Adaptation of Validity and Reliability of Indonesian Instruments of Attitudes towards Suicide (ATTS) with The Rasch Model Approach

Abstract

Background: Indonesia has a suicide rate that continues to increase so suicide attempts need to be made. One of the early prevention steps is to find suicidal ideation. ATTS is a multi-dimensional measuring tool by measuring Cognitive, Affective and Behavioral aspects that is easy to fill, fast and precise in measuring attitudes towards suicide in the general population through large surveys than other instruments and ATTS is also not yet developed in Indonesia. Therefore, with the Rasch Model that can perform analysis at the item and respondent level, it is expected to produce a psychometric tool in preventing suicide in Indonesia.Purpose: The purpose of this study is to test the validity and reliability of the adaptation of the Attitudes Toward Suicide (ATTS) instrument into Indonesian using the RASCH model approach. Method: The method used is non-experimental with an analytical observational approach which is carried out cross-sectionally using the RASCH model with winstep software. The number of samples used was 243 residents with inclusion criteria of being adults (20-40 years) and knowing about suicide cases in Plered District, one of the sub-districts in Purwakarta Regency which had several suicide cases in 2020. This instrument consists of 73 questions.Result: The results showed a value of scale-level (S-CVI) = 0.89 and content validity ratio (S-CVR/UA) = 0.72. This value indicates a good quality of content validity. Analysis based on RASCH shows very good item reliability with a value of 0.98 but the result of person reliability is 0.27 which is caused by the lack of variation in respondents' answers in filling out the questionnaire. This also has an impact on Cronbach's alpha value = 0.32. The analysis is elaborated into four factors consisting of summary statistics, unidimensionality, item size, and Differential Item Functioning (DIF). The results show ideal results, but the pattern of person responses that are not diverse from the respondents causes a logit distribution that is not too far away and the variance is not large.Conclusion: it can be concluded that the consistency of respondents' answers is weak but the quality of the instrument questions in terms of reliability is good and can still be used to measure ATTS.