Who Prepares the GRE Psychology Test, and How
Most psychology faculty members do not know how the GRE Psychology Test is prepared. In fact, neither of us knew
about the procedures when we were initially asked to serve on the committee. The GRE Program of the Educational
Testing Service (ETS) appoints a committee of six psychology faculty members from U.S. colleges and universities. The
committee members are selected to be diverse in terms of their research specialties, the geographic location of their
universities, and so forth. Furthermore, they must have a broad background in psychology as a whole, in addition to
expertise in just a single area.
Each committee member writes about 15 potential test questions per year. The ETS test developers also invite other
faculty members to write additional questions. The committee members individually review the hundreds of questions that
are submitted each year. Then they meet for a three-day session once a year to discuss each question. During this
session, they review whether the question addresses an important issue, whether it is clearly stated, and whether the
question has only one correct answer.
On the GRE Psychology Test, 40% of the questions deal with material in cognition, sensation and perception, learning,
and biological psychology; another 43% deal with social psychology, personality, development, and abnormal behavior.
The remaining 17% cover material such as research methods, history of psychology, and applied psychology. On the
actual test, however, the three types of questions are distributed throughout, rather than appearing in three separate
sections. Your overall score for the GRE Psychology Test is based on questions from all three of these areas.
After each administration of the test--and before scores are reported at ETS--statisticians at ETS calculate extensive
statistics on each question. The information includes the percent of students who answered each question correctly, the
point biserial (that is, the correlation between performance on that question and performance on the test as a whole), and
information on how well the students who selected each choice for a given item performed on the rest of the test. For
example, the results might indicate that the students who chose answer "B"--the correct answer--had the highest overall
scores on the test, whereas those who answered "D" had the worst. These statistics call attention to items with possible
problems, and almost any test has a small number of these items. Mostly, these are cases in which the point biserial is
less than .30, and a large percentage of the best students had marked a choice other than the intended correct answer.
The staff then consults with committee members to decide whether the item might have a second acceptable answer.
Questions that are judged to be problematic are not used in computing scores. The staff at ETS continually reviews the
performance of test questions. If performance patterns on a specific question change, they will stop considering that
question in the scores. Furthermore, they will not use that question in future tests. A change in performance patterns may
occur for many reasons. However, the most common problem occurs when new research either disproves an old theory or
supports a new explanation of behavior.
source:
http://www.virginia.edu/psychology/down ... ystify.pdf
