Friday, January 3, 2014

Putting College Under the MOOC Microscope



Another year draws to an end, but not before yet another"MOOCs aren't as good as college" story slips into the media (NPR, in this case). Amazing insights are present there, like that MOOCs don't provide as much personal, face-to-face interaction as one can potentially get in a college class. Wow, no one could ever have figured that out. Also, a very small fraction of people who sign up for a class (requiring in some cases literally one button-click of a mouse on a website) don't view all the lectures or complete all the assessments. Well, blow me down. And the conclusion in the article? "We have a lousy product."

Lousy??? I'm really fed up with the anti-MOOC movement, especially when it comes from within academia. Despite my snide sarcasm above, I do appreciate that much of this continued MOOC pushback is a response to the MOOC overhype that both preceded and overlapped it. What many MOOC dissenters seem to miss is that most MOOC advocates (including myself) never argued they are a "replacement" for a college education and experience. No way-- not even close. The media and a very few zealots played that line up, and they were wrong from day 1.

But let's turn the tables a bit. Let's put "in-person" college experiences under the microscope used for MOOCs. Before that, we must realize that we cannot compare completion rates for a college class and a free online product that fails to provide credentialing. Especially for introductory-level science courses (the kind I teach in genetics and evolution), the vast majority of students in college attend classes for credentialing rather than to satisfy a keen interest in the specific topics. A few months ago, I asked a room full of college students in a workshop, "How many of you look forward to 2 or more of your classes most weeks?" The answer-- one. Keep in mind all of the students there take 4-5 classes at a time, so the vast majority do not look forward to even half of what they're signed up for. Again, they are signed up for most classes because they're "required", either directly or to fulfill some sort of requirement or credit. If the students fail to complete the "in-person" college class, not only do they fail to fulfill the requirement and fail to get the credit, but they often have the black-mark of an "F" or a "W" on their permanent record. That's simply untrue for MOOCs in all respects-- if you dislike a MOOC, you simply stop watching without consequence.

How can we compare these experiences fairly then? MOOCs are like what students would be willing to look at as "extra," and with no consequence for failing to complete. I looked up some statistics from my on-campus class last spring as a comparison-- every week, I provided online resources (often podcasts or pdfs) that were truly "extra"... the resources were available on the same webpage as required materials for each week, and the resources complemented what was discussed in the lectures. There were 452 students enrolled. The very first such resource was viewed 100 times. How does this (100/452) compare to the MOOC criticism of "About half who registered for a class ever viewed a lecture"? Again, these were students already in a college class on this subject, and it was material pre-identified for them as relevant. If you look at the supplements from the end of the semester, the views are in the low single digits (potentially just reflecting the times I'd open the files to confirm they uploaded). How does this compare with the MOOC criticism of "completion rates averaged just 4%"?

I don't blame these on-campus students for the low uptake at all. They have career aspirations (in my case, mostly pre-med), and frankly, we've placed them into a situation where their grades matter more than what they care to learn about. If they spend time viewing my supplementary materials, that time is not spent studying for organic chemistry or physics. For every B or lower grade they get, their choices of medical schools become more limited, so they need to triage. And maybe they don't even really care about my topics, but they're forced to take my class by major requirements. None of this is true for MOOCs. Further, as I've argued previously, many college classes effectively focus on stratifying students (the essence of a "curve"), and far too little ensuring that all students who want to be engaged and learn are successful in doing so. MOOCs don't concern themselves with stratification at all-- it's all about engaging and learning for an interested audience. I wonder if college was once that way, centuries ago.

Back to MOOCs, let's drop the percentages and look at just the final numbers. I'll use mine as an example, but I suspect you'd get similar numbers in any of them. My MOOC ran twice. Even if we pretend that only those students who completed every assessment and got a passing grade at the end were the only ones who reaped any benefit, that number still comes to ~4000 students. 4000 people from around the world quantifiably learned about genetics and evolution as a result of this MOOC. Presumably there are other students who didn't complete it but found some part of the experience personally rewarding or engaging, and they have a greater appreciation for the topic. And best of all, none, NOT ONE, of those 4000+ "had" to do it-- this was quenching a thirst for knowledge, not jumping through a pre-MCAT or biology-major hoop. I'd like to see more "lousy products" like that in the world. How many of those students enrolled would have gone to a local college instead to satisfy this particular thirst? My guess is less than 1%, if any. Finally, I like the thought experiment of what would happen if I just told my on-campus class from day 1, "You'll all get A's no matter what," (obviating the credentialing)-- how many would still be in my classroom three months later? How many years would I have to run my on-campus class under that condition to get 4000 students to have continued to month 3 of my class?

Yes, MOOCs were overhyped. They are no panacea. They don't have face-to-face interactions with knowledgeable faculty and able other students. They don't invalidate college or provide a serious alternative. They don't provide "education for all." Most of the enrolled students already have higher education, so MOOCs' contributions to equalizing opportunity are limited (if for no other reason than because of variable internet access). And they are misused by some reckless college administrations. But before we cast any more stones at MOOCs for what they "aren't", let's have colleges take a serious look in the mirror themselves at what they've become, and see how badly their faces have broken out.

Personally, MOOCs have helped me see deficits in standard on-campus college experiences. I think the overall college experience needs to be rethought in a big way. It's NOT that I think MOOCs are better or are replacing college, but they highlight college's obsessions with course requirements, with grades, with credentialing, and with hoops of various sorts in the on-campus experience. Unlike on-campus college classes, MOOCs are hoop-free and purely educational: people enroll because they think they want to learn the subject being taught, and they continue in the class if and when they stay engaged in the material and seek to commit their time to it. What a concept that would be for an on-campus introductory science classroom.

Friday, December 20, 2013

Grants and the measure of a scientist's 'worth'



All academic scientists worry about grants. In my weekly lunches with other faculty, that is often one of  the first points they raise as a source of concern. I anxiously await now a score (if any) from my most recent NIH submission. Some colleagues of mine reported either euphoria associated with their recent NSF funding or sadness with recent NSF declines. Part of the worry is about keeping the science going that we want to do. Part is about not laying off good employees. But sadly, a big part is "image" to the university and to peers. Administrations (and often peers) ask frequently about how big particular grants are and then implicitly or explicitly rank the faculty member based on that amount.

We use dollar amounts of grant funding to assess faculty all the time: for hires, yearly performance raises, tenure and promotion, etc. Part of this is reasonable-- grant funding is competitive, so in an ideal world, someone who has good, creative ideas for feasible, high-impact research should acquire grant funding more easily than someone who lacks such ideas. Further, virtually no research is "free"-- one needs salary for themselves and their labs. That research may be subsidized by the university (as part of our "9-month" salary and TA-ships for students), but it's still not free. Hence, we need money to sustain our research.

That said, many readers will agree we've become too grant-obsessed in our assessments at all levels. New faculty members are immediately dubbed "successful" or "hotshots" if they acquire funding early, whereas early publication of high-impact research in a big journal often has a lesser effect. I recall once (I'll be vague for confidentiality) when 2 assistant professors were up for tenure simultaneously. One had multiple papers in the most premier science journal and multiple others elsewhere, and with a consistent publication rate along his years, but had acquired very little funding. The other acquired federal funding early but didn't publish anything until they put out a small number of papers in the year they were up for tenure (and none is journals as prestigious as the first). The faculty tenure vote was more strongly favorable for the second than the first, citing "sustainability" as a concern for the first.

Let me use an analogy. Funding is like gas to make the engine of research run. However, comparing faculty based on grant dollars is like comparing two cars in how far they'll go based on how much gas is in their tank. Many scientists are like Toyota Prius plug-ins (beyond from their interest in reducing emissions)-- yes, they need gas, but they can go very, very far with a small amount (~58mpg). Other scientists may be more like an 8-cylinder Chevrolet Camaro (~14mpg), or even a coach bus (~6mpg). There is even empirical evidence that very heavily funded labs, on average, produce less per dollar than mid-sized labs (ref).

Again, research isn't free, and sustainability is a concern, so we should not ignore funding. However, I will argue that IF we are to use grant dollars as part of a measure of evaluation, we should simultaneously consider that investigator's track record of "research efficiency per dollar" (like gas mileage). How many impactful new insights have they published per dollar? Shouldn't we be in awe of those who publish high impact work while using up less taxpayer money (and thus leave more for other research)? Shouldn't we consider research sustainability not just by how much you have, but how well you'll do in the inevitable situation in which you have little research funding? There are multiple ways to publish in PLoS Biology, Science, or Nature--  two are "scale" (you do something that's slightly creative but on a grander scale than anyone else has done before-- clearly expensive) and "innovation" (you come up with a truly creative idea and test it-- perhaps not expensively). It's time we spent more effort giving attention and reward to the latter of those two approaches, especially now when grant dollars are preciously limited.

Monday, December 2, 2013

Grades (What are they good for?)



Teachers (college or K-12) always complain about grading, and perhaps even more about student whining about grades (see this example). Biology professors, for example, often complain about students who intend to go into medicine being"obsessed" with grades. Given the challenges of the grade-awarding process, I've been reflecting on why we grade lately, and I welcome thoughts from all of you. Personally, I find that this question segues into a more fundamental question of the purpose of formal education.

Most teachers would quickly suggest that we give grades to assess student understanding of the material covered. Some students appreciate virtually all the nuances of the material (and thus get an "A"), others have a very basic understanding (perhaps getting a "C"), and still others fail to understand the material (grade "F"). The grade thus provides feedback to the student and to the institution about how well they grasped the material covered. Fair enough.

So, let me follow with another question-- why do teachers teach material to students? Presumably, it's because the material is worthwhile, and it is thus desirable for the student to learn it. If the purpose is for students to understand and appreciate the content, then an "F" indicates a failure not just of the student but of the teacher's purpose as well. If we desire students to learn something and they fail to do so, then both student and teacher roles have failed (irrespective of whose "fault" that failure was). In this regard, our system is counterproductive to its purpose in that, if one or more students fail to learn material covered, the response is to stick an "F" label on the student and simply move on. Given there may be numerous reasons the student failed to grasp the material (including bad timing or perhaps a teaching style that did not work well), why would we not let students take more opportunities to learn a given body of material, assuming learning the material is indeed valuable?

When we talk about "tests", we think of tests in schools with grades. Here's a different example-- a driver's license test. This test is worthwhile-- it provides training that may even save the life of the awardee and gives certification of their ability. There are no grades to it-- a student passes or fails to get the certification only. If they fail two times and later master the material to pass, there is no consequence of the original failed attempts, since they are irrelevant-- all that matters is the student has now mastered the valuable material.

Our "grade-obsessed" system has an entirely different purpose-- the stratification of students. This stratification may reflect effort or ability, though we can never be certain of the relative weighting of the two in the outcome. Some of the stratification may be arbitrary, too, as some students may have been ranked low directly as a result of having one particular teacher (whose teaching style did not work for them) and not another.

Coming back to the example of premedical students, it's again unquestionable that medical schools use grades as one of their most prominent criteria for admission (along with others, such as MCAT score, rigor of coursework, letters, etc). By awarding grades, undergraduate professors facilitate their stratification of applicants. I think it's safe to argue that, all else being held constant, every non-A reduces an applicant's probability of admission to top-tier medical schools, even if only slightly. The same truth holds for undergraduate admission-- all else being equal, every non-A in high school reduces the applicant's range of schools to which they may get accepted (and the associated financial aid). How can we blame students for seeming grade-obsessed when faced with this reality?

Basically, I think the current system focuses too heavily on innate ability and luck, and gives too little to people who are willing to strive hard but were incompletely successful in their first attempt, the latter of which I think is a big predictor of eventual success. I see no reason why, like driver's license tests, we don't let people try to re-learn and re-test, as those people may in the end actually understand the topic just as well or better, but have demonstrated perseverance. In fact, with the current system, there's frequently virtually no reward to going back and trying to understand better what you didn't understand in the first place-- totally contradictory to our stated goals.

I find these facts to be very disturbing. I did not enter the educational enterprise for the purpose of stratifying students-- I would prefer that students actually learn what I teach. Some colleges allow grades to be optional for some or many classes, but even some of the more famous examples people cite (e.g., Reed College) still record grades in the end.

Can the situation be fixed? I think any solution would involve a radical change in how education works. My first thought was that we'd follow the driver's license example and report specific competencies. For example, students in a transmission genetics course could get certified for competency in their understanding of meiosis, recombination, genetic mapping, heritability, Hardy Weinberg genotypes, etc. However, that approach merely moves the problem-- what if someone only grasps these concepts at the most basic level, and then moves on as though certified with full understanding/ competency?

Honestly, I think the solution (which itself has numerous problems-- see below) is to separate the process of teaching from that of assessment/ stratification. This solution may be more feasible now than in years past, given the growth of resources available electronically. We can have still assessments in classes, but they'd be more for the students to self-assess and not for permanent records. A student would finish any genetics class they like (live, online, self-taught from books, whatever), and when they feel they are adequately prepared, take a "for-the-record" assessment. These assessments may only be taken once every semester or once every year, so they can't just keep taking it weekly. However, students can retake the assessment after the waiting period, up to some maximum number of times (maybe 3-5).

What are strengths of this approach? For teachers, they focus on teaching and not on grades. They are no longer involved in the stratification process-- their only goal is to help students learn the material. Students would better accept that "we're on the same side" with respect to learning with such a change. Again, teachers should still provide extensive in-class assessments for students to practice, but the grades of those tests are informational only. For students, there are two large benefits. First, they can learn however they feel works best for them. Those who prefer live, standard classes can do those. Those who prefer online classes can take those. Second, it provides students with a "marketplace" of opportunities. Some teachers may be known to focus on particular subsets of the material (specialties or areas of research). They can learn those areas from those teachers, and go to other teachers to learn other specialties within the scope of the assessment.

The approach has major weaknesses, though. Students would spend a lot more time researching class options and outcomes rather than just taking "the genetics class offered this semester at my school." They may also be sick or upset on the day of the test and have to wait a year to repair a weak grade from a single test (though this may already be true for heavily weighted final exams). For teachers/ professors, they give up control of tests. Much as we complain about grading and grade complaints, I suspect we'll complain more about the standardized test not focusing on what we think is most relevant. We'll probably also get pressure from students (and administrators) to match course coverage to that of what's likely to be on the test, and professors will immediately scream that their academic freedom to teach whatever and however they like is being impinged upon. (K-12 teachers already encounter this issue with state scholastic requirements.) Finally, there's the question of who actually makes these tests. I don't see that this solution is feasible, honestly, as the negatives are huge.

Are we stuck with the current system, where teachers' roles often devolve to presentation, assessment, stratification, and moving on*? Or are there alternatives? I welcome feedback from readers.

* Footnote: I realize that many teachers do a lot more than "presentation", including but not limited to one-on-one mentoring of students outside the classroom, and including on material no longer being covered in class.

Sunday, November 3, 2013

Should societies facilitate making scientific meeting presentations publicly available?



The answer always sounds simple and obvious initially, yet it never seems to pan out that way, and the problems always come down to money and career advancement. The value of science comes not from doing experiments but from relaying the results and interpretations of completed studies. Science has no value while it sits in a lab notebook or on a laptop's hard drive-- it has value when others can see it, use it, test it, and expand upon it, so knowledge is shared. Shouldn't we make our results as publicly available as possible, and as early as possible? Further, our research was sponsored by federal funds ultimately deriving at least in part from taxes-- it was done on the backs of our friends and neighbors. How can we justify not maximizing its speed of impact?

You may think I'm talking about open- or free access to refereed scientific journal publications, or to deposition of scientific pre-prints to services like bioRxiv, arXiv, or Haldane's Sieve. The argument certainly can apply to both of those activities, but I'm actually talking about something different. I'm talking about scientific presentations at conferences.

I have a proposal that I'd like to raise with one of my scientific societies in a few weeks-- one that was discussed in an earlier blog entry. It sounds simple-- the society should facilitate recording (with permission from the presenters) contributed talk presentations and put them in YouTube or equivalent. This has been done successfully before for smaller meetings (see this conference). Let's also facilitate having presenters (again, with their consent/ permission) upload posters to a free online site (e.g., this one). What better way to maximize impact and speed of sharing? Anyone in the broader community can watch the talks and learn the exciting new ideas, potentially long before a final manuscript is ready, much less a refereed publication. And for conferences with multiple concurrent sessions, this allows the possibility for attendees to see talks they'd missed-- who wouldn't want this?

It's not contentious in some sense, since it'd all be voluntary. If you don't opt-in, it doesn't happen. If you don't want to share for whatever reason, then fine, don't.  In talking with several colleagues, though, it seems the idea of a society even facilitating this effort is at least potentially contentious. Many concerns were raised:

1) Would such presentations online constitute "publication" and thus prevent the author from being able to publish the final version of these results in a refereed journal? Most journals in my field would not be so ruthless, but some journals especially in biomedical areas, are ruthless (see this list of journals/ publishers and their policies onprepublication of manuscripts). From the publisher standpoint, they want the press associated with the findings (hence "money"), and they thus manipulate scientists by limiting their ability to translate the work into refereed journal publications (hence "career advancement"). The concerned ask-- by providing this service to facilitate recording, is the society potentially hurting some of its members?

2) Would such online presentations potentially open the facilitating society or conference organizers to damages or lawsuits if the presenters fail to consider copyrights? Many scientists are careless in grabbing images from Google image search (irrespective of copyright) or failing to get proper permissions to show figures from work published in non-open-access journals. If the society or conference organizers "sponsor" conference recordings, could they be sued (hence again, "money")? Or perhaps a junior scientist may make an off-hand negative comment about an established scientist's work, and be penalized in some way (hence again "career advancement" concerns)?

3) Would allowing a fraction of the conference to be posted online make it so fewer people elect to attend the conference in person, thereby devaluing the conference ("career advancement" for all participants) and potentially also causing the sponsoring societies to incur a financial loss ("money")?

4) Would ruthless others attempt to "scoop" results by copying the work or rushing a related experiment, particularly affecting junior scientists who generously opted to have their presentations shared ("career advancement")?

5) Would people see yet misuse the results without fully understanding the context the way the presenter does?

Again, a seemingly simple idea, but many complications. The discussion is worth having. I'll give my answers to the five criticisms above, though there are others I won't go into.

On #1-- I have two replies. First, it's rare at least in my subfield. Second, I think the community should take a strong stance of not publishing in journals that elect to have such ruthless behavior and demonstrate such an antagonism to scientific dissemination (their purported goal).

On #2-- I think there is reason to be cautious and to give guidance to presenters on proper attribution/ permissions. That said, I find it hard to believe that, short of airing clips of popular movies currently in theaters, there would ever be retribution sought more severe than a demand that something be removed from the internet in a specified timeframe (which can be done). On the junior scientist issue, that issue is potentially "better" if the content is taped, since he/she is likely to be correctly quoted rather than potentially misquoted as having said something worse than they actually did.

On #3-- The biggest benefit of conferences is not sitting passively in talks-- far from it (see this blog). I'd never skip a conference because a fraction of the talks were online. Further, I can't imagine, at least in the near-term, that more than 20% of a very large conference would opt-in given the concerns associated with #2 above and #4 below.

On #4-- This is an issue for every conference presentation, irrespective of whether online or not. People have cellphones with cameras, so if they are so ruthless, they can snap a picture of the poster. They can take detailed notes at talks. I totally acknowledge that it's a greater concern when the material is made so much more accessible (though it also potentially allows clear documentation of when they had presented this work and thus their primacy). So yes, this is a concern, and merits consideration by participants in whether to consent to taking this risk in exchange for sharing their results. But I think they should be allowed to make this decision for themselves rather than having the society paternalistically decide not to permit the option.

On #5-- People misinterpret the context of results from fully documented peer-reviewed journal publications all the time-- it's hardly a unique problem. This concern was raised repeatedly in the context of trying to stop the now-mandatory sharing of data used in published papers in my field. While I'm sure this happens, bluntly, I've always found this criticism as a reason for not sharing to be uncompelling, and condescending to one's colleagues.

But I invite readers of this blog to submit their own thoughts. Maybe I'm totally wrong in my responses, or maybe there's even zero interest in this as a service. Is offering the OPTION to record talks at a conference and make freely available a good idea? Same for submitting posters to an open repository? Do the concerns outweigh the benefits, and am I just being audacious by pushing this idea? Comment very welcome!