Quality of life for individuals and their communities is greatly affected by the degree of altruism expressed when misfortune occurs. The present study investigated the construct validity of an individual differences variable (i.e., unsupportive attributional style) linked to helping behaviour. Unsupportive attributional style (i.e., the tendency to view others' misfortunes as controllable by the victims) is assessed across a number of negative life outcomes of others using the Reasons for Misfortune Questionnaire (RMQ). Modest evidence of unsupportive attributional style at an intermediate level of situation specificity suggested an empirical examination of the situational referents (negative life outcomes of others) of the construct. The present study revealed large variation in the perceived causal controllability of the negative life outcomes on the RMQ. Confirmatory factor analysis of RMQ data (N = 705) revealed that an excellent fit was provided by an attributional style model that included controllable and uncontrollable situation-types. Thus, when perceived controllability of the negative life outcomes of others was included in the definition of unsupportive attributional style, the estimation of individual differences in controllability perceptions was refined considerably. By linking empirically the situational referents for unsupportive attributional style to the construct definition, the present findings demonstrated the ongoing nature of the process of construct validation. It is clear from the present findings that if systematic variation in the situational referents of attributional styles is unaccounted for in construct definition, individual differences in controllability perceptions (i.e., attributional style) will be underestimated.