�Assessment drives curriculum� is an educational maxim. What you test becomes what you teach---if only because it becomes what your students really pay attention to. Any of us who have been good students know that the key question has always been, �Will this be on the test?� Some politicians have taken this truth to mean that testing all the children in public school will improve curriculum. What they either fail or refuse to understand is that a test is not a test. What kind of test you give matters tremendously to whether the testing will improve curriculum or merely undermine public schooling.

Understanding tests, unfortunately, means understanding some boring basics. I guarantee that if you try to explain the difference between a criterion-referenced test and a norm-referenced test, the average politician�s eyes will glaze over. So will the eyes of the average member of the public. (Actually, I wouldn�t be surprised if some of the people reading this have their eyes glazing over. Consider me to have rapped a virtual ruler on your desk to get you to pay attention. There will be a test on this material.)

Most standardized tests, including the Iowa tests, the Terra Novas, and the new state-mandated tests such as the Wisconsin Knowledge and Concepts tests are norm-referenced tests. A norm-referenced test is one that grades on a curve rather than an absolute standard. In other words, if a norm-referenced test is designed properly, half of the kids tested with it will perform below the average. Fewer than half of the kids in a particular group, be it school district or state, may score below average but only if some place exists where more than half of the kids score below average. All of the children of Lake Woebegon can be above average but, by definition, not all the kids taking the Iowas or the Terra Novas can be.

A criterion-referenced test is different. It is not graded on a curve. It is graded against an absolute standard. You�ve seen these tests. You�ve taken these tests. They are the math tests on which you get an A if you get at least 23 out of 25 right and there are as many As as there are kids who get at least 23 out of 25 right.

As more and more states and school systems go to standards and benchmarks, one would expect that their results would be measured against these standards and benchmarks. We pretend that is happening, but it is not. Instead of criterion-referenced tests, we give norm-referenced tests because they are readily available in the marketplace and therefore cheaper. Each state or each district does not have to pay for individualized test development.

The way that designers of tests report the scores tends to add to the confusion. They fool people into thinking that we have given criterion-referenced tests. We tell the public and parents that our students performed at a minimal level, a basic level, a proficient level, or an advanced level. Despite how it sounds, the cut-offs for these levels are not based upon comparing how much the kid knows against how much we expect him to know. They are based upon comparing how much Billy knows against how much Susie knows. Billy can score at the minimal level even though he knows everything we have decided he should know as long as Susie knows more.

As you can see, these norm-referenced tests cannot drive curriculum without turning curriculum into a contest between districts and states rather than into a discussion about what kids need to know in this world. The tests change the content of the discussion from what Billy should know to whether Billy has learned as much as Susie. For a district to �win,� it must not only teach what it believes it should teach; it must also teach more than the district next door.

Worse, the political effects for the public schools are profound. While all children can learn, background matters in norm-referenced tests. On the average, a child from a stable, financially-solid home with better-educated parents will know more than a child from a less stable, poorer home with less-well-educated parents. My children, for example, have large vocabularies in part because I have a large vocabulary. If doing a good job means having every student be at least �proficient� and if �proficient� means above average, school districts with more students with less ideal backgrounds will be labeled failures. Even if these districts teach everything we ask them to, their students are likely to score lower comparatively. These districts, therefore, will always �fail.�

If the group we standardize our test on is the students of this country and any school that is below average is considered a �failing school,� then, by definition, half the schools in this country will be failing schools. Similarly, if the group we standardize our test on is the students of this country and any child who is below average is considered a �failing student,� then, by definition, half the children in this country are �failing students.� Having all these �failing students� and �failing schools� can keep politicians busy reforming education for generations to come.

Unfortunately, assessment not only drives curriculum, it also serves a political agenda. If so many schools and students are failing, why not consider dismantling public education? After all, our assessment shows that public education does not work. (We�re not sure what it shows about private education or homeschooling because we don�t apply the same assessments to them. We do not require those students take the same tests as we require in the public schools. Whether though stupidity or conspiracy, our politicians seem to have guaranteed that they cannot be caught in the comparison game.)

Once again, the wand is waving the wizard. We are measuring one thing and pretending to measure something entirely different. Maybe public schools in general do not work as well as private schools in general but you sure can�t prove it by these tests.

|