The Northeastern University’s institutional review board (IRB) approved the research protocol. Parents provided written informed consent, and children provided written informed assent. Consent and assent forms were both delivered to families of eligible children via email, and the paper forms were collected when parents brought the children to a university laboratory for the study. Data were collected between July 2018 and October 2018. Two research assistants (RAs) conducted the data collection with all participants.

Children were recruited from a large, diverse, urban neighborhood in the northeast United States. RA placed printed flyers in the neighborhood afterschool program rooms and local public libraries. Participants from a previous research pool who were interested in this study were also contacted.

Inclusion criteria required the participants to (1) be aged between 8 and 12 years, (2) speak and understand English, (3) have not previously played the selected AVG, and (4) have no physical limitations with respect to AVG play. Exclusion criterion was the inability to complete an AVG session because of a medical condition or physical limitation.

The 8- to 12-year age group was targeted because children younger than 8 years have cognitive limitations in responding to survey questions [26] and children older than 12 years have entered early adolescence and will be subject to many physical, mental, emotional, and social changes that may make their needs and responses different from those of younger children [27]. In addition, without intervention, obese children in this age group (8-12 years old) are highly likely to become obese young adults [6].

After signing an informed consent form, parents completed a questionnaire about their demographic and socioeconomic information (gender, race, household income, education, and number of adults and children in the household).

Children’s height (nearest 0.1 cm) and weight (nearest 0.1 kg) were measured using a stadiometer (ShorrBoard, Weight and Measure LLC) and a calibrated scale (SECA GmbH), which were then used to calculate BMI. Children’s age- and sex-specific BMI percentiles were obtained from the Centers for Disease Control and Prevention BMI-for-age growth charts [27].

A random list function (MS Office Excel 2019) was used to randomize eligible children to watch either an NV or an NNV before AVG play.

The participants had a full and detailed explanation of the study procedures. A Polar heart rate (HR) monitor (Polar) was attached below the participant’s chest, and an ActiGraph GT3x (ActiGraph) was attached to their nondominant hip.

With all apparatus set, participants first sat comfortably before a television screen to watch an 11-min animated video (either NV or NNV). The RA stayed behind a curtain and remotely turned on the AVG console after the video. Children were asked to play for as long as they wanted.

Rate of perceived exertion (RPE) was assessed with Borg scale before and after the AVG play session. HR and PA were monitored during the entire AVG play session using the attached devices. The ActiGraph accelerometer device has been widely used for assessing individual levels of PA across different age groups, including children and young adults [17,31,32]. The ActiGraph triaxial accelerometers measure accelerations from the subject’s intensity and frequency of movement in 3 individual axes (anterior-posterior, vertical, and medial-lateral) and were initialized at 30 Hz sampling. The ActiLife software version 6.13.2 (ActiGraph) was used to download the data from the activity monitors and to convert acceleration data into the 3 axes activity counts, which quantify the amplitude and frequency of detected accelerations at a 1-second epoch dataset. A 1-second epoch has been found to be the most appropriate epoch length to detect short bursts of intense PA and may be the best choice for data processing and analysis in AVG studies examining intermittent PA [31]. We used Evenson et al’s [33] activity cut points to estimate the amount of time spent in light (26-573), moderate (574-1002), and vigorous (≥1003) PA.

After the AVG session, participants completed questionnaires to assess narrative immersion and social desirability. Narrative immersion was assessed using a 9-item Likert-type scale (response range 1-5) [11], where 1=disagree and 5=agree. Sample items included “I could see myself in this story,” “I wanted to see how the story ended,” and “The story influenced my feelings.”

Social desirability (the tendency of the reporting person to give positive self-descriptions) was assessed using a 9-item Likert-type scale (response range 1-5) [34], where 1=disagree and 5=agree. Sample items included “I like everyone I know,” “I tell the truth every single time,” and “I never say thing I should not.” Social desirability of response has been observed in the self-reports of both adults and children. As this may impact the accuracy of the evaluative responses of AVGs, the validity of the study may be affected. The instrument has shown good reliability and validity in children across a variety of ethnic groups.

After completing the questionnaires, each participant received a US $25.00 gift card and was thanked for their participation.

Normality and homogeneity of the data were checked using Kolmogorov-Smirnov and Levene tests, respectively. Independent t tests were performed to detect between-group differences in age and anthropometrics. Repeated measures analysis of variance was used to detect within- and between-group interactions for HR and RPE. The reliability of the questionnaires was assessed by using Cronbach alpha. In addition, Pearson correlation coefficients were calculated to detect the associations between the variables.

The mediation effects were tested for any associations we detected between the animated video condition and MVPA. The SPSS macro PROCESS was used for this analysis [35]. This macro uses bootstrapping to estimate the mediated effect and bias-corrected accelerated 95% CI of the effect. For each simulated sample, the macro estimates the product of paths b (from the independent variable to the mediator) and c (from the mediator to the dependent variable). These products represent a distribution approximating the sampling distribution of the indirect effect in the population. This technique directly tests the significance of the indirect effect (the b×c product). This technique is considered superior to the causal steps method, which does not directly test a mediated effect [36]. This model determines whether the association between the video type (independent variable) and MVPA (dependent variable) was mediated by narrative immersion (mediation variable).

A priori sample size was calculated to detect a significant difference between the 2 conditions, with 95% power (1−beta) to detect a small effect size (EF=0.3). Owing to some sample loss, the final sample size (n=22) provided a power of 92% (1−beta) for a large EF (ie, 0.8). Although less than expected, it was still above 80%. We set the significance level at 5% (alpha≤.05). We performed our analyses using Statistical Software for the Social Sciences (IBM SPSS version 25).

A total of 22 participants (NV: 12 vs NNV: 10) completed the required research protocol, and none of the participants dropped out. Approximately half of the participants were minority children, and their parents were college graduates (Table 1). We tested our randomization and found that children from NV and NNV groups were not statistically different in age, height, weight, BMI percentile, or socioeconomic status (Ps>.11).

All constructs were averaged to form their respective indices after Cronbach alphas were calculated. Narrative immersion had a Cronbach alpha of .62. We found a statistically significant difference in immersion between NV and NNV groups (mean 3.5, SD 0.6 vs mean 2.9, SD 0.6; P=.04); we did not identify any differences for social desirability (mean 3.4, SD 0.6 vs mean 3.2, SD 0.9; P=.40). Our estimate of Cronbach alpha for the social desirability questionnaire for our participants was .84 (Figure 1).

The PA data from the accelerometers indicated a difference in MVPA between the NV and NNV groups, although we did not identify any differences for the total playing time or the number of levels played between the NV and NNV groups (Table 2).

Participants’ HR and RPE responses to AVG showed a significant time effect for both NV and NNV (Ps<.01), although we did not identify either a group effect or a time × group effect for HR (group: P=.30; time × group: P=.17) or RPE (group: P=.64; time × group: P=.79; see Figure 2).

Narrative immersion was positively and moderately correlated with MVPA (r=0.52; P=.01) and average HR during AVG (r=.43; P=.05). MVPA was also positively and moderately correlated with total time playing AVG (r=0.56; P=.01) and average HR during AVG (r=0.60; P=.001; see Table 3).

The mediation analysis to test whether narrative immersion acted as a mediator variable between animated NV (independent variable) and MVPA (dependent variable) is shown in in Figure 3. Our results indicated that the effect of animated NV on MVPA was mediated by narrative immersion. In the first step (model 1: nonmediated direct effect), the regression coefficient of animated NV on MVPA indicated a positive association (beta=11.61; P=.02; see step a in Figure 3). In the second regression step (model 2: mediated direct and indirect effect), animated NV was positively related (beta=.59; P=.03) to narrative immersion (step b). In the last regression model, narrative immersion was marginally positively associated with the dependent variable (beta=6.95; P=.09); when narrative immersion was included in the model, the regression coefficient was significant (indirect effect: steps b and c), but the relationship was slightly attenuated (step a’: beta=7.51; P=.01). Finally, the indirect effect (step a’ plus b and c) was significant, confirming the mediation role of MVPA in this model (Figure 3).

This is the first study to compare NV and NNV added to an AVG play session. The main finding of this study is that AVGs with added narratives elicited more narrative immersion and narrative engagement, leading to more minutes in MVPA. More importantly, narrative immersion acted as a mediator between narratives and MVPA for AVG, also increasing HR after play without more RPE. Nevertheless, it is noteworthy that there is no interaction effect (group×time) for HR; both the groups increased their HR across time, but there are no differences between the groups. This suggests that although the addition of narratives did not result in longer periods of playtime, the players played the narrative versions of AVGs with more bodily engagement and movement, which increases their physiological response but does not necessarily increase their perceived exertion in the task. As a result, the inclusion of good immersive narratives could be a key factor in inducing children’s MVPA play time, enhancing their engagement in AVG without incurring negative physiological feelings such as exertion.

This is the first report showing a mediation effect of narrative immersion on MVPA. The mediation results showed that immersion is a key factor in increasing MVPA during AVG play. Considering that story immersion is an integrative process in which the cognitive and affective resources of the player are concentrated in the task [14], it is reasonable to hypothesize that the NV elicited more game engagement, more commitment to the game tasks, and more movements, thus increasing their PA and physiological demand, without necessarily increasing their exertion. Other AVGs (without narratives) have been shown to not sufficiently motivate children [37], which may be because of increased exertion and boredom. With narrative animations, it may be possible to prevent children from choosing sedentary games and instead motivate them to play AVGs, thus increasing PA.

Previous evidence indicated that narratives’ immersive properties could make children less self-aware of their physical effort during an AVG when they are immersed into the narrative [9], the very concept of immersion [10]. Thus, they can focus more on the outcome of the AVG play instead of the potential uncomfortable feelings associated with the exercise. Therefore, narrative immersion could be an effective method to promote healthier MVPA behaviors in young children [10] and older children [23].

Narratives enabled children to increase immersion and thus MVPA when playing AVGs. The finding that HR (during AVG), but not RPE, is moderately correlated with narrative immersion and MVPA suggests that despite the cardiovascular demand, being more physically active while being immersed in a narrative during an AVG does not increase the perceived exertion. If players perceived more exertion after AVG play, they might be discouraged from playing it again or, worse, engage in sedentary gameplay instead of active gameplay. In other words, children did not feel more tired after playing an AVG with narratives and spent more time participating in MVPA during the play instead.

The results are not without limitations. The small size of our study population (n=22) only allowed us to detect a large EF (d>0.8), and small and moderate differences may have been undetected. Most (60%-70%) of our participants were male. Children’s fitness level before the AVG play was not assessed. The laboratory setting for AVG playing may limit the external validation of the results.

Narrative immersion acts as a key mediator in increasing a player’s MVPA in AVG after watching a narrative animated video. An NV also induced more MVPA with an increased HR after playing but not RPE. Future research should be conducted to investigate the chronic effect of narratives within AVGs on physiological parameters, PA, and playing time in children.