A survey and an experiment were conducted about the promotion of behavioral intentions for preventing sexual assaults by strangers. The study focused on the "identifiable victim effect", where contextual descriptions of a single victim resulted in the collection of more donations than statistical descriptions of victims. A survey of 42 police websites that made recommendation for preventing sexual assaults revealed that only 3.3% of the threat messages described identifiable victims, whereas 50.0% referred to statistical figures. Additionally, only 9.5% of the coping recommendations mentioned the effectiveness of the recommendation. In the experiment, 207 female university students were randomly presented with one of four threat messages featuring victims (statistical, identifiable, combination, or none), followed by recommendation of high-efficacy or low-efficacy response. ANOVA indicated interaction effects of threat type and response efficacy on behavioral intentions. The effect of threat messages was statistically significant only when a high-efficacy response was recommended. The difference in behavioral intentions for different types of threats partly replicated the findings in donation. These results suggest that current threat appeals for crime prevention need refinements regarding both the threat description and response the efficacy.