The evaluation of the matching method

A critical comparison of the methods available for determining the statistical significance of results obtained by the method of correct matchings. The comparison is based in part on an analysis of the assumptions involved in the use of the various methods, and in part on the results of statistical experiments that simulate the matching procedure. Chapman's table of probabilities ( Amer. J. Psychol. , 1934, ^W46^n, 287-298) should be used when (1) the average number of correct matchings is small, (2) the number of judges is small, and (3) the investigator merely wishes to determine whether a result is significantly superior to chance probability. The calculation of the |c's of the correct matchings is frequently the simplest and most legitimate method. However, the contingency coefficient, C, and its p.e. are the most satisfactory for use in many complex matching problems. The author discusses the application of C to studies in which there is matching of unequal numbers of elements and to studies in which second or further choices in matching are permitted.