An Attempt to 6
observations. Therefore, it was rational to
double random sample the data for the reliability and
validity of observation. First, the number of the
sessions(1-37) were randomized and then the
randomized order was fifty-fifty randomized. After
that the queues were multiplied that resulted in 13
sessions for assessing the reliability and validity
from the matrix of all the category frequencies. The
discrete time was in the columns.
Corelations were calculated between the
columns. The correlations in the matrix were
squared. A procedure that resulted in common
variances in the off-diagonal cells when nulls
replaced the ones in the diagonal cells. Application
of the formula (Nunnally, 1967, p. 195,6-23)to the
common variance matrix minus the diagonal sum of the
ones gave a value of 0.94 for the coefficient of
reliability of observation.
On the contrary, a validity measure was a more
complex one. The question was much about: are the
frequencies in the categories where they are to be.
So the maximum differentiation between the categories