Reforming Public School Testing

My wife was a high school teacher in math and science in West Sacramento when we married.  (West Sacramento is in Yolo, not Sacramento County, and is considerably poorer than Sacramento.)  She is forever grateful that I took her away from “all that,” and allowed her to have a life where she homeschools our three biological and five adopted children.  She does a great job, and I assist her where she requests, usually where it involves advanced math.  Having public school teaching experience is an impediment to schooling at home, but my wife has adjusted well.  (In general, what you learn about teaching at college makes you less capable to teach.  My oldest daughter just went through that, and didn’t believe us when we told her that in advance, but agrees with us now after going through the courses.)  Our educational system would be better if we eliminated teacher credentialing, and just hired motivated people who know the subject areas, who love to work with kids of the appropriate ages, and have grit — a determination to succeed.  The present system favors those who are capable at dealing with the paperwork that state and federal governments demand.

For the first two years we were married, I was quiet or sympathetic to my wife’s fellow teachers.  There is a pathology of sorts where because you have gone to school, and have seen a lot of teachers, therefore you understand teaching.  Teachers often do not like amateurs offering their opinions.  I did not volunteer my opinions to them, in deference to my wife.

That said, I would like to offer my opinion on testing students and using those scores to set teacher pay, or district transfer payments.  Remember, this doesn’t affect me because we homeschool; we pay our state/county taxes though we get less from it than most.

(Before I continue, I would like to note that our dealings with the Howard County school system have been unfailingly pleasant; they are a model for how the public schools and homeschoolers should interact.)

Dedicated teachers will say, “We don’t want to teach to the standardized test.  We want to impart real knowledge, rather than the ability to grind out answers to multiple choice questions.”  That’s a good goal.

In my life, I have taken many exams.  I have passed handily with almost everything, but the toughest exams were essay exams where you had to show your reasoning.  This is true in Math, Science, English, Social Science, etc.  This applied to the second half of my actuarial exams, and also roughly half of my CFA exams.  It was true for virtually all of the exams I had in college and graduate school. (As an aside, the actuarial exam multiple choice problems are much, much harder than normal multiple choice problems… schools should adopt their techniques if they do multiple choice because the methods really test people.)

But for the actuarial and CFA exams, those were my experience with broadly administered exams that were mostly essay.  To grade such exams, the test writers create a grading outline and assign points for each point made.  If someone makes a valid point that the grading outline did not account for the grading outline expands to accommodate it.  Typically two people grade each paper, and the average is the grade, unless they differ by too much, in which case a third grader comes in, and the average of the closest two grades is the final grade.

This is real testing and real grading, and guess what?  It is expensive to do, but generates real results.

With both the actuarial and CFA exams, you could not study to the test.  You could only study.  You could not tell year-to-year what would be on the test.  It varied considerably.

The Payoff

That is the way the public school examinations should be.  The tests should be essay-based (after a certain age, like 10), randomized (not everyone gets the same test), unpredictable, but relevant to what should be taught to students at a given age.  Where warranted, the ability to write should be a factor in scoring, because that is a large predictor of how well you have learned your verbal skills.  Partial credit would come to those that show reasoning capability, even if they didn’t get the right answer.

This would be an expensive way to administer exams, but hey, maybe more college graduates without jobs could get some part-time work as a result.  Or, college students could grade them over a summer, lessening the economic cost of college.  But it would be an effective way of encouraging real learning, and measuring it.

Best of all, the essay exams would be a lot like the life they are going to face.  It’s not predictable.  You don’t know what you are going to face.  Face it bravely, or you have no chance.  Prepare now, study hard, or your odds of success diminish.

The same applies to the teachers.  Teach the most important stuff.  Stress reasoning.  Try to create thinkers, not mere memorizers.  Motivate the material in the best way that you can to inspire students to think.  Teach them to write well, and encourage them to read widely.  I may be conservative in my views, but I have read widely, including many things that I disagree with a lot.  Teach them to look at both sides of a question (when they are old enough), and understand the arguments of those that disagree.

And then, let the teachers live with a system where the test is unpredictable, and they cannot affect its outcome directly.  Will they like that?  I don’t know.  What I do know is that if America is serious about reforming its schools, it will adopt an approach like this, and pay the money necessary to have standardized essay tests graded.

But once teachers can’t affect the results of the testing directly, they will teach their students the best that they can, because if they don’t and the scores of their students sag, they may be gone after some time.  This would empower real accountability, and eliminate the deadwood among teachers that can’t do the job.  Remember, aside from the best, most teachers peak in their effectiveness after year six.  As a result, tenure for public school teachers is really a bad idea.

Education improves by removing unmotivated and untalented teachers.  The best teachers love their subject, their students, and love what they do.  They have grit, and are determined to make a difference.  With 35-45 years behind my public school education, I can still name my teachers who were motivated, against those that went through the motions.

PS — to my regular readers, I know this was an unusual post.  I’m not planning on doing more of these.

4 Comments

  • tapertaper says:

    On a practical note, you’ll need a way to deal with a lot more failures in the system.

    Which means you’ll need to tackle the social problems an honest testing system will ruthlessly expose.

  • somrh says:

    A couple of biologists developed an interesting test style (along the lines you’re describing) for introductory undergrad biology courses.

    http://www.bioone.org/doi/abs/10.1641/0006-3568%282006%29056%5B0066%3AANMFAC%5D2.0.CO%3B2

    The basic test structure could be easily implemented for other subjects. Obviously there was a time factor.

    If you can’t access it, I may be able to send you a copy.

  • duran.perkins says:

    Oh, don’t apologize! We enjoy these “off-topic” posts!
    The method of testing you propose is close to the method already at work on AP (Advanced Placement) tests. These tests give the fairest shake of any standardized test, in my view. Interestingly, the newly reworked SAT also integrates a writing sample into a students’ score. My students are the most worried about this section of the test! A physics teacher I knew once did a correlative study of the best method of predicting success in physics. A high verbal score correlated more highly than a high math score, contrary to his expectations. I agree with the previous respondent that honest testing will expose a myriad of problems that schoolmasters would like to gloss over. Impoverished households tend to be impoverished in more than monetary ways: the middle class child has a much larger vocabulary by age 3.
    Current NCLB testing suffers from a lack of accountability. State officers are free to make the tests easier year-by-year, to produce apparent gains. Most apparent gains in NCLB testing appear to have been illusory, and I predict, based on some small-scale analyses that I’ve seen, that the increases will not stand up to statistical scrutiny. (We could summarize: test administrators cheat.) Not only them, but as soon as tests are linked to school funding or teacher pay, teachers cheat, as has been demonstrated in Atlanta, Camden, Philadelphia. (Other cities will follow. This is not a small problem.)
    It is difficult to imagine a test design that is honest, helpful to teachers, students, and administrators alike, or that is free from class-based bias. Still, it’s worth the effort: we pay attention only to those things that we bother to measure. (My year teaching in a dysfunctional charter school demonstrated that lesson very well. The poorly designed NCLB state tests were the only thing that kept that particular school honest.)

  • toptick says:

    You say “most teachers peak in their effectiveness after year six”. That seems completely plausible — do you have any research citations for that?
    Thx