Appendix A: Comparing behavioral frequencies
To see if types of play differed in their presence across positive- (n = 108) and neutral-affect (n = 69) interactions, the authors focused on the three most frequent play routines they observed—fetch, tease, and tug—which comprise more than 40% of the play routines observed over 178 videotapes with evident human affect (pp. 781-783). (Only one of these showed negative affect.) In the 2 they use to answer this question, they provide an N of 171, indicating that these frequencies are not independent: each dog-human pair could be noted as playing all three routines, or two of them. In addition, theyincluded frequencies of only the presence of the routines in each interaction; however, because the frequencies of positive- and neutral-affect interactions were not identical, the frequencies of interactions in which the three routines did not occur needed to be included for both affect contexts. Thus, to evaluate their data, either the authors must employ a different statistic, or they should examine the relations between affect and the types of play separately. When the latter is done (using data from pp. 782-783), using a test of independence 2 comparing the presence or absence of the play activities in the two affective contexts, teasewas more frequent (2 (df = 1, N = 177) = 9.18, p < .003), and fetch less frequent (2 (df = 1, N = 177) = 14.29, p < .001), in positive- than in neutral-affect contexts, as the authors claim. But contrary to their claim that tug was more frequently in positive- than neutral-affect contexts, tug was equally prevalent in both contexts (2 (df = 1, N = 177) = 0.54, ns).
A further problem in comparing frequencies of fetch, tug, and tease is that tease is a quite different category than are fetch and tug. Although they are not specified, the behaviors enacted in tug and fetch seem likely to be quite specific: for fetch, the interaction begins when the dog gets an object (or when the owner tosses it?), and ends when the dog brings it to the person; for tug, the interaction begins when both owner and dog get hold of an object, and ends when at least one lets go. Unlike tug and fetch, tease (which includes feigning toss, hide and seek, spider fingers, etc.) incorporates multiple and diverse forms of behavior. Thus, by comparison with the other play behaviors and interactions described, teasing seems a supercategory—and thus not at the same level of analysis as fetch and tug. Comparing the frequencies of fetch and tug to those of tease is like comparing the frequencies of hitting (a single action) and aggression (multiple possible actions).
As support for combining the diverse play activities into a category of teasing play comparable to fetch and tug, the authors point to the fact that none of the play activities included as teasing seemed to be a main focus in the interactions they observed: they were “not ritualized, nor did any stand by itself as the entire play episode” (p. 782). The brevity of their videotapes may account for this. In our observations and descriptions of more extended dog-human play interactions, two behaviors that fit their definition of teasing play, fakeout and object-keepaway, were frequently re-enacted by players during an episode, and indeed were the focus of play interactions for some pairs (see Mitchell and Thompson 1991, pp. 198-199).