Validity of the Connor Davidson Resilience Scale (CD-RISC) in adolescents and young adults in Sweden: a think aloud approach combined with a Rasch analysis.

backgroundResilience is defined as the ability to adapt to adversity, most widely assessed with the self-reported questionnaire Connor-Davidson resilience scale (CD-RISC). While previous studies have demonstrated CD-RISC psychometrically sound, it has not yet been validated in a young Swedish population. Therefore, the aim was to evaluate validity evidence based on test content, response processes and internal structure of the Swedish CD-RISC-25 and the 10-item combination among adolescents and young adults.

methodsThis study is divided into two phases. To ensure validity based on test content eight think-aloud interviews were conducted in phase 1. The results guided refinements prior to phase 2. In phase 2, 1500 16-30-year-old individuals, randomly sampled from the general population to participate by completing an online questionnaire. A Rasch rating scale model analysis was performed.

resultsThink-aloud interviews identified difficult wordings in two items, clarified in the survey before phase 2. The Rasch analysis (n = 325) indicated that the response categories were well-functioning, and local independence was demonstrated among all items. Six items in CD-RISC-25 and one item in CD-RISC (10 items) displayed misfit to the chosen model. The iterative process resulted in shortened 19- and 9-item versions. Person-response validity and unidimensionality were acceptable for the shortened CD-RISC (9 items). No floor or ceiling effects were detected. Person-separation index showed that both versions could differentiate between three different levels of resilience in the sample. Differential item functioning (DIF) was observed in one item related to gender in CD-RISC (9 items). In shortened CD-RISC-25 DIF was found in five items related to gender and in two items related to age.

conclusionsWhen used in a young Swedish population, validity evidence based on test content of CD-RISC is lacking due to difficult wordings and may benefit from additional clarification regarding two items. Shortened versions of CD-RISC (19- and 9-item combinations) demonstrated generally acceptable validity evidence based on response processes and internal structure, but only CD-RISC in the 10-item combination exhibited unidimensionality and met set criteria for person-response validity. For assessing resilience in a young Swedish population, CD-RISC in the 10-item combination appears to be more suitable.