Teaching to the Test
A recurring criticism of tests used in high-stakes decision
making is that they distort instruction and force teachers
to "teach to the test." The criticism is not
without merit. The public pressure on students, teachers,
principals, and school superintendents to raise scores
on high-stakes tests is tremendous, and the temptation
to tailor and restrict instruction to only that which will
be tested is almost irresistible.
Although many view teaching to the test as an all or none
issue, in practice it is actually a continuum. At one extreme,
some teachers examine the achievement objectives as described
in their state's curriculum and then design instructional
activities around those objectives. This is done without
regard to a particular test. At the other extreme is the
unsavory and simply dishonest practice of drilling students
on the actual items that will appear on the tests.
In addition to offending our moral sense, teaching the
actual items on a test (what James Popham calls "item
teaching") is counter-productive for the very practical
reason that it makes valid inferences about student achievement
almost impossible. There is nothing special about the set
of words that happens to appear on a given vocabulary test.
We assume that the words are a sample from a larger population
of words, and we want to infer something about the students'
knowledge of this larger set, their general vocabulary.
In like manner, we want to infer that students can solve
not only the particular set of math problems on a test,
but that they can solve an entire class of problems. Drilling
students on a specific set of test items destroys our ability
to generalize to this larger domain.
But is teaching to the test all bad? Emphatically not.
Consider the coach who drills young athletes on the very
skills they will perform in competition, or the typing
instructor who teaches students precisely the finger arrangements
and keystrokes that will be used in typing. These practices
are not seen as unethical or unsavory for the simple reason
that in these two domains instruction and assessment merge
into a single activity. Indeed, instructing students on
anything other than the actual test itself seems illogical.
The above two examples are so obvious as to be trivial.
But more significant illustrations of the issues are easy
to find. In the ambitious New Standards Project, a national
initiative that regularly brought teachers together from
around the country to learn techniques for integrating
instruction and assessment, participating teachers learned
to literally merge these two activities in such a way that
they were indistinguishable. Lauren Resnick of the University
of Pittsburgh, one of the visionaries behind the project,
noted that rather than bemoan the inclination to teach
to the test, we should take advantage of it. We should
make exercises so compelling, and so powerful as exemplars
of a domain, that honing one's ability to solve them represents
generalizable learning and achievement. Viewed in this
light, teaching to the test is no longer vaguely disreputable
because the skills and knowledge are themselves general
and are the very things we wish students to acquire.
In his senior level psychology course on learning at the
University of Nebraska, professor Dan Bernstein (now at
the University of Kansas) was disappointed in the level
of understanding of key concepts that his students displayed.
He decided that the fault might not lie entirely in his
students, but in the way he approached both instruction
and assessment. Over the next few years, he changed his
assessment from short abstract essay questions to problems
that asked students to apply concepts in new contexts;
added out-of-class questions about the readings to free
up class time for discussion; and provided web-based examples
of responses to test problems, so that students could learn
to identify what makes some answers better than others.
In short, Professor Bernstein merged instruction and assessment
in such a way that "teaching to the test" became
an integral part of his craft. The reader is invited to
examine his approach in detail at his online
In its program of advanced teacher certification, The
National Board for Professional Teaching Standards encourages
certification candidates to practice putting together portfolios.
They urge candidates to get suggestions and critical feedback
from their colleagues and from others who have gone through
the process. Candidates are encouraged to study excellent
teachers and how they think, write about, and reflect upon
their work. The National Board advises candidates to take
several videotapes of their own teaching, to think about
and write critically and reflectively about what they see.
Teachers are encouraged to anticipate the difficulties
students will have with various concepts and how to structure
and sequence instruction to minimize these difficulties.
In essence, the National Board encourages teachers to practice
and hone the very things they will be tested on.
There is a lesson here for teachers and assessment specialists
alike. The tension between the instructional and assessment
communities, as well the pejorative connotations that "teaching
to the test" entails, will continue unabated so long
as testing and assessment are seen as something quite apart
from instruction and learning, rather than an integrated
reflection of what was intentionally taught. To paraphrase
A. G. Rud of Purdue University, what is needed is a deliberate
attempt on the part of all parties to link curriculum,
instruction, assessment, and standards in a more generative
and even transparent way.
Carnegie Perspectives is a series
of commentaries that explore different ways to think
about educational issues. These pieces are presented
with the hope that they contribute to the conversation.
We invite your response at CarnegiePresident@carnegiefoundation.org.
Join the Carnegie Perspectives email list by sending
an email to CarnegiePresident@carnegiefoundation.org with "Subscribe" as
the subject line.