In practice it may be found advisable in some cases to analyze the unclassified reactions with a view to ascertaining to what extent each of the various types is represented among them. But one here treads on slippery ground, and one must be continually warned against the danger of erroneous conclusions.

Sec. 5. ORDER OF PREFERENCE.

After having developed the classification here proposed we found that there was still considerable room for difference of opinion in the placing of many reactions, owing to the circumstance that in many cases a reaction presents features which render it assignable under any one of two or more headings. To leave the matter of preference in grouping: to be decided in each case according to the best judgment of the experimenter would mean introducing again the play of personal equation, and would thus court failure of all our efforts to accomplish a standardization of the association test. Therefore, the necessity of establishing a proper order of preference for guidance in the application of the classification became to us quite apparent.

In the arrangement of the order of preference we were guided mainly by two principles, namely: (i) as between two groups of unequal definition, the one which is more clearly defined and which, therefore, leaves less play for personal equation is to be preferred; (2) as between two groups of equal definition, the one which possesses the greater pathological significance is to be preferred. In accordance with these principles we have adopted the order of preference shown in Table III., placing every reaction under the highest heading on the list under which it may be properly classed.

TABLE III

1. Non-specific (common). 2. Doubtful reactions. INDIVIDUAL REACTIONS. 3. Sound reactions (neologisms). 4. Neologisms without sound relation. 5. Repetition of preceding reaction. 6. Reaction repeated five times. 7. Repetition of preceding stimulus. 8. Derivatives. 9. Non-specific reactions. 10. Sound reactions (words). 11. Word complements. 12. Particles of speech. 13. Association to preceding stimulus. 14. Association to preceding reaction (by frequency tables). 15. Repetition of previous reaction. 16. Repetition of previous stimulus. 17. Normal (by appendix). 18. Association to preceding reaction (without frequency tables). 19. Unclassified.

Sec. 6. ERRORS INVOLVED IN THE USE OF ARBITRARY OBJECTIVE STANDARDS.

It may readily be seen that such definiteness and uniformity as this classification possesses results from the introduction of more or less arbitrary criteria for the differentiation of the various types of reactions. The question might arise, To what extent do the distinctions thus made correspond to reality? To consider, for instance, our rule for the placing of sound reactions (50 per cent of the sounds of the shorter word to be present, in the same order, in the other word): when a given reaction (_man--minstrel_) is in accordance with the rule assigned under the heading of sound reactions, can it be assumed that sound similarity and not some other relationship is the determining factor of the association in question? Or when in, a given instance (_cabbage--cobweb_) the sound similarity falls somewhat short of the standard required by the rule, can it be assumed that sound similarity is not, after all, the determining factor?

Similar questions may, of course, arise in connection with other subdivisions.