The question in the title is immensely subjective and broad, and so I would like to narrow it to an answerable question:

What measures are in common use by administrators and researchers in determining the quality of an instructor's teaching? What, if any, is the correlation between these measures?

I ask because I've had difficulty answering this question for myself. I originally thought that test scores would be the best answer, but I have seen the same teacher have the calculus section with lowest scores in a department and the calculus section with the highest scores in the department in the same semester.

I have also seen teachers that I would identify as good teachers have bad test scores, and vice-versa.

I've had theories as to why test scores may not be correlated to good teaching; for instance, poor students may prefer a good professor, and may be interested in joining such a professors class (this effect could be determined by comparing pretests and posttests, but I have not carried out this analysis).

This has led me to wonder if "good teachers" have any positive effect on their students at all, compared to "average teachers". That's why I have asked the highlighted question above; I would like to find out what others have looked at when evaluating teaching, to help me understand what a "good teacher" is.

  • 1
    $\begingroup$ check out the Danielson Framework, it is a new, "hip" method to evaluate teachers, specifically secondary educators, that has caught on in a lot of districts around the country. There are other similar frameworks, but right now Danielson is the most popular $\endgroup$ – celeriko Apr 15 '15 at 18:15
  • 1
    $\begingroup$ While it suffers from the overreliance that you mention on test scores as the final measure of teacher effectiveness, the article theguardian.com/education/2015/mar/11/… quotes a study by Eric Hanushek that found (in the article's words) that "the difference between a good and a bad teacher is worth a whole year." (However, this is at the secondary level, and it looks like you may be interested in the college level.) $\endgroup$ – LSpice Apr 15 '15 at 19:58
  • 5
    $\begingroup$ One measure that my department has proposed, in response to a university-wide initiative for non-(student evaluation) means of evaluation, is interviews with students after their degrees, to determine the effectiveness of their undergraduate education in preparing them for their later career. We rejected this as unworkable on a large scale, but it is probably the closest thing I have ever heard to any realistic approach at evaluating 'real' success (not just test scores). $\endgroup$ – LSpice Apr 15 '15 at 20:00
  • $\begingroup$ See also Jim Belk's excellent answer to my question matheducators.stackexchange.com/questions/858/… $\endgroup$ – Chris Cunningham Apr 21 '15 at 16:58

I think it's important to distinguish between performance evaluation for the purpose of professional development, and for employment related decisions (hiring, firing, and wages).

The rigor needed for professional development is much lower. A teacher (and their supervisor if needed) can develop goals using a variety of metrics, but the most useful tool I have experience with has been peer observations. Having another teacher from my department sit in on my classes to help me address an issue has been very valuable. Sometimes it has been something specific, so I asked a teacher who I thought had a good style with that issue, and sometimes its been more general. I've also gone to observe other teachers, if after a discussion I thought they had something I really wanted to learn from them. This requires the support of the department and administration, but is very effective.

When it comes to employment decisions the feedback needs to come from a supervisor, it's not fair to expect employees to create critical evaluations of their peers. My only experience with this has been in class observations, which were not particularly effective for this purpose. However, we are changing to incorporate data from the Smarter Balance test, and a regional test that we use. What we're currently negotiating for is to use historical data from the same student, instead of the same class, for comparison. So the same student, but a new test. The idea is that this should show student growth.

Meaning: Andy scored in the 75% in 8th grade and then the 75% in 11th grade, that's great Andy has learned all of the new material between 8th and 11th grade.

Barbara scored in the 75% in 8th grade and then 80% in 11th, not only did Barbara learn all of the new material, she has actually developed a greater understanding of math as the material became more complicated.

Chris scored in the 75% in 8th grade and then 70% in the 11th. So he has lost ground and an intervention is required to teach the newer material.

I personally feel better about this as a method for incorporating student data into teacher evaluations. I feel comfortable teaching students with diverse learning needs, given that what I'm going to be assessed with is that those students learned something in my class. I am not comfortable being assessed by comparing this year's class with last year's class, and I think the class-to-class method disincentives inclusion of special needs students. Nothing is going to be perfect, but I do understand the need to incorporate some quantitative data.

  • 2
    $\begingroup$ "it's not fair to expect employees to create critical evaluations of their peers" - well, as far as higher education is concerned, we are used to do this for research, so it seems one needs more to rule it out. $\endgroup$ – Benoît Kloeckner Apr 17 '15 at 7:35
  • 1
    $\begingroup$ It's not at all clear that the numbers cited (70, 75, 80) are evidence of improvement or not. In any kind of reasonable model of student performance, one expects variations of this sort due to random variation. The situation is far more complicated when the data come from different classes taught by different teachers. Deciding what is genuinely evidence of improvement seems a hard problem, and it seems indefensible to use numbers like these without solving it. I trust a purely subjective assessment made by an expert teacher far more than this sort of numerical measure. $\endgroup$ – Dan Fox Apr 17 '15 at 11:02
  • $\begingroup$ @DanFox Sorry I should have been more clear, the numbers are pure fiction and not the threshold with which we decide there has been significant change. We have not established yet exactly what will constitute a significant change in a child's performance, and for evaluation purposes it will probably involve the aggregate of a class if not all classes for a teacher. $\endgroup$ – BBS Apr 17 '15 at 12:09
  • 2
    $\begingroup$ @BBS: Think of it as a sampling procedure. The samples are fairly small. Assuming uniform distribution of homogeneous students among identical classes taught by identical teachers (all quite unrealistic assumptions) what is the probability that, among all the teachers, there is some some teacher for whom aggregated measured performance declines several years in a row? That's not very well said, but the gist should be clear. It seems unlikely that the analysis is easier with realistic assumptions. $\endgroup$ – Dan Fox Apr 17 '15 at 15:26
  • $\begingroup$ @DanFox I understand your point now, and that's absolutely true. The sample size for any teacher even over the course of several years is small and is prone to error. However, I think that the public is demanding quantitative measures of teachers. Provided that these are only one part of a teacher's evaluation, and it is a piece of a more holistic approach, I feel that this is a fair compromise between the interests of teachers and the public. I think it is less prone to error than measuring this year's class against last year's class. The merit of using quantitative measures is another topic $\endgroup$ – BBS Apr 17 '15 at 18:43

How can you determine the quality of your teaching, or someone else's?

  1. Give the students questionnaires to fill out (anonymously) to evaluate the teacher. The great thing about this is that it doesn't just grade the teacher, it also gives directions for improvement.

  2. At regular intervals during the class, observe who is talking. If it's the teacher, put a tick mark in one column. If it's a student, put a tick mark in the other column. Count the tick marks in both columns. The best teachers draw students out.

  • $\begingroup$ As far as articulating #2: It may be worthwhile to consider who is talking using methods not as quantitatively simple as counting those tick marks. For example, a mathematics course may (or may not) wish to avoid the Inquiry-Response-Evaluation (IRE) form of discourse that characterizes many so-called "discussion" classes. Personally, one of the ways that I determine the quality of an individual lesson of mine is by considering whether three students ever spoke in a row (without interruption by the teacher) meaningfully. You may find such a phenomenon very rare in most mathematics courses... $\endgroup$ – Benjamin Dickman Apr 21 '15 at 8:26
  • 1
    $\begingroup$ @BenjaminDickman Very nice! And sadly -- yes, it is rare. $\endgroup$ – aparente001 Apr 23 '15 at 3:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.