3. Mean trust over all 18 items was 4.47 (SD = .50, range 2.50–5.00). The theoretically driven 4-factor model failed to converge in CFA. This was probably due to collinearity, as indicated by between-item correlations as strong as .8. Moreover, even when collapsing response categories 1 and 2, the distribution of trust scores over response categories remained uneven, including empty or near-empty cells. A one-dimensional model resulted in an acceptable model fit (SBχ2(137)=200.73SBχ2(137)=200.73, p < .01, and RMSEA = .05) [23]. Standardized item loadings on this factor were strong (mean: .80, range: .58–.91) [23]. Post hoc
exploratory factor analysis, to check if a one-dimensional this website model fit would be confirmed when no assumptions were made about the data, further established the one-dimensionality of the TiOS. Very strong internal consistency of the TiOS was suggested by Cronbach’s α of .94 [24]. Item-scale correlations were acceptable (range .43–.81) [25]. Inter-item correlations ranged between .2 and .8. As expected, mean selleck compound scores on the TiOS correlated significantly with known correlates of trust, i.e.,
satisfaction with the oncologist (PSQ: rs = .62), willingness to recommend the oncologist to others (rs = .59), number of previous visits with the oncologist (rs = .21) and trust in health care (rs = .33). All correlations in the exploratory analyses were non-significant. In this study, the English version of the 18-item Trust in Oncologist Scale (TiOS) was validated. Mean trust scores were invariably high. Strong internal consistency, inter-item correlations
and item-scale correlations suggest sufficient reliability. Construct validity was confirmed by strong correlations of TiOS scores with satisfaction and moderate correlations with number of previous visits with the oncologist and with trust in health care. Importantly, we found TiOS scores to be one-dimensional, Etomidate indicating that these patients do not distinguish between different aspects of trust, i.e., competence, fidelity, honesty, and caring. Although this distinction was slightly stronger among Dutch patients, we still concluded that trust was best considered as a one-dimensional construct. The present findings confirm this suggestion of one-dimensionality. The even weaker distinction between dimensions of trust by Australian patients could reflect a more homogeneous composition of this sample. Even though mean trust was equally high in both samples, the Australian data lack sufficient variation in trust scores. Very few patients reported weak trust in their oncologist. This lack of variation may be due to Medical Ethical Committee regulations, prohibiting the random and direct approach of patients by mail as employed in the Dutch sample. Recruitment via the participating oncologists may have resulted in selection bias towards including only strongly trusting patients.
Related posts: