One of the many things I wish I had had time to blog about during the just-completed term was the big New York Times article on attrition in science majors. This generated enough commentary at the time that people are probably sick of it, but I haven't seen anything that exactly matches my take, so I'll belatedly throw this out there.
The big point of the article is that lots of students who enter college planning to major in Science, Technology, Engineering or Math (the "STEM" fields, in an awkward but now inescapable acronym) end up graduating with degrees in something else:
But, it turns out, middle and high school students are having most of the fun, building their erector sets and dropping eggs into water to test the first law of motion. The excitement quickly fades as students brush up against the reality of what David E. Goldberg, an emeritus engineering professor, calls "the math-science death march." Freshmen in college wade through a blizzard of calculus, physics and chemistry in lecture halls with hundreds of other students. And then many wash out.
Studies have found that roughly 40 percent of students planning engineering and science majors end up switching to other subjects or failing to get any degree. That increases to as much as 60 percent when pre-medical students, who typically have the strongest SAT scores and high school science preparation, are included, according to new data from the University of California at Los Angeles. That is twice the combined attrition rate of all other majors.
This sounds like the sort of thing that I ought to be all over, as I'm on record saying repeatedly that everybody can do science. I'm weirdly ambivalent about this, though, because as someone who sees a lot of first-year college students who think they want to major in a STEM field, I'm not sure that all that attrition is a Bad Thing. Some of it, particularly on the pre-med side, is probably a Good Thing, moving out people who really shouldn't be in those majors in the first place. Not because they aren't "smart enough" to do it-- many of them will go on to be very successful in other fields-- but because they don't have any idea what they're signing up for.
I see a lot of this because we spend a lot of time teaching introductory physics to first-year engineering majors, many of whom are coming in as engineering majors because they know that engineers make a lot of money, and either they or their parents (or some linear superposition of the two) believe it would be a good career path. The problem is, that's often the only thing they know about engineering. They don't have the foggiest idea what engineers actually do, just that they get paid well, and it involves math.
A lot of those students get driven out early in the curriculum because those intro classes are their first exposure to what real science and engineering involves, and they turn out not to like it. The mental processes involved in science and engineering are things that any human is capable of doing, but it takes a certain personality type to enjoy doing them. If you're not the type of person who enjoys playing around with things to see how they work (as opposed to always following step-by-step instructions), or figuring out ways to make graphs and charts comparing odd things in a quantitative way, you're probably not going to enjoy being a scientist or an engineer.
And there's absolutely nothing wrong with that. Everyone is capable of doing and even casually enjoying activities that they would absolutely hate to do professionally. I really enjoy reading fiction and history, but I would be absolutely terrible as a scholar in the humanities, because I don't have the right personality type to do the sort of thing that they do in those disciplines. I have very little patience for raising questions without answering them, which makes many of my interactions with my colleagues on the other side of campus very frustrating for everyone. I'm happy to roll with it for the occasional convention panel or faculty colloquium, but I could never do that for a living.
I suspect that similar processes are at work in the really high pre-med attrition rate-- a huge number of students come into college thinking that they're going to be doctors, because they and their parents know that doctors make a lot of money. They don't find out until they get to college just how obsessive you need to be to succeed in the pre-med track, and as a result, they switch to something else. Which is definitely not a bad thing, because you really do want your eventual doctor to pay attention to details.
I suspect that there's probably a similar level of attrition for would-be lawyers, but it's harder to track as lots of different majors can serve as a starting point for law school. It's not as critical to keep track of the pre-law contingent, so it's a bit harder to identify those students who entered college with a vague idea of becoming a highly paid lawyer, but dropped that once they learned how much writing is involved. (See also this post about grades and legal careers, but this is getting off track.)
So, attrition in science and engineering majors is not in an of itself a Bad Thing. It's absolutely true that everybody can do science, but that doesn't mean everybody should do it for a living. It's perfectly fine to have students sort themselves into whatever fields are best suited to their personality-- that's part of what college is for in modern American society, after all-- regardless of what sounded like a good idea in their high school guidance counselor's office.
Of course, there's an important caveat, here: the sorting by personality is fine provided it's done on a rational basis. If we're driving out people who aren't temperamentally suited to being professional scientists and engineers, that's probably a good thing on balance, but if we're driving out people who have the right personality type to be good scientists and engineers because the intro classes are boring and useless, that's a bad thing. This is the idea driving the various reform curricula in physics, and ideas about active learning, peer instruction, and the rest. Badly done traditional lectures will drive away some people who would be really good scientists, but who have a low tolerance for pointless drudgery.
The Times piece doesn't really make it clear to what extent the problem really is with the presentation of the introductory material, as opposed to the unrealistic expectations of students. This sort of thing is damnably difficult to sort out, and is probably better measured by attitudinal surveys (CLASS, MPEX, VASS, and so on) than enrollment statistics, student grades, or anecdotal interviews with education reporters.
Badly done traditional lectures will drive away some people who would be really good scientists, but who have a low tolerance for pointless drudgery.
While this statement is, as far as it goes, true, be a bit careful about extensions to it. I dare say that even the best (on its own terms) interactive group-based conceptual clicker-driven buzz-word buzz-word buzz-word class could persuade a certain type to change majors, if they find that they have no interest in spending their days making graphs and tables to model the motion of a ball on an inclined plane.
Now, some of those conceptual approaches give less emphasis to math, and if that increases retention, that is not necessarily a good thing.
OTOH, if the reformed, progressive approaches simply make it more interesting to learn the same core skill-set, and enable better mastery, and that improves retention, that is of course a good thing.
Are you sure everybody can do science? It might be just a matter of temperament and interest, but ability probably plays a role as well. There could be tradeoffs like: at X level of ability you need interest level Y to succeed. It appears this relationship could be nonlinear for physics and math: below a certain ability level (say, 90th percentile math ability), you need > 99th percentile motivation or work ethic to master the undergraduate curriculum.
PS If I were to rephrase your argument replacing "everyone can do science" with "everyone can pole vault" or "everyone can run the 110m high hurdles", would you still advocate that the major factor was interest/motivation as opposed to a combination of ability + interest/motivation?
Interesting. I come at this from the math side, as a statistician. I'm not sure how varied the science backgrounds of your students are, but the math backgrounds of my new students are immensely varied. At the high end are students who could, I think, be successful at any institution but want a small school experience as their first exposure to college. At the other end of the scale were a couple students who were astounded that statistics required more than college algebra to master.
In the middle, most fall by the wayside because (from evaluations) "the classes take too much time,", "we shouldn't have to worry about writing well in math classes," and so on. A lot of math is difficult: we haven't done a good job of making that point.
Lighter side: this "...dropping eggs into water to test the first law of motion..." At the end of my high school physics class we had the chance to earn extra credit by creating packaging for an egg that would prevent egg breakage when the package was dropped from the extended ladder of a fire truck. One of my classmates, call him "Steve", insisted on going last. Our teacher agreed. Finally it was his turn. Since we weren't allowed to climb the ladder, Steve's package was given to the fireman. As he held it he got a strange look on his face and said something to our teacher. They opened it to find: a hen, looking stunned and a little embarrassed. When called on it Steve said "I know there's an egg in there." My favorite line: the teacher replied "Steve, you are not going to drop a chicken off the top of a ladder."
Steve Hsu @ 3:
If I were to rephrase your argument replacing "everyone can do science" with "everyone can pole vault" or "everyone can run the 110m high hurdles", would you still advocate that the major factor was interest/motivation as opposed to a combination of ability + interest/motivation?
Hey, I can do both of those, so long as a 2.5 foot high vault and a 5:00 min hurdle time are acceptable...
Regarding non-elite schools having better retention, I offer two speculations:
1) Elite schools usually (with all exceptions duly noted) cost a lot. There's a strong incentive to try something else if your first few science and math classes don't work out. I went to an elite school, and many of the people who were in STEM as freshmen switched to something else as sophomores. OTOH, I teach at a non-elite regional state school. Here, students will try and try again. The five year plan is a less costly proposition here than at a more elite place.
2) A significant part of the value of a degree from an elite school comes from a combination of signaling (you got a degree from a place full of smart people) and networking (you join a network of successful alumni). At a non-elite school, the name of the school might have less value than the name of the major. Now, I'm well aware that plenty of liberal arts graduates do quite well career-wise, but many students (and the parents paying for that 5-year plan) perceive (rightly or wrongly) STEM to be more lucrative. If you aren't getting a diploma that has a fancy school name on it, you'd better at least get an employable major (or one perceived as such) stamped on it.
Hence, at a non-elite school, struggling in STEM seems (rightly or wrongly) like the best option. At an elite school, doing whatever it takes to get that diploma seems like the best option.
I don't claim to have anything other than anecdotal observations to back this up. However, it's worth noting that the NYT article segued from observations on retention to observations on teaching methods without establishing that the schools with better retention are more likely to use better teaching methods. If anything, I would note that major curricular overhauls usually require resources (release time for the redesign, equipment for teaching labs, sometimes even building renovations to convert a lecture room to a room for group projects), and the more elite places (usually) have more resources. That's not to say that curricular overhauls don't happen at non-elite places, and certainly some elite places are (not surprisingly) quite conservative, but at the very least I don't know that elite places are (on average) any less likely to use more progressive methods.
The fact that the elite schools wash out more STEM majors than non-elite schools was interesting to me. Assuming that "temperament" is a relevant variable, is it that elite schools value and promote ONLY PhD level research scientists as the logical outcome for all STEM majors (MDs excepted), and non-elite schools do a much better job of portraying/preparing students for a broad array of careers?
In which case, folks are washing out in the elite schools because they don't have the temperament to be their professors, not because they don't have the temperament to pursue a STEM career.
Which brings me to my next point- as a student, one tends to judge whether one is "temperamentally" suited to a certain career by whether one's personality matches one's available role models. The way systematic biases produced skewed backgrounds of current professors (including race, gender, SES, ect.) are thus perpetuated (even in a hypothetical world in which systematic biases are gone for the current crop of students), unless professors and others go above and beyond in providing role models.
Alex- I like your second explanation. I also think it's worth noting the broader issue- students at elite institutions usually have (or perceive that they have) more options. Not just as a matter of 'brand name' university, but because the types of students that get to those places often have more resources. The risks of an English Literature degree is lower when you have the safety net of parents to move back in with.
In addition, students at elite institutions may be conditioned to success. Some people aren't willing to spend 8 hours one one homework problem, not because they can't handle that, but because it makes them feel dumb when one or two stars finish it in half an hour (even if most students do not and 8 hours/problem represents a pretty successful student).
"If you're not the type of person who enjoys playing around with things to see how they work (as opposed to always following step-by-step instructions)"
And that is why I am Highly Suspicious of a "temperament" argument- I don't buy most people's descriptions of the required temperament as being useful ways of categorizing people.
I call bullpockey on the notion this "playing around with things to see how they work" is not a fundamental relatively universal human activity. How can anyone watching young children doubt that the vast majority of people can take great pleasure in figuring out how things work ???
There are people who have the *patience* to follow instructions, and people who don't. There are people who are *efficient* at playing around with certain types of systems, and people who have yet to learn enough to do that (and who, therefore, have more incentive to read instructions). But the only people who don't like to figure things out are people who have forgotten how much fun it is and/or had the fun sucked out of it by e.g. bad schooling.
Everyone is a scientist as a child, the trouble is how to remain one when you grow up?
Alex: OTOH, if the reformed, progressive approaches simply make it more interesting to learn the same core skill-set, and enable better mastery, and that improves retention, that is of course a good thing.
That's the claim made by most reform advocates, and it seems plausible given my limited experience with the reform methods. I certainly didn't see any significant decrease in problem-solving ability during the past term's experiment in active learning. This has a lot to do with the fact that the baseline level was pretty low, but still...
Steve: Are you sure everybody can do science? It might be just a matter of temperament and interest, but ability probably plays a role as well. There could be tradeoffs like: at X level of ability you need interest level Y to succeed.
You're comparing two things that aren't quite alike, here. When I say that "everybody can do science," I don't mean that everybody can succeed as a science major, I mean that everybody can do the mental tasks needed to be a scientist. That doesn't mean they will be able to do them professionally, or that they will enjoy them enough to major in the subject.
If forced to, I could learn to pole-vault, because I have the requisite innate physical abilities. I'll never be competitive as a pole-vaulter, for reasons of temperament and body type, but that doesn't mean I can't do it at all.
dean: I come at this from the math side, as a statistician. I'm not sure how varied the science backgrounds of your students are, but the math backgrounds of my new students are immensely varied. At the high end are students who could, I think, be successful at any institution but want a small school experience as their first exposure to college. At the other end of the scale were a couple students who were astounded that statistics required more than college algebra to master.
That's about where we are, too. We get a handful of students who are sophisticated enough mathematically to be uncomfortable with some of the swashbuckling physicist tricks we pull (converting discrete changes into continuous derivatives in a very cavalier manner, for example), and we get a handful who honestly believe that 1/(a+b) = (1/a) + (1/b), and everything in between.
Alex again: Hence, at a non-elite school, struggling in STEM seems (rightly or wrongly) like the best option. At an elite school, doing whatever it takes to get that diploma seems like the best option.
I think this is an excellent point, one that I saw made in a few places shortly after the NYT piece came out.
becca: I call bullpockey on the notion this "playing around with things to see how they work" is not a fundamental relatively universal human activity. How can anyone watching young children doubt that the vast majority of people can take great pleasure in figuring out how things work ???
I'm not saying that. I'm saying that there are some people who like that more than others, and that people who are more comfortable with that will do better as scientists and engineers. Everybody likes figuring out how things work, but only some people like it enough to do it professionally.
To take a somewhat less loaded example, think about crossword puzzles. That's something that, in principle, any native speaker of English can do, and pretty much anybody who figures out the answer to a crossword puzzle clue will feel some pleasure at figuring it out. But not everybody will enjoy the process enough to regularly do crosswords in their daily paper or on the web, and only a tiny fraction of the population will enjoy them enough to get involved in competitions and that sort of thing.
Does that mean that the people who don't regularly do crosswords are somehow deficient as human beings? No, not at all. It just means that they don't enjoy doing crosswords enough to take it up as a hobby. But they probably enjoy other activities that the crossword-puzzle-doers don't, and that's also fine.
While I think it would be great to have more scientists in the world, I don't think it's a bad thing that there are people who wash out of the serious science courses. Those people then go on to other fields carrying at least a basic understanding of science. I think the world would be a much, MUCH better place if more people understood the scientific process, even if they aren't all full-fledged scientists. Even with my lib arts degree, I still appreciate the deeper understanding of things I got from my required basic courses in Chem, Biology, etc.
I think I was the other side of the coin: I was excellent at grueling college-level science and math classes, but not so temperamentally suited to actually being a working theoretical physicist, as I discovered in graduate school.
Matt@10 brings up an important point: The skills needed to do well in undergraduate STEM coursework are not necessarily the skills needed to do well as a practicing scientist. I'm sure most people who do university research encounter this type at some point during their career: the undergraduate who performs brilliantly in class but is useless (or worse) in a lab setting. At least these students find out they are not suited for professional science before they hit grad school. I have heard comments similar to Matt's from some of the theoretical professors here: that undergraduate coursework gives a misleading view of how to be a professional theoretical physics. Unlike with lab work, there is no obvious way for would-be theoreticians to find out what theoretical physics is really like until they hit grad school.
That attrition is higher for pre-meds than other STEM majors is interesting when you consider that at most universities pre-meds can in principle pursue any major. While biology is the default choice at schools which do not have a distinct pre-med track (mainly because the med school admission syllabus prescribes certain coursework which bio majors would have to take anyway), people apply to med school from various other majors, including chemistry, physics, engineering, and even humanities/liberal arts majors. I suspect the difference is that engineers at least go in knowing that they have to be able to do math and basic physics (and many discover that they can't hack those subjects, while other would-be engineers are scared off by those requirements before they even start). Many pre-med majors don't find out until they get to college that they have to learn some math and physics.
Let's not forget the TE (technical and engineering) in STEM. My undergrad (GA Tech) is renowned for producting working engineers. It's also got a good reputation as a research university and breeding ground for researchers. But the order is important - and I think the volume should be recognized as well. There's a much greater demand in the world for working engineers than for research scientists. Unfortunately, there's a quirk in our current educational system that means that most/all of the working engineers are trained by research scientists.
From a personal and professional perspective, I think the washout rate is hugely important. I work in STEM's nasty back alley - computer science - as a working engineer and project manager. I can not begin to express the frustration I experience at the number of 'IT' majors with no analytic or critical thinking skills. It makes me want to scream. Each and every one of those folks that were unwilling or unable to make the leap from memorizing a pile of facts and passing a certification to actual understanding and analysis should have been washed out of anything technical well before they cross my path as an 'IT professional'.
[BTW, just to uphold some level of credibility, I also hold an undergrad in mathematics.]
As a final note, I need to convey one of the hardest tasks I ever had to perform. I had to tutor my wife's cohort of education majors through their undergrad 'physics for liberal arts majors'. There is nothing in the world that has been more difficult or frustrating to me than trying to help folks understand physics, than trying to do it without any mathematics. There is so much richness, so many connections, so much relationship that is utterly lost without using real mathematical tools. This was also incredibly difficult because without the connections the topic lost all its vibrancy - I really understood why the innumerate find physics and chemistry deathly boring.
My preferred solution - we need to find a way to set equations and physics to songs/soundtracks. With a good, rich score, it may well be feasible to engage pattern-building cognitive elements without flogging them to adapt to mathematics. And expose to our innumerate brethern some of the vitality and exaltation that we all know lives in the sciences.
Too many undergrad courses are taught by TA's for whom English is a second language at best. How can anyone be expected to learn this stuff if they can't understand the lecturers? This smacks of deliberate washout quotas. I don't begrudge foreign nationals coming here for an education, but daggummit, colleges should teach, not set up obstacle courses.
"Unlike with lab work, there is no obvious way for would-be theoreticians to find out what theoretical physics is really like until they hit grad school."
I don't really agree with the first part of this. Yes, undergraduates do "experiments" in lab, but I'm not sure these are any closer to what real experimental research is like than doing calculations and modeling in class is to what real theoretical research is. Both cover the basic lab skills but leave out much of what research really is.
MRW @14: Most universities I am familiar with (including every one I have ever been affiliated with) make it standard practice to involve undergraduates in research. This is true of SLACs like Chad's institution as well as research universities (it counts toward the "broader impacts" requirement in your NSF proposals). It is even quite common to require a thesis or some equivalent research experience of undergraduates (this was true of my undergraduate department). But it is difficult to assign an undergraduate a research topic in theoretical physics, because most students will not have the background to understand the work until they have a year or more of grad school under their belts. Not so with lab work, where the student can hope to understand the specific part of the work she is doing even if she doesn't understand the bigger picture. So an undergraduate physics major has a realistic shot at finding out what life in the lab is like before she applies to graduate school.
MRW: I don't really agree with the first part of this. Yes, undergraduates do "experiments" in lab, but I'm not sure these are any closer to what real experimental research is like than doing calculations and modeling in class is to what real theoretical research is. Both cover the basic lab skills but leave out much of what research really is.
This is another area where reform curricula are trying to improve things: by making labs more "inquiry-based" and less about following directions carefully. The idea is that rather than giving students very specific instructions on what to do for the lab, you provide them with some apparatus, and ask them to devise their own measurement procedures and so on. In theory, this gives them a better idea of what real science is like. It's really hard to do well, though.
Eric: But it is difficult to assign an undergraduate a research topic in theoretical physics, because most students will not have the background to understand the work until they have a year or more of grad school under their belts. Not so with lab work, where the student can hope to understand the specific part of the work she is doing even if she doesn't understand the bigger picture. So an undergraduate physics major has a realistic shot at finding out what life in the lab is like before she applies to graduate school.
This is not universally true. I have colleagues who are theorists who are very successful at involving undergraduates in their work-- the trick is finding the right sort of problems. It's really, really difficult to do pencil-and-paper theory with undergrads because of the math background required, but it's much easier to do computational theory. One of our local theorists routinely takes students between their first and second years, teaches them a little bit about programming, and gets them involved in her research from a very early stage.
You're not going to do string theory with sophomores, but they can do simulations pretty well, and get a real idea of what that sort of theory is like.
I had to tutor my wife's cohort of education majors through their undergrad 'physics for liberal arts majors'. There is nothing in the world that has been more difficult or frustrating to me than trying to help folks understand physics, than trying to do it without any mathematics.
Teaching ed students basic statistics, or nursing students simple concepts in bio-statistics, can be quite, uh, interesting as well. A disconcertingly high percentage of students in both of those areas are justifiably unsure of their mathematical ability, and are also unjustifiably sure they will never need any math and certainly not statistics.
I'm a theorist at an undergrad institution. I farm out most of my numerical work to bright undergrads. I also believe that numerical work can teach things that are useful in pencil-and-paper theory: You have to think of the system that you're trying to model and ask how to construct a step-by-step procedure to get to a quantitative answer to your question. Even if you don't do the sort of theory that generates a lot of numbers, at some point you have to ask what aspect of the system can be tackled analytically, and by what procedure (if any) that analytical answer could be used to tell us things about the rest of the system. Also, you get good at analyzing curves qualitatively, interpreting what they tell you, etc. Recognizing, say, the significance of a peak or a node or whatever is important in analytical theory as well as computation, e.g. the existence of a peak tells you that something is unstable. Seeing that all of your simulations give curves with similar shapes (up to some rescaling of the variables) tells you something VERY important in analytical theory, if you think deeply.
My high school physics students had to do all motion problems from velocity-time graphs. I never told them the equations. When some actually looked into the text they would say, "They used this equation to solve a problem like this one." At that point I would show them the vel-time graph that corresponded to each equation. Most of them stuck with the pictures, especially when I gave them problems the equations couldn't handle. The ones who went on to engineering and stayed in were the ones who liked problem-solving and explaining to their classmates.
My tutoring group was the moms in their 40's and 50's who went back to school and had to pass math classes. Usually it was just algebra and simple stats, but some went through calc, chem and phys. As each one made it through, I got another through the grapevine.
I counseled my students about engineering if they did not have an engineer at home. If my crazy way of looking at things made sense to them they had a shot. If they thought I was just a crazy hairy man, they would see a lot more like me before their training was through.
I find your take on that article interesting. I'm probably repeating some things I have posted elsewhere, but your observations triggered a few new thoughts as well.
1) Way too many pre-med students don't fully grasp that doctors deal with people who are critically ill, or that one minor error on a homework problem can kill someone.
2) In the end, what is the harm when students with a reasonably solid grounding in high school math go into non-STEM careers? We need that also.
3) I'm with dean @4 and in followup comments: Much of the problem I see is with weak algebra skills. Disorganized "work" must never get corrected before college, and there seems to be a lot of equation grabbing taught in place of problem solving. Worse, the slow pace of AP classes creates the illusion that college will not require much work. And that overlooks the level of attrition before they get to a real science or math class -- which is, frankly, stunning. You don't see it because it happened in HS before they got close to your classroom door.
4) Your comment @8 about what Alex wrote makes the point that reformers "claim" certain things about the reform curricula. However, all I ever see is a demonstration that students either pass physics at a higher rate or do better on the FCI, not that they survive the junior year of engineering at a higher rate. Your place is small enough that you might be able to look at the true impact of your course on retention. I'm convinced that a LOT less would be more.