Call for more social science modeling for Social Security

An op-ed in the New York Times explains how poorly financial forecasts for Social Security are made and suggests social scientists can help:

Remarkably, since Social Security was created in 1935, the government’s forecasting methods have barely changed, even as a revolution in big data and statistics has transformed everything from baseball to retailing.

This omission can be explained by the fact that the Office of the Chief Actuary, the branch of the Social Security Administration that is responsible for the forecasts, is almost exclusively composed of, well, actuaries — without any serious representation of statisticians or social science methodologists. While these actuaries are highly responsible and careful and do excellent work curating and describing the data that go into the forecasts, their job is not to make statistical predictions. Yet the agency badly needs such expertise.

With considerable help from the actuaries and other officials at the Social Security Administration, we unearthed how the agency makes mortality forecasts and uses them to predict the program’s solvency. We learned that the methods are antiquated, subjective and needlessly complicated — and, as a result, are prone to error and to potential interference from political appointees. This may explain why the agency’s forecasts have, at times, changed significantly from year to year, even when there was little change in the underlying data.

We have made our methods, calculations and software available online at j.mp/SSecurity so that others can replicate or improve our forecasts. The implications of our findings go beyond social science. As the wave of retirement by the baby boomers continues, doing nothing to shore up Social Security’s solvency is irresponsible. If the amount of money coming in through payroll taxes does not increase and if the amount of money going out as benefits remains the same, the trust funds will become insolvent less than 20 years from now.

Sociologists seem to be looking for ways to get involved in major policy issues so perhaps this is one way to do that. It is also interesting to note this op-ed is based on a 2012 article in Demography titled “Statistical Security for Social Security.” Not too many articles can make such a claim…

Also, I’m sure this doesn’t inspire confidence among some for the government’s ability to keep track of all of its data. Does the federal government have the ability to hire and train the kind of people it needs? Can it compete with the private sector or political campaigns (think of what the lauded 2012 Obama campaign big data workers might be able to do)?

Rahm Emanuel: Chicago the model for pro-growth policies

Chicago Mayor Rahm Emanuel had an op-ed in the Washington Post on Friday where he explained how his city could show America the way toward growth:

While infrastructure improvements have been neglected on a federal level for decades, Chicago is making one of the nation’s largest coordinated investments, putting 30,000 residents to work over the next three years improving our roads, rails and runways; repairing our aged water system; and increasing access to gigabit-speed broadband. We are paying for these critical improvements through a combination of reforms, efficiencies and direct user fees, as well as creating the nation’s first city-level public-private infrastructure bank. Democrats should champion these kinds of innovative financing tools at a national level.

If we want to build a future in which the middle class can succeed, we must continue the push for reform that the president began with Race to the Top, bringing responsibility and accountability to our teachers and principals.

Chicago has adopted its own Race to the Top for early childhood education, allowing public schools, Head Start, charters and parochial schools to compete for dollars by improving the quality of their pre-kindergarten programs. In addition, this year Chicago Public Schools put into effect a 30 percent increase in class time, which means that when today’s kindergartners graduate high school, they will have benefited from 2½ more years’ worth of education.

In partnership with leading private-sector companies, we reengineered our six community colleges to focus each on skills training for jobs in one of Chicago’s six key growth fields. Democrats can be the party that closes the nation’s skills gap by making our community colleges a vital link between people looking for jobs and companies looking for skilled workers.

The strength of these investments is proven in the number of people we’re putting back to work: Chicago is first in the nation in terms of increase in employed residents, and for several months we have led the nation in year-over-year employment increases. We added 42,500 residents to the workforce in the past year alone — 8,000 more than the next highest U.S. city…

If Democrats develop innovative policies that help Americans compete in a global economy, we will outperform Republicans on Election Day. It’s that simple.

I’ve made this argument before (see here): Rahm Emanuel is more of a pro-business Democrat. As he notes in this article, he is in the mold of Bill Clinton who was willing to do what it takes to add jobs and fuel growth (illustrated by his recent push for digital billboards on city property alongside busy highways). And thus far, Emanuel has been able to push through his agenda in Chicago.

However, two things might hold back his arguments on the national level:

1. How much do Democrats and other Americans want government  to work closely private firms and corporations? Emanuel is a fan of public-private partnerships but people on both sides may not like this idea much.

2. Critics will charge that Chicago is hardly a model for others to emulate. Crime? Residential segregation? Massive budget issues? Battles with local unions? Underperforming schools?

I imagine some other big-city mayors might argue their cities could provide better models for the whole country. It would be fascinating to see a number of them respond with different visions.

(One last question: how much of this argument is simply boosterism from the mayor of the city’s third largest city?)

Civilization II a good “sociological simulator”? I say no

I was amused earlier this week to see a report from a guy who has been playing the same game of Civilization II for ten years. Here is a little bit of his report on the state of the Civ II world:

  • The world is a hellish nightmare of suffering and devastation.
  • There are 3 remaining super nations in the year 3991 A.D, each competing for the scant resources left on the planet after dozens of nuclear wars have rendered vast swaths of the world uninhabitable wastelands.

While I loved playing Civ II (and I think the gameplay was superior to later versions of the game), I’m scratching my head at how much attention this report has received in the media. Does it really tell us anything about the world’s possible future? Here is one overview from the BBC that I think goes too far:

A man who has been playing the computer game Civilisation II for ten years describes the year 3991 AD as a hellish nightmare of suffering and devastation.

Daniel Knowles, from the Telegraph and a fan of the game, says the game has certain assumptions built in to it about what will happen if there is a nuclear war or if you stop producing green technology.

“It’s a kind of sociological simulator… a giant economical model” he told the Today programme.

He believes gamer James Moore “would not still be playing it if he had reached an Utopia”.

Civilization II is a “sociological simulator”? I doubt it. Granted, the game is intended to replicate real-world nation-building and interaction. As you build your society, you have to make decisions like what kind of government to have (for example, in latter stages of the game fundamentalism is quite effective when waging all-out war), what to build and produce in individual cities, how to move certain units (military and otherwise) around, and pursue scientific and technological advancements. But, all of these types of games (and I’ve also been a fan in recent years of Age of Empires III) are only as good as what they account for. In other words, this is a low-level simulator of anything. The real world is far more complicated and many more moving pieces that games like this can allow. Indeed, these sorts of games seem geared toward all-out war between nations even as some would argue the international scene is getting more peaceful.

We are still far from a true “sociological simulator” that could account for all of the human variability in real life. This hasn’t stopped some scientists from trying – there was news recently of a group trying to put together a “Living Earth Simulator.” But, we need to remember what Civ II really is: it is a fun game with some modeling of human behavior but it really tells us very little or nothing about what the world might look like in 3991 AD.

The “gravity law” vs. the “radiation model” in predicting intercity mobility

Here is an overview of two ways to model intercity mobility: the “gravity law” and the “radiation model” which was just recently proposed in Nature:

The reigning model of intercity mobility, used to predict patterns of movement from commuting to the spread of infectious disease, is called the “gravity law.” It was developed in the early 1940s by a Harvard lecturer named George Zipf and is, of course, based on Newton’s law, which says gravitational force increases when the mass of two objects is great and the distance between them is minimal.

In that same spirit, Zipf’s “gravity law” of mobility assumed that movement between two cities would be most frequent when their populations were large and their separation small. In reality, however, the “gravity law” doesn’t do a great job estimating the intercity movement it was intended to predict. While Zipf’s law frowns on the notion that people travel frequently between distant cities, recent research on so-called “super-commuters,” outlined by our own Richard Florida, shows that a considerable subset of urban populations is actually willing to commute quite far…

The “radiation model,” as the new idea is called, makes several assumptions the gravity model does not. For starters, it downplays the distance between two cities and emphasizes not only the cities themselves but the density of the areas surrounding them. That enables the model to estimate the number of jobs in a region more accurately. It also accounts a bit more for actual human behavior: while the radiation model presumes that people choose a job based on a balance of proximity and benefits, it recognizes that they’re willing to make long commutes if few jobs in their region satisfy their requirements.

As a result, the radiation model out-predicts the “gravity law” in direct competition. As an example, the researchers looked at mobility between two pairs of counties in Utah and Alabama. Both counties of origin had similar populations, as did both destination counties, and both pairs are more or less equidistant from one another. Actual Census data shows that 44 people make the commute in Utah, while six do in Alabama.

This sounds very interesting and required advances in data collection on this topic as well as modeling social networks and demographics. The main finding seems to be this: distance is not the only factor that matters in looking at trips between cities. As the case of the super-commuters suggests, people will live one place and work in another place far away in the right circumstances. Perhaps we should have already known this because of the relative importance of different cities: world-class cities or cultural centers or centers for certain industries (New York City, Los Angeles, and San Francisco, respectively) would draw people from longer distances compared to “average” big cities (St. Louis? Denver?). Or, if we put this in world systems theory terms, certain cities sit at the center of American urban life and businesses and industries tend to concentrate within them while other cities are in more peripheral positions.

I would be interested to know whether the “radiation model” can suggest whether the number of super commuters will increase in the long-term and how this is affected by the strength of the overall economy and housing market.

Sociologist discusses the Living Earth Simulator

A sociologist explains a little bit more about the Living Earth Simulator that aims to model society:

Time travel: probably not going to happen any time soon. At least, not in the physical, “Back to the Future” sense. But that doesn’t stop us from trying to peek into the future. In the December issue of Scientific American, writer David Weinberger chats with Dirk Helbing, a Swiss physicist and sociologist who is pitching a project called the Living Earth Simulator, a billion-euro computer system that would absorb vast amounts of data, use it to model global-scale systems — economies, governments, etc. — and predict the future.

Well, maybe. Weinberger speaks with researchers who point out the roadblocks. While it’s possible to model small systems, such as highway and pedestrian traffic, getting a read on the economy, the environment and public health all at once is a much more complicated process. For instance, how would you account for feedback loops in the system — that is, what happens when the computer model’s conclusions alter the situation that it’s modeling? And if you can’t understand the process through which the model generates an answer, the whole thing is just a giant Magic 8 Ball, anyway. The computer may call upon world leaders to “set fire to all the world’s oil wells,” writes Weinberger. “That will not be actionable advice if the policymaker cannot explain why it’s right.”

So data mining will not be encouraged or will the model’s supervisors insist that every discovered pattern come with an explanation?

Interestingly, Helbing is also featured in a recent article in The Economist about pedestrian traffic:

In 1995 Mr Helbing and Peter Molnar, both physicists, came up with a “social force” computer model that used insights from the way that particles in fluids and gases behave to describe pedestrian movement. The model assumed that people are attracted by some things, such as the destination they are heading for, and repelled by others, such as another pedestrian in their path. It proved its worth by predicting several self-organising effects among crowds that are visible in real life.

One is the propensity of dense crowds spontaneously to break into lanes that allow people to move more efficiently in opposing directions. Individuals do not have to negotiate their way through a series of encounters with oncoming people; they can just follow the person in front. That works better than trying to overtake. Research by Mr Moussaid suggests that the effect of one person trying to walk faster than the people around them in a dense crowd is to force an opposing lane of pedestrians to split in two, which has the effect of breaking up the lane next door, and so on. Everyone moves slower as a result.

Two quick thoughts:

1. Combining physics and sociology to explain social behavior seems to be growing in popularity. Here is what I assume: the physics side brings experience in dealing with complex models and a more naturalistic way of explaining human behavior while sociologists bring more theories and knowledge about human contingencies. (But I could be wrong.) It does seem like the combination of these two disciplines could uniquely bridge the gap between the natural and social sciences.

2. Overall, I assume there will be many more projects like this. Getting the data is not so much a problem and we have the computing power to calculate complex models. If this does increase, this will mean some changes within the discipline of sociology: a shift toward mathematical sociology (making regression look relatively simple), thinking about “natural laws” in a way that sociology has generally avoided, and viewing the world in a different way (individuals operating within complex systems).

Living Earth Simulator to model social world

Here is an interesting project, the Living Earth Simulator, that hopes to take a lot of data and come to conclusions about social life:

Described as a “knowledge collider,” and now with a pledge of one billion euros from the European Union, the Living Earth Simulator is a new big data and supercomputing project that will attempt to uncover the underlying sociological and psychological laws that underpin human civilization. In the same way that CERN’s Large Hadron Collider smashes together protons to see what happens, the Living Earth Simulator (LES) will gather knowledge from a Planetary Nervous System (PNS — yes, really) to try to predict societal fluctuations such as political unrest, economic bubbles, disease epidemics, and so on.

Orchestrated by FuturICT, which is basically a consortium of preeminent scientists, computer science centers around the world, and high-power computing (HPC) installations, the Living Earth Simulator hopes to correlate huge amounts of data — including real-time sources such as Twitter and web news — and extant, but separate approaches currently being used by other institutions, into a big melting pot of information. To put it into scientific terms, the LES will analyze techno-socio-economic-environmental (!) systems. From this, FuturICT hopes to reveal the tacit agreements and hidden laws that actually govern society, rather than the explicit, far-removed-from-reality bills and acts that lawmakers inexorably enact…

The timing of EU’s billion-euro grant is telling, too. As you probably know, the European Union is struggling to keep the plates spinning, and the LES, rather handily, will probably be the most accurate predictor of economic stability in the world. Beyond money, though, it is hoped that the LES and PNS can finally tell us why humans do things, like watch a specific TV show, buy a useless gadget, or start a revolution.

Looking at the larger picture, the Living Earth Simulator is really an admission that we know more about the physical universe than the social. We can predict with startling accuracy whether an asteroid will hit Earth, but we know scant little about how society might actually react to an extinction-level event. We plough billions of dollars into studying the effects and extent of climate change, but what if understood enough of the psychology and sociology behind human nature to actually change our behavior?

I don’t know about the prospects of such a project but if the BBC is reporting on it, perhaps it has a future.

A couple of statements in the description above intrigue me:

1. The simulator will help uncover “the tacit agreements and hidden laws that actually govern society.” Do most social scientists think this is possible if we only had enough data and the right simulator?

2. The comparison between the natural and social sciences is telling. The portrayal here is that the natural sciences have come a lot further in studying nature than social scientists in studying human behavior. Is this true? Is this a fair comparison – natural systems vs. social systems? How much “unknown knowledge” is really in each realm?

3. The coding for this project must be immense.

4. The article makes no mention of utilizing social scientists to help develop this project and analyze the data though the group behind it does have some social scientists on board.

Predicting and preventing burglaries though statistical models in Indio, California

In January 2011, I wrote about how Santa Clara, California was going to use statistical models to predict where crime would take place and then deploy police accordingly. Another California community, Indio, is going down a similar route to reduce burglaries:

The Indio Police Department with the help of a college professor and a wealth of data and analysis is working on just that — predicting where certain burglaries will occur.The goal is to stop them from happening through effective deployment or preventative measures…

The police department began the Smart Policing Initiative a year ago with $220,617 in federal funding from the U.S. Department of Justice…

Robert Nash Parker, a professor of sociology at the University of California, Riverside and an expert on crime, is working with Indio.

On Friday, he shared his methodology for tracking truancy and burglary rates.

He used data from the police department, school district, U.S Census Bureau and probation departments, to create a model that can be used to predict such daytime burglaries.

Nash said that based on the data, truancy seems to lead to burglary hot spots.

A few issues come to mind:

1. Could criminals simply change up their patterns once they know about this program?

2. Do approaches like this simply treat the symptoms rather than the larger issues, in this case, truancy? It is a good thing to prevent crimes or arrest people quickly but what about working to limit the potential for crime in the first place?

3. I wonder how much data is required for this to work and how responsive it is to changes in the data.

4. Since this is being funded by a federal agency, can we expect larger roll-outs in the future? Think of this approach versus that of a big city like Chicago where there has been a greater emphasis on the use of cameras.

Using cell phone data to research social networks

Social network analysis is a growing area within sociology and other disciplines. The Wall Street Journal reports on the advantages of examining cell phone data:

As a tool for field research, the cellphone is unique. Unlike a conventional land-line telephone, a mobile phone usually is used by only one person, and it stays with that person everywhere, throughout the day. Phone companies routinely track a handset’s location (in part to connect it to the nearest cellphone tower) along with the timing and duration of phone calls and the user’s billing address…

Advances in statistics, psychology and the science of social networks are giving researchers the tools to find patterns of human dynamics too subtle to detect by other means. At Northeastern University in Boston, network physicists discovered just how predictable people could be by studying the travel routines of 100,000 European mobile-phone users.

After analyzing more than 16 million records of call date, time and position, the researchers determined that, taken together, people’s movements appeared to follow a mathematical pattern. The scientists said that, with enough information about past movements, they could forecast someone’s future whereabouts with 93.6% accuracy.

The pattern held true whether people stayed close to home or traveled widely, and wasn’t affected by the phone user’s age or gender.

The rest of the article then goes on to talk about a lot of interesting research on topics like social contagions (see an example of this research here) and social relationships using this data.

Some may be concerned about privacy, particularly with recent reports about iPhones and iPads containing a file that records the movements of users. I have a few thoughts about this:

1. Compared to other possible data sources (surveys, time diaries, interviews, ethnography), this seems like a treasure trove of information. The article suggests that nearly 75% of people in the world have cell phones – what other data source can compare with that? Could the research potential outweigh individual privacy concerns? In thinking about some of these research questions, it would be very difficult to use more traditional methods to address the same concerns. And just the sheer number of cases a researcher could access and work with is fantastic. In order to build more complex models of human behavior, this is exactly the kind of data one could use.

2. I would be less concerned about researchers using this data than companies. Researchers don’t particularly care about the individual cases in the data but rather are looking for broad patterns. I would also guess that the cell phone data is anonymized so that researchers would have a difficult time pinpointing specific individuals even if they wanted to.

3. How much of a surprise is it that this available data is being used? Don’t cell phone carriers include some sort of statement in their contracts about using data in such ways? One option here would be to not get a smart phone. But if you want a smart phone (and it seems that a lot of Americans do), then this is the tradeoff. This is similar to the tradeoff with Facebook: users willingly give their information to enhance their social lives and then the company can look for ways to profit from this information.

h/t Instapundit

Finding the right model to predict crime in Santa Cruz

Science fiction stories are usually the setting when people talk about predicting crimes. But it appears that the police department in Santa Cruz is working with an academic in order to forecast where crimes will take place:

Santa Cruz police could be the first department in Northern California that will deploy officers based on forecasting.

Santa Clara University assistant math professor Dr. George Mohler said the same algorithms used to predict aftershocks from earthquakes work to predict crime.”We started with theories from sociological and criminological fields of research that says offenders are more likely to return to a place where they’ve been successful in the past,” Mohler said.

To test his theory, Mohler plugged in several years worth of old burglary data from Los Angeles. When a burglary is reported, Mohler’s model tells police where and when a so-called “after crime” is likely to occur.

The Santa Cruz Police Department has turned over 10 years of crime data to Mohler to run in the model.

I wonder if we will be able to read about the outcome of this trial, regardless of whether the outcome is good or bad. If the outcome is bad, perhaps the police department or the academic would not want to publicize the results.

On one hand, this simply seems to be a problem of getting enough data to make accurate enough predictions. On the other hand, there will always be some error in the predictions. For example, how could a model predict something like what happened in Arizona this past weekend? Of course, one could include some random noise into the model – but these random guesses could easily be wrong.

And knowing the location of where crime would happen doesn’t necessarily mean that the crime could be prevented.

Obesity as “social contagion”

Findings in recent years that certain medical conditions, such as obesity, are strongly influenced by social networks have seemed to shake up thinking about how such conditions spread. A new model that explores obesity as a “social contagion” suggests obesity will eventually “spread” to at least 42% of the US population:

The social contagion hypothesis garnered widespread attention in 2007 when researchers from UC San Diego, documented that obesity can spread through a social network — just like viruses spread — because people “infect” each other with their perceptions of weight. That study was published in the New England Journal of Medicine, and subsequent research has confirmed the validity of the social contagion theory.

In a paper published Thursday in the journal PLoS Computational Biology, Harvard scientists applied a mathematical model of social contagion to 40 years of data from the Framingham Heart Study, a study that has followed more than 5,000 adult residents of Framingham, Mass., since 1948 to assess their heart health. Among the participants of that study, obesity increased from 14% in the 1970s to 30% in 2000 and continues to increase. Based on that data, the rapid upswing in obesity rates is due largely to social-network influence, said the authors of the new study. But, they noted, any subsequent rise in obesity rates will be slower than it has in the past. It may take another 40 years to reach that 42% obesity mark.

Interesting findings and I bet it is a complicated (but interesting) method by which such models are constructed.

If similar models were applied to other conditions or diseases, what are the results? In addition to obesity, what conditions are strongly influenced by social networks? Additionally, how can social networks be changed or altered so that an issue like obesity is combated?