717 ADVERTISING STRATEGIES FOR CHARITIES Results and discussion An ANCOVA was run using moral identity, altruism, age and religion as covariates. The dependent variable was calculated by subtracting the intention to donate time from the intention to donate money. Before running the main ANCOVA, an initial test was done to test the assumption of homogeneity of the regression slopes. Twelve interaction effects (3 independent variables × 4 covariates) were calculated and two interaction effects (needy vs helped beneficiaries × religion, statistical vs identifiable x religion) were significant (both p < 0.05). The covariate religion was removed to create a more concrete model. Table 3 presents the results. The same results appeared for appeals and beneficiaries. Respondents showed greater intention to donate money than to donate time for the egoistic appeal and greater intention to donate time for the altruistic appeal (M = 0.264 (s.d. = 1.23) for egoistic appeal, M = –0.491 (s.d. = 1.11) for altruistic appeal. Concerning the beneficiaries, showing helped beneficiaries attracted greater volunteering intention than monetary donating intention (M = –0.357 (s.d. = 1.15) ), and showing needy beneficiaries attracted more monetary donating intentions (M = 0.103 (s.d. = 1.26) ). For the third hypothesis, the result of this study differed from previous studies results. In previous studies, the third hypothesis was not supported. However, the current result showed that portraying statistical victims generated more monetary donations (M = 0.043 (s.d. = 1.36) ), that portraying identifiable victims generated more volunteering intentions (M = –0.310 (s.d. = 1.04) ) and that the relationship was statistically significant (F = 12.15, p = 0.001). This difference seems to come from the use of different victims in the experiments. Studies 1 and 2 used foreign children as victims, for whom respondents may find volunteering difficult, even though they were told to hypothesise that the charity does not prefer either resource and is willing to accept both. Another strong possibility is that foreign children are more difficult to identify as identifiable
