Black Lives and Power

I have been thinking about the increased visibility of power as a concept for months, now. It’s the product of the steadfast focus on power in Black Lives Matter movement. Through this focus BLM achieves an exceptionally consistent sublimation of this broad theoretical concept into specific, everyday reality. I’m sure in the past people have done the same thing with “freedom.” Now we have the same regular concretization of the concept of “power.”

It’s addictive, thinking more and more about power, all the time. I know it’s not everyone’s thing, but as for me, I like it so much I traded years of my young, potentially-profitable life going to school for it.

Thinking about power is at the heart of political science. When we study politics, we study the social distribution of resources — that is, “who gets what, when and how.” The research we do aims to explain “who gets what”; who gets the majority of votes, the favorable trade deal, the tax break, or the prison sentence. This focus on “who gets what” can also be expressed as “who has the power to achieve what outcome.” Describing the nature of power, then, becomes very important to being able to understand the thing you’re studying. For people in academics studying legislative votes, you wrestle with Robert Dahl’s equations. For people in academics studying organizations, you think dig Max Weber’s description that power depends on defeating social resistance. (And for people in academics who use Google to discover there’s a whole encyclopedia devoted to different definitions of power, you suddenly realize the definition of power itself is the site of power-struggle and enjoy the moment of irony.)

Power is such a critical concept because it’s a word that stands in for an explanation of outcomes from competition–interpersonal, intergroup, or against an inanimate object or force. When two people or two groups have different preferences and one gets their way, we use the word “power” to describe that outcome. Conflict over resources–like money, status, or security–which leads to a distribution of those resources is the very essence of politics. Power is the measure of difference across the resources people obtain.     

As a political movement, Black Lives Matter’s sustained focus on the concept of power has been an important aspect of their work. They’ve made it a regular part of public conversation across a range of incidents and events, creating new interpretive openings into things white America had previously not viewed as ambiguous. One example of this is in the activists’ repeated observation: “Watch whiteness work.” Activists apply this phrase in cases where we see a white person experiencing judicial or social lenience– a lenience, which we know from repeated examples, is unlikely to be extended to similarly-situated black people.

When an activist points to whiteness “working,” she invokes a person’s race silently pushing the bad judicial or social outcome away. The whiteness displaces the opposing force. Like gravity, or electricity, it is invisible power. It is the achievement of a non-action in a case where there would otherwise be an action. With its implicit reminder of the contrasting black experience, what “watch whiteness work” does is show us something that we can’t ordinarily see: the arrest not made, the threat not perceived.

When we don’t see this difference, we are not aware of the role of power in the interactions we have with each other, with law enforcement, with policymakers, with our current and potential employers, with our teachers, with our landlords, with our judges and juries.

Once it’s dialed in as a focus, though, you can see it in everything.  

(Next time! The rhetorical value of “privilege” over “power.”)

Posted in Uncategorized | Leave a comment

Understanding the 2014 Maine gubernatorial vote (update: now including omitted bear-iables)

Now that the election dust is settled — and while the policy dust is being kicked up — what can a closer analysis of the November results tell us about who wants what in Maine?

I’m going to use the election results, plus existing contextual data, to talk about the two big questions: 1) what drove voter turnout? and 2) who were Gov. LePage’s new supporters? Both of these questions remain highly relevant for legislators considering the governor’s policy plans.

A third question concerns the political impact of changes to municipal revenue sharing. The governor has been an avowed foe of municipal revenue sharing since arriving in office and, just as he did in 2012, he is currently advocating for the elimination of the program. As I’ve spent a fair amount of time thinking about municipal revenue sharing, I was curious to discover if the towns that faced the hardest consequences of that change – either in terms of reduction of direct revenue from the state or in terms of property tax increases  – responded by voting against the governor.

By combining data from a variety of sources, we can see relationships between the political, economic, and demographic features of Maine municipalities.

The first thing to observe is the clear power of incumbency. Support for Gov. LePage improved across the board, with only 15 of the 457 municipalities I studied giving him a lower level of electoral support. Votes in 2010 strongly predicted votes in 2014.

vote for lepage

Beyond this, however, there is some evidence that the prediction that municipalities with a larger proportion of older residents are not more supportive of Gov. LePage. In the 2014 elections, there was a weak negative relationship between votes for LePage and percentage of municipal residents over age 62.

62 plus

This relationship is difficult to see because of the highly visible and significant political counterexamples of Portland and Orono, both with very low median ages (36 and 21!) relative to the rest of the state.  However, characterizing older Mainers as being uniformly conservative looks to be fairly inaccurate; meanwhile, younger sections of western and southern Maine are far from uniformly liberal.

Wealth was another interesting variable to examine. Towns with a higher percentage of older residents also tend to have lower median household incomes, suggesting higher incomes and support for LePage might travel together. At the same time, Maine’s higher income areas tend to be in areas with higher levels of support for Democratic candidates.

Comparing the 2010 vote with the 2014 vote provides insight here. While, on average, the higher a community’s average median household income the lower its level of support for LePage in 2014, this was much less true than it was in 2010.  Either because of the collapse of the Cutler candidacy–or because they had come to like him as a candidate–a number of residents of higher-income municipalities switched to supporting LePage in 2014. Wealthy towns moved further towards LePage in 2014.

At the same time that we see the emergence of a break between older and poorer and younger and wealthier Maine, the interesting and contradictory element suggested by the election results is that while older voters were not more likely to support LePage, they certainly were more likely to turn out for the election. For every additional year in median age of municipal residents, turnout increased .6%. This was literally visible in the numbers for our youngest and oldest municipalities: moving from the municipality with the youngest median age (Orono) to the oldest (Ogunquit), we see an increase in turnout from 38% to 68% of the towns’ registered voters. The higher voting levels among older voters is consistent with the “traditional voter” hypothesis, which states that older voters vote as a matter of routine and therefore are most likely to turn out for the more local, non-presidential elections.

As for the third question, I was surprised to find that neither the loss of municipal revenue sharing nor increases in property tax appeared to have electoral consequences at the municipal level. Despite the strong effort by actors like the Maine Municipal Association to link the governor’s reduction of municipal revenue sharing to harder times for municipalities, those efforts were not visible either in the relationship of lost municipal revenue to votes or increased local taxes to votes. It would be useful to know more about this dynamic, to find out how people in towns facing the greatest loss of state support felt about those changes– or if they are even aware of them. Certainly as advocates gear up for a fight over new tax proposals, they should be aware that current strategies appear to not have fully got traction with voters (at least as of last November.)

In case you were interested in looking at more potential relationships I am including the dataset I used to explore these relationships. If anything interesting stands out to you in the numbers below, please share it in the comments!

UPDATE: I cannot believe I forgot to look at the bears. With municipal level data it is much more possible to identify the effect of interest in Question 1, the bear-baiting measure, than it was with the county-level data. So what was the effect of the highly contentious ballot measure on turnout and support for LePage? Was it a game-changer?

It totally was. If 2010 vote for Gov. LePage predicts nearly two-thirds of the 2014 vote, and increased warmth towards LePage in wealthier towns added a bit more, adding in votes to support bear-baiting adds still ADDITIONAL predictive strength. Controlling for people in the town who voted for LePage in 2010 and the power of median municipal household income, every four votes against Question 1 predicted an increase of 1 vote in support of Gov. LePage.

Opposition to bear-baiting also drove turnout. While the age of municipal residents was the variable with the highest power for predicting 2014 turnout, and municipal median income was also important, percent opposing Question 1 (and supporting bear baiting) predicted an additional and substantial piece of turnout: 17% additional turnout, moving from the condition of highest support to the highest opposition to Q1. (An additional effect appeared to come from the lack of interest by independents. Municipalities with the highest proportion of “Unenrolled” voter registrants were much less likely to turn out to vote, controlling for the above-mentioned variables.)

Now, while it seems that attitudes about appropriate methods of killing bears may in fact have had a very substantial impact on the vote, it is not immediately easy to tell if it is actually responsible for the governor’s margin of victory. In order to figure it out, one would need to calculate the increased turnout for each municipality by first estimating an effect size for each municipality and then converting that into actual number of votes relative to each municipality’s population size. I think the spreadsheet contains all the information you would need to do that, so if anyone feels like taking it on that task, please feel very free to do so! (And let me know if you do, of course.)

Posted in Uncategorized | 1 Comment

Obama, Third Parties and Bears, Oh My!


The dust from the victory parties has settled. The governor has made some predictably gubernatorial statements. The election results have been fully experienced, if it will yet be some number of weeks before they actually show up on the election results site.

Luckily, we have the Bangor Daily News and Portland Press Herald unofficial election results. Putting these together with an official data record of the state’s voting, hunting, and demography, what can we learn about exactly why Tuesday’s vote turned out the way it did?

I’m interested in two outcomes: the vote for LePage which was not well predicted by polls, and Maine’s high voter turnout, relative to the rest of the country.  (Because there’s nothing like seeing a national paper validate your prejudices about exactly which state is the Best State in America, I hereby forgive Reid Wilson for not citing my earlier post.) While there are lots of potential causes for both of these outcomes, I wanted to see what the evidence for them looked like.

Let’s take the governor first. I identified three most likely causes for the substantial increase in support he enjoyed this election: anti-presidential sentiment; the impact of Cutler/anti-Democratic Party “Third Party”-type preference; and, of course, bears.  (Please indulge me in putting aside the argument from incumbency for the sake of this post – it’s obviously important, but it does not explain the details of variation in the governor’s support.)


The impact of Maine’s referendum on bear baiting was the most interesting potential reason discussed for increased support for LePage, theoretically achieved through the mechanism of increasing turnout in more Republican areas. There’s an endogeneity problem with exploring this, however, since how do we know that energized LePage voters aren’t simply more interested in supporting bear baiting?

Luckily, some of the old, wonderful DataShare site is still up and running.  One of the more popular datasets on the site happens to be IFW Kills, a record of all of the registered hunting successes of the 2010-2012 period. This dataset lists all of the wild animals legally killed by date and location, giving us a very good picture of the geography of Maine hunting frequencies. I looked at the number of animals killed by county, and then divided each by population to get each county’s per-capita hunting frequency.

ifw 2

Even though not all of the successful hunting trips up in Piscataquis, Somerset and Aroostook counties are conducted by people living there, we can anticipate that this is a pretty good representation of locations where a larger proportion of people would identify as hunters. And, in fact, counties’ per capita reported hunting kills correlate pretty well with Tuesday’s vote on Ballot Question 1. If there’s at least one registered kill for every ten people in your county, then your county soundly rejected the measure.


However, was this relationship determinative for the governor’s victory? Let’s look at where Tuesday’s new votes came from.

new votes

Gov. LePage’s vote total on Tuesday, according to the BDN’s unofficial count, was about 32,000 higher than Rep. Michaud’s. Even though the governor gained votes in all of the counties, there just aren’t enough voters in the serious hunting counties to create that difference.

Adding up new votes for LePage in 2014 from the two southern counties of Cumberland and York alone, however, nearly gets you all the way there.

Third Parties

Although Maine’s most visibly liberal city is in Cumberland County, there are an increasing number of people in southern Maine (and in the former Democratic stronghold of Lewiston/Auburn, and in the area around the University of Maine, Orono) voting for a Republican governor.

rep vote

This brings us to our second possibility: that Maine’s “Independent” voters category has some meaningfully different preferences from those of Maine Democrats, and some of Cutler’s southern supporters ultimately preferred a Governor LePage to a Governor Michaud. Despite the strong public narrative around Eliot Cutler’s appeal to Democrats, it’s important to consider the possibility that Cutler was also appealing to people looking for a socially liberal, but economically conservative, alternative to the Democratic party.  Then, when Cutler was no longer viable, they voted for the candidate who would enact their economic preferences.

I compared two competing possibilities for explaining the governor’s improved vote share. If the governor were gaining more votes from people who share his social conservatism, we should expect to see even more votes from the socially conservative counties which opposed the 2012 citizens’ initiative to legalize same-sex marriage. Meanwhile, if the governor were gaining more votes from people who share his economic conservatism, we should expect to see more votes from places that, say, opposed the 2012 higher education bond issue.


2012 county level results on both of these measures do a pretty good job predicting the county level 2014 LePage vote, but rejection of the 2012 higher ed bond is a closer match. When we look at the same-sex marriage vote against LePage vote there are outliers, like Aroostook County which strongly opposed same sex marriage but supported Michaud at 43%, or Cumberland County which had fewer than 35% opposing same sex marriage but over 40% supporting LePage. For the economic question, there are no real outliers.

Looking at the past twenty years of gubernatorial elections in Maine, it becomes pretty clear that Maine has a persistent interest in candidates who occupy the socially liberal, economically conservative space.   Totaling all third-party or Independent candidate votes together in each election, we have each of the Republican, Democratic, and 3rd Party/Independent categories averaging out to about a third of Maine’s gubernatorial electorate.



This brings me to my final theory. What was the impact of national sentiment towards the president?

(Don’t worry, there’s no chart for this one.)

Looking at county vote share for Obama in 2012, I did find a small negative relationship between county vote for Obama in 2012 and an increase in votes for LePage in 2014 – in other words, there might be some direct anti-Obama sentiment in the vote for LePage – but it seems more likely that if there was an effect it was indirect and not directly corresponding to the 2012 presidential vote.


I’ve now covered the question of the increase in votes for LePage. I’m moving to my second question. What about our levels of turnout in general? Were there specific and identifiable causes increasing county movement to the polls?

Not that I could figure out.

People, this is still a mystery to me. I looked at bears, I looked at 2012 Obama support, I looked at social and economic conservatism, I looked at demographic change in counties between 2000 and 2010. Pretty much all of the counties increased their turnout between 2010 and 2014, with the exception of Knox and Oxford which decreased very slightly. Counties in both north and south were way up in terms of turnout. Central Maine was maybe a little less excited, but still turned out far above their 2010 levels. It was a good year for voting all around Maine.

And what’s interesting is that it has, since 1998, been a better year for voting every midterm election in Maine. Despite national midterm turnout being flat over the last decade or so (with a substantial dip lower in 2014), Maine has been regularly increasing its voting turnout every midterm election since 1998.

turnoutFor most states this would mean that the electorate was growing along with the state’s population, but Maine’s population isn’t growing! According to the US Census, the state might even be shrinking a little.

So I leave it to you. What is it that explains Maine’s steady increase in midterm voting? And did I miss something obvious that connects it to the election results?

(Update: If you’re interested in seeing the data I used for this post, I’ve collected it in a single spreadsheet here. My sources are linked in the second tab.)

Posted in Uncategorized | 1 Comment

When do people turn out? When their votes matter.

When you’re from Maine, it’s hard to watch people pooh-pooh this year’s turn-out. Mainers turned out in extremely high numbers for a non-presidential year, topping this year’s election turnout contest. Why did they come out in those numbers, despite having just been through a ridiculously unseasonal snowstorm with inch-accumulation in the double-digits?

Why WOULDN’T you come out, would be the Mainer’s answer, given the closeness of most of the state’s major races? Not only did the state have a gubernatorial election that multiple sources touted as the tightest election in the nation, but there was an extremely tight race for Maine’s second Congressional district — a representative who constitutes half of Maine’s delegation to the US House — and then there was this unfortunately timed, recurring, yet extremely hotly contested ballot measure about bear hunting. (Personally, I am now considering this to be the election where the bears finally figured out how to get back at us.)

Competitiveness is a classic reason why people get excited about turning out for elections. When people know that their vote could be decisive, they have a greater interest in participating. This is a big reason why political scientists think gerrymandering depresses voter turnout.

So I knew personally about Maine. But how well does competitiveness predict the scale of other states’ turnout figures?

To find out, I used the last week of polls posted at Real Clear Politics. Since we’re talking about state-level turnout, most individual US House seat races wouldn’t be effective to count, but I did include all polls for state races covering at least half of the state. This added up to 66 races for governor, US senator, or US representative for states with two or fewer US House reps like Maine and New Hampshire. It did not include bear baiting or any other exciting state-level ballot initiative, because RCP didn’t include those polls.

I sorted all polls taken in the last week of these 66 races and, where there was more than one poll, chose the poll with the narrowest projected victory. I then scored all of these polls on a 0-5 scale according to their competitiveness: I gave a score of “5” to projected ties, “4” to races where the front runner was a single point ahead, and so on down to “0” for a projected vote spread of 6 points or more.  This is what that looked like (click to sort.)

competitiveness table

Because several states (like Maine) had multiple competitive races, I totaled their competitiveness scores together into a composite competitiveness index. This helped reveal that while states could have a few big statewide contests, it was possible that none of them was competitive. This was the case in South Dakota, for example, where the closest of the three statewide races was projected to be won by 11 points. Other states, meanwhile, had multiple competitive statewide races.

Here’s what the ranking of states according to the competitiveness index looks like:

competitiveness indexBased on a highly scientific eyeball test, this set of scores bears a passing resemblance to the initial state turnout rates collected by Michael McDonald and visualized at FiveThirtyEight. States with competitive elections were more likely to turn out in higher rates.

The correlation is certainly not perfect but it looks to explain a good chunk of the variation.

The biggest outlier in terms of high competitiveness–but relatively low turnout–was Georgia. The state has come under attack for “losing” some 40,000 voter registrations, making it the focus of the major election day condemnation from the Lawyers’ Committee for Civil Rights Election Protection project.

Georgia’s outlier status therefore doesn’t disprove the voter interest theory behind increased turnout. Instead, it suggests that voting is a two-way street. Even if you’re interested in turning out to vote, your state also has to be interested in permitting you to do so.

Posted in Elections | 1 Comment

Brunswick vs. Cape Elizabeth – and why campaign contribution numbers matter

Two lovely coastal towns in Maine. What could they possibly have to fight over?

I personally would not have picked them to represent poles in our current gubernatorial election campaign, but some evidence suggests otherwise.

What evidence? Evidence available in information available as of last night, when the gubernatorial candidates’ 11-day pre-election filings were made available online at the Maine Ethics Commission Public Disclosure site. (When I first started looking at them this morning, they were just in PDF form. By afternoon, they were all integrated and available as downloadable structured data – including even the new filings from today. Let’s have a round of applause for the efficient and hard-working folks at the Maine Ethics Commission!)

Last week, in an attempt to observe any signs of a 2010-style surge in Eliot Cutler’s support, I observed trends using information about the past several months of donations to the gubernatorial campaigns. The trends that I observed — that Mike Michaud had deeper reservoirs of committed Maine supporters as revealed through numbers of campaign contributions, and that they existed across a broader swath of the state — continue to be true for donations to the candidates made between September 17 and October 21.

number maine donations 11 day

There was a massive increase in the number of donations coming in during this final major filing. Paul LePage nearly tripled his donations from Mainers, collecting 1088 individual donations. Mike Michaud more than doubled his previous monthly record of 901, collecting 1916 donations in the 9/17-10/21 period. Eliot Cutler did see a small pre-election increase in donations, but not enough to match July, the month in which he collected the largest number of Maine-based donations.

These data confirm my observations from last week. If Cutler is going to see any surge, it certainly isn’t showing itself in people trying to donate to his campaign in October. Meanwhile, Michaud and LePage are both witnessing a significant increase in the number of Mainers willing to take out out their checkbooks in order to make sure their candidate’s the one to occupy the Blaine House in 2015.

With the larger number of records, it’s also easier to observe concentrations of support for the candidates.

In my last post, I observed that Michaud was doing better along the midcoast than Cutler. The numbers that lay behind those maps that I made looked like this: zip codes, and then the quantities of individual donations from each zip code.
me donations sept

If you click on that table, it will bring you to a Google Fusion Table that lets you sort the columns by clicking on them. (Not sure why I can’t embed it in WordPress post but I can’t – if anyone wants to let me know how to fix that, I’d be grateful.)

The September numbers revealed the most general fundraising success for Michaud lying in the Portland area, but the largest number of Michaud donors from a single zip code came from Brunswick — 44 Michaud donations in September from 04011. Despite beating Mitchell by 10 points there in 2010, Cutler received just 7 donations with a Brunswick zip code. LePage received 6.

Cutler’s base of Maine donations in September, meanwhile, came most from Cape Elizabeth (33 from 04107), Yarmouth (29 from 04096), and Falmouth (25 from 04105). In all three of those towns, he outpaced donations to Michaud during early September. Although he also got many September donations from Portland’s West End and Deering neighborhoods (30 from 04102), Michaud got 47 September donations from 04102.

How did these dynamics change over the past month?

me donations sept-oct

(Again, click on the numbers to go to a live Fusion Table where you can sort it by clicking on the column headers.)

Looking at the new month’s records, we can observe very serious competition for the affections of Cape Elizabeth. In this final large pre-election filing, LePage received the most attention from 04107 donors, with 40 donations. Michaud tripled his number of donations from 04107, receiving 38. Cutler received 34 donations from 04107, which, although it was his largest single number of donors from any one zip code, put him behind the other two candidates with regard to October’s Cape Elizabeth donations.

As opposed to the split opinions in Cape Elizabeth, Brunswick residents provided over 100 donations to Michaud’s campaign. This represented five times the 21 Brunswick residents giving to the surging LePage campaign, and twenty times the 5 Brunswick residents giving to Cutler. Michaud’s lead in donations from Portland residents (04101, 04012, 04103) and and the northern Portland suburbs also dramatically increased.  Michaud received more donations from Yarmouth than either of the other two candidates in October. Michaud and LePage tied for donations from Falmouth with 35 each, while Cutler received 28.

Also of interest: in October, many more donations came from the town of Scarborough. Although Cutler carried Scarborough handily in 2010, Michaud raised significantly more 04074 zip code October donations (52) than either LePage (27) or Cutler (12).

Now, a last question. Does any of this matter at all? The demographic profile of campaign donors, especially with an incumbent in the race, is certainly quite different from voters at large. For example, I expect LePage’s election-day support to be many, many times larger than what we’re seeing in campaign donations, though the massive jump in number of donations during this last filing helps fill in some of the places where we would expect him to do especially well. (For example, in addition to the large increase in wealthy parts of southern Maine, we see lots of donations to LePage from Skowhegan in this filing.)

We can’t, unfortunately, seek to compare the relationship of campaign donations to votes in the last Maine gubernatorial election. In 2010, Libby Mitchell ran as a Maine Clean Elections candidate, which meant that she raised donations in only a very cursory initial way. Furthermore, thanks to the US Supreme Court rulings that took important limits off of campaign finance, the role of campaign donations has skyrocketed in terms of its frequency, amount, and importance in the last several years’ political campaigns. Maine’s 2010 gubernatorial election simply doesn’t offer a good comparison for understanding how campaign donations relate to votes.

I did try to think of a recent case where I would have access to both detailed campaign finance information and detailed vote information. I knew that the New York City Campaign Finance Board provides very detailed records that allow you to sort donations by date and location. New York City also had a very recent important election: their 2013 mayoral race to replace Mayor Michael Bloomberg. Since the Democrats were strongly favored to win this election, the interesting test lay not in the general election but in the city’s 2013 Democratic mayoral primary.

I looked at donations going from residents of the city’s five boroughs to the four most popular candidates running in the Democratic mayoral primary:  Christine Quinn, John Liu, Bill Thompson and Bill de Blasio.  I looked specifically at July and August, the last two months before the September primary.

I totaled the number of donations each candidate received for those two months, and then determined what proportion they had received of the total donations to all four candidates. Finally, I looked at the number of votes each of these four candidates received in the September primary, and determined what proportion each had received of the total votes to the four candidates. (I couldn’t use their proportion of votes received overall, because there were nine candidates running and a number of candidates were written in on submitted ballots.)

Here’s what I found:

2013 NYC Democratic primary

And rendered visually,

pre-election donations and votes

In other words, there was a great deal of correlation between New Yorkers’ donations and July and August and who they ended up voting for.

This is hardly rigorous proof of the validity of using local campaign donations to predict elections. This technique is highly speculative and still exploratory, and I think it might anyways be best suited to open races where nobody starts with the advantage of incumbency.

However, the New York correlation does show, I think, that local campaign donations in the post-Citizens United age can play a new role in understanding the flux and scope of local preference. In the Maine case, for example, I think you’d have to show that Cutler’s supporters are less likely than Michaud or LePage supporters are to donate to their candidate in order to say that there’s no predictive value here.

We’ll certainly soon have an opportunity to know much more, in very specific detail, about the relationship between local campaign donations and voting behavior in Maine.

That, of course, is not by a long shot the most important outcome of this election. However, as a present resident of the District of Columbia, it’s the only outcome that’s left for me to reap directly. I know it’s not for many of you. Nonetheless, at least by providing this information, I feel like I have done some civic due diligence and offered what insights I can.

Posted in Uncategorized | Leave a comment

Watching the electoral winds in Maine

In 2010, at just about this time I was conducting a poll of Maine’s gubernatorial race at Thomas College with the help of about 40 political science students. We all learned a lot. What many of the students learned was that they never wanted to have a career that involved cold-calling people at home. What I learned, meanwhile, was a little bit more exciting and immediately consequential.

I learned that Eliot Cutler had pulled even with Paul LePage.

We were among the last of the polls in the field that year and maybe it was my students’ persuasiveness, or maybe it was just luck of the draw, but our results ended up spotting the Cutler surge more effectively than any other poll.

gov race

Though my media skills were yet insufficiently developed to get this out into the local news, I immediately called my Democratic friends and family and told them to vote for Cutler. If they wanted to support an alternative to Paul LePage, I told them that the mass of non-LePage supporting voters had turned away from Mitchell and that they should switch their vote as well.

I was serving as an informal coordinating mechanism. And in this case, I was right. Cutler achieved an unprecedented 10 point increase between late October and November 2.

Why did I do that?

I felt the need to help like-minded voters coordinate. We desperately need coordinating mechanisms in a state like Maine, where races frequently see more than two strong candidates. Without an election method like ranked choice voting which lets voters choose their true preference without a negative consequence, our current system punishes sincere voters who vote for the third-ranked candidate by effectively removing their votes from the top-two candidate they would otherwise support. In other words, voting your top choice may help get your last choice elected.

Because of this unpleasant reality, voters mostly understand that they need to vote strategically. But how do you vote strategically if you don’t know which two candidates have the most support? Without that information, voters are unable to tell whether their true preference is in the top two, or if they should instead vote for the least-bad alternative.

After Cutler’s unprecedented surge at the end of the 2010 race, we have an especially strong need for this information since we know that the answer isn’t always obvious.

The question Maine’s Democrats and left-leaning voters need answered this week is: are we seeing a surge for Cutler similar to what we saw in 2010?

Without ranked choice voting, we must come up with imperfect ways to compare the candidates’ likely support before election day. This is why polling Maine’s gubernatorial race is so vitally important.   Since I am not in Waterville this week to conduct this personal civic duty, I tried to come up with an alternative mechanism.

We can take a look at how people are voting with their wallets.

The huge increase in the amount and significance of campaign donations in Maine’s elections since 2010 is generally very depressing, but in this case it might come in handy as an indication of how much committed support the candidates have across the state. While the profile of Maine campaign donors is not a great match for Maine voters more generally, it’s a reasonable way to think about the range and depth of commitment across at least a meaningful and politically active segment of Maine’s electorate.

In particular, it is useful as a gauge when we look not at how much money the candidates are raising, but how many Mainers are contributing. For this kind of use of campaign data it doesn’t really matter how much an individual can give, because it’s just a proxy for committed candidate preference. People might be able to give more or fewer dollars, but they can only vote once.

Finally, while we can use the Maine Ethics Commission data to look just at Maine donors, we can only know about those who gave more than $50, since the totals for those giving less than $50 don’t come with address information.  These small donors probably include a large number of Mainers, but without address information we can’t be sure. (The totals of donations under $50 made to the candidates in 2014 were: LePage $12,060; Cutler $20,369, and Michaud $76,971.)

So how many Mainers gave more than $50 in the last few months to the gubernatorial candidates?

number maine donations

The numbers vary substantially, but for the last four months Michaud has generally received twice the number of donations made by the other two candidates.

This is not immediately obvious from looking at campaign donation totals. The average Michaud donation is substantially smaller than the average Cutler donation or average LePage donation.

ave maine donation amount

This may lead to campaign contribution amount totals being a misleading measure of local support, since it suggests a lower number of contributors.

I also looked just at where donors were coming from in September and mapped the number of September donors in each zip code using Google Fusion Tables. This gives a sense of where support for each candidate is strongest, here close to the end of the race. (You can click on the maps to enlarge them.)

Cutler September 2014 donors

Number of Cutler donors, September 2014

Michaud September 2014 donors

Number of Michaud donors, September 2014

LePage September 2014 donors

Number of LePage donors, September 2014

New donations will be filed with the Ethics Commission on 10/24, and I’m looking forward to taking one last look later this week at how support changed during October.

So what do campaign contributions tell us about how the wind’s blowing in Maine this year?

They tell us that Michaud has witnessed the highest number of Maine donors in the last four months. They tell us that both Michaud and LePage have a stronger base of support across the state than Cutler, and that Michaud has the highest density of southern Maine support across the three candidates. This is a telling difference from 2010, when Cumberland and Sagadahoc Counties provided Cutler with a critical base of support. All up the coast, in fact, where Mitchell did more poorly than average in 2010, Michaud appears to have support.

So what’s causing this difference?

2010 and 2014 began in very different ways. The 2010 gubernatorial primaries on both sides were fiercely contested, with astonishing levels of rancor and spending. The fact that three independent candidates joined the partisan candidates in vigorous campaigns throughout the race revealed the breadth of disagreement over the state’s political future.

This time, meanwhile, the field settled quickly, with two partisan candidates who already enjoyed strong statewide support. The Republican party quickly confirmed Governor Paul LePage as the 2014 nominee, despite some early chatter about primaries. (Even with the governor’s challenges, as national odds are strongly in his favor: since 1980, almost 80% of incumbent governors win re-election.) Maine Democrats, meanwhile, made up for their party’s divisive 2010 primary by selecting Congressman Mike Michaud, a nominee whose six successive elections to represent Maine’s second Congressional district demonstrated his electoral strength in the part of the state traditionally less friendly to Democrats.

Because of the strength and broad name-recognition of this slate of candidates, recent polls are showing that only between 3 and 5% of voters are currently undecided, and that suggests — contrary to 2010 — that if a large shift happens at the end of the race it will require a large number of people to vote away from their current preference.

It’s not a poll, but I don’t see a large shift in the direction of candidate support happening this year.

Posted in Uncategorized | 1 Comment

Shopping for Diversity

Come with me as I daydream about real estate.

Real estate is excellent fodder for daydreaming. Entire professional fields are built on developing the glamour of homeownership, working off of the aspirational escapism of all of us poor hamsters stuck on the human wire wheel of status attainment. For recreational hausfrauen like yours truly, real estate websites offer the same seratonin hit that flipping through a J.Crew catalogue used to produce twenty years ago.

Me being me, my very favorite way to engage in this pastime is by exploring all of the websites and web gadgets that real estate companies have built up in the past few years using local data. Zillow and Trulia are the most developed examples of this I’ve seen, and what makes them so addictive is the illusion offer that you can specify exactly what kind of life you’d like to be able to lead in your new house.

My slice of demographic loooooves schools. Zillow’s property finding tool excels at allowing you to specify schools that you’d prefer — on one very specific dimension. Through an integration with the Great Schools rating website, Zillow surfaces the ranking of local schools on standardized tests. Although I know only too well that these rankings are essentially simply reports on the local economic environment of these schools, my lizard brain nonetheless enjoys looking for 9s and 10s and exploring the local housing stock.

But perhaps you are not interested solely in schools. Would you like to specify the number of reported crimes you’d be willing to tolerate in your neighborhood? How about your commute distance, by car, public transport or bike? Trulia lets you filter for properties meeting a standard you choose for both of those variables. It definitely tries to maximize the number of qualities you could choose for, offering to find you zones of lower earthquake and fire risk, proximity to (or distance from) restaurants and nightclubs, and median neighborhood property price. (It also has a school filter, but it doesn’t work as well as Zillow’s for some reason.)

But for all of these choices, none of the real estate tools I’ve seen let me search for a feature that I really want in a neighborhood: racial diversity.

This is a huge potential selling point for housing in the DC area, and yet the real estate websites have yet to make this search option available.

Taken all together, the DC metro area is incredibly racially diverse – substantially more racially diverse than the US as a whole, in fact. Unfortunately, like many metropolitan areas, that aggregate diversity hides an on-the-ground reality of neighborhoods which are largely segregated by race. As events in Ferguson, Mo. tragically underscore, the interplay of race, mutual social isolation, and power is still a core American issue, and continued de facto racial housing segregation helps to contribute this dynamic.  While no longer an explicit policy, decades of racially discriminatory housing law are effectively continued through the racialization of class and the dramatic difference in income for different kinds of city-based work.

You can see this de facto segregation in visualizations of DC’s racial demographics, such as CUNY’s Center for Urban Research maps using 2000 and 2010 data from the U.S. Census. A line runs roughly through the center of Washington DC, with most whites living on one side and most blacks living on the other.

That said, the DC metro area is getting better, at least as concerns black-white residential segregation. While the average white or black DC metro resident in 1980 lived in a neighborhood that was 70%  made up of members identifying with their race, in 2010 that figure was down below 60%. (Meanwhile, though the DC area’s Hispanic/Latino population increased over the last 10 years, people identifying in this category became more likely to live in more racially isolated neighborhoods.)

So what does this mean for casual real estate window-shoppers like myself? Well, some of us would like to be able to choose to filter not just for good schools and low crime, but also for diversity: for the neighborhoods and suburbs where our neighbors would reflect the true full range of people — and cultures, historical backgrounds, and experiences — that make up this place.

Trulia looks like it might have been exploring this option a bit a couple of years ago. In 2012, the company partnered with Forbes to look at “America’s Most Diverse Neighborhoods and Metros,” where the list of top-ranking neighborhoods linked to Trulia real estate listings. The next step will be linking that data with the rest of their data, so that potential customers can find the combination of factors that allows them to best maximize across preferences, including distance to work, transport options and school availability.

Once a “shopping for diversity” option exists, it will also need to include a consideration of whether a neighborhood’s diversity is supported through policy or local tradition and therefore seems likely to persist, or whether its just a temporary artifact of a neighborhood moving quickly from ownership dominated by one race to another (like through gentrification created by rapid new property development.)  This isn’t to say that there’s only one way to arrive at a diverse neighborhood, but if you’re looking to support the existence of racially diverse communities then the duration of diversity in a place over a period of time might signal that it’s a pretty stable quality of the location.

Anyway, even WITHOUT a dedicated tool (ahem!) I am able to use the excellent mapping tools that do exist elsewhere in order to casually explore the area. Using, I can look around and notice Germantown:

Or the environs of Silver Spring:

urbanresearch silverspring

Lots of possibilities!

Anyway, I am using this opportunity to speak directly to the real estate brokerage community out there. I know you’re keeping an eye on me, since several of you out there seem to have purchased my email address. (And seriously, Asif Qadir, do I look like someone who’s a good target for million dollar listings? Your automailer needs a few tweaks.) We should have a way to build a preference for diversity into our everyday choices, including housing.

Even if we’re just window-shopping.

Posted in Uncategorized | Leave a comment

While we wait for ranked choice voting


It’s like deja vu all over again. 

Just like in 2010, three main contenders vie for Maine’s governor’s seat. Just like in 2010, it is looking unlikely at present that the majority of Maine voters will support our next governor. The cost of this outcome will again be a certain degree of delegitimation of our electoral outcome.

The League of Women Voters has been trying hard to get Maine to adopt ranked choice voting – also known as instant runoff voting – following Portland’s successful use of this system for their 2012 mayoral election. In ranked choice voting, people rank the candidates for individual seats in the order in which they prefer them. If you had a most-preferred candidate, but in addition there were one (or more) candidates whom you would be fine with electing, ranked choice voting allows you to communicate that.

The largest benefit is that your vote continues to count even if your top choice lacks sufficient support to win. In fields of more than two candidates, this means that the winning candidate will have gotten support from a majority of the total electorate, at least at some level on the voters’ rankings. Unfortunately it’s a very different approach to ballot-marking–  and ballot-counting — so I think there’s going to be a period of education and effort before they see success.

So while I agree with the Press Herald that ranked choice is better, the 2014 elections are upon us so it’s important to see what we have in the tool box already.

We’ve seen a couple of ways that people use procedures with broad current availability to vote strategically. Just as ranked voting allows people to throw support behind their second-favorite candidate rather than lose their opportunity to affect the outcome, other methods of strategic voting allow people to weight the probability of their preferred candidate’s success in casting their vote.

The two existing methods that people use to vote strategically are open primaries and frequent opinion polling.

Open primaries are in widespread use across the US, with 11 states using them for all elections and an additional 24 states having open rules for some primaries. Open primaries have also been the recent beneficiary of some good press. In its recent report, the Bipartisan Policy Center proposed the use of open primaries as a tool for reducing gridlock in federal politics. The Republican primary for Mississippi’s  2014 US Senate election, meanwhile, made for a pretty supportive backdrop to this report’s opening day. In elections where the district or jurisdiction is overwhelmingly captured by a single party, members of the minority party are all but guaranteed to be unable to affect the outcome of the election. This lack of support for the inevitable victor means that representatives are less likely to see those voters as people they need to represent accurately, leading to long-standing problems in democratic representation.

In Mississippi, the fact that the Republican primary run-off used the open primary rules meant that all voters, not just registered Republicans, were allowed to vote. As a result, Senator Cochran was able to reach beyond the small hardcore set of partisan primary voters in his run-off campaign and appeal to a much broader range of voters who would have an interest in supporting the more centrist candidate.

The effect of the open primary in this case was to effectively re-enfranchise all of the voters who would not have otherwise had a meaningful impact on the electoral outcome. Cochran’s perception of his constituency will almost certainly be affected by the results of his run-off election and his gratitude to those who supported him will probably now affect his lawmaking. Race is a particularly important part of this story since, well, Mississippi, and also because of the intensity of racial partisan polarization across the South. The open primary managed to bridge this substantial gap. As a result, people who would not have had a champion in Congress are now more likely to have one.

So how did Democratic voters know that it was important for them to come out for this election, which meant doing something unusual and voting in an opposing party’s primary? Polls revealed that the outcome in this election would be extremely close. The primary election itself even more convincingly showed how close the margin was for Cochran, providing the critical information necessary for strategic voting.

In general, while we don’t get two elections like this in a row, we now do have a similar ability to understand probable outcomes through frequent public opinion polling. Polling aggregators like HuffPost Pollster show how many polls are being taken in a multitude of races across the country, providing voters a pretty good idea of how things are trending for their preferred candidate.

Polling was very important to the story of Maine’s  2010 gubernatorial race. Real Clear Politics aggregated 16 polls taken in Maine between June 2010 and the November election. The final set of these polls can be seen in the image below:


There was no massive scandal, or glorious new program introduced in the final weeks of October. Instead, a number of Democrats realized that Mitchell was not going to be able to win over additional voters, and support for Cutler surged 25 percentage points: from a low of around 11% throughout early October to a final vote share of 36% of the electorate.

That kind of enormous change in the revealed preferences of the electorate demonstrates the substantial power of polling to allow an electorate to collectively estimate probabilities for their preferred candidates. While it’s not ranked choice voting, it is the same essence of strategic choice at work.

With increased access to polling data and analysis of polling data, voters are able to learn more about the preferences of the other members of their district. If a candidate is close to winning, they know that it’s worth backing that candidate. If a candidate trails persistently, that is also informative. In the case of Maine’s 2010 gubernatorial election, voters who preferred a candidate other than Paul LePage were able to make a collective last minute determination that Libby Mitchell could not gain sufficiently on LePage to win.

While we are seeing a version of this story playing itself out again in Maine, the trendlines are different from where they were at this point 2010. Nonetheless, the real difference is our improved ability to learn what the electorate prefers, through even more frequent polling, and through the established recognition that complicated elections require thoughtful strategic voting.

An imperfect solution next to ranked choice voting for sure, but something that we have the power to use right now.

Posted in Uncategorized | Leave a comment

Improving Democratic Side-Payments

With apologies to the 7% of you out there who are still satisfied with how our Congress is working, I have to say that I have no expectations that our country’s most consequential legislative body is going to produce anything of value anytime soon. I count myself as part of the majority of Americans whose highest expectation for Congress is that it manage to avoid passing bad laws. Luckily, they seem to be following that advice on their own. Unluckily, the consequence of this is a failure to deal with a number of issues that matter a great deal to the country at large.

To address this – even if it comes at the cost of having more Congressional law made – I would like to take this opportunity to consider the excellent observation made by several political scientists and journalists that some of the blame for this inaction is due to the ban on earmarks. The decline of the use in earmarks, resulting from a series of agreements created in 2007 and 2011, corresponds to a decline in the number of laws passed, and maybe indirectly to the further decline in Congressional popularity.

Earmarks — or pork, if you are especially opposed to the practice, and want to create a good visual — served as the grease that allowed for legislators to achieve non-ideological compromises. This lack of ability to compromise is something that some drown-it-in-the-bathtub-types favor, because they would prefer that the federal government avoid doing much at all. By reducing the ways to get to agreement, the earmark ban helps achieve that purpose. It doesn’t actually reduce spending. It does increase executive power to make decisions formerly made by Congress. It does reduce transparency by taking these conversations out of the bill text and putting them into inaccessible bureaucratic records. But if one’s primary goal is to hobble Congressional agreement, well then mission achieved, I guess.

By trading their vote on a bill, legislators achieving earmarks brought material benefits to their districts. In itself, is this really such a bad thing? These are people who were elected by the specific people of a specific district. Maybe some of those people care a little less strongly about certain large ideological battles than they do about having jobs, or a nice bridge or something. Does anyone ever ask them that? They really should.

Outside of the “pork” context, this type of arrangement is sometimes viewed favorably. In game theory, the concept that someone might be open to additional material trades in exchange for cooperation is known as a “side payment.” Side payments can indeed be associated with illegal transactions like under-the-table payoffs and bribery. However, side payments can also be seen as a legitimate part of the negotiator’s toolbox, such as in cases like the negotiation of international conflict. While the side payment necessarily does not directly relate to the main issue being negotiated, it recognizes that people might have different degrees of commitment to that issue – or might have a different need that’s not currently being met, but which could be resolved in the context of the existing negotiation.

Thinking about our concern about earmarks issue in greater depth, the biggest ethical issue they raise does not lie in specifically targeting benefits to a particular district. Specific districts get targeted benefits all the time, for a variety of reasons. If a member represents a constituency that doesn’t care strongly about a particular issue up for vote, but does care strongly about getting funding a local project, seeking an earmark in exchange for a vote might actually be the more accurate representation of that constituency’s preferences.

The big ethical issue comes when earmarks are used to provide private benefits to the member of Congress.  Because earmarks are negotiated by individual members, they create a risk of corruption. Earmarks could be used to benefit representatives’ friends, family or campaign donors. Earmarks are also used as a way to secure votes for re-election, since they represent a material benefit that incumbents can claim as a direct outcome of their work. Congress passed transparency measures that clearly associate individual members with their earmark requests, and this helps to address the first issue, but it doesn’t do much for the second.

An additional issue that I haven’t seen discussed anywhere concerns the issue of whether the sought earmark does represent a good political bargain from the perspective of the representative’s constituency. Would a majority of the representative’s constituency prefer the new local bridge or to contribute to the new piece of legislation? The way that earmark requests developed doesn’t allow a representative to discover this information.

Both of these problems — unfair electoral benefits to incumbents and how to determine whether a constituency would prefer money or ideological victory — could be dealt with through a frank effort to integrate side-payments into political life. If we accepted that political outcomes are sometimes national, ideological victories and sometimes just a pile of money for a bridge, then we’d have a clearer set of conversations to have in our districts.

If we wanted to improve and democratize side-payments, as well as make them less of a personal benefit to sitting representatives, we’d develop a regular practice of polling constituencies with questions about their preferences.

90% of Americans have cell phones. Survey research is by this point a regular and highly routinized practice. More surely than at any time during the past, you can call people and just ask them to press 1 if they’d prefer to vote against healthcare reform, but press 2 if they’d rather obtain $800,000 in federal research funds.

Meanwhile, the idea that representatives should not collect opinions, and should instead substitute their own judgment, is a concept of democratic governance that becomes less supportable the cheaper and easier it becomes to collect public opinion.

Indeed, you may ask, why stop with the question of side-payments? Why not poll each district’s constituency for every vote?

I personally think that’s a very exciting possibility. Even if representatives refused to do this themselves, they couldn’t easily ignore well-designed and frequent surveys of their district. Democratic accountability through the hammer of frequent, granular constituent feedback.

Better living through survey research!

Posted in Uncategorized | Leave a comment

Dismantling public higher education, Maine style

The mass layoffs currently coming to pass at the University of Southern Maine are the final moment of metal collision in the slow-motion train wreck we’ve been watching over the last couple of years. The USM faculty have been nauseously swaying out in the wind over the last handful of years. After finally gaining a contract, it appears that any security offered by this contract is fairly illusory.

Some of the professors USM plans to terminate are tenured. Tenure represents the golden ring of academia: the promised state of professional security which leads all younger academics to put in effort and hours far beyond what would otherwise be rational for their pay levels. As the American Association of University Professors put it,

Tenure is a means to certain ends; specifically: (1) freedom of teaching and research and of extramural activities, and (2) a sufficient degree of economic security to make the profession attractive to men and women of ability.

Unfortunately for tenured USM faculty on the chopping block, the guarantee of tenure technically protects only individuals from being removed from existing positions; it does not protect people in positions which are altogether eliminated. 

Nonetheless, we absolutely see a violation of the principles that tenure is intended to protect in the execution of the faculty-layoff plan. The layoffs disproportionately impact women: nine of the eleven faculty laid off are women, in an institution where women make up less than 50% of the faculty.  The disproportionate disregard of women’s tenure makes it pretty questionable whether women considering working at this institution could truly consider themselves free to teach or research in the way they’d see most fit.

Similarly, this action makes it difficult for potential professors to view USM as a reliable academic employer in the future. To the extent that a capable individual with a substantial training investment already loses money when she chooses a job in academia over industry, it’s important for academic institutions to do what they can to demonstrate “a sufficient degree of economic security” in order to make that difference worth it. USM is presently cutting two economists, who generally do quite well in the open market. What rational economist would come to work for USM in the future?

The whole situation is just awful. The state into which USM has been permitted to fall is a terrible tragedy. The university and state should be making much better use of the campus they have situated in the most populous, vibrant, and livable city in Maine. Why are they looking to further weaken this institution?

The state of Maine has certainly led the way to this outcome, reducing 20% of its funding to the university system between 2002 and 2010 and further reducing since then. As a result of having less and less to work with, the cutting of positions was more or less inevitable.

However, here we are. The present cry of protesters, seeking to save the professors, has been to ask the university to “chop from the top.” What does that mean, exactly? How much would that save, relative to other measures?

As a publicly funded institution the University of Maine is required to publish all of their salaries, so you can, in fact, evaluate this yourself. Public institutions are unique in this country in being required to provide their salary information – even where the state only pays for about 30% of the bottom line, as is the case for UMaine. (Even where non-profit and for-profit organizations receive tax breaks or enormous public contracts, those organizations don’t face this sort of transparency requirement, although they are also partially paying their people with public money. It is honestly quite unfair.)

After converting the pdf to Excel to work with the numbers, I looked at groups of workers by “bargaining unit” in order to think of people meaningfully by groups. Here’s the list of bargaining units identified in the document, which includes non-represented groups:

  • AFUM – Associated Faculties of the Universities of Maine
  • UMPSA – Universities of Maine Professional Staff Association
  • COLT – Assoc. Clerical, Office, Laboratory and Technical Staff of the Universities of ME
  • Service and Maintenance – Teamsters Union Local #340, Service and Maintenance
  • Police – Teamsters Union Local #340, Police Unit
  • PATFA – The Maine Part Time Faculty Association
  • Non-Represented Salaried
  • Non-Represented Faculty
  • Law Faculty – Non-Represented
  • Non-Represented Hourly

How does the entire University of Maine workforce break down along these categories? Here are two pie charts to explore that: the first describes the proportion of the 4,794 listed employees in each of the official bargaining-or-non-represented groups; the second provides categories that simplify several of the bargaining groups.

total employees

You will probably be unsurprised to learn that although more than half of the employees employed by the University of Maine are union-represented staff, that group is not the best remunerated. Here are the average salaries by bargaining group:

salaries by bargaining unit

These salary averages, put together with the total number of employees in each category, gives us a sense of the relative weight of each group of employee on the balance sheet.  The top four categories are three faculty categories – law faculty, non-represented faculty, and union faculty – plus “non-represented salaried,” a category made up primarily of administrators and management.  The remaining seven categories, although collectively constituting the largest number of employees, have substantially lower average salaries.

The pie charts below describe the relative proportion of each group of employees in making up the university’s total salary expenditure. Again, the first pie chart uses the given bargaining or non-represented groups and the second contains my simplification of these groups into more conventional categories.


As a final cut at this, I looked at the proportion of each of these categories in the list of 270 University of Maine employees making $100,000 or more.


Although “non-represented salaried” workers make up about 10% of the total employees, they make up 16% of the university’s total salary cost and about one-third of the 270 University of Maine employees making $100,000 or more.

What should USM do? I think the state is the one that should be paying more, since it is clearly their reduction in funding that has made the biggest change in university revenues.

However, I’d also be looking to administrators who seem not to be able to do their jobs in securing that funding to be offering funding up from their own departments, rather than hollowing out the value of their very product. Instead, this move they’ve taken has destroyed the attractiveness of being employed by – or enrolled at- USM for the foreseeable future.

Posted in Schools | 1 Comment