Apparently healthy (in this case, narrowly defined as the absence of physical impairments) young adults, aged 18-30 years, were recruited to voluntarily complete an open online cross-sectional survey via email, Web-based advertising (eg, websites, social media, notification via the Vrije Universiteit Amsterdam [VU University] online communication platform), printed flyers with a link to the questionnaire (eg, universities, fitness centers, cafes), and personal approaches (eg, asking participants personally to complete the questionnaire) at both VU University and several secondary vocational education schools. Participants were informed that they would receive an incentive for their participation (ie, an activity tracker worth 90 Euros). Participants were eligible for inclusion if they met the age criteria (18-30 years) and if they did not report physical impairments that limited their physical activity (eg, a doctor’s order to not participate in any sports or physical activities); incomplete surveys (n=78) were excluded (Figure 1). A total of 260 individuals agreed to participate, but only 182 completed the survey in April 2014 (a completion rate of 70%). Data for 3 participants who had physical impairments were removed as this most likely influenced their amount of physical activity, leaving data for 179 eligible participants for analyses. The Medical Ethical Committee of VU University Medical Center approved the study. All participants gave informed consent. Seventy-eight participants did not complete the questionnaire, probably due to time constraints or technical issues.

The questionnaire was primarily based on existing validated instruments [34-37]. When no Dutch version of a questionnaire existed, it was translated into Dutch and back-translated into English by different translators to ensure correct interpretation of questions. The survey was pilot tested among 10 master’s students at VU University who met the inclusion criteria of the study. The online survey was administered through Survey Monkey [38] and data were downloaded to SPSS 20.0 (IBM) and filtered for survey completion and eligibility criteria, according to the protocol for online questionnaires (Checklist for Reporting Results of Internet E-Surveys, CHERRIES) [39]. Personal data were deleted to prevent unauthorized access and to ensure privacy.

The 108-item questionnaire assessed demographics, personality traits [34], exercise self-efficacy [36], social support for physical activity [40], exercise self-identity [41-43], and physical activity levels [37]. Questions were included to indicate participants’ ratings of BCTs applied in a mobile phone physical activity app according to the 93-item taxonomy [22].

Multiple binary logistic regression analyses were used to estimate the association between the different personality characteristics (personality traits, exercise self-efficacy, exercise self-identity, and meeting the Dutch physical activity guidelines) and positive ratings (no, yes) of BCTs addressing “goal setting and goal reviewing” and “feedback and self-monitoring.” Multiple linear regression analyses were used to explore the association between personality characteristics and the preference for the BCTs addressing “social support and social comparison.” Potential effect moderation was evaluated for meeting the Dutch physical activity guidelines. The models were evaluated for potential confounds of total physical activity (min/week) and body mass index (BMI, kg/m²). For all associations, P≤.05 indicated a statistically significant association, except for the interaction terms, which were considered significant at P≤.10.

Sample characteristics and mean scores of the independent and dependent variables for the total study population are presented in Table 2. Overall, most participants were female, highly educated, and physically active. Table 3 describes preferences for BCTs in physical activity apps, and Figure 2 describes the physical activity app functionalities desired by participants.

Table 4 shows that few personality characteristics were significantly associated with high ratings of the BCTs addressing goal setting and goal reviewing. Meeting the Dutch physical activity guidelines did not significantly moderate the association. “Agreeableness” was significantly positively associated with high ratings of the BCTs addressing goal setting and goal reviewing (OR 1.60, 95% CI 1.06-2.41), indicating that respondents who scored 1 point higher on agreeableness (range 1-7) were 1.60 times more likely to rate this BCT category as important.

Table 5 shows that no significant associations were found between BCT feedback and self-monitoring and exercise self-identity and meeting Dutch physical activity guidelines. A significant negative association was found with “neuroticism” (OR 0.76, 95% CI 0.58-1.00) and a significant positive association was found with exercise self-efficacy (OR 1.06, 95% CI 1.02-1.11), indicating that respondents who scored 1 point higher on neuroticism (range 1-7) were 1.32 times less likely to rate this BCT category as important, and respondents who scored 1 point higher on the 60-point exercise self-efficacy scale were 1.06 times more likely to rate “feedback and self-monitoring” as important BCTs. Meeting the Dutch physical activity guidelines did not moderate the associations.

Data obtained from the linear regression analysis between personal characteristics and ratings of BCTs addressing social support and social comparison (range 4-20) are presented in Table 6. No significant associations were found and meeting the Dutch physical activity guidelines was not a significant moderator.

This study examined young adults’ ratings of BCTs applied in a mobile phone physical activity app aimed at improving self-efficacy and social support for physical activity. Furthermore, a number of possible correlates of such ratings were explored. It should be noted that participants were asked to rate BCTs based on their experiences and wishes or requirements of a hypothetical mobile phone physical activity app and not of an existing app. In this study, the ratings of all BCTs addressing self-efficacy were relatively high, but the BCTs addressing social support were not. Respondents who scored higher on agreeableness were more likely to rate BCTs addressing “goal setting and goal reviewing” positively. Furthermore, respondents who scored higher on neuroticism were less likely to rate BCTs addressing “feedback and self-monitoring” positively, while higher scores on self-efficacy were associated with more positive ratings of “feedback and self-monitoring.”

This study’s finding that BCTs addressing “goal setting and goal reviewing” and “feedback and self-monitoring” were positively rated is in line with previous research, in which app features for (automatic) tracking of behavior, setting goals, monitoring behavior, and receiving feedback were evaluated positively in different studies [27,29]. In this study, the most preferred BCTs by young adults were “goal setting on the outcome of behavior,” “self-monitoring of behavior,” and “self-monitoring of the outcome of behavior.” The study of Ehlers and Huberty [30] supports this finding, by indicating that self-regulation techniques (eg, tracking physical activity, goal setting, and receiving feedback) are valuable features of health behavior apps. Middelweerd et al [28] pointed out that participants, overall, preferred a combination of a (virtual) coach with goal setting and that participants would like to receive personal feedback. This study provides further support for positive ratings of BCTs, including goal setting and receiving personal feedback. Earlier findings that social support features were less appreciated among middle-aged women were replicated for young adults, based on the lower preference for social support techniques in apps found in this study [30]. It may be that the high physical activity and exercise self-efficacy levels of the participants in this study caused them to perceive social support as unnecessary. By contrast, about half of the participants preferred a personal coach, which could be seen as another form of social support. Additional analyses (not shown) indicated that those preferring a personal coach did not significantly differ with respect to their preferences for social support from those who did not prefer a personal coach.

The association between personality traits and BCT categories has not been previously reported or investigated. This study found a significant positive association for agreeableness and BCTs addressing “goal setting and reviewing.” Agreeableness is characterized by having the tendency to be kind, cooperative, and trustworthy [37]; thus, participants who were categorized as being cooperative and trustworthy were more likely to be open to app features like goal setting and goal reviewing. This study found an inverse association between neuroticism and BCTs addressing “feedback and self-monitoring.” Neuroticism is characterized by negative affect and emotional instability [48]. Rhodes and Smith [32] noted that avoidance of physical activity or cancelling physical activity plans may be a logical extension of this personality trait, which can also make them less likely to use app features like goal setting, goal reviewing, receiving feedback, and self-monitoring of physical activity behavior. These preliminary results indicate that personality traits could be considered when designing app-based interventions. Previous research showed that tailoring advertising messages to respondents’ personality traits increased their motivation to use a product. Thus, tailoring BCTs to participants may increase the effectiveness of an app-based intervention. However, future research is needed to examine how BCTs should be tailored.

Furthermore, this study found an association between exercise self-efficacy and BCTs addressing “feedback and self-monitoring.” It may be that when a person perceives that one can successfully engage in physical activity, monitoring physical activity and receiving feedback on physical activities affirm and ratify a positive feeling toward physical activity behavior. Moreover, when participants’ exercise self-efficacy is low, monitoring their physical activity behavior may elicit unpleasant feelings and, therefore, generate less appreciation for self-monitoring app features. This suggests that app developers should consider designing app-based interventions tailored to personality and exercise self-efficacy.

Although this study did not find any associations between participants meeting the Dutch physical activity guidelines and ratings of potential effective BCTs, Middelweerd and colleagues [28] found some differences in ratings of app features between young adults who met and who did not meet these guidelines. For example, the latter preferred a goal setting feature and a personal coach, whereas participants meeting the physical activity guidelines reported goal setting as unnecessary and preferred highly detailed training information [28]. Furthermore, differences may also exist between men and women, as men appear to have a preference for team-based, competitive activities, while women do not [49]. The lack of association with participants’ physical activity level could also be explained in terms of a high overall level of physical activity in this study, as more than 80% of participants met the guidelines; thus, this study may not have had sufficient power to detect differences in preferences for BCTs between those meeting and not meeting the Dutch guidelines.

This is the first study to explore associations between personality characteristics and ratings of BCTs applied in mobile phone physical activity apps. Given the current lack of adequate physical activity and preference for apps to support and monitor physical activity, scientific evidence to inform app-based interventions is needed.

Several limitations need to be considered in the interpretation of the findings of this study. The first important limitation is the cross-sectional design, so no causal inferences can be made. Second, the sample was rather homogeneous in gender, physical activity level, and education level; therefore, the results cannot be generalized to more heterogeneous samples. The fact that participants were highly active and had high levels of self-efficacy may have influenced their preferences for certain BCTs. Reasons for the homogenous sample in this study could be that the participants comprised a convenience sample, and the study’s topic (physical activity and apps) and the activity tracker incentive may have attracted more physically active participants. An additional incentive to participate in the study was a prize draw in which one activity tracker could be won. Such an incentive was probably more attractive to participants who already were physically active and who may already have had a preference for tracking their behavior. This may have influenced their preferences for certain BCTs. Consequently, the high levels of preferences for self-monitoring BCTs might be somewhat biased. The time and effort needed to complete the survey may have led to lower participation of less educated participants [50], despite efforts to actively recruit participants at secondary vocational education schools to reach participants with lower education. However, it is well-known that those with lower education and who are physically inactive are hard to reach. Furthermore, men are less likely to participate in lifestyle-related research [51-55]. The fact that our sample was mainly female, with higher education, and more likely to be physically active than the general population of Dutch young adults may have biased the results, and the BCTs identified may not be reflective of the population at large. Furthermore, this study examined a relatively large number of correlations, but only 3 associations were found to be significant, resulting in increased risk of Type 1 error. These associations should, thus, be interpreted with caution and regarded as exploratory findings. Finally, the participants were asked to rate BCTs based on a one-sentence description, and not on actual experience. Consequently, for some participants, past experiences might have positively or negatively influenced their ratings, whereas for others the ratings remained hypothetical.

Future research should focus further on indicating potential differences in the ratings of BCTs between active and inactive participants and using more representative samples. Self-regulation BCTs are potentially effective and were highly appreciated among the young adults in this study. Therefore, these techniques may be considered by physical activity app developers, who should implement these BCTs correctly; the apps should be subsequently tested for their effectiveness in improving physical activity motivation, self-efficacy, and behavior [56]. Although the literature indicates social support as an important correlate of physical activity, young adults in this study did not appreciate this in a mobile phone app. This may indicate that social support should be provided through different or traditional interventions. In addition, this study suggests the value of studying in detail the tailoring of BCTs to participants’ personality characteristics.

Future research should examine the effectiveness of BCTs applied in apps. In previous research, it has been shown that the BCTs included in this study can effectively change behavior; however, because of lack of evaluation research, little is known about their effectiveness in apps [14]. Even if BCTs have been found to be effective in other intervention methods, effectiveness in app operationalization should be tested, as effectiveness of BCTs may be dependent on the actual intervention.

To conclude, ratings of various self-regulation BCTs in a mobile phone app were high in a selected group of highly educated and physically active young adults. BCTs addressing social support were less appreciated. Differences in ratings of BCTs due to differences in personality and exercise self-efficacy between young adults should be taken into account.