Despite it being an open secret in academia that Teaching has near-zero impact on the career success of professors, I realize now in retrospect after leaving academia for 4 years that this was the single most important skill that I learned during my 8 years as a professor. There is only a limited amount of time and mental capacity for any one individual, and to have impact more than a sole proprietorship small business means that a business leader must be able to effective teach his team. However, ironically, most professors at universities do NOT learn how to teach, due to either (1) lack of will, (2) lack of skill, or (3) lack of feedback.
First, lack of will: This is relatively self-explanatory, because STEM professors at universities are judged based on only 2 metrics: number of funding dollars brought in, and number of top tier publications. In fact, when I was just starting as an Assistant Professor, my department chair at the time privately give me the quantitative bar for tenure: 60 Impact Points and $2M in federal grant funding. 60 Impact Points refer to the sum of the Impact Factors of the journals that my (corresponding author) manuscripts are published in, for example Nature is 50, Nucleic Acid Research is 15, and PLoS is 3. So within 5-6 years, one needs 1 top publication + 1 good-in-field publication, or 4 good-in-field publications, or 20 minimally acceptable publications. Federal research funding was also easy to strategize around: Either get a $2M+ NIH R01 award, or 3-5 NIH R21’s / NSF awards worth $400k to $600k each. NOTHING ELSE MATTERED, except that you don’t severely piss off your department chair or university.
Given these incentives, it was not difficult to understand that, even though teaching was required for all tenure-track faculty, a good 50% of pre-tenure faculty basically did not try. Reading off Powerpoint presentations inherited from former professors or borrowed from friends at other universities became a great “minimal effort” way of checking off the teaching box. And even though Teaching Ratings given by students were theoretically supposed to impact tenure and raises as well, those were easy to hack by giving out more A’s and being more flexible regarding students requesting extensions. Thus, many professors don’t even TRY to learn how to teach.
Second, lack of skill: This is probably more prevalent a problem in STEM than in humanities/social sciences, because let’s be honest, a good majority of top STEM scientists (including myself) have some degree of autism or Asperger’s. Many top scientists are good at thinking about objective facts and hypotheses, but not as proficient at theory of mind — trying to place themselves in the shoes of their students to see if their teaching is engaging or understandable. If there are one or two outspoken students in a lecture hall who ask the majority of questions, then the professor would naturally assume that they are representative of all other students in the course — an assumption that is more likely wrong than right for most situations. Finally, in 1-on-1 mentorship with graduate students, professors tend to focus on the research projects (because research projects lead to publications and funding dollars), and generally teach their students only the immediate skills needed to advance the research project. More general/valuable skills like how to pick a research project, how to design a series of experiments, and how to decide whether to abandon a project, are generally conducted only in the professors head, leaving it a black-box for the Ph.D. student.
Third, lack of feedback: As alluded to in the previous two sections, even professors who genuinely want to learn how to be a better teacher or want to do right by their students to teach them valuable skills, often struggle with improving because they do not receive timely and accurate feedback on their performance. Teaching Rating, for example, is calculated based on the average of the student’s stores given at semester-end. The students are not professional reviewers who are providing feedback to the professor to help him/her improve teaching skills, they are giving single “how good did I feel” score, which is completely uncorrelated to whether the teaching was effective. The Teaching Rating reflects more the professor’s humor, agreeableness, and even looks, as compared to teaching ability/effectiveness.
How I learned to teach: Being a good strategic gamer, I realized early on that academic success didn’t depend on teaching, so initially I had no interest in being a good teacher. However, the course I was assigned to teach (Biostatistics), happened to be one that I knew only a moderate amount about and was relevant to my research on diagnostics. So I prepared the lecture materials based on my own learning the area. I would create synthetic datasets to statistically analyze via Matlab to test my own comprehension, and these naturally became examples to go through in class and to assign as Problem Sets.
I didn’t start out very effective as a teacher, because I’m naturally an impatient person and a fast talker, and I had more background in statistics than most of my students. So I ended up with some of my first lectures being much shorter than the 75 minutes they were supposed to take up. I had the intuition that (1) I was going to piss off the other professors if I consistently ended classes 30-45 minutes early, and (2) I was going to piss off the students if I tried to cram too much more material into the classes. So the solution that I came up with is to add “filler” — adding true historical examples and made-up stories with tangential relevance to the statistical techniques as the ones we were covering, as well as on-the-fly programming programs to be done during class. For example, why a cancer test with 80% sensitivity and 95% sensitivity could result in causing more cancer (via unnecessary CT scans) than save people from cancer. Unsurprisingly, this actually ended up being a much better way to teach the students, because of the power of Story and the power of Doing.
Teaching then became almost like a roleplaying game for me, I entered each class more interested in giving a good storytelling performance than effectively teaching statistics. After telling a story, I would assign them a 5-10 minute coding problem while I mentally previewed my next story to deliver. But ironically, I think many of them retained more key statistical concepts with them for longer than a more “normal” class.
Dave Zhang's Startup Blog 