Purpose Psychological elder abuse (PEA) assessment is described with different thresholds. The purpose of this paper is to examine how the prevalence of PEA and the phenomenon’s characterisation varied using two different thresholds. Design/methodology/approach Participants from the cross-sectional population-based study, Aging and Violence (n=1,123), answered three questions regarding PEA. The less strict measure considered PEA as a positive response to any of the three evaluated behaviours. The stricter measure comprised the occurrence, for more than ten times, of one or more behaviours. A multinomial regression compared cases from the two measures with non-victims. Findings Results show different prevalence rates and identified perpetrators. The two most prevalent behaviours (ignoring/refusing to speak and verbal aggression) occurred more frequently (>10 times). Prevalence nearly tripled for “threatening” from the stricter measure (>10 times) to the less strict (one to ten times). More similarities, rather than differences, were found between cases of the two measures. The cohabiting variable differentiated the PEA cases from the two measures; victims reporting abuse >10 times were more likely to be living with a spouse or with a spouse and children. Research limitations/implications Development of a valid and reliable measure for PEA that includes different ranges is needed. Originality/value The study exemplifies how operational definitions can impact empirical evidence and the need for researchers to analyse the effect of the definitional criteria on their outcomes, since dichotomization between victim and non-victim affects the phenomenon characterisation.