Gas station without pumps

2013 October 5

Balancing fun and fundamentals

In Computing is a Liberal Art » Automatons and Entertainers, Keith O’Hara writes

One oft-cited axiom in the MOOC debate is that Math and CS are easier to MOOC-itize than other fields. This is one fact that crosses the intellectual aisle. MOOC-leaning scientists and anti-MOOC humanists both take for granted that the teaching of math and programming should be the first to be automated. As you might have guessed, I whole-heartedly disagree. If you can’t automate the teaching of writing you can’t automate the teaching of math. You can automate multiplication drills, but you can also automate spelling drills. Humanists don’t consider spelling writing, Mathematicians don’t consider multiplying math. And for the record, computer scientists don’t consider programming language syntax computer science. Spelling is necessary to write. Multiplication is necessary to do math. Writing grammatically correct programs is necessary to study algorithms. But those aren’t the interesting parts of those disciplines, for exactly that reason, they can be automated. Academics are concerned with new knowledge, and that necessarily lives on the boundary of what is known and what is unknown. And if we can automate something, we know it very well.

I agree with him whole-heartedly.  The Java syntax courses that pass for first programming classes are a lot like spelling, grammar, or arithmetic drills.  They are essential skills, but boring as hell. Details matter, but details are not the whole picture.

One approach that gets used a lot in K–12 education is to drop the drills and concentrate on the “fun” parts.  This has dominated English teaching at the elementary and secondary school levels for a while now, so that many students entering college cannot spell and have only the vaguest notions of what a grammatical sentence is.  They’ve also only written self-reflections and literary analysis—styles of writing that have little existence outside English classrooms. Some math curricula have gone the same way, and students are entering college unable to multiply or add fractions and with only vague ideas about algebra, trigonometry, or complex numbers.  I can’t support a system in which fundamental concepts are ignored in this way.

Another approach is not to allow kids to do the fun stuff until they have mastered the fundamentals (the finish-your-spinach-or-no-dessert approach).  Unfortunately, the result is that many students never get to the fun stuff, and end up believing that huge swaths of knowledge are inaccessible and uninteresting.  Note: this approach dominates many engineering schools, which do a bottom-up approach teaching years of math, physics, and “fundamentals” before getting to engineering design, which is the heart of the field—the result is often a very high attrition rate and “engineers” produced who can’t actually do any engineering.

I think that a balanced approach, that mixes fun stuff in from the beginning but continues to teach the boring details, is essential to effective teaching in any field.

I am trying to create such courses for the bioengineering majors at UCSC: the applied circuits course, for example, and a new freshman design seminar. The bioengineering major at UCSC probably has the least engineering design of any of the majors in the School of Engineering (except for Technology and Information Management, which I don’t think belongs in the School of Engineering).  I want to fix that flaw in the curriculum, but it is hard to overcome the “you have to know all this before you can do anything” attitude of both students and faculty.  Fitting in all the prerequisite chains for math, physics, chemistry, biology, programming, and statistics makes it very difficult to schedule courses for freshman—students need to get prereqs done early enough that they can finish in 4 years.

I think that the computer engineering department at UCSC does a good job of mixing in engineering design down to at least the 2nd year courses (I’ve not looked at their freshman courses lately—they may be doing well there also).  I’m less impressed with what computer science has done, though the large sizes of their courses make good teaching and design content harder to incorporate.  Their game-design major does get into design much sooner than the standard CS course sequence, I believe. I’m decidedly unimpressed with EE, where there is no design content at all until the 3rd or 4th year, even where it could have been easily incorporated.

Note that engineering design is damn hard to MOOCify.  The essence of design is that the answers are not known in advance—there are many ways to achieve desired design goals.  There are also many different tradeoffs to make in setting the design goals.  Students not only have to come up with designs, but have to build and test them—the real world is very important in engineering, and simulation is rarely an adequate substitute. (Note, I’m not saying that engineering students should not use simulations. Learning how to use simulators properly and what their strengths and limitations are is an important part of engineering education—but simulation-only education is not sufficient.)

One problem I am facing in trying to improve the bioengineering curriculum is that most of our bioengineering students are in the biomolecular engineering track.  Molecular engineering is decidedly slower and needs more science background than most other fields of engineering (which is why it is mainly a graduate field elsewhere). It is particularly hard to provide freshman with design experience in molecular engineering.

UCSC has one honors course that attempts to provide this experience to freshmen, but the capacity in the course is only about 20 (shortage of wet-lab space and teaching resources), and probably only 3 or 4 bioengineering students qualify for the honors course—what do we do with the other 50–100 bioengineering freshmen?

The freshman design course I’ll be creating this winter will be able to handle maybe half of them, but it will be focusing on designing low-cost do-it-yourself instrumentation, not molecular engineering.  (I’m hoping that we can entice more students into the bioelectronics and rehabilitation tracks, and reduce the load on the biomolecular track.) The course is just 2 units, not 5, so that students can add it to a nominally full schedule, without delaying any of their required courses. That was all I thought I could get students to take with the current curriculum.

I’d like to have a required 5-unit course with substantial engineering design in the freshman year, and not just a 2-unit optional course, but I don’t currently see how to fit that in even with a revised curriculum—it would require reducing the chemistry requirements for the degree substantially, and the Chem department is unlikely to create a faster route to biochem.


2013 September 20

Some MOOC retention rates amazingly low

Filed under: Uncategorized — gasstationwithoutpumps @ 23:42
Tags: , ,

Mark Guzdial, in his post Lessons Learned From First Year College MOOCs at Georgia Tech (and SJSU), points out the astonishingly low completion rates for some MOOCs (about 1% of registered students, 2–6% of those who did the first assignment):

Karen Head has finished her series on how well the freshman-composition course fared, published in The Chronicle. The stats were disappointing—only about 238 of the approximately 15K students who did the first homework finished the course. That’s even less than the ~10% we saw completing other MOOCs.

Georgia Tech also received funding from the Gates Foundation to trial a MOOC approach to a first year of college physics course. I met with Mike Schatz last Friday to talk about his course. The results were pretty similar: 20K students signed up, 3K students completed the first assignment, and only 170 finished.

Note that the composition course had only 3 major assignments, one of which was to produce a video, so it probably did not do a lot for the writing skills of even the students who completed it, compared to a normal freshman composition class with feedback from experienced instructors.

I can’t help thinking that for the cost of creating the MOOCs, a lot more students could have completed conventionally taught courses. After all, 238 students is only 10 sections of normal freshman comp, at about $6k per section or $60k, and 170 physics students is one lecturer and 6 discussion sections, or about $50k. (And a real physics class would have had an associated lab.)

I’m sure that the MOOCs cost a lot more than $60k—or they wouldn’t have needed Gates Foundation grants to run them.  If the courses had really managed to get a significant number of students to finish, they might have been worth the huge price tag (despite the arguably lower quality of the education).  But having fewer completions than higher quality, cheaper traditional education does not make for a convincing argument in favor of MOOCs.

2013 September 9

MOOC failures in remedial math

Filed under: Uncategorized — gasstationwithoutpumps @ 07:10
Tags: , ,

By now, probably everyone interested in the MOOC debate has read one or more articles about San Jose State’s experiment with Udacity last Spring.  The best article I’ve found about why San Jose State terminated the pilot is from the San Jose Mercury News: MOOC mashup: San Jose State University — Udacity experiment with online-only courses fizzles. That article explains that the experiment was stopped for the same reason that clinical trials are often stopped—there was clear evidence that the new treatment was worse than the standard one:

In the Udacity Remedial/Developmental Math course there was a disappointing 29 percent pass rate compared to an 80 percent pass rate in the regular face-to-face SJSU course. Only 12 percent of non-SJSU students in the Udacity version of the course passed, including students from Oakland Military Institute, the college-prep charter school.

Likewise, in the online College Algebra course, only 44 percent of San Jose students achieved the required C pass rate compared to a 74 percent C pass rate in the face-to-face version. Here again, only 12 percent of non-SJSU students in the online version achieved a C.

Finally, in the statistics class, which Udacity expected to produce far superior results, only 51 percent of students achieved a C pass rate, in stark contrast to the 74 percent C pass rate students accomplished in the face-to-face version of the same course.

I’ve been planning to post this for almost 2 months now, but I didn’t have much to add to the Mercury News article.  Today I was reading another blog post on Daniel Collins’s Angry Math blog, Reasons Remedial is Rough, which explains some of the reasons why the failure rate in remedial math classes is so high. Unless MOOCs can address some of these problems, they will not be able to reduce the failure rate.

  1. Lack of math skills from high school. Many students simply don’t have the requisite skills from high school, or really junior high school (algebra), or in many cases even elementary school (times tables, long division, estimations, converting decimals to percent, etc.). It’s hard to make up many years of deficit in a single semester.
  2. Lack of language skills from high school. What’s dawned on me in the last year or so, in the context of applied word problems, is that many students may actually be worse at English than they are at the basic math. Grammar is not taught any more, so students can’t parse a sentence in detail, can’t identify the noun or verb in a sentence, and so forth. This cripples learning the structure of any new language, algebra included.
  3. Lack of logic skills from high school. Basically, no one is taught basic logic anymore, so students can’t parse If/Then, And, Or, Not statements, which form critical parts of our mathematical presentations and procedures.
  4. Lack of study skills or discipline. Almost none of my students do any of the expected homework from our textbook. (On the one hand, I don’t collect or award points for homework, so you might say this is unsurprising; but my judgement is that the amount of practice students need to do greatly exceeds the amount of time I could possibly have to mark or assess it.)
  5. Lack of time to study. Certainly most of our community college students are holding full-time jobs, or caring for children, or supporting parents or other family members. The financial aid system actually requires a full-time course load for benefits; combine that with a full-time job—really, the equivalent of two 40-hour jobs at once—and you get a very, very challenging situation. (Side note: In our lowest-level arithmetic classes, I find that work hours are positively correlated with success, but not so in algebra or other classes.)
  6. Learning disabilities like dyslexia and dyscalculia. All I can do is speculate as to what proportion of remedial students would exhibit such problems if we could institute comprehensive screening. But I suspect it’s quite high. When students are routinely mixing or dropping written symbols, then disaster results. Unlike other languages, concise math syntax has no redundancies to enable the “you know what I meant” safety net.
  7. Emotional problems or contempt for the class. I put this last, because it’s probably the least common item in my list—but common enough that it usually shows up in one or two students in any remedial classroom; and a single such student can irrevocably damage the learning environment for the whole class. Some students who actually know some algebra start the course thinking that it’s beneath them, and become regularly combative over anything I ask them to do, sabotaging their own learning and that of others. It’s pretty self-destructive, and the pass rate for “know-it-all” students like these seems to be about 50/50.

Dan does give the caveat that these are personal observations from teaching remedial math, not large studies by a sociologist.

Of these problems, MOOCs only address the last one (disruptive students), since there is no classroom for them to disrupt—though MOOCs that rely on forums can be easily have the environment destroyed by a couple of trolls, so they don’t even solve this problem.

I think that the key observation is that “It’s hard to make up many years of deficit in a single semester. “ The whole premise of remedial math education is flawed—we are spending enormous amounts on trying to rescue students damaged by poor prior education, with a very low success rate.  It would be far better (and probably cheaper) not to damage the students in the first place.

Of course, it is easy to say “first do no harm”, but that is hard to achieve in practice.  It is very rare for a teacher or school to deliberately harm a student, but many are coming out of the system harmed anyway.

The purist ideological positions that people staked out during the math wars are probably contributing to the harm—students do not benefit from pure drill-and-kill nor from discover-all-of-math-by-yourself (to use the pejorative descriptions of the two extreme positions).

Math teachers need to use enough drill in the early years that students can have “automaticity”—a terrible eduspeak word that means that they are fluent with basic arithmetic and don’t have to think about it.  This is point 1 above—students need to develop the prerequisite skills to the point where they don’t need to think about them any more, and can work on the higher level thinking skills.

But math teachers need to teach more than just skill drills.  Students who can do routine algorithms when told to can still struggle mightily when trying to figure out which algorithm to apply. And it isn’t just the math teachers responsible for teaching this. As described in both points 2 and 3 above: students need to be able to read and comprehend fairly sophisticated English in order to do algebra.  The discarding of grammar from most English instruction in the US means that many students have only vague ideas about how sentences are constructed and interpreted, and so can’t translate what they read into the precise concepts needed for doing math.

Points 4 and 5, the lack of study skills and the lack of time, go hand in hand.  Students with little time and no idea how to use that little time efficiently are going to struggle to learn anything unfamiliar.  Having a regularly scheduled study group of students of roughly the same ability can help with time management—it is easier to block out time for a regularly scheduled meeting than to handle multiple priorities and make sure that studying comes to the top of the list sufficiently often.  The social pressure from a group of fellow students who are seriously trying to learn can also make a big difference in how much time is spent studying and how effective that studying time is.  This may be one of the biggest advantages that in-person classes have over MOOCs.

Learning disabilities and attitude problems are always going to be difficult to handle, and may require one-on-one attention, not just remedial classes.

In summary, I think that elementary schools, middle schools, and high schools need to look carefully at their programs—especially if a large fraction of their graduates are requiring remedial math in college.  Do the elementary students and middle school students have sufficient facility with arithmetic and fractions to be able to use those skills freely when taking secondary school math?  Do they have enough grammar to be able to pick apart a sentence and translate it precisely into mathematical terms?  Have they practiced doing so?  Secondary schools need to be sure that they are remediating the flaws of their feeder schools, and not just kicking the can down the road to the colleges.

The sooner problems are fixed, the smaller the fixes needed.  MOOCs do not seem to be a rational attempt to fix the real problems.

2013 August 4

Lose the lecture

Filed under: Uncategorized — gasstationwithoutpumps @ 16:03
Tags: , ,

Joe Redish in his blog  The Unabashed Academic had a MOOC-related post a year and half ago that I only just came across today: Lose the lecture.  The post basically makes the same point that I have made in posts like Teaching by hand and What online education cannot teach:

But the transformations that are increasingly pressed on us by Deans and Provost focused on this year’s bottom line, take us further from that value that only we can add and move us towards delivering education that is increasingly equivalent to what the on-line colleges can deliver. This is a recipe for disaster. Brick-and-mortars can’t compete financially with online institutions on their own turf. We have too many maintenance costs.

Those academic Chairs, Deans, and Provosts who think that the new technology will make it cheaper to deliver their product with fewer faculty (and larger classes) are undermining the future of their own universities. We should be moving in the opposite direction, providing students with more faculty interaction, more group learning environments, and more hands-on activities.

I doubt that I could convince our current dean of this—he manages as if he wants us to be a research institute in Silicon Valley rather than a university in Santa Cruz.  I’m sure that research institutes are profitable than public universities, but the swing away from teaching in the past decade has been rather extreme.

2013 July 30

What online education cannot teach

Filed under: Uncategorized — gasstationwithoutpumps @ 00:05
Tags: , ,

In response to yesterday’s MOOC Roundup, one of my fellow faculty members sent me a pointer to an article by Jennifer M. Morton in The Chronicle of Higher Education that was just published today: Unequal Classrooms: What Online Education Cannot Teach.  I would have included the article in the roundup, if it had come out a little earlier.  Here is a paragraph from the middle of the article:

MOOCs would seem like a promising way to increase access to education for those who cannot afford the steep price of a liberal-arts education. And indeed, my students often end up sitting in crowded lecture halls being lectured at by a professor who doesn’t even know their names—as is the case for many students across the country. Many of my students also work, some full time, or have families of their own, and they struggle to fulfill the course requirements for graduation. However, the adoption of online education by large public universities threatens to harm the very students for whom a college education is an essential leg up into the middle class.

Prof. Morton makes a point that several other skeptics of MOOCs have made—that a big chunk of college is the interaction in the classroom (between students as well as between faculty and students), and that is mostly missing from MOOCs.  For students from poverty-stricken communities, college may be the only place to learn the “practical skills to navigate middle-class institutions”, and MOOCs deprive them of this learning.  The learning may not be an explicit part of the curriculum of the colleges, but it is implicit in the use of a college degree as a ticket into the middle class.

While I do not regard social mobility as a primary goal of college education, for many people that is the main justification for public universities.  It seems a bit unlikely that MOOC-based education will serve that purpose well, even if it manages to convey curricular content adequately (which has not yet been convincingly shown either).

« Previous PageNext Page »

%d bloggers like this: