan lean. I rank a race as "Solid" if I believe there is simply no evidence that that party can lose that seat. "Likely" means I acknowledge the race is in play, but in my heart of hearts I don't see it going to the other party. "Leans" means I believe a race could reasonably go either way, but I have identified enough of a tilt that I'm comfortable making a call. And, finally, "Tossup" means I simply have no freaking idea.

Each new set of ratings will look like the lieutenant-governor chart below. Each race occupies a row and a column—the row denoting my ranking of the competitiveness of the race, and the column depending on which party currently holds the seat. While columns are binary—a seat is either Democrat-held or Republican-held—rows are a spectrum, going from the most Democrat-favoring race at the top to the most GOP-leaning at the bottom. This allows you to skim down the progressively more competitive races while also seeing, at a glance, which party is playing defense where. Totals in the bottom row and rightmost column give you a summary of the national picture.

With that, finally, here is how lieutenant governor breaks down. Of the 50 states, 45 have lieutenant governors. Two are not popularly chosen, however, leaving 43 lieutenant-governor elections. Twenty-one of these are conducted jointly with gubernatorial elections in both the primary and general, and an additional five LGs run on their own in the primary but fused on a ticket in the general. That leaves 17 states where lieutenant governors are their own, separate elections.

Four of these stand for election only in presidential years, while two more stand in odd years. That leaves 11 states holding lieutenant-governor elections in 2014 that are separate from their states' gubernatorial elections and that thus require their own ratings. (I consider the 19 additional lieutenant governors being elected this fall on tickets as already ranked by other handicappers' gubernatorial ratings.) The states are Alabama, Arkansas, California, Georgia, Idaho, Nevada, Oklahoma, Rhode Island, South Carolina, Texas, and Vermont.

Most of these races are snoozers, with no Leans races and only one Tossup. Republicans are heavily favored to win most of the seats, but that has a lot do with the fact that they hold eight of them already. One that maybe they shouldn't hold is Vermont, where Progressive Party candidate Dean Corren is getting $200,000 in public financing (that's a lot for a campaign in Vermont) and where the voters are already among the most liberal in the nation. But incumbent Phil Scott is an old-fashioned New England Republican, and he's beloved in the Green Mountain State; just 4% (!) disapproved of him in a 2012 poll.

Arkansas looked interesting for a time, but ever since Republican Congressman Tim Griffin jumped into the race—and Democrat John Burkhalter faced accusations of workplace violence—the momentum has shifted to the GOP in polls, a microcosm of how Arkansas Democrats have seen their ability to cruise to statewide wins suddenly evaporate in recent cycles. With Griffin's $433,980-to-$42,541 cash-on-hand lead, this one is barely on the fringe of competitiveness now. If we're judging by fundraising, there are actually two red-state races that Democrats have better shots in. In South Carolina, 30-year-old Bakari Sellers (D), the son of a local civil-rights activist, has raised almost as much as Republican Henry McMaster and has 2.8 times his cash on hand. In Texas, State Senator Leticia Van de Putte (D) is as good a candidate as Dan Patrick (R)—who previously called himself Dannie Goeb and hired undocumented immigrants to work at his bar, despite today referring to immigration as an "invasion"—is a bad one. Van de Putte has raised a million dollars and has $1.2 million on hand (outpacing Patrick in both measures) but has struggled to gain traction in polls (she was most recently down 46% to 26%).

In the end, the only truly exciting LG race—and it is quite exciting indeed—is the one in Nevada. Democratic Assemblywoman Lucy Flores and Republican State Senator Mark Hutchison are vying for the open seat; it's universally agreed to be the most important downballot race in the country because a Republican victory could encourage Governor Brian Sandoval to challenge Majority Leader Harry Reid in 2016. There has been no polling in the race since the primary, but the candidates together had raised almost $900,000 through only June 6 (the next campaign-finance report is due October 14). Flores has the benefit of a great story and the backing of Reid's political machine, but Hutchison has Sandoval's coattails. The race is unpredictable right now, and given Nevada's close political divide, it's an obvious candidate for "Tossup" status.

Tuesday, September 16, 2014

How Andrew Jackson Convinced America to Wage War on the Islamic State

For the second straight September 10th, Barack Obama faced the bright lights and television cameras and made his case for intervention in Syria. “Our own safety—our own security—depends upon our willingness to do what it takes to defend this nation and uphold the values that we stand for,” the president concluded. “Timeless ideals that will endure long after those who offer only hate and destruction have been vanquished from the Earth.”

As with most things this president does, the predominant sentiment among the chattering classes was skepticism. But there was a remarkable difference from the last time Obama had made a plea to get involved in the Middle East. On September 10th, 2013, a deep reluctance pervaded Capitol Hill—as well as the voting public—following Obama’s speech arguing that America must react with force to Syrian President Bashar al-Assad’s use of chemical weapons. His inability to corral support soon led to an abandonment of the more bellicose elements of his plan. But by September 10th, 2014, congressional leaders had made a 180; many were fearful that Obama’s plan to fight the Islamic State in both Iraq and Syria did not go far enough in eradicating the emergent terrorist group.

With fatigue over the wars in Iraq and Afghanistan, as well as a heightened urgency surrounding domestic issues (i.e., the economy), American public opinion has made a well-documented turn toward isolationism in the past decade. So what explains the sudden turn toward gung-ho intervention this summer?

Perhaps a 13-year-old book called Special Providence. In this influential international-relations text, political scientist Walter Russell Mead seeks to answer the age-old question of how to characterize and classify the American foreign-policy tradition. He argues that American opinion on foreign policy is a mixture of four schools of thought, each embodied by a different great American: Alexander Hamilton, Woodrow Wilson, Thomas Jefferson, or Andrew Jackson.
  • Hamiltonians believe in a strong federal government with an active foreign policy. Their primary concern is America’s economic empire; they are realists who seek to build a world order that maximizes commercial opportunities and strengthens American business both here and overseas.
  • Wilsonians think that America has a moral responsibility to the rest of the world. The original Wilsonians were Christian missionaries, but today they believe that peace, democracy, and human rights are values all nations must share for a better world—and thus American foreign policy should actively promote these values.
  • Jeffersonians, in contrast, disdain foreign engagements in favor of preserving democracy at home. To achieve their main goal of protecting Americans’ freedoms, they support only deescalating, diplomatic solutions that deflect war. As strict constitutional constructionists, they also put their faith in Congress, not an overstepping executive branch taking unilateral action.
  • Jacksonians want to be left alone, and they believe the best way to do so is to be well armed. They dislike international entanglements, but they care deeply about defending the homeland. Even though they are suspicious of big government, they love a strong military, and, if America is threatened, they do not hesitate to unleash its full force to indiscriminately destroy our enemies.
In short, Hamiltonians and Jacksonians believe American interests are paramount, while Wilsonians and Jeffersonians focus on American values. Yet Hamiltonians and Wilsonians agree that America should take a global outlook to foreign policy, while Jeffersonians and Jacksonians are more insular. The schools convincingly resonate with what we know about how foreign-policy decisions were made in the past—and what motivates people today.

Although one school has frequently been enough to shape an administration’s entire foreign policy, one school is rarely enough to garner the majority support of the public. Such was the case with Syria in 2013. The case for intervention a year ago was a Wilsonian one: acting as the “policeman of the world,” the U.S. felt it had a responsibility to enforce the international norm against chemical-weapons use and to provide humanitarian relief to the Syrian populace. There was no economic incentive to entice Hamiltonians; there was no clear and present danger to convince Jacksonians.

With one quarter of schools to justify it, support for intervention was mired around one quarter of respondents in polls. In a CBS News/New York Times poll conducted September 19–23, 2013, just 26% said “the United States has a responsibility to do something about the fighting in Syria.” An ABC News/Washington Post poll conducted the same month put support for airstrikes at just 30%; Gallup measured support for “military action” at 28%.

But the same polls a year later tell a different story. In an ABC News/Washington Post poll conducted August 13–17, 2014, a full 54% of Americans favored airstrikes against the Islamic State in Iraq, compared to 39% who were opposed. The same month, the Pew Research Center and USA Today found 54% who approved of airstrikes and 31% who disapproved. Pew went on to find a sharp increase in the number of Americans who believe the U.S. does too little to solve world problems. In November 2013, 17% said so and 51% said the U.S. does too much; in August 2014, 31% said too little and 39% said too much.

Last year’s dilemma and this one have important differences that explain Americans’ sudden hawkishness. In 2013, we would have fought Assad; in 2014, we are bombing the Islamic State. In 2013, we would have been inserting ourselves into Syria; in 2014, the question is over a country we got at Pottery Barn. And 2013’s intervention would have been humanitarian, while 2014’s is more about national security.

These differences have won over adherents of multiple schools, explaining the popularity of airstrikes today. Wilsonians still support intervention, since the Islamic State threatens the well-being of non-Sunnis, including the now-famous Yazidis whose confinement on Sinjar Mountain was the original impetus for the strikes. Hamiltonians now have a reason to get involved, since the Islamic State controls copious oil wells. But, with 67% of Americans saying the Islamic State is a “major threat,” the beast that has truly been awakened is the public’s Jacksonian strain. With their one criterion (danger to the homeland) fulfilled, Jacksonians are now more than ready—they are itching—to bring full American military might down on the Islamic State.

Better than any other school, the Jacksonian influence can actually be pinpointed in the polls. Jacksonians tend to be anti-elite populists with a strong sense of honor, self-reliance, and military pride. According to Mead, “Jacksonians are today’s Fox News watchers,” and in their constant hatred of big government they often see conspiracy theories about elites trying to build a world government. In other words, they are Tea Party Republicans!

According to each poll’s crosstabs, Tea Partiers account for most of the shift in public opinion. In the ABC News/Washington Post poll, 63% of “conservative Republicans” supported airstrikes (+9 points from the overall sample). In the Pew poll, the GOP is supportive by a whopping 71% to 14% (+17 points on the overall sample). Tea Party Republicans specifically are nearly unanimous (91%; +24 points) in saying that the Islamic State is a major threat. And on the question over U.S. involvement in the world, the Tea Party completely reversed itself: in November 2013, they said the U.S. does too much global problem-solving 54% to 22%; in August 2014, they said the U.S. does too little 54% to 33%. Democrats and independents also shifted toward interventionism, but by just 10 points each, meaning they still comfortably say the U.S. does too much, even today.

Paradoxically, then, Obama’s newest foreign-policy agenda could be saved by his biggest adversaries. He may very well have drawn it up this way; the president’s words in last Wednesday’s speech were no accident. Before referring to upholding our values and “timeless ideals,” he appealed to “our own safety,” “our own security,” “our willingness to do what it takes to defend this nation.” Perhaps learning from his 2013 failures, he has successfully built a coalition of Wilsonians and Jacksonians who are giving him the political capital he needs to see a mission through in the Middle East. Perhaps the difference in the interim was he read a little Walter Russell Mead.

Saturday, September 6, 2014

What's Wrong With Rafael Soriano?

Last night, Nationals closer Rafael Soriano blew a three-run lead, allowing two home runs to the Phillies (to add insult to injury, one was hit by Ben Revere). The outing before that, he blew another save. The outing before that, he got the save, but gave up a run in a rocky ninth inning.

Soriano's recent struggles have Nats fans screaming bloody murder about removing him from the closer's role. But they're failing to see Soriano for what he is: one of the game's best closers going through a rough patch of infinitesimal sample size.

What's wrong with Rafael Soriano? The short answer is nothing. The longer answer is random variance, factors out of his control, and the rest of this blog post.

In the first half, Soriano had a 0.97 ERA and a 0.811 WHIP. Since the All-Star break, he has had a 6.98 ERA and a 1.759 WHIP. These numbers measure results—which are important, to be sure. But pitching involves a lot of luck; it's been shown that pitchers have little to no control of what happens to a ball after it makes contact with a bat. So to measure the actual quality of Soriano's pitching, we need to fall back on what are known as his peripheral stats.

The main thing pitchers do have control over is throwing balls and strikes, and by extension walks and strikeouts. In the first half, when he was pitching lights-out, Soriano allowed 2.68 walks per nine innings and 8.76 strikeouts per nine innings—both quite good, if not elite, numbers. So far in the second half, Soriano has allowed 3.26 walks per nine innings and 8.84 strikeouts per nine innings. As a function of innings, then, his walks have increased slightly—which is not ideal, but 3.26 BB/9 is still above average—and his strikeouts have actually increased—which is obviously desirable.

Perhaps a better way to measure it is the percentage of batters Soriano has walked and struck out. In the first half, Soriano walked 8.2% of batters and struck out 26.7%. That's an average walk rate and an excellent strikeout rate. In the second half thus far, he's walked 7.5% of batters—an improvement—and struck out 20.4% of batters, which still rates as pretty good.

So Soriano isn't pitching significantly worse when it comes to balls and strikes. What has been worse (from Soriano's perspective) is hitters' contact with those pitches. In the first half, opposing batters had a .153 average, .222 on-base percentage, and .226 slugging percentage off Soriano and hit 0.24 home runs off him every nine innings. Since the All-Star break, opposing batters have hit .321/.387/.530 and 1.4 HR/9.

What accounts for this huge difference? As it turns out, mostly luck. We can use two stats to measure how unlucky a pitcher has been: BABIP (batting average on balls in play) and HR/FB (home runs allowed divided by fly balls allowed). These stats jump around a lot among players and among seasons for a given player, but with no discernible pattern: in other words, they are random variance, and they tend to normalize at a league-average rate over time. (This is because things like the quality and positioning of fielders is more important to deciding whether a ball drops in for a hit, and wind or the dimensions of a ballpark can turn a lazy fly ball into a home run—or vice versa—fairly easily.) Typically, batters "should" hit .300 on balls in play, and 9.5% of fly balls "should" be home runs. Truly talented pitchers can skew these numbers downward—particularly HR/FB—but not significantly.

Soriano's BABIP in the first half was .207 and his HR/FB was 2.2%. These numbers are almost impossibly good and suggest that Soriano was due for some regression; a player can't sustain a 0.97 ERA when it's built on unsustainable secondary stats like those.

But here in the second half, Soriano's BABIP has been .387, and his HR/FB has been 10.0%. Those numbers are also outliers, just like his first-half numbers—just in the opposite direction. A .387 BABIP is particularly unsustainable, and there is no way hitters can keep being this lucky off Soriano in the long term. Meanwhile, the HR/FB rate is also above the rate at which batters normally hit home runs—and it's also well above Soriano's career HR/FB rate of 7.7%. As we saw with the 1.4 HR/9, a big problem for Soriano in the second half has been the increased number of home runs off him, but if we dig deeper we can see that this is atypical for MLB hitters and downright aberrant for batsmen hitting off Soriano.

A final stat that can affect pitchers' luck is the rate at which they strand runners on base. League average LOB% (left-on-base percentage) is about 71% of runners stranded; it too can bounce around a lot for individual seasons for no apparent reason, but it always ends up around 71% in the long haul. In the first half, Soriano stranded 90.9% of runners, which is really lucky—many more of those runners should have crossed the plate. But in the second half, he has stranded just 62.9% of runners, which is fairly unlucky. A lot of those runners had no business scoring.

For Nats fans, it's certainly torture to see Soriano struggling like this—and doubtlessly it has been for Soriano too. But he can take solace in the knowledge that it's not really his fault—from the pitching side of things, he's continued to do things about as well as he's done for his entire career. When luck enters into the equation as much as it does with pitching, though, there are going to be stretches when even the best execution leads to crummy results. Small sample sizes (Soriano has pitched just 19.1 innings during the second half) can produce huge outliers; that's why baseball plays 162 games a season.

Although he has materially pitched the same throughout the year, the disparity between Soriano's extremely lucky results early on and his unlucky ones today is naturally going to catch the eye—and ire—of the fan base. Advanced stats show that, no, Soriano was never as good as he appeared to be in the first half—but he's also not nearly as bad a pitcher as he looks right now. The real Rafael Soriano is somewhere in between: a solid relief pitcher whose true value indeed lies around the average of his first and second half performances. Need proof? His ERA as it stands today (3.04) is almost a dead ringer for his FIP (3.16).

Wednesday, September 3, 2014

Primary Endorsements Show Where Massachusetts Dems Are Drawing Their Battle Lines

As any member of an organized political party will tell you, they're really not all that organized. My home state of Massachusetts is one of several places where one-party rule has caused the political divides that really matter to be intra-party ones. (Kansas is another example.) When outsiders make cracks about how liberal and/or monolithic Massachusetts is, they're showing how little they know about the Democratic Party's extreme factionalism there. In actuality, the party's seemingly giant legislative majorities (36–4 in the State Senate, 125–29 in the State House) are more the result of British-style coalition building between several smaller parties, sometimes with very different identities and policies but yet all, for reasons practical to historical to programmatic, considering themselves Democrats.

Hoping to pen a more ambitious analysis of these subparties in my home state, I canvassed every endorsement issued in the Massachusetts Democratic primary this cycle for the four contested statewide races: governor, lieutenant governor, attorney general, and treasurer. (In a state with one-party rule, after all, the primary dictates which clan sits on the throne for the November coronation.) Unfortunately, the data ended up being too thin to expose the detailed fault lines I wanted to map out, so my original idea for this post fell by the wayside. However, there's still a place for the data that do exist, so I'm leaving them here for Massachusetts political junkies to play with.

Any Democratic politician or organization that endorsed this cycle is listed in this Google doc—also embedded at the bottom of this post. You can see whom they endorsed in each of the four major races. Most only endorsed in one or maybe two races (that's why the data was too thin), but if they endorsed in three or four, their full voting slate is laid bare in that spreadsheet.

Just for fun, we can cross-analyze how much endorsements line up with each other. Do all Tom Conroy for treasurer supporters also support Warren Tolman for attorney general? (Yes, it turns out, but not all Tolman supporters support Conroy.) This gives us a tease of some of the aforementioned Massachusetts Democratic factions—witness, for instance, how endorsements for progressives Don Berwick and Maura Healey line up pretty well—but mostly it's just a confusing mess.

In the chart below, the rows represent the universe of people we are considering to calculate the percentages (i.e., the denominator), and the columns represent the characteristic being measured within that universe (i.e., the numerator). Read the chart by asking, "What percentage of [row] are [column]?" So, for instance, you can find what percentage of Steve Grossman endorsers also endorsed Steve Kerrigan by going to the Grossman row and the Kerrigan column. Percentages are out of the people who endorsed in both the former and latter races being considered, not just in the former race; as a result, percentages all add to 100% and there is no cell for "[Candidate] endorsers... ... supporting no one." Endorsements of withdrawn candidates, like Joe Avellone, don't count for chart purposes (even though they're listed in the spreadsheet for comprehensiveness).

In the chart, read down a column to see the percentage support that a given candidate got among universes of different endorsers (and compare it to the share of the race's total endorsements received by that candidate, displayed in the bottom row); read through a row to see how endorsers of a given candidate split their support in the other three races (useful if you're undecided in one of the races and want elite support to guide your vote).

Spot any interesting patterns? Leave them in the comments!