The data were entered into SPSS for statistical analysis. Descriptive statistics were used to describe the sample demographics as well as situations and actions the sample perceived to be abusive and to determine which barriers were most important in considering reporting PA. Independent t-tests were used to examine differences between the groups (senior center and meals-on-wheels). Logistic regression analysis was used to evaluate change in reporting after introduction of barriers. For the logistic regression analysis the data from OAPPAS were transformed to dichotomous responses (definitely/probably abuse = yes abuse, definitely not/probably not abuse = not abuse; definitely/probably would report = yes I'd report, definitely not/probably not report = no reporting). The dichotomous responses were used in Fig. 1 and Fig. 2.