In this research, treatment participants showed significant gains with large effect sizes for primary outcomes relevant to a scale for PA and for minutes of PA, and for secondary outcomes supporting constructs associated with SCT constructs (i.e., knowledge, attitudes, and self-efficacy) and TRA (i.e., behavioral intention). Medium to small effect sizes were measured for other primary and secondary outcomes. Taken together, the results suggest that the intervention positively influenced the PA levels of a sample of sedentary workers, a group that might be expected to be difficult to affect. The fact that 61% of treatment participants walked some distance in a very large building and climbed stairs to use the Web site again without payment is a testament to its motivational impact. The exertion involved would be especially challenging for sedentary individuals, and it may have been a barrier to additional visits. Although results from Internet PA research have been mixed,8–10
our results add strength to those studies with positive outcomes.
Defining and identifying a sedentary population is an inexact science, but the BMI values suggest that the participants were overweight to obese and therefore appropriate to test the intervention.45
Being overweight does not in itself identify an individual as sedentary, but being overweight2,7
and having elevated BMI46
are associated with insufficient PA.43,47
Thus, those who tested the Get Moving
Web site seemed to have the characteristics of users for whom it was designed.
Participants who used the Web site repeatedly showed greater intervention effects, suggesting a dose response. Internet PA interventions with higher levels of program use or dose response tend to have more positive outcomes,9
but not always.44
Expected or adequate dosage is often undefined,9,23
and some research indicates the potential for significant effects from only a single exposure.48,49
Further, degree of familiarity with using an Internet program might explain some of the variance in observed treatment effects23
and could influence the dose response.
Overall, this research suggests the potential efficacy of a Web site designed to increase PA among sedentary workers. The results are even more persuasive considering that the intervention was not supported by a work-site campaign to motivate and support participants.20,21
Although employees were restricted to using a computer lab in our research, easier access to a Web site from desktop workstations or from home would likely enhance the number of repeat visits. This type of readily accessed, stand-alone automated PA intervention could be put to use in medical clinics or community centers or on agency Web sites to promote change in PA across a large population. This change could produce important public health effects, particularly if it occurs among those who are most sedentary.7
Limitations of this research include use of self-reported PA data, several single-item measures, and the short follow-up period. Results using more validated scales and a 6-month to 1-year follow-up would have provided stronger evidence for the maintenance of positive outcomes. Moreover, the sample may not be representative because only a small percentage of employees at the factory participated, the computer lab was not accessible to all employees, and most employees who participated were salaried Caucasians with some college education. Finally, participant program-use data were not collected, which negated the potential to identify efficacious program components and mediators responsible for the intervention effects.
Questions remain not only about efficacious program components, but also about optimal structure and even programmatic philosophy. Research suggests that Internet PA interventions might stand alone as we have shown or be enhanced by supplemental components that provide education, coaching, and monitoring.12–19
Our research used a browser instead of a tunnel navigation architecture,50
which structures the order of program presentation. The programmatic philosophy accepted any minimum commitment to start a personal PA plan, even if it didn’t add up to recommended minimums for beneficial PA.43
Positive outcomes resulted, but the impact of our programmatic design decisions is unknown.
Future research using methodology adapted for eHealth could address these concerns.23
For example, an Internet smoking cessation intervention study51
used the Multiphase Optimization Strategy model to empirically identify the active components of the smoking cessation program.52
This type of creative thinking and sophisticated analysis might well be applied to Internet PA research to improve our understanding of how to design more effective Web-based interventions.
SO WHAT? Implications for Health Promotion Practitioners and Researchers
What is already known on this topic?
Recent research has explored the efficacy of various computerized approaches for PA interventions, but the most effective PA mediators are still unknown. Internet PA interventions are appealing because of their potential to provide cost-effective individualized behavioral programs, but research on this approach in real world settings has been limited.
What does this article add?
A fully automated stand-alone Web site to improve PA was well received by sedentary factory workers. It had at least short-term benefits, with large to moderate effect sizes. This research thus supports use of the Internet for delivery of PA interventions.
What are the implications for health promotion practice or research?
Results from other Web-based PA interventions have been mixed, however, so potential users of this technology should be cautious about their expectations. Internet interventions also add a new layer of issues for researchers related to the design of a Web site’s architecture and decision-rules to maximize efficacy.