12.07.2015 Views

Speaking to and About Patients: Predicting Therapists' Tone of Voice

Speaking to and About Patients: Predicting Therapists' Tone of Voice

Speaking to and About Patients: Predicting Therapists' Tone of Voice

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Judges <strong>and</strong> RatingsTwelve (6 male <strong>and</strong> 6 female) undergraduates at HarvardUniversity were each paid <strong>to</strong> rate all three tapes (beginning,middle, <strong>and</strong> ending segments) <strong>of</strong> therapists talking <strong>to</strong> patients,<strong>and</strong> another 12 undergraduates (6 male <strong>and</strong> 6 female)were each paid <strong>to</strong> rate all three tapes <strong>of</strong> therapists talkingabout patients. Judges were r<strong>and</strong>omly assigned <strong>to</strong> one <strong>of</strong>three counterbalanced conditions representing the orderin which they would rate the three tapes (i.e., ABC, BCA,CAB). Judges were <strong>to</strong>ld that they would hear therapiststalking <strong>and</strong> that they would be able <strong>to</strong> underst<strong>and</strong> onlythe speaker's <strong>to</strong>ne <strong>of</strong> voice. Consistent with our earlierexperience with this type <strong>of</strong> study, judges were given nospecial training for their task (Rosenthal, 1982). All judgesrated all the segments (98 beginning, 98 middle, <strong>and</strong> 98ending segments) on 10 dimensions—not warm-warm,not hostile-hostile, not anxious-anxious, not dominantdominant,not empathic-empathic, not competent-competent,not optimistic-optimistic, not pr<strong>of</strong>essionalpr<strong>of</strong>essional,not honest-honest, <strong>and</strong> not liking-liking. Eachsegment was played once, <strong>and</strong> all judges were given 15s<strong>to</strong> make the 10 ratings. Each rating was made on a scalerunning from 1 (not at all warm) <strong>to</strong> 9 (very warm). Becauseratings <strong>of</strong> the beginning, middle, <strong>and</strong> ending segmentswere substantially correlated (median r ?= .49 for the shortestclips), these ratings were combined prior <strong>to</strong> subsequentanalyses (see Blanck, Rosenthal, Vannicelli, & Lee, 1983,for a detailed discussion <strong>of</strong> the various types <strong>of</strong> phase-<strong>to</strong>phase<strong>and</strong> judge-<strong>to</strong>-judge reliabilities).Principal Components AnalysisThe mean <strong>of</strong> all judges' ratings <strong>of</strong> the <strong>to</strong>ne <strong>of</strong> voice inwhich therapists spoke <strong>to</strong> <strong>and</strong> spoke about their patientswere intercorrelated separately, <strong>and</strong> a principal componentsanalysis was computed for each <strong>of</strong> these 10X10 correlationmatrices <strong>of</strong> mean judges' ratings. For the ratings <strong>of</strong> therapists'<strong>to</strong>ne <strong>of</strong> voice in speaking <strong>to</strong> their patients, thisanalysis yielded four interpretable fac<strong>to</strong>rs after varimaxrotation: (a) pr<strong>of</strong>essional-competent (consisting <strong>of</strong> competent,optimistic, ind pr<strong>of</strong>essional), (b) warmth (consisting<strong>of</strong> warm, not hostile, not dominant, empathic, <strong>and</strong> liking),(c) anxious, <strong>and</strong> (d) honest. Each new fac<strong>to</strong>r-based variablewas denned as the mean rating <strong>of</strong> the variables includedin that fac<strong>to</strong>r with the sign <strong>of</strong> the loading taken in<strong>to</strong> account.(Because the variances <strong>of</strong> these variables were so homogeneous,st<strong>and</strong>ardizing was not employed prior <strong>to</strong> computingmeans <strong>of</strong> ratings.) For example, the pr<strong>of</strong>essionalcompetentvariable was denned as the mean rating <strong>of</strong> competence,optimism, <strong>and</strong> pr<strong>of</strong>essionalism in <strong>to</strong>ne <strong>of</strong> voice.In this way, the 10 variables <strong>of</strong> <strong>to</strong>ne <strong>of</strong> voice in speaking<strong>to</strong> patients were reduced <strong>to</strong> four fac<strong>to</strong>r-based variables forsubsequent analyses. A detailed description <strong>of</strong> the fac<strong>to</strong>rstructure <strong>of</strong> therapist's <strong>to</strong>ne <strong>of</strong> voice in talking <strong>to</strong> patientsis described in Blanck, Rosenthal, <strong>and</strong> Vanicelli (1983).For the ratings <strong>of</strong> therapists' <strong>to</strong>ne <strong>of</strong> voice in speakingabout their patients, the principal components analysisyielded seven interpretable fac<strong>to</strong>rs after varimax rotation:(a) pr<strong>of</strong>essional-competent (consisting <strong>of</strong> pr<strong>of</strong>essional <strong>and</strong>competent, (b) warmth (consisting <strong>of</strong> warm, empathic,<strong>and</strong> liking), (c) not hostile, (d) anxious, (e) dominant, (f)optimistic, <strong>and</strong> (g) honest. Each new fac<strong>to</strong>r-based variablewas denned as the mean rating <strong>of</strong> the nonverbal variablesSPEAKING TO AND ABOUT PATIENTS 681included in that fac<strong>to</strong>r with the sign <strong>of</strong> the loading takenin<strong>to</strong> account. For example, the warmth variable was dennedas the mean rating <strong>of</strong> warmth, empathy, <strong>and</strong> liking in <strong>to</strong>ne<strong>of</strong> voice. In this way, the 10 variables were reduced <strong>to</strong>seven interpretable fac<strong>to</strong>r-based variables for subsequentanalyses. The fac<strong>to</strong>r structure <strong>of</strong> the therapist's <strong>to</strong>ne <strong>of</strong>voice in talking about patients is described in detail inBlanck, Rosenthal, <strong>and</strong> Vannicelli (1983).Reliability <strong>of</strong> Judges' RatingsThe reliability <strong>of</strong> the judges' ratings <strong>of</strong> the <strong>to</strong>ne <strong>of</strong> voicevariables was computed by means <strong>of</strong> intraclass correlations(Rosenthal, 1982). The effective reliability <strong>of</strong> the mean <strong>of</strong>the 12 judges' ratings for the 30-s clips (three clips <strong>of</strong> 10s each) <strong>of</strong> therapists speaking <strong>to</strong> patients ranged from .22<strong>to</strong> .82, with a median r <strong>of</strong> .59. The effective reliability <strong>of</strong>the mean <strong>of</strong> the 12 judges' ratings for the 60-s clips (threeclips <strong>of</strong> 20-s each) <strong>of</strong> therapists speaking about patientsranged from .47 <strong>to</strong> .83, with a median r <strong>of</strong> .64.In evaluating these median reliabilities <strong>of</strong> .59 <strong>and</strong> .64,respectively, one should keep in mind that they are basedon ratings <strong>of</strong> content-filtered speech <strong>of</strong> only 30-s <strong>and</strong> 60-s duration, respectively. For our present purposes the obtainedreliabilities are more than adequate; however, hadwe increased our segment lengths <strong>to</strong> 10 min, we couldhave increased our median reliabilities <strong>to</strong> .97 <strong>and</strong> .95,respectively (Rosenthal, 1982).Results <strong>and</strong> DiscussionWe employed three types <strong>of</strong> analyses <strong>to</strong> addressthe question <strong>of</strong> whether talking aboutpatients was predictive <strong>of</strong> talking <strong>to</strong> patients:(a) simple correlations, (b) multiple regressions,<strong>and</strong> (c) canonical correlation. We describeeach analysis in turn.Simple CorrelationsFor our first analysis we correlated each <strong>of</strong>the 10 variables that had been rated by judgeswhile therapists talked <strong>to</strong> patients with each<strong>of</strong> the corresponding 10 variables that had beenrated by judges while therapists talked aboutpatients. The first column <strong>of</strong> correlations <strong>of</strong>Table 1 shows the results <strong>of</strong> this analysis. All<strong>of</strong> the correlations were positive, most weresignificant, <strong>and</strong> half were significant at p

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!