I am interested in developing confidence intervals for reporting results
on a State Assessment. A typical test is constructed thusly:
60 items, with 55 mult. choice questions and 5 constructed response
questions. Each mult choice que. is scored 0 or 1, and the constructed
response items are scored 0, 1, 2, or 3.. The constructed response items
are scored by two raters; the final score for each item is the ave. of the
two raters.
I would like to compute the standard error from the standard deviation
and the reliablity estimate of the test, but i am unsure how to compute a
reliability estimate that accounts for the inter-rater variance.
Any ideas/suggestions?
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|