The Tradition vs. Statistics War is Coming to Education

The most interesting non-hurricane story of the week has been the pushback by certain members of the political media establishment against Nate Silver’s election predictions. As Ezra Klein points out, at the heart of it is the conflict between traditional modes of analysis and statistical analysis. The traditional “come up with a theory and write about it” model of election prediction cannot coexist with one that strictly relies on statistical models. Only one method can be the best.

A similar conflict will eventually rock education. At the moment there is a large coalition that opposes relying on formal assessment data to evaluate student achievement. A lot of of the opposition stems from legitimate concerns with standardized tests, but within it you can sense the dedication to traditional methods of student evaluation. Many steadfastly believe that a teacher’s evaluation of a student (e.g. tests, homework, class participation, personal conversations, etc.) is the best measure of student achievement. They will not accept new people coming along and saying that state-mandated formal assessments are a better way.

The big battle has yet to take place because the statistics side still doesn’t have its act together. However, at some point people will create growth and value-added measures that are significant improvements. With many of the reasonable fears allayed, those who cling to traditional measures will be isolated in their opposition to statistical models. And then, as with political journalism, there will be a fight over who knows best.


How Recessions Can Influence Views On Social Spending

During economic downturns it’s understandable if certain people become more stingy about the way public money is spent. When the pie is smaller, you want to make sure pieces of it aren’t being given away to those who are undeserving. But what makes somebody undeserving? Can recessions influence whether people are seen as worthy of help, and thus influence preferences for resource distribution beyond what one might expect?

New research (pdf) by a group of TCU psychologists suggests that the answer is yes. In two experiments white participants were primed with thoughts of hardship or prosperity and then shown photos of biracial faces. The researchers found that participants primed with thoughts of economic hardship were less likely to categorize the faces as white.

The results suggest that in a bad economy people are less likely to think of somebody as part of their in-group. For example, during an economic boom you might feel that 64% of Americans are part of your in-group — hard working people who deserve government help if they need it. But when the economy tanks you may begin to feel that only 53% of Americans are part of that in-group. (Or put another way, that 47% of Americans take no personal responsibility.)

Because our tribal evolution has left us with a desire to protect our own groups, perceiving your group to be smaller means you’re less likely to support policies that help a broad range of people. This could partly explain why austerity can become popular when the economy is bad. Not only does scarcity make people people want to keep public funds out of the hands of the out-group, it makes people more likely to see that out-group as larger.
Rodeheffer, C., Hill, S., & Lord, C. (2012). Does This Recession Make Me Look Black? The Effect of Resource Scarcity on the Categorization of Biracial Faces Psychological Science DOI: 10.1177/0956797612450892

Why People Who Want “None of the Above” Should Vote For Obama

In the coming days there will be a flood of articles about the “none of the above” voters — people who don’t want either major party candidate to be president. Yet all of these articles will overlook the fact that if voters truly don’t want Obama or Romney to be the U.S. President, the logical thing to do is vote for Obama.

This year the “none of the above” voters are screwed. Either Obama or Romney is almost sure to be elected president. So the goal of these voters should evolve into ensuring that neither Obama nor Romney is elected president in 2016. Here’s where the decision become clear. If Obama wins this year there’s almost no chance that Obama or Romney will win the 2016 election. But if Romney wins there’s better than a 50% chance that Obama or Romney will win the 2016 election. For people who claim they don’t want either Obama or Romney to be president, a 2012 Romney victory is a disaster because it ensures that in 2016 one party’s nominee will be somebody they already disapprove of.

I’m not sure there’s any logical case that somebody who equally dislikes Romney and Obama can make for voting for Romney. Many of these “none of the above” voters will say that regardless of who wins this year there’s almost no chance they would approve of the 2016 Democratic or Republican nominees. Still, an unknown scenario is better than helping pave the road for the favored nominee to be somebody you already know you disapprove of.

An Accurate First Impression Is a Good First Impression

What’s the best way to make a good first impression? The question has dogged college freshmen and Project Runway contestants for ages, but some new research suggests the answer may lie in that old parental advice, “Be yourself.”

The study examined how the accuracy of first impressions influences the development of long-term relationships. Impression accuracy was quantified by measuring “distinctive self-other agreement” — i.e. the degree to which the way that a person is viewed matches their own self-description. The researchers found that over time an accurate first impression promotes more social interaction and better relationships.

Assessing accuracy as distinctive self-other agreement, we found that more accurate personality impressions of new classmates were marginally associated with greater liking concurrently, and significantly predicted greater interaction throughout the semester and greater liking and interest in future interactions by the end of the semester. Importantly, greater distinctive self-other agreement continued to promote social interaction even after controlling for Time 1 liking, suggesting that these positive effects of accuracy operate independently of initial liking…In sum, independent of the benefits of biased impressions, forming accurate impressions has a positive impact on relationship development among new acquaintances.

Why does this happen? The researchers speculate that accurate impressions help promote familiarity and comfort, and that they can make a person more likable by creating the sense that the person is easy to understand.

Though the study deals with specific circumstances involving the reciprocal relationship between two people, the mechanisms through which accurate first impressions foster relationships make it easy to see how the findings could apply to other situations — for example, the  “relationship” between a candidate and a voter. Perhaps the way we talk about a candidate being “genuine” or having the ability to “connect” is simply a function of positive evaluations that results from giving off accurate first impressions.
Human, L., Sandstrom, G., Biesanz, J., & Dunn, E. (2012). Accurate First Impressions Leave a Lasting Impression: The Long-Term Effects of Distinctive Self-Other Agreement on Relationship Development Social Psychological and Personality Science DOI: 10.1177/1948550612463735

A Bipartisan Argument For Increasing the Size of Congress

The idea that we should increase the size of Congres is not new. There’s often speculation it could help representatives stay in touch with constituents, create more diversity in government, and inch the country closer to the citizens-to-representatives ratio our founders intended.

An intriguing new paper by Nels Christiansen in the Journal of Public Economics offers another reason for making Congress bigger: More representatives will lead to more of the budget being spent on public goods.

Christiansen’s analysis is based on “strategic delegation” — the idea that it may be in a district’s best interest to elect a representative who doesn’t share the district’s preferences. For example, a district’s voters may prefer the Democratic candidate in a year when the GOP will control Congress. Because a member of the majority has more influence, the district may get more priorities taken care of (e.g. funding for a new hospital) by electing the Republican.

The model Christiansen uses in the paper is based on a simplified system in which representatives are either “hogs” (they want to maximize the pork projects brought home to their district), or “pubs” (they want to maximize a single public good.). Their jobs are to divide up a fixed set of money, but the money must go to the public good or to a specific representative’s pork project. Christiansen examined a variety of situations and found that districts preferring hogs will sometimes be best off if they elect pubs and reap the rewards of a well-funded public good. On the other hand, districts preferring pubs will never be better off if they elect hogs. Because hog districts can elect pubs but not vice versa, there will always be at least as many pub representatives as pub districts. The result is that increasing the number of districts can increase the proportion of pubs but not hogs. Thus, increasing the number of districts ought to increase the proportion of the budget that goes toward the public good.

We show that voters have an incentive to strategically delegate to affect how the budget is divided at the legislative level. When voters’ preferences for pork are not too strong, the incentives for strategic delegation exist to appoint representatives who will direct more money to the public good and not to pork projects. This generally results in at least as many representatives as districts that favor the public good. The comparative statics predict that when strategic delegation occurs, increasing the size of the legislature increases the fraction of the budget spent on the public good.

The reality of the American political system is obviously more complex than the model in the paper, but the implication is that increasing the size of Congress might make spending more efficient. This should thrill liberals because it gives people more government services, but it should also thrill conservatives because spending on public goods is generally less wasteful than spending on a specific district.
Christiansen, N. (2012). Strategic Delegation in a Legislative Bargaining Model with Pork and Public Goods Journal of Public Economics DOI: 10.1016/j.jpubeco.2012.10.001

Don’t Wear Red to a Job Interview

I doubt society is erring on the side of too little concern about personal appearance, but psychology research has established that the colors you wear can influence how you’re perceived. The most well-known effect is that wearing red makes your more sexually desirable, but a new study suggests that if you’re looking for employment, wearing red is a bad decision:

We conducted three experiments examining the influence of red on the evaluation of male target persons. In Experiment 1, participants viewing red, relative to green, on the shirt of a person presented on a photograph perceived him to be less intelligent. This effect was strongest in a job application context compared to other contexts. In Experiment 2, focusing solely on the job application context, participants viewing red, relative to blue, on an applicants’ tie perceived him to have less earning and leadership potential. In Experiment 3, participants viewing red, relative to green, on a job applicants’ tie rated him as less likely to be hired, and perceptions of ability and leadership potential mediated this effect.

Studies on impression formation tend to get a lot of attention, but the question with this genre of research is whether the effects hold once people acquire better information. For example, an employer looking at a photo of a job candidate wearing red may rate him as less competent, but when the employer observes how the candidate talks and thinks newer judgements may replace those that were influenced by the candidate’s clothing. In other words, I’m not convinced that somebody’s opinion of Mitt Romney on election day is even marginally influenced by the red tie Romney was wearing when the person first saw him on TV.

It would be interesting to see a follow up experiment examine the effects of clothing color when participants watch a video of job candidates saying something substantive. Although first impressions do persist, my inclination is to think that studies using photos lead people to overestimate the medium- to long-run impact of clothing color. Of course if you happen to be speed dating/interviewing, by all means load up on the red/anti-red outfits. There’s also the question of what to do if  you’re romantically interested in the person who’s going to be interviewing you. My advice is to wear blue and hope you have a good personality.
Maier, M., Elliot, A., Lee, B., Lichtenfeld, S., Barchfeld, P., & Pekrun, R. (2012). The influence of red on impression formation in a job application context Motivation and Emotion DOI: 10.1007/s11031-012-9326-1

The Hypothetical Situation Driving Education Reform Animosity

Reihan Salam’s latest piece on education reform is worth reading in full, but he does a particularly good job of articulating the thinking behind one of the least discussed, but most promising ways of altering the way we educate children — increased specialization:

Instead of simply increasing the number of teachers, Hess and Meeks propose shifting teaching from a profession built around generalists – people who teach reading and fractions and supervise bus-loading and monitor the cafeteria and grade papers – to one built around specialists. Just as the Mayo Clinic has specialists working on discrete medical problems (cardiologists here, neurosurgeons there) and support staff who enable them to do their work, schools could “unbundle” the job of teaching. We don’t find it strange or scandalous that highly trained obstetricians don’t also clear bedpans. In the same vein, schools should rely more heavily on support staff to load the bus, monitor the cafeteria and grade exams while letting teachers who are really great at teaching fifth grade geometry focus on teaching fifth grade geometry. Like medical specialists, specialist teachers in the rarest, most demanding fields should expect more compensation. School employees with skills that aren’t quite as uncommon, meanwhile, could be paid less without sacrificing quality.

This particular scenario is a key cause of the animosity and vacuousness that plagues education policy debates.  Think about what an era of specialization would mean for teachers. Many who don’t earn “lecturer” status would receive pay cuts or be relegated to jobs they felt were beneath them. Others might need new training or credentialing in order to stay employed. Overall, while it would unequivocally be better for kids, it would be a disaster for unions.

One reason more fire is being trained on unions is that they’ve failed to effectively respond to these hypothetical, but realistic situations in which union priorities conflict with student priorities. The standard response is to reaffirm that anything bad for teachers has t0, by definition, be bad for kids, a point that’s incredibly shortsighted given the potential for technological advances.  What’s worse, unions are bent on destroying charter schools, the 4% of our education system where schools can experiment with some of these new ways of instruction.

Faced with this apparent close-mindedness, many have decided that the optimal reform strategy is to eschew any specific policies and simply do everything possible to destroy unions. The result is that unions see opponents who are only concerned with union destruction, and they respond by vehemently opposing everything these opponents might conceivably favor. This union recalcitrance then confirms everything their opponents initially believed. The cycle continues until we’re left with one side claiming it’s bad when rich people spend money to educate poor kids and the other side claiming that teachers are lazy, selfish, money-grubbing jerks.

The larger consequence is that for all the hype that certain current events (e.g. CTU strike, NYC mayoral election) are huge battles for the future of the education system, the real battle royale will take place a few years from now when there’s enough data from from Rocketship, School of One, and Teach to One that we’ll have a real answer to questions about computer-based learning. At that point it will be up to “reformers” to prove to the public that it’s possible for something that’s bad for teachers to be good for kids, and if they can, for unions to come up with a legitimate response.