Response productivity in verbal content analysis: A critique of Marsden, Kalter, and Ericson.

Abstract

While agreeing with the contention of Marsdcn, Kalter, and Ericson that use of percentage scores to control for productivity in verbal content analysis can yield erroneous conclusions, exception is taken to their sweeping generalizations in this regard. We particularly refute their inclusion of the Gollschalk-GIcscr scales as typical of the unreflecting use of percentage scores. Furthermore, limitations and contradictions are pointed out in the methods they recommend to correct for response productivity.

Further comments on response productivity.

“…Another general criticism raised was our treatment of the residual score method of attempting to control for productivity. Gleser and Lubin (1976) correctly recapitulate our argument that any residual score only accomplishes part of what a partial correlation does. They go on to claim that "whether the linear effect of T should be removed from Z depends on what question is being asked" (p. 510).…”

“…While agreeing with us that the use of X/T, the division of the frequency of a content analysis category by total number of units, can result in erroneous conclusions and is therefore an inadequate control for response productivity, Gleser and Lubin (1976) raise a scries of criticisms about our previous article (Marsden, Kalter, Ericson, 1974). These concerns fall into basically three domains: general issues, criticisms of the methods we suggest for controlling for productivity, and a claim that the Gottschalk and Gleser (1969) work was inappropriately cited in our article.…”

“…1. Gleser and Lubin (1976) criticize us at several points for not emphasizing the importance of examining the relationship between productivity, as a variable in its own right, and other variables. We are puzzled by this criticism on two counts.…”

Content scaling of affect in adolescent speech samples

“…The correlations between number of words spoken and the affect scores are all nonsignificantly different from zero, indicating that the correction for productivity applied in our scoring formula is successful in removing productivity as a factor (Gleser and Lubin, 1976). Anxiety is positively correlated with all other affect scores, while Overt Hostility Outward, Covert Hostility Outward and Ambivalent Hostility form one correlated cluster, and Hostility Directed Inward, and Social Alienation and Personal Disorganization form a second cluster.…”

Content analysis of verbal behaviour in psychotherapy research: A comparison between two methods

“…An additional step involves extraction of square root for normalizing the distribution. Gottschalk-Gleser affect scales do not form a set of exhaustive categories covering all thema, so that cases of no codable content are possible Gleser & Lubin, 1979). Although the same holds true for ATD, the probabilities for no scorable content are lower (single word method).…”