Monthly Archives: November 2011

What Big Sort?

Political scientists Mo Fiorina (Stanford) and Sam Abrams (Sarah Lawrence College) have done work analyzing and ultimately critiquing Bill Bishop and Robert Cushing’s popular Big Sort.

Synopsis of Big Sort: Bill Bishop claims that we are increasingly self-sorting ourselves into neighborhoods politically and only associating with like-minded political neighbors with all kinds of horrible consequences.   Much of Bishop and Cushing’s evidence about the corrosive effect comes from psycho-sociological experiments like Asch‘s where group pressure causes people to behave immorally (a la Lord of the Flies or the Stanford Prison Experiment), or to censure their own dissonant voice even when they originally believed those views to be  correct. [Note: Fiorina has made quite a name for himself on how the political elites in America have become ever more polarized and the masses have over time sorted themselves out more reliably into political parties but the masses views' have not become any more extreme, so obviously the Big Sort doesn't square with his other research that uses ongoing surveys like the General Social Survey, the American National Election Studies, etc.]  There is a wonderful cartoon that the New York Times did about the Big Sort.

While Bishop and Cushing try to look a wide variety of evidence, among them voting records, patent applications, IRS income data, advertisers’ data, etc., Fiorina asserts that the backbone of Bishop’s evidence compares two closely fought presidential elections — 1976 where a moderate Republican Gerald Ford took on a moderate southern Democrat Jimmy Carter vs. 2004 when a Texas born-again Republican George W. Bush took on a liberal northeastern Democrat John Kerry.   Bishop observes that there was an increase of 22 percentage points in the number of “landslide” counties from 1976-2004 (defined as a county that went for a candidate by more than a 60/40 margin).

Fiorina thinks that this comparison in and of itself is skewed since presidential campaigns are all about personalities and one can’t simply compare one against another and assume that one is witnessing changing behavior of voters.  Furthermore, he thinks because of the contestants in those contests, there are many reasons to expect more landslide results by county in 2004 when voters were faced with a starker choice.

Nonetheless, he and Sam Abrams have searched for a measure that proxies well for voter preference but measures against a more steady yardstick than votes.  They look at partisan political registration by county (which they say predicts voter choice according to other scholarly work).    Comparing counties in 1976 and 2004, even if one dramatically lowers the threshold of “landslide” counties to ones where a simple majority of registered residents are one political party (e.g., Republicans), there has been a drop in such counties from 75% of counties in 1976 to 40% in 2004.  This doesn’t show sorting at all.  For sure, there has been a significant increase over this same time in voters registering as independents, but that itself is an undermining of the “Big Sort” hypothesis, since independents’ vote choice is much more volatile according to Fiorina. Fiorina is doing another project on independents:  they are almost never just weak identifiers with a party, but either break with a party over one significant issue or have a much more esoteric alignment of political values.  He says that looking at independents over time one sees that there may be as low as 35% of Independent voters from one presidential election to the next consistently saying they are Independent, voting Democratic.

Fiorina also says that even if there were a “big sort” going on, and the data found increasing polarization at the neighborhood level (his data show nothing like this happening at the county level), he’s not convinced it would have a big impact on politics for three reasons:

  1. Neighborhoods aren’t such an important center, especially in the age of media and blogs and where 2/3 of Americans only know at most 25% of their neighbors’ names.
  2. Neighbors don’t talk to each other all that much: a Howard, Gibson and Stolle 2005 CID study found that 55% of Americans never talk about politics with neighbors and Putnam’s Bowling Alone showed how interactions with neighbors has sharply declined over last generation;
  3. Politics is simply not that important a topic of discussion or way in which we identify ourselves.  The three most important ways in which people identify themselves are family (51%), occupation (16%) and religion (10%).  Even if you go down to people’s third most important factor, politics only registers 2.7% of people listing that as the third most important factor.

Questions: one person asked Fiorina about the Bischoff-Reardon study showing increased income residential segregation over the last generation (at the census tract level); since income itself predicts being Republican, she wondered how those findings are consistent.  Fiorina hadn’t seen the study so didn’t want to comment.

Another asked how one knows whether Americans really are moderate or like to portray themselves that way. Fiorina said that any survey data is subject to such doubts but that highly volatile results, like the recent contrasting results in Ohio criticizing Obamacare while supporting the rights  of unions, with many voters voting yes on both are consistent these data.  Fiorina also noted that one has to look back to the late 1800s for 4 consecutive elections that show the level of political instability that exists today.  [2004: All Republican control of president and both houses of government; 2006 Republican president, democratic control of both houses of Congress; 2008 democratic control of President and both houses of government; 2010 democratic presidency, republican House and Democratic Senate.]  We’ve had four elections each with a distinctive result, and the next election, if current Intrade predictions pan out could show a 5th result and a flip from 2006, with a democratic President (Obama) re-elected and republican control of both houses of Congress.  See also David Brooks’ interesting related column “The Two Moons.”

Fiorina who is working on Americans Elect, believes that the way this could change is for things to get bad enough that a “younger, saner Ross Perot emerges” as a third party candidate (quoting David Brooks).  While this is not predictable, Fiorina cited Sid Verba who noted that before the Berlin Wall fell, no one saw this coming, and afterwards everyone could identify the reasons why this was inevitable.

He thinks Obama’s most promising re-election strategy is to assert that he’ll be the bulwark against likely control of both houses of Congress by the extremist Tea Party-led Republicans and a bulwark against the political extremism among political elites.

Fiorina believes that although trust of Congress is at all all-time low of 9%, turnout is not down because the political parties are providing a much stronger ground game and a much higher percentage of voters now indicate they’ve been contacted by the political parties.   [It may also be a function that more voters see an increasing difference between the two political parties and the media and others may make stronger appeals that the stakes are ever more consequential.]

Fiorina also commended the recent research by Jim Stimson and Chris Ellis  and a forthcoming book that indicates that most liberals truly are liberals whereas white conservatives are a blend of different things.  26% of conservatives are movement conservatives who really do have conservative values (what Ellis/Stimson call “constrained”); 34% are traditional-symbolic conservatives (like Mike Huckabee), many of whom are recruited through churches but don’t necessarily know the conservative party position or have consistent conservative beliefs (what Ellis/Stimson call “moral” conservatives); slightly less than a third are what Fiorina calls “clueless” conservatives (what Ellis/Stimson call “conflicted” conservatives), many of whom are younger, who actually hold liberal positions but think that the conservative label conveys greater respect (like a military official in uniform); and 10% of conservatives are libertarian (just wanting less government in general, whether it is for making marijuana legal and eliminating an army, or doing away with food stamps).  Fiorina agrees with the book that when one says that 40% of Americans are “conservative” it is misleading since a far smaller percentage of them uphold conservative positions across the board.

See also this earlier post about the “Big Sort.”

Only you can stamp out gerrymandering (UPDATED 7/30/13)

Gerrymandered 4th District

Frustrated by a closed process that results in gerrymandered districts, Michael McDonald (George Mason Univ. turnout expert) and Micah Altman (Harvard) together with thinkers like Norm Ornstein have initiated the quite interesting Public Mapping Project to enable citizens’  input on district boundaries for political seats.

Background:  The WSJ on 7/30/13 reported on the consequences of having legislators draw their own district boundaries (akin to the fox guarding the chicken coop): “Of 435 districts in the Republican-controlled House, the nonpartisan Cook Political Report rates only 90 as competitive, meaning those seats have a partisan rating that falls within five points of the national average. The rating measures how each district votes relative to how the country as a whole voted in the most recent presidential election. The number of competitive districts [is] at its lowest since Cook first started the partisanship rating in the 1998 election cycle.”

But many states (VA, MI, OH, NY, AZ) and Philadelphia have used this new software developed by Michael McDonald and others to offer redistricting competitions where citizens compete to design the best districts.  The software helps evaluate these maps along various criteria and prizes are awarded for the best maps.  These public maps can become a reference against which the traditionally closed deliberations for redistricting are judged and to refute notions that there were not other better alternatives.  This software can also be used by advocacy groups to weigh in against the redistricting commissions.  In some cases students have designed much better maps than “experts” and is a vivid example of crowdsourcing, how incorporating the wisdom of the general population rather than relying on a small number of experts can lead to much smarter outcomes.

Norm Ornstein of the American Enterprise Institute thinks that this open-sourcing redistricting is more like a wonk-fest, since in general one has to be somewhat specialized and an avid political junkie to participate effectively.

Of course what makes a good map is itself contentious:  Democrats may care more about helping to craft districts that help minorities get their own candidates elected than Republicans do.  But clearly both parties are interested in trying to maximize the number of districts that are “safe” for their party or “lean” towards their party.  To the extent that redistricting commissions reflect the party in power, or to the extent that the current legislative must approve it,  even with better crowdsourced maps, it will not take the politics out of the process.  And the irony of partisan-leaning districts is that they protect against smaller public opinion movements away from their party,  but by creating fewer completely “safe” districts, it can put many more seats potentially at risk in there is a large storm surge against that party.

See a Brookings Institution discussion on the Congressional Redistricting effort on July 18, 2011 (Michael McDonald appears from about 21:00-31:00 in the talk).

Test out the software here and see other people’s maps.

Dramatic growth in social capital scholarship

The attached graph from Google Lab’s Beta “Books Ngram Viewer” lets you chart the mention of various words over time.  It’s quite fascinating and thought-provoking.

One interesting comparison is the rise over time in the discussion of “human capital” vs. “social capital” as depicted in the following chart from 1900-2008 (blue is “human capital”, red is “social capital”.  [The data comes from the 5.2 million books that Google has digitized as part of their book project.]

Basically it took “social capital” about a quarter of the amount of time to become as dominant a concept in academic as human capital.  [Scholarly attention to social capital from 1993-2003, 10 years, advanced it to the point that it took “human capital” 40 years to achieve, from 1963-2003.

If you want to see a better image of the graph, click here.

Google Labs Books Ngram Viewer

Increasing US urban residential segregation and decline of middle-class communities

Flickr photo by OldOnliner

A new report by Kendra Bischoff and Sean Reardon for the Russell Sage Foundation and Brown University found that, as a likely consequence of widening American income inequality, fewer and fewer Americans  live in urban middle-class neighborhoods and urban communities are instead increasingly polarized into rich and poor neighborhoods.  They call this increased “income segregation” or “family income segregation.”  Their report studies 117 metropolitan areas with a population of 500,000 or more in 2007 and examines these patterns at the census tract level, covering roughly two thirds of the US population.

This is bad news for the opportunity to build bridging social capital (social ties across race or social class), bad for building any sense that we’re all in this together, and by insulating the rich increasingly from the poor, makes it less likely that the rich will want to take action to help the poor (in the same way as the rich become less interested in public education investment if they send their kids to private schools or become less interested in safe streets if they live in a gated community with a private police force).  This research shows how children are less likely to grow up socializing with and playing with children of other socio-economic backgrounds, especially in an era where mandatory school-busing has come under attack.

These trends are all prior to the 2008 great recession, so it is impossible until 2013 to know whether that exacerbated these patterns or ameliorated them somewhat, and even then we probably won’t know about specific neighborhoods.

Bischoff points out that these segregation indices would change during the recession only if foreclosures or job losses force people to move, then income segregation could change.  For instance, if low- and moderate- income families need to move to lower-income neighborhoods, urban residential segregation would increase (more clustering).  Alternatively, if middle income families lose income, but remain in their homes (or neighborhoods), then residential income segregation would decrease as neighborhoods as increased family income volatility leads neighborhoods to become more diverse in income terms. The American Community Survey may never be able to resolve what happened at low levels of geography.

Excerpt:

As overall income inequality grew in the last four decades, high- and low-income families have become increasingly less likely to live near one another. Mixed income neighborhoods have grown rarer, while affluent and poor neighborhoods have grown much more common. In fact, the share of the population in large and moderate-sized metropolitan areas who live in the poorest and most affluent neighborhoods has more than doubled since 1970, while the share of families living in middle-income neighborhoods dropped from 65 percent to 44 percent. The residential isolation of the both poor and affluent families has grown over the last four decades, though affluent families have been generally more residentially isolated than poor families during this period. Income segregation among African Americans and Hispanics grew more rapidly than among non-Hispanic whites, especially since 2000. These trends are consequential because people are affected by the character of the local areas in which they live. The increasing concentration of income and wealth (and therefore of resources such as schools, parks, and public services) in a small number of neighborhoods results in greater disadvantages for the remaining neighborhoods where low- and middle-income families live.

Key findings, based on Census American Community Survey data:

  • From 2000 to 2007, family income segregation grew significantly in almost all metropolitan areas (in 89 percent of the large and moderate-sized metropolitan areas). This extends a trend over the period 1970-2000 during which income segregation grew dramatically. In 1970 only 15 percent of families were in neighborhoods that we classify as either affluent (neighborhoods where median incomes were greater than 150 percent of median income in their metropolitan areas) or poor (neighborhoods where median incomes were less than 67 percent of metropolitan median income). By 2007, 31 percent of families lived in such
    neighborhoods.
  • The affluent are more segregated from other Americans than the poor are. That is, high-income families are much less likely to live in neighborhoods with middle- and low-income families than low-income families are to live in neighborhoods with middle- and high-income families. This has been true for the last 40 years.
  • Income segregation among black and Hispanic families increased much more than did income segregation among white families from 1970 to 2007. Notably, income segregation among black and Hispanic families grew very sharply from 2000 to 2007. Income segregation among black and Hispanic families is now much higher than among white families.

Read “Growth in the Residential Segregation of Families by Income, 1970-2009” (Sean Reardon, Kendra Bischoff).

See NY Times story, “Middle-Class Areas Shrink as Income Gap Grows, New Report Finds” (11/16/11, by Sabrina Tavernise) which also shows this pattern for Philadelphia, which showed the biggest increase in income segregation over this period as well as the overall decline in middle-class neighborhoods and the rise of poor neighborhoods.

See earlier blog post: “Stalled upward social mobility in the US”

See somewhat related post by Liberty Street Economics (the blog of the Federal Reserve Bank of New York) that shows both that median wages were growing fastest in the high-skilled occupations, and also that job growth was fastest in both the high-skilled and low-skilled occupations and slowest in middle-class job occupation.

Social capital and disaster recovery

I recently heard a talk by Daniel Aldrich (Purdue).  He has been pursuing a handful of projects over the last 5-6 years looking at how local social capital (at the neighborhood or zip code or prefecture) predicts more resilient disaster recovery. Aldrich points out that people are far more likely to be hit by a disaster in their life than be the victim of a terrorist attack and asserts that the number of disasters is increasing in recent years.

Aldrich has studied 4 different disasters (1923 Tokyo earthquake; 1995 Kobe earthquake; 2005 Katrina disaster; 2004 Indian Ocean Tsunami).  I think he is currently doing some work on the recovery from the recent Japanese tsunami (early thoughts by him here).  Aldrich measures social capital with local measures like: voting and participation in rallies (1923 Tokyo); non-profit organizations per capita (1995 Kobe); number of funerals attended in past year (Indian Tsunami); and voter turnout (Katrina). His outcome variables for economic recovery are things like population growth (1923 Tokyo; Kobe) in an area or amount of aid received (Indian Tsunami) or ability to keep FEMA trailers out of an area (Katrina).  [It wasn't clear to me that this last measure is a measure of disaster recovery as much as NIMBY-ism, a topic that Aldrich has also written about.]

At one level, Aldrich’s findings are not surprising since places with low social capital tend to wait for the state to repair devastation and places with high social capital take more immediate self-action to repair.  This is reflected in Emily Chamlee-Wright’s recent book “The Cultural and Political Economy of Recovery: Social Learning in a post-disaster environment” and Robert Putnam observed this about Italian recovery from earthquakes: in places with high social capital one was unaware there had been an earthquake there several years later, whereas in low social capital places, the results of an earthquake were apparent 30-40 years later and residents were still blaming government for not adequately responding.

Aldrich’s work is very interesting and will appear next year as a U. Chicago press book “Building Resilience: Social Capital in Post-Disaster Recovery”.  [Brief presentation of his work here.] I would find his work even more interesting if he examined whether it is only the more political forms of social capital (like voting or protesting) that help in disaster recovery or whether it extends to “schmoozing” type variables at well (e.g., number of close friends, or knowing neighbors).   He might also be able to use volunteering data gathered by the CPS since 2002 to test that as a predictor or use datasets gathered by Rick Weil on social capital in New Orleans.  He also talks about the various types of social capital (bridging, bonding, linking) but his work doesn’t help sort out whether one type of social capital is more important than another for disaster recovery.  Also, given that social capital always rises after disasters and then most typically rapidly tails off, it would be useful if he tracked local social capital by neighborhood after a disaster since the shape of this drop-off in social capital need not be the same across communities; one might have more of a sustained burst of social capital than another.

His case study work does suggest that social capital is more important in disaster recovery than physical capital, physical infrastructure, or financial capital and more important than the conventional explanations that experts claim predict disaster recovery: amount of aid (positively predicting recovery); governance (stronger governance increases recovery); amount of devastation (less predicts greater recovery); wealth (positively predicting recovery); and population density (negatively predicting recovery).  He controls for these factors in his model and finds consistent and robust effects of social capital on post-disaster recovery.

Aldrich’s colleagues have also done some experiments of paying people to participate in focus groups, of giving people local “scrip” if they volunteer (which can be spent locally at farmers’ markets) and found that these built social capital and helped partially “inoculate” communities from the effects of disasters.  He didn’t present in any detail the methods or the results of these mini experiments.  He also recommends that post-disaster if we need to move survivors, we do them conscious of the clustering in their social networks,  so that they minimize the hit they take to their social capital.

Against this good news for social capital, there are three studies that find negative findings in the short-term, after disasters on outcomes like stress, health, etc. [I should note that Aldrich in his book addresses some negative outcomes of social capital in recoveries, for example, groups blocking certain castes from getting aid, or the Japanese promoting vicious attacks on Koreans after the 1923 Tokyo earthquake or ostracizing mercury victims in Minamata Bay from the late 1950s onward.] Basically the story of these other scholars is either that greater commitment to a community or greater social ties lead to worse ST outcomes, either because you feel it is really costly to leave or you are besieged by social and financial requests from other victims, which puts great strain on you unless you are wealthy.  The Rhodes et. al paper finds over longer term, people with more social capital do better, so this is a short-term finding only.  Weil and Lee, to my knowledge, have not looked at longer-term impacts.

See Jean Rhodes, Christian Chan, Christina Paxson, Cecelia Rouse, Mary C. Waters and Elizabeth Fussell. (2010) “The Impact of Hurricane Katrina on the Mental and Physical Health of Low-Income Parents in New Orleans.” American Journal of Orthopsychiatry 80(2):233-243. Not sure that their longer-term findings have been published.  See also manuscript from their project by Lowe, S. R., Chan, C. S., & Rhodes, J. E. “Pre-disaster social support protects against psychological distress: A longitudinal analysis of Hurricane Katrina survivors.”

Community Attachment and Negative Affective States in the Context of the BP Deepwater Horizon Disaster” by Matthew Lee and Troy Blanchard (LSU @ Baton Rouge) American Behavioral Scientist 55(12).  October 3, 2011

Weil, Frederick, Shihadeh, Edward, and Lee, Matthew. “The Burdens of Social Capital: How Socially-Involved People Dealt with Stress after Hurricane Katrina” Paper presented at the annual meeting of the American Sociological Association, 2006.

See also earlier blog post on disaster recovery from 2011 Japanese tsunami.