Cardace Fabio, Rubel Julian, Altmann Uwe, Merkler Martin, Schwartz Brian, Deisenhofer Anne-Katharina, Paulick Jane, Schoenherr Desiree, Strauß Bernhard, Lutz Wolfgang
Klinische Psychologie und Psychotherapie, Universität Trier,Trier, Deutschland.
Abteilung für Psychotherapieforschung, Justus Liebig Universitat Giessen, Giessen, Deutschland.
Psychother Psychosom Med Psychol. 2022 Feb;72(2):59-67. doi: 10.1055/a-1519-7259. Epub 2021 Sep 13.
The Liebowitz Social Anxiety Scale (LSAS) and the Social Phobia Inventory (SPIN) are established measures in the investigation of social anxiety. Furthermore, the subscale Interpersonal Sensitivity of the Brief Symptom Inventory (BSI-53) is frequently used to screen social anxiety. All three scales claim to capture the same construct, which raises the question of the convergence of these scales. To make research findings comparable by a cross-questionnaire factor (common factor), an item response theory (IRT) linking approach is used in the present study.
64 German-speaking psychiatric patients and 295 healthy subjects completed the three questionnaires. Different IRT models, including Graded Response Models (GRM), were constructed, and their model fit compared. Regression analyses were performed based on the best-fit model. The common factor was predicted from the questionnaire total values.
The relationship between the different scales was best explained by a bifactor GRM with one common factor and three domain-specific factors (RMSEA=0.036, CFI=0.977, WRMR=1.061). Based on the results of the regression analyses, three equations were derived for the transformation of questionnaire's total values.
The IRT linking approach allows the derivation of a general factor of social anxiety, taking into account commonalities and differences between the instruments used. This has advantages for both research and practice. A replication of this study as well as the implementation of further instruments are recommended to verify the validity of this approach and to generalize the results.
利博维茨社交焦虑量表（LSAS）和社交恐惧症量表（SPIN）是社交焦虑调查中常用的测量工具。此外，简明症状量表（BSI - 53）的人际敏感分量表也经常用于筛查社交焦虑。这三个量表都声称能捕捉相同的结构，这就引发了这些量表的收敛性问题。为了通过跨问卷因素（共同因素）使研究结果具有可比性，本研究采用了项目反应理论（IRT）链接法。
64名说德语的精神科患者和295名健康受试者完成了这三份问卷。构建了包括等级反应模型（GRM）在内的不同IRT模型，并比较了它们的模型拟合度。基于最佳拟合模型进行回归分析。从问卷总分预测共同因素。
不同量表之间的关系最好由一个具有一个共同因素和三个特定领域因素的双因素GRM来解释（RMSEA = 0.036，CFI = 0.977，WRMR = 1.061）。根据回归分析结果，推导出了三个用于问卷总分转换的方程。
IRT链接法考虑到所用工具之间的共性和差异，能够推导出社交焦虑的一般因素。这对研究和实践都有好处。建议重复本研究并使用更多工具来验证该方法的有效性并推广结果。
