One group of Ss (N = 50) had a test-retest interval of only two
or three minutes; another group (N = 50) had an interval of one day:
and the remaining Ss (N = 336) had an interval of one week. The
length of the test-retest interval, within these limits, made no appreciable
or statistically significant' differences on any of the scores. All
scores showed some slight practice effect, regardless of the time intervening
between test and retest. Therefore all the data were combined.
The means and standard deviations for all the scores on each of the
two administrations are presented in table 2, along with the reliability
coefficients estimated for a single administration. The reliability was
determined by the intraclass correlation (R1) between the first and
second administrations. The reliability of the two administrations combined
can be obtained, of course, by means of the Spearman-Brown
formula, boosting the length of the test by a factor of two.
Two points should be especially noted in table 2: (a) some scores
show greater "practice effects" from the first to the second administration
than do others; (b) the reliabilities of the various scores differ