The current study was a quasi-experimental trial using stratified cluster randomization to recruit participants. A total of 4 floors in residence were asked to take part in this pilot study. There were 2 campus residence buildings that participated, each having 1 floor randomly (by toss of a coin) assigned to the intervention condition, and 1 floor selected to the comparison condition. Early in the fall semester, 198 eligible students living on the selected floors were invited to participate. Written consent was obtained from 146 potential participants. These students were sent a link to the baseline questionnaire, which included measures of demographic characteristics, psychosocial variables of attitudes, subjective norms, perceived behavioral control and intentions, and physical activity behaviors. Baseline data was obtained from 91 (59 female) students, with most students in their first year of study. Following the pilot 6-week intervention (described in detail below), participants that completed the baseline questionnaire were sent a link to a follow-up questionnaire with the same measures of physical activity cognitions and behavior. Further attrition resulted in a final sample of 65 students (44 female; mean age 18.51, SD 0.92) completing both baseline and follow-up questionnaires. The study protocol was approved by the Research Ethics Board at the University of Toronto.

For pragmatic reasons, and given the pilot nature of the study, a true control group was not included. A minimal-contact condition or comparison condition was used instead, where students were provided access to, but were not prompted to use, the intervention website. Given that the effectiveness of Internet-based interventions are greatly enhanced with the use of additional methods of participant interactions such as email messages [14], participants in the intervention condition were sent weekly emails prompting students to access the intervention website. The emails provided participants with a synopsis highlighting a weekly intervention topic for each of the 6 weeks. For instance, the topic for week 2 of the intervention was focused on the many student-specific benefits of maintaining a physically active lifestyle. To illustrate, students were sent the following email:

An external link accompanied the email message that linked participants to the website with more details about the student-specific benefits of being physically active.

Hosted within a university portal (Blackboard), Active Transition was a password-controlled website developed to be an informational forum specifically targeting psychosocial mediating variables based on Ajzen’s [17] theory of planned behavior (TPB), and educating students around self-regulating/self-monitoring techniques. Findings from a recent review reported that Web-based TPB interventions are more effective than other theory-based interventions (eg, social cognitive theory, transtheoretical model) [14], and that control theory techniques (eg, goal setting, action planning) are most effective for changing physical activity behaviors [18]. Designed to be 6 weeks in length, Active Transition was a passive intervention, delivering weekly topics on behavioral, normative, control beliefs in physical activity, as well as goal-setting, action planning, relapse prevention, and behavioral maintenance.

The first 2 weeks of the intervention targeted students’ motivation (ie, attitudes, perceptions of control), including student-specific benefits of being physically active (eg, improving concentration, helping with social life), and coping strategies to deal with salient barriers that students typically face during their transition into university. In an attempt to bridge the gap between students’ intentions and subsequent behaviors [9], topics for week 3 and 4 focused on behavioral modification techniques. The intervention material included information on how to effectively set goals, as well as providing sample schedules of how to action plan (Figure 1). Action planning and goal setting are two strategies that have been found to be effective in translating people’s intention into behaviors [19,20]. The topics for the final 2 weeks of the intervention focused on relapse prevention and behavioral maintenance, encouraging students to maintain behaviors despite the normal disruptions that occur while at university. In addition to the highlighted topics, the website included additional resources for students, including a discussion board to facilitate social networking (eg, finding other people with similar interests), links to available leagues and facilities on or nearby campus, mapped out running/biking routes (Figure 2), as well as the opportunity to contact a physical activity expert. Prior to the intervention, Active Transition was pre-tested with students and physical activity experts. A few minor changes were made regarding the text and material; however, there was consensus that the intervention was both acceptable and user-friendly [21].

To examine the feasibility of the website-delivered intervention, study compliance was first examined. Among a total of 198 eligible participants living in the selected residences, 65 of the students completed both baseline and follow-up questionnaires. Initially, 74% (139/198) of all the eligible students had provided written consent, expressing interest in the physical activity intervention and participating in the study. Baseline data, however, was obtained from only 91 (59 females) of those students, representing a 62% response rate; and of the 91 participants that completed the baseline questionnaire, 65 completed the follow-up questionnaire (44 females), representing a 71% (65/91) retention rate. One-way ANOVAs revealed no significant differences between adherers and dropouts for baseline physical activity levels (F _1,89=1.67, P=.17), or on any of the physical activity cognition measures (P’s >.05); thus the final sample of 65 participants was available for subsequent analyses. A comprehensive breakdown of participant responses is shown in Figure 3.

To determine the level of engagement that participants had with the intervention, usage data (the number of times participants logged into the intervention website) was also examined. Importantly, it uncovered that 41% of the participants assigned to the comparison condition were actually users of the intervention (11/27, defined as logging into the website 2 or more times), and 53% in the intervention condition were non-users (20/38, defined as logging in once or not at all). More broadly, usage results indicated that compliance to the intervention was low, with only 6% of participants (4/65) entering the website on an average of one time per week (ie, 6 or more occasions), and only 45% of participants (29/65) logging onto the intervention website on more than 2 occasions. Similarly there were only 5 participants that used the discussion board during the intervention period, and only 1 student had contacted the physical activity expert. Given that this was a pilot study that resulted in unexpected usage within both study conditions, subsequent comparisons in physical activity cognitions and behaviors are made between intervention users (29/65) and intervention non-users (36/65).

The descriptive statistics showing mean scores (SD) for the users and non-users at baseline and follow-up are presented in Table 1. Results of an initial one-way ANOVA found no significant differences in baseline MVPA between users and non-users of the intervention F_1,63=0.13, P=.72). Differences between groups and over time for each of the measured variables were evaluated using the following equation: 2 (time: baseline/follow-up) X 2 (usage: user/non-user), with repeated measures ANOVAs treating the first factor as a within-subjects variable and the second factor as a between subjects variable. Results revealed significant main effects for time, in MVPA (F _1,63=18.1, P<.001, η_p ²=.22), attitudes (F _1,62 =55.19, P<.001, η_p ²=.47), and perceived behavioral control (F _1,62=17.56, P<.001), η_p ²= .22). The main effect of time for intentions (F _1,62=2.91, P=.08, η_p ²=.06) also approached significance. Main effects for usage conditions were all non-significant (P’s>.05).

A significant interaction effect was observed between usage and time on perceived behavioral control (F _1,62 =5.13, P=.03, η_p ²=.08), indicating that intervention users maintained higher levels of perceived control over time, while steeper declines in these perceptions occurred among non-users from baseline to follow-up. While the interaction effects were non-significant between usage and time on attitudes (F _1,62=2.02, P=.16, η_p ²=.03) and usage and time on intentions (F _1,62=1.97, P=.16, η_p ²=.03), mean scores suggest that intervention users were also able to better maintain positive attitudes and higher physical activity intentions while non-users showed greater negative changes in these perceptions over time. The interaction effect between intervention usage and time on physical activity (F _1,63=1.54, P=.22, η_p ²=.03) was not significant. Additional analyses, however, uncovered significant differences in the proportion of students exhibiting declines in their MVPA during the intervention period comparing intervention users and non-users (χ ² ₁=6.11, P=.01, Cohen’s d=.64). The results found fewer intervention users (15/29, 48%) declined in their MVPA scores from baseline to follow-up compared to non-users (8/36, 78%). Findings did not differ when intention-to-treat analyses were conducted, treating dropouts as non-users.

Overall, the results of this pilot study were largely mixed. Positive results found 75% (139/198) of the students living in the selected residences had initially expressed interest in the study (ie, being a part of a physical activity intervention), and that implementation of the website-delivered physical activity intervention on campus is feasible and of interest to students. Participant engagement, however, was highly problematic, as response, retention, and compliance rates were all low. In particular, it was alarming that only 45% (14/38) of the participants in the intervention condition were considered users of the intervention, despite a liberal categorization of usage (ie, considered users if logged in 2 or more times over 6 weeks). At minimum, participants should have been accessing the intervention on a weekly basis. The low usage appears consistent with the notion that students may be generally ambivalent about their physical activity levels [10]. However, with intervention usage, there is some evidence that Active Transition may be helpful in attenuating some declines in students’ physical activity cognitions and behaviors.

Adherence to intervention protocols is essential for physical activity interventions to be successful [24], and our results clearly demonstrated the need to identify innovative strategies to both engage participants and maintain their interest over the duration of the intervention. The development of Active Transition was constrained by limited financial resources. With adequate funding, future interventions with appropriate resourcing should incorporate other Internet-based tools, such as social media, social networking, and/or enhanced visual prompts (ie, video-messaging), and increase participant interaction (eg, providing students with immediate feedback) [25,26]. In an effort to better engage the student population, Active Transition for example, can be augmented to incorporate elements of social media (ie, Facebook, Twitter), and/or integrate the use of short message service (SMS) and photo/video text messaging. There is empirical evidence to suggest that intervention effects are markedly improved if it is theory-based and integrated with personalized contact [13,14]. Given that technology has profoundly changed the way information is being delivered, and how people connect with one another, it appears imperative that attempts are made to integrate these emerging technologies with intervention efforts aimed at behavior change. However, it may also be necessary to first better understand what appeals to students, and how they would want technological tools incorporated to help them facilitate greater physical activity participation.

Consistent with the research literature [7-9], there were significant declines in MPVA among the university students between baseline and follow-up, both for intervention users and non-users. Despite a non-significant interaction between average MPVA and usage, a significantly greater proportion of intervention users were able to maintain or even increase their physical activity behaviors. It is noteworthy that users of the intervention reported engaging in 30 minutes of MVPA more each week during the 6-week intervention period, and if maintained over time, this could have some clinical significance. One plausible reason, however, for why the intervention did not have a statistically significant interaction effect on the mean physical activity scores is because Active Transition specifically targeted MVPA declines through psychosocial mediators (ie, TPB variables). MacKinnon [27] suggested that if the intervention is specifically developed to target social cognition variables, it is very likely that there would be a delayed effect on the behavioral outcome. In other words, the length of the intervention may have been too short to see meaningful differences in behavior. It is necessary to ensure that future implementation of the intervention targets these proposed mediators with a higher fidelity [28].

Findings from this pilot also uncovered substantial decreases in students’ attitudes, perception of control, and physical activity intentions over the course of the first semester. Given that most participants were first-year students, decreases in physical activity cognitions may be a direct reflection of the adaptations required for students entering a new environment, consisting of more barriers to physical activity compared to at high school [10,29], entering a volatile period requiring constant changes and corresponding adjustments. There was evidence to suggest that intervention users were better able to maintain their perceptions of control. Maintaining strong perceived control and intentions towards activity is critical, as they are considered to be the most proximal determinants of behavior [16]. Larger efficacy trials are needed to determine the extent to which perceptions of control are maintained over the course of a calendar year at university. Future work is also required to replicate these findings, and to determine whether the attenuations in students’ physical activity cognitions can lead to significant attenuations in the declines in physical activity typically seen during students’ transition into university.

There are several important limitations to acknowledge. First, a self-report measure was used to assess physical activity, and such measures are susceptible to recall errors and social desirability bias. Second, the study was limited by a small sample size. Given that the purpose of the study was to pilot the feasibility and efficacy of a newly developed intervention, power calculations were not conducted a priori. However, a post-hoc power calculation was conducted, confirming that we were indeed underpowered (1-ß_{err prob}=0.69). To be sufficiently powered, the sample size would require 43 participants in each group. Third, the study design excluded a true control group, resulting in unanticipated usage for some participants initially assigned to the comparison condition. Lastly, it should be acknowledged that the study design had a relatively short follow-up period. Without a subsequent follow-up, it is unknown whether the declines in physical activity cognitions are sustained throughout the academic year.

Notwithstanding the notable limitations, this was the first theory-based website-delivered physical activity intervention aimed at students entering university. Although the institutional portal provided a platform for Active Transition that was potentially advantageous in terms of student access (eg, approximately 70,000 students have access to it at this institution), and its high potential for adoption and diffusion by other colleges or universities, there were major issues with low usage of the intervention. Future work must better target student engagement. Modifications to future interventions should utilize and incorporate other technological and interactive tools that motivate students to continually engage with the intervention. Physical activity decline continues to be problematic during students’ transition into early adulthood, and research must continue to develop innovative strategies for encouraging students to maintain a physically active lifestyle that can be sustained through university and beyond.