Note on computation of bi-serial correlations in item evaluation

Correcting Turnover Correlations: A Critique

“…These faster methods were still incredibly slow by today's standards. Even with his improved procedures, Dunlap (1936) estimated that it still took 250 working hours to compute 3,480 correlations! 5.…”

“…C. Peters & Van Voorhis, 1940). For example, Dunlap (1936) published a 13-step, timesaving process in which test scores were "serialized," meaning they were sorted into intervals based on criterion scores. 4 The means or midpoints for each test item were then "used to represent the cases in each interval" (1.…”

Three Aids in the Evaluation of the Significance of the Difference Between Percentages

“…The results obtained shall be accurate to a degree commensurate with published results. 4. The results obtained shall be subject to a &dquo;standardized&dquo; interpretation.…”

An experimental examination of the Thematic Apperception Technique in clinical diagnosis.

“…Before resuming the series, paragraph (3) (b) above will be repeated. (6) (7) Inquiry: Upon completion of the last picture, S. will be questioned briefly about any obscure or incomplete references in his stories. Inquiry will be limited to non-suggestive questions necessary to clarify specific items.…”

“…Two forms of correlation were employed, the Pearson product-moment correlation for the comparisons of the blind analysis interpretations, and the bi-serial correlation method for the major bulk of the comparisons between TAT response categories and the personality variables of the rating scale. The formula for the bi-serial correlations is taken from Dunlap (6); the standard error for bi-serial r is given by Peters and Van Voorhis (20,.…”