|
SAT Percentile Table
Before presenting
information on examples of standardized tests and different methods for
scoring those tests, I would like to list some of the terms associated
with scores and reliability of standardized tests. Knowing these terms
makes the issue of testing easier to understand.
Confidence Interval-A range extending one standard error of
measurement to either side of the actual test score.
Mean-The average of a set of scores.
Norm-referenced scores-Scores that compare a student's
performance with other students' on the same task.
Normal Distribution (normal curve)-The optimal pattern of score
distribution. The performance level of the characteristic being
measured is on the x-axis and the number of people scoring at each
levels on the y-axis. A theoretical pattern would have the most people
performing at the moderate level and fewer and fewer people scoring at
each end. This is what the graph itself would look like:
Reliability Coefficient
(correlation coefficient)-A statistic that measures how reliable a
test is. A reliability coefficient will range from 0 to 1 (with one
being perfect reliability and very rare).
Standard Deviation-A statistic that measures the variability of a
set of scores. The smaller the standard deviation, the closer the
scores are to each other.
Standard Error of Measurement-The measure of how close a
student's score is to what it actually should be (their true score).
True Score-What a person would score if they could be tested with
complete accuracy.
Validity Coefficient-A statistic that measures how valid a test
is. A validity coefficient will range from 0 to 1 (the higher the
number, the greater the predictive validity of a test). |