JMIR Publications

ABSTRACT

Background:

Companion robots are increasingly introduced in healthcare and daily environments, where users’ emotional and relational expectations play an important role in acceptance. Existing instruments, however, insufficiently capture these attitudinal dimensions. Because companion robots operate through digital interfaces, users’ digital health literacy may also influence how they evaluate such technologies.

Objective:

This study aimed to develop and validate the Companion Robot Attitude Scale (CRAS) and to examine whether digital health literacy contributes to individual differences in attitudes toward companion robots.

Methods:

A two phase web-based survey was conducted with 1,000 adults. Exploratory factor analysis (n = 600) refined an initial item pool, and confirmatory factor analysis (n = 400) evaluated structural validity, reliability, and convergent and discriminant validity. Hierarchical regression analysis was performed to determine whether digital health literacy predicted CRAS scores after adjusting for sex and age.

Results:

The final CRAS Scale consisted of 25 items across five sub-dimensions: interaction (six items), enjoyment (six items), ease of use (four items), usefulness (four items), and anxiety (five items). The model fit indices confirmed structural validity (comparative fit index=0.927, Tucker-Lewis index=0.920, root-mean-square error of approximation=0.047, and standardized root-mean-square residual=0.054). Convergent validity was confirmed with average variance extracted values above 0.5, and reliability was established with composite reliability values exceeding 0.7. Internal consistency was strong, with Cronbach’s α ranging from 0.82 to 0.94 across subscales. Discriminant validity was also confirmed, as the square root of the average variance extracted values exceeded the factor correlations. In regression analysis, digital health literacy was a significant positive predictor of CRAS scores (β = 0.311, p < .001), whereas sex and age were not significant predictors.

Conclusions:

The CRAS is a psychometrically robust instrument that captures emotional, relational, and functional evaluations of companion robots. The findings demonstrate strong structural validity and internal consistency, supporting its use in research and practical assessment. The predictive role of digital health literacy suggests that individual digital competencies contribute to how users form attitudes toward companion robots. The CRAS offers a comprehensive framework to guide future investigations and user-centered development of companion robot technologies.