Coming Soon to a Classroom Near You?

Some rambling thoughts on a fascinating set of articles about measuring teaching.

Today's NYT carried two stories -- on on page 1 -- about new techniques being used to evaluate K-12 teachers.  The news in the stories concerns two things: existence of a very large program for measuring educational effectiveness in schools and the central role of video-taping teachers teaching in that program.

Local readers' radar might ponder the resonance between programs like this and higher education assessment and higher education "learning and teaching centers" and the individuals and organizations who live off, rather than for, education.

The first story ("Teacher Ratings Get New Look, Pushed by a Rich Watcher") highlights Bill Gates' (via the Gates Foundation) interest in a gigantic project measuring the "value added" by teachers through multi-mode assessment. Among other tools : videos of instruction that are scored by experts.

"Interesting" is the fact that one of the movers and shakers in the project is none other than Educational Testing Services. And so this represents yet another opportunity for that organization to live off, rather than for, education in the U.S. Other contractors are mentioned in the story too -- as has been true of the assessment movement more generally, a big part of the driving force seems to be entrepreneurs who, after persuading you that you need to do something are more than happy to sell you the equipment needed to collect the data and then expertise to evaluate it.

The second article, "Video Eye Aimed at Teachers in 7 School Systems," describes some 3,000 teachers who are a part of the first phase of this search for new methods to evaluate teachers. Each will have several hours of teaching video-taped and the tapes will be assessed by experts using a number of carefully validated protocols.

The first article, describing the scope of the project, notes that the rating of 24,000 video-taped lessons will come to something like 64,000 hours of video watching. On a full-time basis that represents 32 person years of work. At 180 days/year, that's about 44 years of teaching.  The article suggests the costs to a school district will be about $1.5 million up front and then $800,000 per year.

I wonder if anyone has assessed the value of the information produced.

In the middle of the report there is a line about how this is a step forward because rather than having the principal observe once or twice during the year, outside experts (using scientific protocols) can observe up to a half dozen times.  This suggests an interesting phenomenon: in the name of standardization and objectivity, we deskill and depersonalize (among other things).

In one paper on value added modeling (VAM), by an ETS staff person (Braun 2004, 17), one finds this argument: (1) quantitative evaluation of teaching is here to stay; (2) evaluation of gains is preferable to just measuring year-end performance; (3) we have to think what would get used if not this; (4) therefore, use VAM even if it has real limitations. Another, by a Michigan State University economist concludes (about VAM):

We are looking at the educational system through a poor quality lens. The real world is probably more orderly than it appears from the analyses of noisy data (Reckase 2004, 7).

Resources

Amrein-Beardsley, Audrey. 2008. "Methodological Concerns About the Education Value-Added Assessment System." Educational Researcher, Vol. 37, No. 2, pp. 65–75

Braun, Henry. 2004. "VALUE-ADDED MODELING: WHAT DOES DUE DILIGENCE REQUIRE?"

Rand Corporation. 2007. "The Promise and Peril of Using Value-Added Modeling to Measure Teacher Effectiveness"

Reckase, Mark D. 2004. "Measurement Issues Associated with Value-added Methods"

Wikipedia. "Value Added Modeling"

Closing the Loop in Practice: Does Assessment Get Assessment?

At a liberal arts college with which I am familiar, the administration recently distributed "syllabus guidelines" with 34 items for inclusion on course syllabi. Faculty leaders balked and asked for clarification: which of the 34 items were mandates (and from whom on what authority) and which were someone's "good idea"? The response was that guidelines are merely guidelines and most of the content were indeed good ideas. Most were.

A subsequent examination of a sample of syllabi revealed that most syllabi did not contain all 34. More specifically, there was not universal inclusion of several that, apparently, are important for accreditation purposes. 

The semester has begun.  The syllabi are printed.  The administration disseminated the guidelines -- their obligation is fulfilled.  If faculty choose not to comply, that's their decision. Overall, the situation is alarming because the school could appear to be non-compliant to its accreditors.  And it's the faculty's fault.  And folks are wondering how to fix it.

THIS COULD BE TURNED INTO SOMETHING POSITIVE, a shining example of assessment, closing the loop, and evidence-based change.

But first, WAIT A MINUTE! Do faculty get to say "We told them what to do; if they can't comply and don't learn, it's not my fault."?  Of course not.  If students aren't learning, faculty are doing something wrong.  Lack of learning = feedback, and feedback must lead to change.

Here we have a case of an institution ignoring unambiguous feedback. The feedback is simple: the promulgation of a list of 34 things one should do on a syllabus does not produce the uniform inclusion of the small handful of actually really important things to include on a syllabus.  That's it; that's what the evidence tells you.  It doesn't tell you faculty are bad; it tells you that this method of changing what syllabi look like was ineffective.

Never mind that any good teacher knows that you cannot motivate change with a list of 34 fixes.

The correct response? Close the loop: listen, learn, change the way syllabus guidelines are handled.

The unfortunate thing here is that folks who know (faculty) brought this immediately to the attention of the folks in charge. Faculty noted that the list was too long, its provenance ambiguous, its authority unclear, its applicability variable, its tone insulting. A solution was suggested. All this was met with, basically, a brush off -- they're just guidelines not requirements, what's the big deal?

And, it turns out, that is precisely how faculty understood them. No need for alarm.  Some adjusted their syllabi to some of the suggestions in the guidelines. But apparently, the faculty didn't all implement a few of the guidelines that really do matter (to someone).  Arrrrrrrgh.

And now for a little forward looking fantasy of what the outcome of this situation COULD be.

Since administrations and the assessment industry are apparently NOT really ready to adopt the underlying premise of assessment -- pay attention to feedback and change accordingly -- the faculty will.

From now on, only the faculty will disseminate syllabi guidelines. They will very clearly distinguish between legally mandated content, accreditation relevant functionality, college-specific custom and standards, and good pedagogical practice in general. They will invite all parties who become aware of syllabi-related mandates (or new good ideas) to communicate them to the faculty's educational policy committee for consideration for inclusion in their next semester's guidelines.

Those guidelines will explicitly articulate general goals (exactly which ones to be determined) such as syllabi are to be interesting documents that are useful to students and that permit colleagues to get a sense of what a course is about and at what level it is being taught as well as suggestions of particular features, boilerplate and examples that might be useful, and fully explained required items. They will include examples of an array of syllabi that explicitly demonstrate a variety of forms that meet their standards. And, all suggestions will be referenced where possible and requirements will be documented in terms of on what authority they are an obligation.

For assessment purposes the faculty will adapt* any externally supplied "rubrics" to their own intellectually and pedagogically defensible standards and practices and encourage our colleagues to make use of these college-specific tools in developing their syllabi.

Educators really committed to the stated goals of assessment would see in this affair an opportunity for an achievement they could boast about.  Those committed to one directional, top-down, assessor-centered, non-interactive, deaf-to-feedback approaches will see in it only faculty reluctance to get with the program. 

One lesson learned here is that institutional processes need adjustment. The amount of faculty and administrative time, emotional energy, and the augmentation of frustration and mistrust that this little thing has engendered was a phenomenal waste of precious institutional resources. Alas, accountability for THIS is unlikely ever to be reckoned.

How Academic Assessment Gets it Backwards

In a letter to the NYT about an article on radiation overdoses, George Lantos writes:

My stroke neurologists and I have decided that if treatment does not yet depend on the results, these tests should not be done outside the context of a clinical trial, no matter how beautiful and informative the images are. At our center, we have therefore not jumped on the bandwagon of routine CT perfusion tests in the setting of acute stroke, possibly sparing our patients the complications mentioned.

This raises an important, if nearly banal, point: if you don't have an action decision that depends on a piece of information, don't spend resources (or run risks) to obtain the information.

Consider, for a moment, the trend toward "assessment" in contemporary higher education. A phenomenal amount of energy (and grief) is invested to produce information that is (1) of dubious validity and (2) does not, in general, have a well articulated relationship to decisions.

Now the folks who work in the assessment industry are all about "evidence based change," but they naively expect that they can, a priori, figure out what information will be useful for this purpose.

They fetishize the idea of "closing the loop" -- bringing assessment information to bear on curriculum decisions and practices -- but they confuse the means and the ends. To show that we are really doing assessment we have to find a decision that can be based on the information that has been collected.

A much better approach (and one that would demonstrate an appreciation of basic critical thinking skills) to improving higher education would be to START by identifying opportunities for making decisions about how things are done and THEN figuring out what information would allow us to make the right decision. Such an approach would involve actually understanding both the educational process and the way educational organizations work. My impression is that it is precisely a lack of understanding and interest in these things on the part of the assessment crowd that leads them to get the whole thing backwards.

Assessment and Evaluating Student Work

It's ironic, given it's centrality, how little that's sensible and defensible has been said about the relation between grading and assessment.  To my mind, it's a lost opportunity to offer constructive criticism of grading in general as well as a failure on the part of the assessment industry to demonstrate and convey clear thinking and to develop useful tools for teachers.

And so here is part one of working through a relationship between grading and assessing.

My students always want to know "how much does this count for" and so my syllabus always says something like

Exam 1	30%
Problem Sets	20%
Final Essay	40%
Participation	10%

When I grade any particular item -- assignment or paper -- I make use (at least implicitly) of a similar decomposition of the grade.  If it is an essay I may be evaluating the quality of the writing, the use of evidence, the structure of the argument, the use of sources, and so on.  If it is an exam, the questions can usually be separated into a finite number of groups, each "testing" a particular skill or understanding of a particular concept (but see fn 1 below).  Let's imagine a class in which five skills or concepts, J,K,L,M, and N, make up the content.  And let's imagine my graded activities from above can be described this way

Exam 1	25%J	25%K	25%L	25% critical thinking
Problem Sets	20%J	20%K	20%L	20%M	20%N
Final Essay	20% Writing	30% J-N	30% Argument	20% Scholarly conventions
Participation	33%  J-N	33% Staying up with material in course	33% Poise,  verbal skills, etc.

Now let's look at how all of the things I've graded fit together.  In the table below, the rows represent skills or learning outcomes that I want students to demonstrate.  The columns show me the evaluative tools I've used and which of these each one included.

Substance	Exam 1	Problem Sets	Final Essay	Participation
Concept J	+	+	+	+
Concept K	+	+	+	+
Concept L	+	+	+	+
Concept M	-	+	+	+
Concept N	-	+	+	+
Writing	-	-	+	-
Argument	-	-	+	-
Scholarly Conventions	-	-	+	-
Critical Thinking	+	-	+	-
Keeping Up	-	-	-	+
Verbal Skills	-	-	-	+

Next let's suppose that I graded each of these items on an A-F scale and that I've made some attempt to put on paper how I "operationalize" the grades "excellent," "good," "satisfactory," etc. I might, for example, have let students know that I consider an excellent use of concepts in the final essay to be when

Essay employs 3 or more of main concepts from the course in a manner that's appropriate to the subject at hand and that demonstrates a strong understanding of what they mean and how they can be useful.

And finally, let's assume that my program goals include concepts K and M and the school as a whole includes writing and critical thinking as goals.

I can simply take scores on concepts K from all four evaluations and M from the last three and then take the writing score from the final essay and the critical thinking scores from the first exam and the final essay and, oila, I've got my assessment.

fn 1  Two things require mention: 1) not every skill/concept that we expect to be learned is measured on every exam/exercise -- exams are samples; 2) many "items" will depend on more than one skill or concept.  More on these issues later.

Responding to Student Writing

In a piece called "ABOUT RESPONDING TO STUDENT WRITING," Peter Elbow writes:

The fact is there is no best way to respond to student writing. The right comment is the one that will help this student on this topic on this draft at this point in the semester -- given her character and experience.My best chance for figuring out what is going on for any particular student at any given point depends on figuring out what was going on for her as she was writing.

I found this document on our schools website under "Teaching Resources" on the Provost Office page. It sounds like pretty good advice.

It also sounds different from the advice I find on another page of the institution's website. On that page I find a "rubric" for assessing student learning in essays. It gives me six categories (overall impression, argument, evidence, counter evidence, sources, citations) and wordy descriptions of different levels of achievement in each. It's pretty unclear from the document how it is intended to be used, but basically, it's a grading scale.

Here's my question: which kind of teaching students to write does my boss want me to use?

Spellings' Flawed Metaphor

An interesting article and even more interesting responses in Inside Higher Ed from a few years back: "The Flawed Metaphor of the Spellings Summit"

Six Things to Beware of, Grasshopper...

A lot of such talk as there is about innovation and change in higher education these days shows up in the general orbit of assessment. Having recently listened to or read a lot of material from assessment experts, I jotted down a few cautions.

Beware the tyranny of software. I like software. I write computer programs. But, at the risk of sounding Asimovian, software has to serve education, not the other way round. For the last ten years we have repeatedly adjusted the way we educate to the needs of the software ("Banner won't let you do that..."). Banner and its ilk are just hammers. No right-thinking carpenter changes the way she builds a house because of hammer limitations.

Beware the fetishization of uniformity. Healthy ecosystems, organizations, relationships, families, and individuals entertain a healthy dialectical tension between sameness and difference, uniformity and irregularity, standardization and improvisation. Whether you are trying to be a virus that can outwit immune systems (or an immune system that can shut down a virus), or building a firm that can weather economic ups and downs, or running a college that produces excellent graduates from a stunning variation of inputs, the key is to cultivate order and chaos simultaneously.

Beware projection and other forms of X-o-centrism. We are all subject to our own version of Saul Steinberg's classic "New Yorker's view of the World." What works for me, or in one course, or in my department, or in our division, or in one school I know about, is surely good for you. Whether it's a analogy or metaphor, an algorithm, a social form, or a paradigm, or a homeomorphism, context and local history matter.

Beware foolish numberers and their arrogant misquotations of Lord Kelvin -- if you can't measure it, it does not exist -- and mindless adherence to quantification as an end in itself.

Beware saviors, those who in the face of skepticism and critique fancy themselves the new Galileo or who too readily imagine they are members of a new Vienna Secession or Salon des Refusés.

Beware assurances that complex things can be done with little effort or in far less time than you think. Most things that are easy and simple and beneficial have already been done. Things like the valid measurement of educational outcomes are not simple. Getting it right takes time and effort.

Most of all, beware a movement that cannot apply its own techniques to itself.