Monday, November 07, 2011


Grading and Choosing

You know that feeling when you’ve suspected something for a long time but couldn’t prove it, and then someone proves it for you?

This article had that effect. It’s about how student attrition in STEM majors is actually higher in more selective institutions than in less selective ones. It brought back vivid memories of my days at Snooty Liberal Arts College, and even of late high school.

The article is based on a study by the College Board that suggests that grading is substantially harder in STEM majors than in most others, so students who don’t immediately hit it out of the park in STEM classes tend to gravitate towards the more-welcoming liberal arts and business classes. But that tends to be less true at less selective colleges, oddly enough.

I like it a lot because it explains a number of disconnected impressions I’ve picked up over the years. For example, in my student days, I recall noticing that even though the STEM classes (we didn’t use that term then, but still...) were “harder,” they also had much flatter grade distributions. It was easier to pass a history class than a chemistry class, but easier to get an A in chemistry than in history. The history classes had bell-ish curves; the STEM classes had flat lines. They were easier to fail and easier to ace; the “squishier” subjects were the land of B’s and C’s. Even in my wheelhouse, I was the master of the A-minus; full A’s were basically unicorns. In physics and chemistry, the top students finished with GPA’s above 4.

So saying one is “easier” is kind of misleading. It’s easier to pass, yes, but harder to really nail.

(One underappreciated variable, I think, is the level of consensus in the field. As far as I know, there’s remarkably little controversy in the scientific community -- I’m open to correction on this -- about the material that gets taught in the first couple years of the undergraduate major. That’s certainly not true in the humanities and social sciences. When consensus is missing, it’s harder to definitively nail a subject.)

Students noticed. Those who didn’t much care what their degree was in, as long as it had the SLAC name on it, clustered into English or history. (SLAC didn’t have a business major.) Those who cared strongly about their own “squishy” specialties had the mixed blessing of a bunch of classmates who had taken the courses as second choices. The idea -- accepted as gospel by all -- was that you were either a science person or you were not. If you were, you stuck with it and did great; if you weren’t, you did something else.

(This was made explicit in Organic Chemistry, which was pitched unapologetically as the pre-med weedout course. Difficulty wasn’t a bug; it was a feature. The idea was to winnow the herd, and to leave only the truly worthy still standing.)

In the cc world as I’ve experienced it, that assumption isn’t widely held. Here the idea is that the community (broadly defined) needs more STEM majors, and it’s our job to make that possible. Rather than weeding out, the goal is to bring people in.

If you start with that assumption, then of course your approach will change.

It isn’t just about easier grading. It’s about the purpose of a given class, and therefore the approach to it.

If I were appointed guru of American higher ed, one of the edicts I would issue would be that theory should be taught inductively. It rarely is, which, I’m convinced, is why so many undergrads spit the bit. (This is probably why I bombed geometry, but never mind that.)

Theory is easiest to learn when there’s a context for it. When you know why you need to know something, you’re much more likely to get it. That’s partly a function of motivation, but it’s also a trick of memory. A theorem derived on a board by someone with his back to you is far less memorable than something that comes with the force of “eureka!,” solving a problem with which you’re engaged.

At its base, I suspect, theory is based on pattern recognition. And pattern recognition is easiest when you’ve seen a whole bunch of examples. If you can get a student to the point at which a theory comes as a solution, rather than as an edict, you’ve won. And if you can get students to test theories against each other, you’re raising the cognitive level of what they’re doing and engaging them much more fully.

Instead, science is too often taught from theory to application. Worse, in the selective settings, it’s taught with a clear goal of thinning the herd. At least at the community college level, we don’t consider ourselves to have failed if a significant portion of a chemistry class does well. At this level, we actually pay attention to the teaching itself. That’s not to deny that there’s much more work to be done -- no argument there -- but at least we’re attacking the right problem. The goal shouldn’t be to keep science pure by keeping the great unwashed out of the lab; it should be to keep science accessible by forcing it to be true to its radically democratic roots. Data are no respecters of rank.

If we’d like to direct more of our bright young minds into STEM fields -- a goal I find absolutely worthy -- the elites may actually have something to learn from the community colleges. Hey, Harvard: you’re welcome.

Don't you get different students taking science courses from CC? Many pre-meds at my elite college took some portion of their science classes at their local CC over the summer, especially physics. History, not so much. If these science courses are not required to transfer into a two-year, then your observations can be explained entirely by selection.

Many of these schools have low grades in their intro science classes, but relatively flat grades once people have dropped. Same phenomenon, selection.
Well, you are wrong about that. Science is not a collection of facts, but an organizing principle. With your approach one is doomed to teach the history of science, not science
What EliRabett said

I hated classes that went context-to-theory - I'm a holist by instinct, and what I needed from class was the framework to fit examples into, not more specific cases - too much of that at school already.

Science is a process, and that process is NOT purely inductive. Having only inductive, serialist type classes does neither the subject nor the students any favours. At least in the context where I teach, which isn't a CC, so maybe your student population has different needs?
If I were appointed guru of American higher ed, one of the edicts I would issue would be that theory should be taught inductively. It rarely is, which, I’m convinced, is why so many undergrads spit the bit. (This is probably why I bombed geometry, but never mind that.)

Noted. I'll do my best to make sure that someone who doesn't remotely understand science is never appointed the guru of American higher ed.

The time I had the most overlap between courses was actually one of the most interesting- I had philosophy, cultural geography and history at the CC all intersecting on some common ideas/themes/topics. It made things easy to learn, without getting boring (well, except for bits of cultural geography, which was was structured to emphasize memorization/multiple choice test).
I did NOT "Get" math, in large part, until I got to the science it was useful in. Physics made calc make SO much more sense (well duh, calc was invented to make physics sensible. it's not shocking the inverse works).

In science it's not so much overlap of ideas/themes/topics that is useful so much as overlap of cognitive problem solving strategies.

As an aside- I wouldn't sneer at the history of science- the best molecular biology course I ever had was a guy who explicitly went through things in historical perspective.
He told us what the thinking/knowledge BEFORE the key experiments were, then WHAT the key experiments were. It was about "how do we know this" not just "what do we know" and it was much more about the process than the facts than many other classes.
Most memorably, he did it for "how do we know DNA is the genetic material". Working forward from Mendel (i.e. we know there is a *something* transmitted from parent to offspring that determines traits, what is it?), through Avery-MacLeod-McCarty and the transforming principle*, he made it clear how the experiments approached the question. It made the scientific facts more memorable, and more importantly, instilled a good appreciation for how to ask scientific questions. Science via history of science can work very well.

*My prof was a microbiologist, and he treated Watson and Crick as basically an afterthought (which, truthfully, they are, unless one is a structural biochemist). It's just that no one wanted to believe that what was true for microbes was true for us for a while.
I think the effect at elite universities has its origin in HS They have had more AP classes, which take a year to do one semester of college work, so they enter with the illusion that college will be like those classes. IMHO, they would do better to take those classes at a CC like I did. However, they might just quit earlier when they realize they can earn more money with less work if they leave STEM. The nation "says" it wants more STEM majors, but the value placed on those jobs says something else.

As for other issues....

Science is not taught with the goal of thinning the herd, it is taught with clear, objective outcomes that require hard work and some skill to achieve. This has the effect of thinning the herd under any conditions. The difference you note between CC pass rates and Uni pass rates, if verified by a case control study (matched SAT scores) that tracks how the students who pass do in their major years later, reflects an emphasis on teaching in small classes run by experienced professors rather than huge lectures with newbie grad students as TAs.

As for pedagogy, the key to success is engagement. There are many ways, some old, some new, to do this. Several new approaches that have come out of science ed studies show that hands-on matters. (No surprise, actually.) This has hints of a constructivist approach, but it is heavily guided and truly depends on first providing the framework for processing what is observed. It is particularly effective at breaking down misconceptions about nature. I'll second other comments that pure induction is a poor approach. Science is not stamp collecting. Further, few students would come up with Maxwell's Equations in less than a hundred years if that process was the exclusive path to understanding.

Comment to DD (and others):
The "flat" distribution you refer to is shorthand for "we actually give failing grades to students who attend and take every test". The comparison is not to a bell curve distribution, which might exist in some CC classes, but to an exponential distribution. Some of the majors mentioned in that article, most notably education according to data I have in my possession, give twice as many A's as B's, twice as many B's as C's, a few D's, and hardly any F's.
Having been a successful STEM major myself who has done some semi-successful humanities on the side, I think, looking back, that STEM fields tend to have much more of a bright line between getting it and not getting it. Maybe it's a mostly a function of the objectivity of the grading, but my impression is that there's a certain way of thinking, an approach to problem-solving, that is an all-or-nothing kind of thing.

The students that come in with, or can quickly attune to, this pattern of problem-solving will do well on all the problems they are presented with. The ones who haven't locked into the approach will do poorly on nearly all of the problems. There isn't really a cognitive state in which students just *sort of* understand the approach to the problems. Thus the flat or even inverted grade distribution.

Whereas non-STEM fields are often more like layers to an onion. You can gain some level of understanding and then dig in further and peel another layer, etc. Most students can reach a few layers deep. Only a few will plow all the way, or be unable to scratch the surface.

I think most students can succeed in either STEM or non-STEM studies, but for students who have trouble initially with the STEM approach, the rewards of struggling will not gradually accrue. Rather, one day they may just have your epiphany and things start to make sense. Will it happen in time to pass the class or finish the degree? Maybe it's just a matter of taking a different approach to teaching with that struggling student. If CCs can find a way to do that, then great. I'm not surprised that bigger schools don't bother with it, as the "weed-out" courses are often so big that no one will even notice the strugglers.
Inductive learning is important, but it rightfully plays a limited role at the college level, for these reasons:
1. Students already have a lifetime of inductive learning in their memory banks. Life is composed of inductive learning experiences. It's been said many times that the best engineering students used to be farm boys, because they had a lot of practical, engineering-type experiences in their backgrounds.
2. Inductive learning is less efficient than principle-first learning (as pointed out above).
3. Inductive learning is most beneficial when preceded by some instruction on principles. A good teacher then gauges how much inductive learning is needed to solidify the concept. Sometimes that amount is none, or almost none. This is true in STWEM fields and other fields as well.
I aced all my classes at my CC and then discovered upon transferring to Caltech that I was merely average -- at least at Caltech. It was shocking. But I really think my CC eased my transition into higher education and that immediate exposure to the high performance level at Caltech might have discouraged me. Hard to say. (Caltech appears to agree with me, because freshman are evaluated on a pass/fail standard rather than letter grades. It lessens the shock of being high school valedictorians suddenly reduced to mediocrity.)
There's a much larger issue here that no one seems to be addressing:

Why do we need more STEM graduates?

If a genuine shortage of scientists & engineers existed, then wages would make a meteoric rise, and the problem of lack of supply would remedy itself almost instantly.

Is it possible that employers' claims of "There's a shortage of STEM graduates" is actually only half of the real truth of "There's a shortage of STEM graduates who are willing to work for what we're willing to pay?"

This all smacks of the NSF prediction in the early 1990s of a "shortage" of 675,000 scientists and engineers over the next two decades.

Ask anyone who's graduated with a STEM degree over the past 20 years how that turned out.
I like it a lot because it explains a number of disconnected impressions I’ve picked up over the years. For example, in my student days, I recall noticing that even though the STEM classes (we didn’t use that term then, but still...) were “harder,” they also had much flatter grade distributions. It was easier to pass a history class than a chemistry class, but easier to get an A in chemistry than in history. The history classes had bell-ish curves; the STEM classes had flat lines. They were easier to fail and easier to ace; the “squishier” subjects were the land of B’s and C’s. Even in my wheelhouse, I was the master of the A-minus; full A’s were basically unicorns. In physics and chemistry, the top students finished with GPA’s above 4.
Not at the SLAC I am teaching at. Easy to get an A in humanities courses. Science courses, the As are earned by a smaller portion of the students. Yes some of them are able to get As in basically all their science classes but you interact with them and you understand why. They are really, really bright. They tend to do well in the non-science classes they take as well.

When I attended a different SLAC in my youth, I found the experience to be the same. Science classes harder to earn an A in than non-science classes. Pre-meds I knew flocked to non-science majors to boost their GPA (i.e., get lots of As) and have more time to study while taking the science classes required for their medical school applications.

Basically I counter your anecdotal evidence with anecdotal evidence.
I think that your anecdotal evidence is not representative, and at fancy colleges GPA's are lower in sciences than humanities (although clear data seem harder to come by than I thought they would be). Here are some stats turned up by a brief googling, although the second comes from a dubious source.

GPA can make a difference for graduating seniors applying for a first job (or at least I think that is the perception among students) and so many people pick classes where they hope for easy A's.

On a related note, I remember when I was in high school doing college tours people would ask the question "is it better to take an honors class and get a B or a regular class and get an A"? They never got a straight answer, but my observation (again anecdotal) was that students who chose easier classes and got higher grades did better in admissions than the ones who chose harder classes and got lower grades.
"And pattern recognition is easiest when you’ve seen a whole bunch of examples."

Perhaps you meant:
"And pattern recognition is easiest when you’ve worked through a whole bunch of examples."

"Seeing" examples isn't all that useful; going home and working through the examples (ie doing homework every night) is what writes the pattern into your mind.
"There's a shortage of STEM graduates who are willing to work for what we're willing to pay...."

Amen – DD’s idea that a post grad degree should be a green card is already the reality in the sciences, especially in engineering. Well – it’s your ticket to a H-1b visa anyway. Why employers think that a postdoc with 12 years experience doing hands on science should be paid 50k and smile is a mystery to me. We’re good at math in these fields, ya know?

"Seeing" examples isn't all that useful; going home and working through the examples (ie doing homework every night) is what writes the pattern into your mind.

Learning science is like learning music. You have to put a lot of individual practice time to become fluent and good. There's psychomotor skills like pipetting that are learned by practice. There’s pattern recognition that can only be learned by repeated exposure over time (especially in biology). If you are naturally talented, this requires less time but for most of us, hitting the books and hitting them hard was the way to succeed. My impression from taking humanities and social science courses was that while you did have to read, you were building on past reading and writing from K-12. Science majors have very little preparation for their courses in K-12 and in biology spend the bulk of the first year learning the vocabulary.
One way to reconcile our different impressions might be this: Maybe your impression of hard-to-get A's in "squishy" subjects is from the more serious students who were majoring in them, while less serious ones were finding ways to satisfy the minimum requirements for the major with easier classes. I think many departments have different 'tracks' a student can take, and those who are grad-school bound end up in harder classes than those who just want to finish with the minimum.
There's also some other pretty decent research looking at what students say about why they leave the STEM disciplines.

In particular two things that might be of interest:
Seymour, E., & Hewitt, N. (1997). Talking about leaving: Why undergraduates leave the sciences. Boulder, CO: Westview Press.

S&H studied only students who were considered well prepared for college SME majors, having math SAT scores above 649. Half of the students included those switching from SME majors and the other half were persisters.

Reports of poor teaching in S.M.E. classes were by far the most common complaint of all switchers and non-switchers.

Students were very clear about what was wrong with the teaching they had experienced and had many suggestions about how to improve it. They strongly believed the source of these problems was that S.M.E. faculty do not like to teach, do not value teaching as a professional activity, and lack, therefore, any inventive to learn to teach differently.

However, by far the most effective technique for dissipating student interest was the widely-reported practice of reading or copying materials straight from text books. Reports of this teaching method came from every S.M.E. discipline and were reported on every campus...

S.M.E. classes were often faulted for their dullness of presentation. The most tedious classes were those in which professors over-focused on getting students to memorize material

Most S.M.E. undergraduates do not place primary responsibility for their learning difficulties on teaching assistants. … Indeed, undergraduates reported that graduate teaching assistants had a higher level of interest in teaching and a greater willingness to meet the intellectual needs of undergraduates than did faculty.

Linn, M. & Kessel, C. (1996). Success in mathematics: Increasing talent and gender diversity among college majors. In J. Kaput, A.H. Schoenfeld, & E. Dubinsky (Eds.), Research in collegiate mathematics education, 2(pp. 101-143). Providence, RI: American Mathematical Society.

“at many institutions, students report a few examples of college mathematics lecturers who read directly from the textbook” (p. 116). Should the faculty members not read from the text, they write, as described by a student (over 650 SAT-Q): “They just continuously write. And they’re just standing in front of what they write, but just don’t care” (p. 116). Linn and Kessel suggest that many of the students who persist do so for altruistic reasons, for example, “hoping to change the way mathematics is taught” (p. 116). They offer this quote from a student to summarize:
Part of the problem with the math department, I think, is their attitude. I think they realize they’re bad, but they don’t really care. It’s not their problem that their students are failing their course. It’s the students’ problem. (p. 117).

Linn and Kessel suggest that content difficulty and student abilities are not important factors in achieving higher rates of success; instead, “poor teaching and advising may select for the most determined rather than the most talented students” (1996, p. 123). As a result, liberal arts colleges, and especially women’s colleges, over-produce, relative to student factors, the number of students who persist in mathematics courses. They suggest that engendering a supportive learning environment, where support comes from a nurturing relationship with members of the faculty and other students, is the best means of increasing student success and persistence. They then offer some assessment of reform initiatives that attempt to do just that. “Students’ initial reactions to reform are often negative” (p. 125), and as a result, faculty frequently abandon those efforts before they have a chance to succeed.
The first issue, regarding teaching, is simply incentives.

At a university, teaching is only as important as the amount of money a class brings in. This is peanuts compared to a good STEM research grant, so tenure (and after tenure, promotion) is basically based on how much money is brought in, how many papers are published. The incentives at a CC are not nearly the same. It is not rare to find a professor who cares about their students and learning, but it is a rare institution that provides incentive to make teaching a priority.

In other words, in most schools, bad teacher evaluations harm tenure chances while two dollars and good evaluations will get you a cup of coffee.
The second issue is this: what is the point of an undergraduate STEM degree?

At least in physics, the point is create more grad students. In order for that to happen, you have to prepare your students well enough that they do not fail out first semester of grad school.

This requires a certain amount of rigor in one's classes, and a certain amount of self-motivation in the students. Upper division classes provide the opportunity to pick up the knowledge and weed out those without the motivation to learn it.

This is a feature, not a bug. If you want to do physics, you have exactly one option: a physics PhD. As it is, too many people hurt themselves trying to reach this goal when they aren't going to get it.

That's a lot of pain and suffering that could be avoided if we stopped lying about wanting want more STEM-trained people. If we wanted more Americans in STEM fields, we'd pay for it, not just in terms of research grants to pay for their training, but in terms of paying livable salaries to those who earn the degrees.

We don't. On top of that, when 1 in 2 people won't even get a degree, it becomes more kind to push people early to figure out whether they have both the ability and the drive, instead of letting them figure it out during a long, drawn out graduate career.
I'm going to echo what the others have said. Why do we need more STEM majors? There are plenty graduate level scientists in my Northeast area who are under- and unemployed. Not sure why we're still selling that line to undergrads.
If there really were a job market for people with STEM degrees, then you wouldn't be able to hire adjuncts with masters degrees in math at community colleges. Until you see the adjunct market clear in a field, don't believe that there is a shortage of people trained in that field.
This comment has been removed by the author.
I'd also like to counter your anecdote with a few of my own.
* As a chemistry professor, I can tell you that my students grades are anything but flat. My intro students have a definite bell curve centered around the C/C+ border. The shape is less clear in my upper-level classes due to smaller class sizes, but it looks more or less bell-shaped and certainly has far fewer As than Bs or Cs.
* As a student who was a English/Chemistry double major - I had more As in English than in Chemistry.
* I'd love to give out more As, but I'm not willing to relax my standards to do so. Courses build on each other in Chemistry, so I wouldn't be doing them any favors with a false sense of accomplishment. If I don't hold them to a high standard now, they'll just sink in the next class that required them to be experts at the previous material from day 1. As for raising grades by raising student performance - that's the goal, but it's not easy.
* In my department, at least, we don't consider the early chemistry class weed-out classes. In fact, we rely on them to convert students who think they want to major in something else into Chemistry majors (and it works).
Forgot one last comment. We're cordial about it, but the Chem 101 professors here spend about an hour a week just debating what we expect students to know about the next week's topic. We compromise, but we definitely don't all agree. Similarly, we're debating revamping the sophomore level curriculum with some pretty strong feelings on both sides. That is, both sides of whether we should make major changes - I'm sure there will be more than two sides if we decide to make changes on the scale of creating/abolishing required courses.
Yeah, is there a sudden spike in STEM wages outside the computer field?

If not, then why are we bothering? We obviously have plenty -- or, at least, it's pointless from a student's perspective to specialize in a difficult field.
some really great advice, thanks! I took some of these same approaches when I was taking classes at Catholic colleges in PA. I really think it helped me learn in the best ways for myself
After looking into a handful of the blog articles on your site, I really like your technique of writing a blog. I book marked it to my bookmark site list and will be checking back in the near future. Take a look at my website as well and let me know your opinion.
SAT coaching chennai | SAT training centre in chennai
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?