Obama's Science Budget

Wired has a handy chart of which agencies get what in Obama's new budget proposal. The NIH gets an extra $1 billion, NSF, NASA, and the EPA each get an extra $500M, and a few others get smaller increases. Only the CDC's budget gets chopped.

Given that the proposed budget comes with a giant helping of deficit spending it's unlikely that these raises will make it through intact. NIH's proposed budget increase (3.2%) is right on par with inflation (2.7%), so if it gets reduced, expect much wailing and gnashing of teeth.

[0 comments]

Money in Science

John Tierney has a piece in the Times today pooh poohing people who are concerned about the role of corporate money science: Corporate Backing for Research? Get Over It

He may well be right to say

Conflict-of-interest accusations have become the simplest strategy for avoiding a substantive debate. The growing obsession with following the money too often leads to nothing but cheap ad hominem attacks.

but I think that he is overly dismissive of attempts to ensure that people disclose possible conflicts of interest:

The new fetish for disclosing “conflicts” has led some of the best medical researchers to shun drug company money altogether — not because they think it leads to bad research, but because they are tired of that fact being highlighted every time they are identified in a news story, as if that were the most important thing to know about their work.

Corporate research has multiple purposes. Internal research helps companies develop and test products, of course, but research that gets released externally serves different roles:

• First, research released in public legitimizes the corporation as a place to do research and helps to attract smart people. As best I can tell, this is why Microsoft Research set up their Theory Group. Originally it was a handful of mathematicians trying to prove that P &neq; NP via some rather esoteric methods. The group's primary contribution to Microsoft was press releases and papers - in effect, they were a very expensive hood ornament that let people doing more practical things at MSR say, "I work with a Fields Medalist."
• Second, public research serves as a form of compensation for employees - it lets them build their reputations in their professional communities, which is useful for future opportunities. I've seen an econ paper that estimated that researchers in pharma who were able to publish were paid ~20% less than those whose research was all secret, so publishing really does appear to matter in a quantifiable way (alas, I have no idea where the reference is).

Both of these goals have to be balanced with companies' need to protect intellectual property, preserve their brand, and so on - there are limits on what one can publish. Some companies will value the HR benefits more highly than the need for secrecy / brand preservation, so employees will be able to publish fairly freely; others will put a greater premium on secrecy and image control, and employees' work will be highly filtered. Filtering need not come from management, either. Promotion in industry is not a function of the weight of your publications, and there will simply be some kinds of work that are not worth one's time to publish because it won't do one's career much good.

The point: while it may well be true that , "a team of academic researchers (not financed by industry) analyzed dozens of large-scale clinical trials in previous decades and reported that industry-sponsored ones met significantly higher standards than the nonindustry ones," it is possible that subtle (and in some cases not-so-subtle) biases can arise from the various forms of filtering. The work in industry may be of excellent quality, but some kinds of results may never see the light of day - for example, negative results from some kinds of clinical trials.

I don't mean to demonize industry here - I think that all institutions and funders create incentive structures that favor some kinds of work over others. The point is that disclosure of funding sources and interests is important because it provides readers with valuable context and insights into what kinds of questions may not have been asked.

[2 comments]

Math anxiety is contagious

ArsTechnica has a great summary of a study today entitled Female teachers transmit math anxiety to female students. The quick version:

The study found that when elementary school teachers, who are primarily female, displayed a high level of anxiety about math, that skittishness was transmitted to their female students. Those students who spent a year with a math-phobic teacher displayed lower math achievement and an increased belief in stereotypes about female mathematical ability.

This is cool for a couple of reasons.

First, there's the possibility of a (slow) virtuous cycle. Increase the maths skills (and comfort levels) of female elementary school teachers, and you increase the performance of girls in math. This should in turn lead to increased math skills for elementary school teachers down the road when the students become teachers.

Second, the study underscores the importance of applying the scientific method to the teaching of science and then applying the results. There are a lot of subtle things about how we communicate that turn out to be really important determinants of our effectiveness at getting our point across (see stereotype threat). Gladwell spends a chapter on some of these ideas in The Tipping Point and in Blink. Yet the people who teach college level science and math at major universities almost never learn these things - grad students get thrown into classrooms as TAs or instructors and have to learn to teach on their own.

How hard would it be to distill down the most important and relevant findings in education research to a manageable (10-20 page, say) document that you could make sure everybody read before teaching college level courses?

[0 comments]

Network, network, network

I got my job at Google due in large part to having a friend working there in a similar position. My Microsoft job I found through a friend of a friend. My postdoc advisor at Dartmouth had met me previously at a conference. The same for my stint at Rice. And so on. Pretty much the only job I've ever gotten without having any kind of personal connection at the employer was my stint as a bicycle messenger in DC.

One reads such tales about the importance of networking all the time. What's unsatisfying is that they're all anecdotal.

I've been reading Malcolm Gladwell's books recently (they're great!) and came across this in The Tipping Point:

In his classic 1974 study Getting a Job, Granovetter looked at several hundred professional and technical workers from the Boston suburb of Newton, interviewing them in some detail on their employment history. He found that 56 percent of those he talked to found their job through a personal connection. Another 18.8 percent used formal means - advertisements, headhunters - and roughly 20 percent applied directly.

Note that the people in question are professional and technical workers - probably not so different than PhDs looking for industry jobs. So this kind of thing is quite common.

But, curiously, Granovetter found that of those personal connections, the majority were "weak ties." Of those who used a contact to find a job, only 16.7 percent saw that contact "often" - as they would if the contact were a good friend - and 55.6 percent saw their contact only "occasionally." Twenty-eight percent saw the contact "rarely." People weren't getting their jobs through friends. They were getting them through their acquaintances.

Gladwell goes on to discuss the idea of "weak ties" in some detail, and it's pretty interesting. The point for PhDs is that job networking isn't about having tons of friends - it's more about having lots of acquaintances. That means getting out to conferences and making sure that you meet lots of people. But there are some less obvious ways to make a lot of acquaintances, too:

• Participate in open source software projects. It's a great way to showcase your work, you learn a lot of valuable skills, and you can meet people who are outside the academic bubble. That was how I met my Google connection.
• Participate in relevant non-research-related activities - organize a speaker series, participate in your graduate student or postdoc association, take a leadership role with your professional society, blog. My Microsoft connection I knew through science policy activities.
• Find hobbies that bring you in touch with lots of people. I've had a few interesting job leads through one of my running clubs.

[0 comments]

PE for grad students

Another study links aerobic exercise to increased cognitive performance:

Start running and watch your brain grow, say scientists

Running (a lot of it) was found to improve the performance of mice on tests of memory. The mechanism: increased neurogenesis.

There's a fair amount of evidence that exercise boosts cognitive performance in people, too. A few overviews:

• Brain Rules: 12 Principles for Surviving and Thriving at Work, Home, and School - the author gave a great talk at Google that's on YouTube

• Spark: The Revolutionary New Science of Exercise and the Brain

So, grad students and postdocs, lace up your running shoes!

[4 comments]

Be thankful you aren't in the humanities

The title says it all: "Graduate School in the Humanities: Just Don't Go". A particularly grim line:

...some responsible observers expect that hiring may be down 40 percent this year. What is 40 percent worse than desperate?

Something that rings true beyond the humanities is the author's characterization of the reasons that people go to graduate school in spite of poor career prospects afterwards (hello life scientists):

I have found that most prospective graduate students have given little thought to what will happen to them after they complete their doctorates. They assume that everyone finds a decent position somewhere, even if it's "only" at a community college (expressed with a shudder). Besides, the completion of graduate school seems impossibly far away, so their concerns are mostly focused on the present. Their motives are usually some combination of the following:

• They are excited by some subject and believe they have a deep, sustainable interest in it. (But ask follow-up questions and you find that it is only deep in relation to their undergraduate peers — not in relation to the kind of serious dedication you need in graduate programs.)
• They received high grades and a lot of praise from their professors, and they are not finding similar encouragement outside of an academic environment. They want to return to a context in which they feel validated.
• They are emerging from 16 years of institutional living: a clear, step-by-step process of advancement toward a goal, with measured outcomes, constant reinforcement and support, and clearly defined hierarchies. The world outside school seems so unstructured, ambiguous, difficult to navigate, and frightening.
• With the prospect of an unappealing, entry-level job on the horizon, life in college becomes increasingly idealized. They think graduate school will continue that romantic experience and enable them to stay in college forever as teacher-scholars.
• They can't find a position anywhere that uses the skills on which they most prided themselves in college. They are forced to learn about new things that don't interest them nearly as much. No one is impressed by their knowledge of Jane Austen. There are no mentors to guide and protect them, and they turn to former teachers for help.
• They think that graduate school is a good place to hide from the recession. They'll spend a few years studying literature, preferably on a fellowship, and then, if academe doesn't seem appealing or open to them, they will simply look for a job when the market has improved. And, you know, all those baby boomers have to retire someday, and when that happens, there will be jobs available in academe.

[3 comments]

Assumptions

In reading a review of Louis Menand's The Marketplace in Slate today, I came across this interesting piece of history:

...Professors, the people most visibly responsible for the creation of new ideas, have, over the last century, become all too consummate professionals, initiates in a system committed to its own protection and perpetuation.... They have been taught to think of their own work, which is accountable only to the internal standards of their profession, as something pure, something unrelated to the messy business of the world. But this belief itself was only ever dreamed up as a solution to different problems, and once we understand it as a matter of historical contingency, we shall presumably be better able to deal with its consequences.

The dissociation of scholarship from the rest of the world was largely engineered to in order to (1) insulate the faculty from religious interference and (2) to stave off competition from professional schools by inserting a liberal arts undergraduate training as a prerequisite to degrees in law/business/medicine.

So the idea that universities should focus on knowledge for its own sake rather than "applied" things is not an absolute Truth, but rather part of a 19th century liberal arts faculty jobs-preservation bargain.

[2 comments]

Fighting yesterday's battle

In Wired this week: Darpa: U.S. Geek Shortage Is National Security Risk

The reason for the concern:

According to the Computer Research Association, computer science enrollment dropped 43 percent between 2003 and 2006.

Hm. Why might that be? There is a great paper by Richard Freeman that makes a strong case that year-to-year fluctuations in enrollments in various majors are strongly influenced by the relative prospects in the field at the time of major selection.

Students typically choose their majors in their first or second year of college, and it takes 4-5 years to graduate. Students graduating in 2003 would have made the choice to major in CS near the peak of the dot-com bubble - a time when things looked pretty good for CS majors. In contrast, the class of 2006 made their choices around 2003, not long after the crash that followed. So it's no big surprise that enrollments dropped.

Since 2003, things have gotten a lot better in the IT world, and it would be quite surprising if enrollments hadn't picked back up again. It's likely that the problem has already solved itself.

CS in particular offers some quite attractive career options these days. There's a whole world of people doing contract work in Ruby on Rails, AppEngine, Django, etc. who are able to work for themselves or with a small group of friends, choose their own hours, and get paid quite handsomely (I know all too well because I've had to hire some Rails contractors recently). One big downside: buying health insurance as an individual can be pretty tough if you're someone who ever has to use it. It's entirely possible that health insurance reform would do as much for encouraging more CS grads as anything DARPA is proposing.

[0 comments]