Forsberg Lars, Källmén Håkan, Hermansson Ulric, Berman Anne H, Helgason Asgeir R
Department of Clinical Neuroscience, Division of Addiction Research, Karolinska Institutet, Karolinska University Hospital, Stockholm, Sweden.
Cogn Behav Ther. 2007;36(3):162-9. doi: 10.1080/16506070701339887.
The aim of this study was to evaluate inter-rater reliability when using the Swedish version of the Motivational Interviewing Treatment Code (MITI) as an adjunct to MI training, clinical practice and research. Coders were trained to use the MITI for scoring taped sessions. The 4-month basic training had a duration of 39 hours. Following training, 60 audio-taped live interviews were randomly assigned for MITI coding. Mean intra-class correlation (ICC) coefficients were calculated for 7 coders across all pairs of coders. Cronbach's alpha was calculated to estimate the covariance between each pair across their common interviews. Six months later, a second inter-rater reliability test was performed, when 5 coders coded the same 15 randomly selected tapes. At the second reliability testing the mean ICC was 0.81 and the mean Cronbach's alpha was 0.96. However, the ICC varied for different sub-variables of the MITI, ranging from 0.42 empathy to 0.79 for number of Closed questions. In conclusion, MITI shows promising potential to be a reliable tool to confirm and enhance MI training as well as practice in clinical settings and in evaluating MI integrity in clinical MI research. However, coder assessment of empathy and MI-spirit, "global" variables, requires further refinement.
本研究的目的是评估在将瑞典版动机性访谈治疗编码（MITI）用作动机性访谈（MI）培训、临床实践及研究的辅助工具时，评分者间的信度。编码员接受了使用MITI对录音访谈进行评分的培训。为期4个月的基础培训时长为39小时。培训结束后，随机分配60段现场访谈录音用于MITI编码。计算了7名编码员在所有编码员对之间的平均组内相关系数（ICC）。计算了Cronbach's α系数以估计各对编码员在共同访谈中的协方差。6个月后，进行了第二次评分者间信度测试，由5名编码员对随机选取的相同15段录音进行编码。在第二次信度测试中，平均ICC为0.81，平均Cronbach's α为0.96。然而，MITI不同子变量的ICC有所不同，从共情方面的0.42到封闭式问题数量方面的0.79不等。总之，MITI显示出有望成为一种可靠工具，用于在临床环境中确认和加强MI培训及实践，并在临床MI研究中评估MI的完整性。然而，编码员对共情和MI精神（“整体”变量）的评估需要进一步完善。
