Sexual risk behaviors, which can lead to a number of negative health consequences, continue to be a problem on college campuses (e.g., Grello, Welsh, & Harper, 2006; Gullette & Lyons, 2006; Ravert et al., 2009). The theory of planned behavior (TPB; Ajzen, 1985; 1991) has been shown to be effective in predicting sexual risk behavior (e.g., Albarracín, Johnson, Fishbein, & Muellerleile, 2001; Godin & Kok, 1996), as well as a range of other behaviors (e.g., Ajzen, 1991; Armitage & Conner, 2001). The predictive validity of the TPB has been shown to increase when other variables are included and therefore an extended version of the TPB was examined in the current study to aid in the prediction of condom use, contraceptive use, dual use (condom use plus another contraceptive method), and casual sex among college students.
Structural equation modeling was utilized to test both sufficiency models as predicted by the TPB and extended models where six additional variables were allowed to have direct paths to both intentions and behavior. Specifically, the first goal of the study was to examine both the TPB sufficiency models and the extended TPB models for each of the seven behaviors to determine whether the six added variables (past sexual risk behavior, anticipated affect, moral norms, sexual excitation, sexual inhibition, and sensation seeking) had direct relationships with intentions and/or behavior and to assess which of the two types of models provide a better fit to the data.
Given that there is often a discrepancy between intentions and behaviors (e.g.,
Sheeran, 2002), the current study also sought to further the research on factors that predict intention-behavior consistency by exploring a set of variables that may discriminate between 1) those who intended to engage in safe sex behaviors and did so and 2) those who intended to engage in safe sex behavior and did not engage in such behavior. Specifically, the second goal of the study was to determine if characteristics of students' last sexual encounter can aid in predicting whether those students who intended to use condoms, contraception, or dual use did so in their last sexual encounter over the past 2 months with both relationship and casual partners. The seven tested variables were condom planning and preparatory behaviors, mood, sexual arousal, substance use, perceived partner attractiveness, intention certainty, and intention stability.
Approximately eighty-eight percent of the 530 college students who participated in the baseline assessment returned for the follow-up. In relation to the first goal, none of the TPB sufficiency structural equation models demonstrated an adequate fit to the data, whereas the extended structural equation models for casual sex, casual condom use, casual dual use, relationship condom use, and relationship dual use demonstrated good fits for the respective data. In general, the additional paths in the extended TPB models also allowed for more variance accounted for in both intentions and behaviors compared to those in the sufficiency models. None of the models for casual or relationship contraception use were shown to have an acceptable fit to the data. In relation to the second goal, the six discriminant function analyses examining the intention-behavior relationship were able to correctly classify between 74% to 92% of the participants concerning whether or not these intenders engaged in safe sex behaviors or not. The variables that best discriminated between behavior engagement or not over the six analyses were intention stability, intention certainty, and engagement in condom preparatory behaviors.
Implications for future research on the prediction of sexual risk behaviors, the
utility of the TPB, and the intention-behavior relationship are discussed. The implications of these results for sexual risk prevention and intervention research are reviewed.