You are not currently logged in.
Access JSTOR through your library or other institution:
The Number of Alternatives for Optimum Test Reliability
J. Brown Grier
Journal of Educational Measurement
Vol. 12, No. 2 (Summer, 1975), pp. 109-113
Published by: National Council on Measurement in Education
Stable URL: http://www.jstor.org/stable/1434035
Page Count: 5
Preview not available
The expected reliability of a multiple-choice test is maximized by the use of three-alternative items. The proof uses a modified version of the Kuder-Richardson formula 21 and assumes that the total number of alternatives is held constant. This result agrees with an earlier one due to Tversky that shows three-alternative items optimize other test criteria. Some empirical results in the literature either tend to support or not contradict this theoretical result.
Journal of Educational Measurement © 1975 National Council on Measurement in Education