![[ Visit Client Website ]](images/banner.gif)
Schedule:
Sunday, January 24, 2027
Continental 1-4 (Hilton San Francisco Union Square)
* noted as presenting author
Background and Purpose: Intimate Partner Violence (IPV) is a significant public health issue. Despite a rapidly aging population, IPV research has predominantly focused on women of reproductive age, with limited attention to older adults. Additionally, few studies have applied advanced analytic approaches to examine how IPV manifests across the life course or to identify characteristics associated with older adult victimization. This study addresses these gaps by integrating multivariate and machine learning methods to examine age-related patterns in IPV and identify predictors of victimization among older adults aged 50 and older.
Methods: Data were drawn from the Missouri Incident-Based Reporting System (MIBRS), 2021–2023. The analytic sample included 71,839 IPV-related incidents with known victim and offender characteristics. Victim age group (<18, 18–24, 25–34, 35–49, 50+) served as the primary outcome. Predictors included offender age, gender, and race/ethnicity; victim gender and race/ethnicity; victim–offender relationship; offense type; incident clearance; and location. Descriptive statistics were stratified by age group, and bivariate associations were assessed using chi-square tests. Multinomial logistic regression models estimated odds ratios between predictors and age groups. Classification trees and random forests were used to assess predictive performance and identify the most influential variables. A 10‑fold cross‑validated random forest was used to evaluate model performance and tune mtry, while estimating accuracy, κ, and variable importance.
Results: IPV characteristics varied across the life course. Older adults were more likely to be involved in spousal relationships (39.2%), have older offenders, and experience IPV by female offenders at higher rates (35.3%). They also showed demographic differences, including the highest proportion of male victims (36.1%) and White non‑Hispanic victims (66.6%). Incidents most often occurred in residences, increasing steadily with age and peaking among adults 50+ (86.5%), and simple assault was the most common offense. Regression findings aligned with these patterns: older victims had increased odds of spouse‑perpetrated IPV (OR = 2.84, 95% CI = 2.69–3.00) and female offenders (OR = 1.92, 95% CI = 1.83–2.02). The first tree, dominated by Offender Age, separated age groups, whereas the second tree showed older adults dispersed across branches defined by relationship, race/ethnicity, and location, indicating heterogeneous experiences. Random forest performance was modest (accuracy = 43.1%, κ = 0.137), with extremely low sensitivity for adults 50+ (2.7%) despite high specificity (99%), demonstrating frequent misclassification. Cross verification showed modest performance (accuracy = 39.7%, κ = 0.08), an optimal tuning parameter of mtry = 2, with similar results in regard to variable importance.
Conclusions and Implications: Findings suggest that older adult IPV is not easily captured by standard administrative incident data, making victimization in later life less distinct and potentially under‑recognized. The difficulty in identifying older adults through typical incident characteristics indicates that current screening and assessment practices may overlook or misclassify IPV in this population. Results highlight the need for improved detection strategies—particularly for non‑physical abuse and cases involving medically complex older adults—and underscore the importance of developmentally informed, age‑responsive approaches to IPV identification and intervention within social work practice and law enforcement.
Methods: Data were drawn from the Missouri Incident-Based Reporting System (MIBRS), 2021–2023. The analytic sample included 71,839 IPV-related incidents with known victim and offender characteristics. Victim age group (<18, 18–24, 25–34, 35–49, 50+) served as the primary outcome. Predictors included offender age, gender, and race/ethnicity; victim gender and race/ethnicity; victim–offender relationship; offense type; incident clearance; and location. Descriptive statistics were stratified by age group, and bivariate associations were assessed using chi-square tests. Multinomial logistic regression models estimated odds ratios between predictors and age groups. Classification trees and random forests were used to assess predictive performance and identify the most influential variables. A 10‑fold cross‑validated random forest was used to evaluate model performance and tune mtry, while estimating accuracy, κ, and variable importance.
Results: IPV characteristics varied across the life course. Older adults were more likely to be involved in spousal relationships (39.2%), have older offenders, and experience IPV by female offenders at higher rates (35.3%). They also showed demographic differences, including the highest proportion of male victims (36.1%) and White non‑Hispanic victims (66.6%). Incidents most often occurred in residences, increasing steadily with age and peaking among adults 50+ (86.5%), and simple assault was the most common offense. Regression findings aligned with these patterns: older victims had increased odds of spouse‑perpetrated IPV (OR = 2.84, 95% CI = 2.69–3.00) and female offenders (OR = 1.92, 95% CI = 1.83–2.02). The first tree, dominated by Offender Age, separated age groups, whereas the second tree showed older adults dispersed across branches defined by relationship, race/ethnicity, and location, indicating heterogeneous experiences. Random forest performance was modest (accuracy = 43.1%, κ = 0.137), with extremely low sensitivity for adults 50+ (2.7%) despite high specificity (99%), demonstrating frequent misclassification. Cross verification showed modest performance (accuracy = 39.7%, κ = 0.08), an optimal tuning parameter of mtry = 2, with similar results in regard to variable importance.
Conclusions and Implications: Findings suggest that older adult IPV is not easily captured by standard administrative incident data, making victimization in later life less distinct and potentially under‑recognized. The difficulty in identifying older adults through typical incident characteristics indicates that current screening and assessment practices may overlook or misclassify IPV in this population. Results highlight the need for improved detection strategies—particularly for non‑physical abuse and cases involving medically complex older adults—and underscore the importance of developmentally informed, age‑responsive approaches to IPV identification and intervention within social work practice and law enforcement.