Introduction
Physical inactivity depicts one major risk factor for a variety of noncommunicable diseases (Kohl et al.,
2012) while sufficient physical activity (PA) represents an effective primary prevention strategy for noncommunicable diseases throughout the lifespan (Beaglehole et al.,
2011). However, only 32% of the worldwide population reach the PA recommendations of 150 min of moderate or 75 min of vigorous PA or an equivalent of both for adults (> 18 years) and an average of 60 min moderate to vigorous PA (MVPA) per day for children (5–17 years) (Bull et al.,
2020; Hallal et al.,
2012). Hence, effective interventions to reduce physical inactivity and to enhance PA are needed for adults and children to meet their respective guidelines. Today, mobile health (mHealth) interventions are promising tools for health behavior change due to preliminary results for effectiveness, 24/7 availability, extensive coverage, and their assumed cost-effectiveness (Vandelanotte et al.,
2016). Important key facets for effective mHealth interventions are hereby the theoretical foundation, the use of behavior change techniques, interventions’ embeddedness in a social context, and individual tailoring (Fiedler, Eckert, Wunsch, & Woll,
2020). Besides these contextual and cognitive factors, there is a further need to investigate affect-related determinants in individuals assigned to a mHealth intervention targeting PA to identify reasons for uptake, or barriers, of subsequent PA (Dunton,
2017).
Ecological Momentary Assessment (EMA) provides an opportunity to not only deliver interventional content but also to gather real-time within- and between-person longitudinal data throughout the intervention period (Trull & Ebner-Priemer,
2013). This allows the detection of dynamic associations between determinants of subsequent PA on an individual level, which can be considered in personalized behavior change interventions (Conroy, Lagoa, Hekler, & Rivera,
2020). Of particular interest are hereby dimensions of affect that are assumed to be linked to an improved health behavior (Trull & Ebner-Priemer,
2013). There is much contradiction and overlap in the conceptualization of affect, mood, and emotion (for a review, see Ekkekakis,
2013). James Russel (
2003) proposed a framework that establishes interrelationships between these concepts and defined core affect as a “neurophysiological state consciously accessible as a simplest raw (nonreflective) feeling evident in moods and emotions” (Russell,
2003, p. 148). Building on this, different models and dimensions of core affect have been postulated in recent years. According to the three-dimensional model, core affect includes at least three basic intercorrelated affective dimensions that map the complexity of affective states in daily life: valence (pleasure–displeasure), energetic arousal (wakefulness–tiredness), and calmness (relaxation–tension) (Schimmack & Grob,
2000).
Extensive research has been conducted in the past years investigating the relationship between PA and core affect in adult populations (Forster et al.,
2021; Liao, Shonkoff, & Dunton,
2015). Previous research indicates that valence (Carels, Coit, Young, & Berg,
2007; Emerson, Dunsiger, & Williams,
2018; Kanning & Schoebi,
2016; Schwerdtfeger, Eberhardt, Chmitorz, & Schaller,
2010) and energetic arousal (Liao, Chou, Huh, Leventhal, & Dunton,
2017; Niermann, Herrmann, von Haaren, van Kann, & Woll,
2016; Schwerdtfeger et al.,
2010) are positively associated with subsequent PA, while calmness is negatively associated with PA (Kanning & Schoebi,
2016; Reichert et al.,
2016). Although the results seem to be coherent on the affective dimensions, a direct comparison is difficult because the studies analyzed different temporal aspects of subsequent activity (i.e., 24 h, 15 min) and different types of movement (i.e., free-living PA vs. structured exercises) (Forster et al.,
2021). For example, Carels et al. (
2007) and Emerson et al. (
2018) investigated the relationship between affect and PA within a single day and the results indicate that higher ratings of valence in the morning were associated with increased PA over the day. Here, both studies assessed PA by self-report, which may not represent changes within an individual in detail (Reichert et al.,
2020) and often differs from device-based measured PA (Fiedler, Eckert, Burchartz, Woll, & Wunsch,
2021). Comparable results for the relation of PA and energetic arousal alone were also found in children between 9 and 13 years (Dunton et al.,
2014) and for all three affective states in children between 12 and 17 years (Koch et al.,
2018). Despite these findings, it is important to note that the dynamic relationship between affective states and PA has been studied much less in children and that the existing results are heterogeneous (Bourke, Hilland, & Craike,
2021). In addition, parameters of sleep (i.e., perceived sleep quality, duration, efficacy) are further important determinants of health-related behavior that are assumed to be linked with PA (Wang & Boros,
2021). However, a recent meta-analysis including adult samples revealed no direct relationship between sleep on subsequent PA (Atoui et al.,
2021), while a longer sleep duration was associated with improved eating behavior and higher levels of PA in children (Khan, Chu, Kirk, & Veugelers,
2015).
As stated above, the dimensions of core affect and perceived sleep quality can influence PA behavior in both adults and children. Therefore, it is important to investigate these covariates during a theory-based intervention in which key facets of behavior change are implemented. This can help to assess the possible impact of affective states and sleep quality on the main outcome (PA) of the intervention. Here, existing studies have mainly evaluated EMA-measured constructs as time-lagged predictors immediately before PA uptake to investigate their momentary effect (Liao et al.,
2015). However, in the intervention context day-level peculiarity might also be of interest, as intervention studies usually include time intervals of several days to weeks, and the question if EMA-derived variables have an impact on this time scale is important for designing such interventions. Another important point is to take the PA outcome into account. Here, a study by Reichert et al. (
2017) found differences in the relationship of PA to affective states for exercise and nonexercise PA which suggests that there is no uniform relationship between PA and affect. Knowledge of the mechanisms and barriers related to PA uptake during a longer measurement period will also help to anticipate mental health- and sleep quality-related barriers causing physical inactivity which can then be considered for the development of future mHealth interventions (Dunton,
2017).
Hence, the present study aimed to investigate several potential mental health-related covariates of PA including valence, energetic arousal, and calmness as well as perceived sleep quality on a daily level during 3 weeks to predict same-day PA measured by (1) steps, and (2) MVPA, among children and adults during a PA intervention period. These two PA measures were used to account for possible differences in the relationship between an intensity independent (steps) and intensity-related (MVPA) PA measure, and to project two different types of PA guidelines: the step-related guideline of reaching between 7000 and 10,000 steps per day (e.g., Paluch et al.,
2021), which is followed by most people using fitness trackers or smartwatches as a daily goal, and the intensity-related guideline provided by the World Health Organization (Bull et al.,
2020).
Following previous findings on the topic, it is hypothesized that on days where participants report higher than usual valence and energetic arousal, they have greater device-based measured step count and MVPA on the same day while on days where participants report higher than usual calmness, they have lower device-based measured step count and MVPA on the same day (within-persons). Between-person effects of valence, energetic arousal, and calmness on steps and MVPA (e.g., participants who report higher valence on average have higher/lower average device-based measured step count compared to persons who report lower valence on average), and the relationship between sleep quality and PA on a within- and between-person level will be explored.