Rasch Analysis on Video-based Assessment for Detecting Errors in Performing Activities of Daily Living for Older People with Dementia

Objective : This study aims to verify the construct validity of the VADCE-ADL tool through Rasch analysis. Methods : The VADCE-ADL comprised 40 videos of errors frequently seen during ADL in older people with dementia. One hundred and fifteen older Korean adults watched the videos and verbally expressed what the errors were. The understanding level was scored on a 4-point scale. Descriptive statistics, frequency analysis, and reliability analysis were conducted using SPSS 25.0, and participant and item suitability and item difficulty were analyzed using Winstep 5.7.4. Results : The internal consistency of VADCE-ADL was 0.916. Participant suitability analysis found no un-suitable individual. Item suitability analysis found three inappropriate items including ‘no significance for safety,’ ‘difficulty in distinguishing bath products,’ and ‘difficulty in controlling the amount.’ As a result of analyzing the difficulty level of the items excluding the three inappropriate items, the items with high difficulty were ‘errors in managing personal items,’ ‘not flushing the toilet after us-ing the toilet,’ ‘lack of awareness of other people's emotions,’ ‘difficulty finding space in the house,’ and ‘excessive respect for others.’ Conclusion : VADCE-ADL presents a promising approach to assessing error detection abilities in older adults dur-ing ADL, offering a time- and energy-efficient alternative to traditional performance-based tools. However, to maximize its effectiveness, improvements in animation design, scenario realism, and item calibration are essential