The concept of test reliability is examined in terms of general, group, and specific factors among the items, and the stability of scores in these factors from trial to trial. Four essentially different definitions of reliability are distinguished, which may be called the hypothetical self-correlation, the coefficient of equivalence, the coefficient of stability, and the coefficient of stability and equivalence. The possibility of estimating each of these coefficients is discussed. The coefficients are not interchangeable and have different values in corrections for attentuation, standard errors of measurement, and other practical applications.

A table is presented which directly gives phi coefficients accurate to three places when entered by the proportion of one sub-group responding in a specified manner and the proportion of a second sub-group responding in the same manner. The table gives coefficients identical with those obtained by formula if the sub-groups are equal in number. The phi coefficients can readily be expressed, if desired, in terms of critical ratio or chi square. The table is more accurate than the use of abacs and eliminates the use of time-consuming formulas. Accurate determination of item validity on the basis of statistically rigorous techniques can be made more quickly by means of the table than validity determined by less efficient methods which have previously been used to save time.

Under controlled flight conditions, the distance between a navigator's report of position and his actual position is a criterion of success in dead reckoning navigation. Students' logs were evaluated for five separate missions by comparing the students' entries with standards determined by experts. The reliability of this technique is indicated by the fact that mission to mission intercorrelations of error scores were low, while the intercorrelations between legs of the same mission were moderately high. The intercorrelations between the error scores for the different navigation variables were computed and analyzed by using both factor analysis and multiple regression techniques. Both analyses indicated that a major portion of all dead reckoning error could be attributed to errors made in determining magnetic deviation. As a result of these analyses, recommendations were made for changing the instruction in dead reckoning and alterations in the equipment used were suggested.

A technique of analysis utilizing frequencies of differences is described and applied to a hypothetical experiment involving two methods of instruction. A nomograph is provided for computing the chi-square values applicable to the method.

The KG Index described below is proposed for evaluation as one approach to the problem of providing an index giving comparable values for items (1) of equal discriminative power at all levels of difficulty and (2) of different numbers of alternative responses.