Monday, October 26, 2015

The Correct Number of Standardized Tests

The national conversation will now swing around to figuring out exactly how many standardized test should be given in schools. What, we will all wonder, is the correct number of standardized tests necessary for every student to experience in a year, or in an entire school career.

Here's the correct answer.


Zip. Zero. Nada.

Students need standardized tests like a fish needs a bicycle. Standardized tests are as essential to education as a mugging is essential to better financial health.

Is there a benefit to the child to be compared and ranked against the rest of the children in the country, to be part of the Great Sorting of children into winners and losers? No. Having such rankings and ratings may advance the agenda of other folks when it comes to writing policy and distributing money, but those benefits are for those folks-- not the children. The mugger may benefit from mugging me, but it does not follow that I enjoy a benefit.

Are there standardized tests from which a classroom teacher can glean useful information? Sure-- but those tests are best chosen to fit the needs and concerns of one particular teacher and one particular collection of students. A diagnostic test might help me with Chris, but there's no reason to believe it would help me better understand Chris if it were given to every other student at the same time.
Do the poor children of some non-white non-wealthy neighborhood need to take the Big Standardized Test just like the rich white kids so that we have equity? Maybe-- but you know how else we could even that out? We could have all the public school kids do what the very wealthy private school students do-- take no BS Test at all. That would also provide equity.
Can I squeeze some useful information out of some standardized tests? Sure. I can grow and learn important lessons from being mugged, but that doesn't mean that getting mugged is still great and worthwhile. Do not tell me what I can learn from student standardized test results-- tell me what I can learn from those results that I cannot learn in faster, better, clearer, easier, cheaper ways on my own. Getting mugged might teach me not to take a lot of money with me when I leave the house, but are you sure I couldn't learn that lesson without getting punched in the face?

Do we need tests so that teachers, parents and students know "how the student is doing"? Only if the teacher, parents and students are clueless. Parents and teachers who are paying attention and doing their jobs know how the children are doing (and those who don't care still won't care when you wave a test score at them). And the students should be learning one of the most important lessons and skills of an educated person-- how to evaluate and assess yourself, so that you can be a self-directed, self-actuated human being, answerable to your own judgment, goals and assessment. You cannot learn that from a standardized test.

If you want me to inflict a Big Standardized Test on every single one of my students, you need to be able to answer one question:

What will I be able to do to further my students' education that I could not possibly accomplish any other way? If your BS Tests were denied access to my classroom, what benefits would my students be cheated of? If the universal one-size-fits-all BS Tests were banned today, what would my students be missing from their education tomorrow?

Nothing. The number of necessary standardized tests is zero.


  1. Let me suggest a few reasons that standardized tests might be useful.

    In the classroom it might inspire a teacher to reevaluate his or her evaluation of a student. This is especially important for male students. Make students, on average, do better on standardized exams than would be predicted by teacher assigned grades (conversely, of course, female students, on average, do worse on standardized exams than would be predicted by teacher assigned grades). This will become increasingly important for college bound students if standardized test scores are no longer used for college admission, making the gender imbalance in higher education larger.

    At home, it give parents another view of their children's academic attainment. More information is always helpful, and grades that are inconsistent with standardized test scores should help parents start a conversation with teachers and school administrators about what is best for their students.

    In the community, standardized test scores give some insight into how well the school is functioning. Grades are not useful because even in the least resourced, and most chaotic high school in the country, ten percent of the graduating class graduates in the top ten percent of the class. Schools where the top ten percent of the graduating class are getting scores on standardized exams that are the same as the bottom ten percent of other high schools should raise concerns in the community.

    At the national level standardized exams are important because they provide evidence about the causes of poor academic performance. Again, every single high school, from the richest neighborhood to the poorest, has a valedictorian (or likely these days multiple ones) with a perfect GPA. If teacher assigned grades are our only yardstick, we might well conclude that poverty has no impact on academic performance because the distribution of grades is roughly equivalent across high schools irrespective of the communities that they serve.

    1. As a parent of two NCLB era children who have never known school without annual standardized testing, I can tell you with 100% certainty that those test results tell me NOTHING about what my children know and can do. I get much more detailed information from my children's teachers, who actually see my children work and learn. Those pretty little charts tell me things I already knew: my son is not as good at analytical math as he is at reading literature. My daughter scores head and shoulders above most of her peers. Guess what? She does that at school, too. So go ahead and spend more money to tell me something I don't already know. It's good to know I'm right.

      As a teacher in the same district, guess what? I MAYBE get ONE NUMBER. That's it. No line item analysis, no indication of which standards met or failed -- nothing. Guess how I learn about my incoming students? I give my own assessments.

      I received NO numbers about any of my students this year. Mercy me, the sky is falling! How can I teach without that ONE NUMBER?!???

      The tests do NOT reveal that, thanks to insane emphasis on test scores, my district has narrowed the English curriculum so much that my 8th grade students cannot read or write in cursive, they cannot construct complete sentences, they cannot tell nouns from verbs, they cannot think beyond one correct answer. They do not question. They are most comfortable with multiple choice quizzes and tests, and typically excel at those (yeah!). They are obedient little sheep, just waiting for me to tell them what is the right answer so we can move on. This is in a suburban district, by the way.

      No creative thinking. No creative writing. Curriculum modeled after a BS test the state jettisoned after one year.

      So tell me again, how is that standardized test helping anyone?

  2. The male student example you frequently offer simply assumes the answer that you claim to be searching for. I look at the disparity between standardized test scores and classroom grades and conclude that it is further proof that standardized tests don't measure useful things.

    Since standardized tests are generally normed, they are subject to exactly the same fallacy you claim for school grades. Standardized tests do not measure a student against an objective standard, but against other students.

    Your "evidence of the causes of poor academic performance" argument is just silly.

  3. In the early 1980s the State of Alabama got on the objectives/accountability bandwagon. Competencies were identified in language arts, reading, and math. A standardized test was developed. In the middle school where I was serving as assistant principal, that test was given at the sixth grade level.

    The test was given fairly early in the second semester, but the results didn't come back until late in the year. (I'm writing this from memory, more than 30 years later.) They came on green and white striped, tractor feed, fan fold paper. I sat looking at the math test results. The report they had sent was called the non-mastery report. Across the top of the first page were the 15 or so math competencies that had been identified for sixth graders. Down the left hand side of the page were the names of the students. Out from each student was a minus sign under any competency that the student had failed to "master," which I think meant getting 80% of the questions pertaining to that competency correct. If the student had mastered a competency, no mark appeared; this was the non-mastery report.

    I leafed through the pages, wondering what we could make of this information. Surely the teacher knew far more about what any given student knew than would be revealed by how that student did on one test given on one day of the year. Suddenly it dawned on me that these weren't actually multiple pages, but one large page fan-folded. I went into the library where I could open up the sheet on a library table and see the entire report at once. I walked down to the "bottom" of the page, and what I saw were rivers of minuses running down from some of the competencies, and no clear patterns under others.

    This, I thought, was news we could use. The rivers of minuses did not give us answers, but they showed us where to look. Perhaps so many students had failed to master a given competency because it was scheduled to be taught after the time of the test administration. Perhaps we had covered it in September and the students were not fresh on it anymore. Perhaps our text and other materials gave insufficient attention to this competency. Perhaps the way we were teaching the competency was inadequate in some way. In any event, we needed to dig to find explanations for the competencies that had stymied so many of the sixth grade. Further, next fall we needed to do some diagnostic work to determine how much work we needed to do on these before heading off into the seventh grade curriculum.

    This, it seems to me, is a valid and useful application of standardized testing. Add to this the ability to break results down by the NCLB demographic categories and you have a powerful tool to help guide analysis into how your instructional program is affecting students over time. We are not using testing in this analytical way today. We are using it to rank and punish, but that's a subject for another note.

  4. Given your statement about norm referenced exams (exams like the SAT, SAT 2, etc) vs criterion referenced exams (exams like AP, NAEP, TIMSS, PISA, and PARCC, etc), perhaps we should pin down the types of standardized exams that you find objectionable. Is it really all standardized exams of all types?

    It is not the discrepancy between standardized test scores and teacher assigned grades is surprising, but that a student's gender reliably predicts the discrepancy between test scores and teacher assigned grades even at an early age. Here is a link to the working paper that discusses this: (the published version is, I believe, behind a paywall).

    Saying something is "silly" does not constitute an argument, so perhaps you could flesh this one out a bit more. The problem with using teacher assigned grades to evaluate student educational attainment is that those grades have little meaning outside of the student's classroom, and virtually no meaning outside of the school.

    1. For all practical purposes, there is no such thing as a criterion-referenced standardized test. All standardized tests are norm-referenced because they are designed to produce a bell-shaped curve to tease students apart from each other as far as who "deserves" to pass or fail. It's just that the allegedly criterion-referenced exams get to set their pass-fail cut marks after the tests are already taken. If ever there was an exam which all students passed, no matter how much the students actually learned, people would be shouting that the test was obviously too easy precisely because all students passed it.

    2. Dienne,

      If the College Board is trying to norm the AP exams they are certainly doing a very poor job of it. In the Physics 1 exam only 5% received a score of 5 and 39.2% received a score of 1. In the Calculus BC exam, on the other hand, 45.4% received a score of 5 and 14.8% received a score of 1. Neither is very bell-shaped. You can find the scores here:

    3. When states assign cut scores AFTER the tests are taken and graded, those scores are meaningless in every area other than student ranking. States can and do adjust cut scores for purely political purposes.

  5. The College Board absolutely norm-references the AP Physics exams, but it is not trying to fit it into the standard Gaussian distribution. It is intentionally narrowed for high scores in AP Physics 1 and 2 (as is also true for the new Chemistry exam), perhaps to look more "rigorous?" The distribution for AP Physics C is basically flat. The justification for norming the exam is that it is difficult to assure the questions are equivalent year-to-year in difficulty and will yield the same numerical raw scores, so yes, it still relies on norming for year-to-year comparison purposes.

    1. Helene,

      Any thoughts on BC Calc? As far as I can tell, between 45% and 50% of students have earned a score of 5 in since 2011. That seems to be very different from the "bell-shaped curve" that Dienne is concerned about.

    2. I do not know. I have worked with a number of folks involved in the physics exams but not the math exams. Again, I reiterate my point that the distribution is partially historical artifact so students may be ranked across years of tests. My point was that the AP tests are normalized so the distribution is consistent year to year, whatever its shape. Clearly distributions vary from subject to subject (each test is a unique little snowflake?) and to try to pin Dienne down on the shape of the particular distribution seems to me a petty argument that ignores the gist of what she is trying to communicate, that is, the AP tests are normalized using the performance of the cohort of students taking that year's test. They are not purely criterion referenced.
      Dollars to doughnuts, if the College Board undertakes a redesign of AP Calc BC, the distribution will look much more like Chem and Physics 1. There will be a much more exclusive club of kids who score 5's.

    3. Thanks, Helene. Yes, it's the fact that tests are normed to create a desired outcome - by definition that means that students are compared against each other, not against some "objective" criteria. Many such tests (PARCC, SBAC, etc.) are designed to produce a bell-shaped curve. But even those that aren't don't disprove the fact that students are still being compared against each other.

    4. teachingeconomist, if you look at the number of students that took BC calc, you will see that it is much smaller than the number that took the AB. It's a good bet that College Board recognizes that only the most elite students are taking the more advanced tests. Similarly, the Physics C exams, which are calculus-based, have far higher scores than the Physics 1 and 2, which are algebra-based. Again, as expected, fewer students take Physics C than the Physics 1 and 2.

  6. Adults, adults, adults. You come off so smart sounding, and yet, in the end, just like standardization, you are meaningless. Whatever the reason you come up with for testing, grades, or achievement and success even, it will always remain meaningless unless it is decided by the student. Go learn about how the future of "education" is in the hands of students, not adults. Letting them own their school day is but a start. Only then will you hold the seeds of sustainability, peace, justice, and respect.

  7. Common core and its tests are, as many of us know, tools to get the money and power into the hands of corporate interests. The same people behind ed-reform are the same people who have been OK with off-shoring our economy. We can't allow the con to continue. And our book--