Reliability and Validity of Standardized Tests|
Chong Ho Yu, Ph.D.s
An acceptable standardized test should have reliability coefficients of
- .95 for internal consistency
- .90 for test-retest (stability)
- .85 for alternate-forms (equivalency)
- Concurrent validity, which is a form of criteron-related
validity, is often reported. This is a numerical estimates of the extent to
which a test correlates with another established test or tests. A valid
achievement test may have concurrent validity coefficients ranging from .6 to
- It is most important to evaluate the content validity of standard
achievement tests: Does the test items match my instructional objectives?
- Internal consistency is generally high. Items are written
by specialists, pretested, and selected on the basis of the results of
a quantitative item analysis.
High degree of validity in terms of prediction
- Students in an Ohio community took the Iowa Test of Basic Skills in the spring of their eighth-grade year and the Ohio Ninth Grade Proficiency Test (ONGPT) in the following fall of their ninth-grade year.
The scores from both tests, for a period of 3 consecutive years, were correlated to determine the predictability
of passing or failing the ONGPT, based on the standardizedtest scores.
The correlations were found to be significant for reading, mathematics,
writing, and citizenship.
- It was also shown that the percentages of students failing
the ONGPT who scored below the third stanine were high in three areas
(reading--60 percent, mathematics--93 percent, and citizenship--90
percent). Therefore, stanine scores can be helpful predictors of the need for intervention programs.
Standardized tests are impersonal
- Informal evaulation is subject to Hawthrone effect and Halo effect. However, standardized tests are robust against human attitudes.
- During the course of a large-scale study of teacher
preparation for standardized tests in two low-stakes school districts,
the question of whether the attitudes of teachers about standardized
tests are a factor in their administration of mandated standardized
tests in the classroom was addressed.
- Third-, fourth-, and sixth-grade teachers (n=178)
completed a survey of perceptions of standardized tests that also asked
how much time teachers spent preparing their classes for the tests.
- Students in one school district took the Comprehensive
Test of Basic Skills in grades 3 and 6, while fourth graders in the
other district took the Iowa Test of Basic Skills.
- Teachers' perceptions of standardized tests varied, but
were not consistently related either to the effort that teachers put
into preparing for the tests or to their students' test performance. It
was concluded that teachers administered the tests in ways largely
uninfluenced by their personal feelings as reflected in the Likert
items of the survey.