Title : Assessing elderly exposure to violent crimes in India: Evidence from sage waves 1 & 2 using Bayesian modelling
Abstract:
Background: India’s ageing population is rapidly increasing, which is projected to exceed 1S% by 2050, accompanied by social and health vulnerabilities that expose older adults to abuse, neglect, and victimization. Understanding these trends is crucial for informing effective social protection and elderly care mechanisms in a transitioning demographic context.
Objectives: This study examines the changes in reporting elderly victimisation among the Indian elderly population using WHO’s Study on global AGEing and adult health (SAGE) Waves 1 (2007–10) and 2 (2014–15) and identifies key sociodemographic predictors and regional disparities inffuencing victimisation risk by employing Bayesian modelling.
Methodology: A comparative analysis was conducted using data from elderly respondents (aged c0 years and above) across both SAGE waves- 3,S71 (Wave 1) and 4,214 (Wave 2). Predictors included age, gender, marital status, education, residence, wealth, caste, and religion. Statistical analysis was done using RStudio. The results include descriptive statistics, state-level slope graphs, and a thematic India map highlighting interstate variations in victimisation. Bayesian logistic regression models with weakly informative priors were estimated using MCMC (four chains; R-hat = 1.00), and S5% credible intervals were reported.
Results: The self-reported Victimisation rate rose from 2.7c% in Wave 1 to c.3% in Wave 2. The Bayesian posterior means for victimization probability (Table 1) revealed higher risks in Uttar Pradesh, Assam, and West Bengal, whereas Maharashtra and Rajasthan consistently had the lowest estimates. The posterior mean estimates illustrate that victimization likelihood approximately doubled between waves in most states, with the most substantial increase observed in Assam (3.44 % to 12.5%) and Uttar Pradesh (4.2% to 8.S%). Bayesian analysis showed caste significant in Wave 1 (Scheduled Caste: 1.35 [0.11,2.75] ), narrowing by Wave 2 (0.42 [-0.14,1.0c]). Model diagnostics confirmed convergence (R-hat=1.00).
Conclusion: The prevalence of reported elderly victimization nearly doubled between Waves 1 and 2. Socioeconomic deprivation, lack of formal education, rural residence, and widowhood emerged as consistent risk factors. Although gender differences were modest, women and socially marginalised groups exhibited greater vulnerability. Regional variations appeared to reffect both genuine increases in risk and improved awareness or reporting. Despite existing legal safeguards, such as the Maintenance and Welfare of Parents and Senior Citizens Act (MWPSC Act, 2007), awareness and enforcement remain limited, particularly in rural and low-literacy populations.
