Abstract
Sexual behavior among college students is common. College is seen to provide new sources of sexual partners, increased chances for engaging in sexual behavior. With increased sexual behavior among university and college campuses comes sexual aggression and risky sexual behavior that have the potential for many negative outcomes. Contemporary research on the origin of sexual aggression has identified several contributing factors including: (a) early abuse, (b) personality traits (psychopathy), (c) behavioral traits (antisociality), and (d) cognitive variables (sexualization). A unified three-path model of the origin of sexual aggression was developed and previously tested on both adult and juvenile sexual offenders, as well as adult community males providing excellent fit for the data (Knight & Sims-Knight, 2003, 2004). Because of the important role sexual behavior has on college students and much research supporting many of the paths illustrated, this etiological model of sexual aggression was tested. Using data gathered from a computerized interview (the Multidimensional Inventory of Development, Sex, and Aggression, MIDSA, 2011, available at www.midsa.us), this model was employed to predict sexual aggression in a sample of undergraduate college students. Replication of the three-path model was pitted against Malamuth's (1998) two-path confluence model. Along with a replication to support the validity of the etiological model, a revised model was conducted due to the addition of more psychopathy scales since its development. Structural equation modeling revealed excellent fit with the data (CFI = .93, RMSEA = .065) for the three-path replication model, whereas the two-path model fit less well (CFI = .870, RMSEA = .086). These results further support the validity of the three-path model predicting sexual aggression with a college male sample. With additional psychopathy scales, psychopathy was split into two latent traits within the model. The revised model revealed excellent fit for the data (CFI = .93, RMSEA = .059). Comparisons of the replication and the revised model for college students were made with previous three-path models used on juveniles and community males. The revised model was used with risky sexual behavior as the dependent variable. Unexpectedly, while excellent fit of the data was found (CFI = .92, RMSEA = .056), the direct paths to the risky sexual behavior were nonsignificant. Further exploratory analyses of the revised model were performed with a college female sample to examine gender similarities and differences. Discussion of the results cover: (a) the importance of the etiological model, (b) the illustration of psychopathy within the revised model (c) differences within the revised model compared to previous studies (e.g. Knight & Sims-Knight, 2003, 2004), (d) comparisons across gender, (e) risky sexual behavior, (f) limitations and future directions, and (g) contributions of the study..