Get access

QUANTIFYING CONTINGENT RELATIONS FROM DIRECT OBSERVATION DATA: TRANSITIONAL PROBABILITY COMPARISONS VERSUS YULE'S Q

Authors


  • This research was part of the first author's thesis submitted in partial completion of the MS degree at Vanderbilt University. The research was supported, in part, by U.S. Department of Education Grant H325D050102 (Kennedy). We thank John Borrero and a panel of anonymous reviewers for their helpful feedback on this manuscript.

Correspondence should be addressed to Blair Lloyd and Paul Yoder, Department of Special Education, Peabody College Box 221, Vanderbilt University, Nashville, Tennessee 37203 (e-mail: blair.lloyd@vanderbilt.edu or paul.yoder@vanderbilt.edu).

Abstract

Measuring contingencies or sequential associations may be applied to a broad range of response–stimulus, stimulus–stimulus, or response–response relations. Within behavior analysis, response–stimulus contingencies have been quantified by comparing 2 transitional probabilities and plotting them in contingency space. Within and outside behavior analysis, Yule's Q has become a recommended statistic used to quantify sequential associations between 2 events. In the current paper, we identify 2 methods of transitional probability comparisons used in the behavior-analytic literature to estimate contingencies in natural settings. We compare each of these methods to the more established Yule's Q statistic and evaluate relations between each pair of indices. Advantages and disadvantages of each method are identified, with recommendations as to which approach may be most appropriate for measuring contingencies.

Ancillary