THE notion of homogeneity of a psychological test refers to the degree to which the various items in it measure a common attribute. To the extent that a test is homogeneous, three basic conditions should obtain: (1) The average intercorrelation among all pairs of items should be positive. (2) Each item should correlate positively with the total score on the remaining items. (3) The total test variance should exceed the sum of item vari-ances. It will be shown presently that these three statements are mutu-ally redundant. Various statistics are available to represent the degree to which each of these conditions obtains. The purpose of this paper is to point out the mathematical and common-sense relations among sev-eral of the available statistics. Measuring the Correlation Between Two Test Items If two items are dichotomous and scored either 1 or 0, representing
No takes yet. Share an insight, caveat, or question.
William A. Scott (1960) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: