Exploring habit strength, reward value, and context stability of sedentary activities in older adults: An ecological momentary assessment study

Abstract

Sedentary behavior (SB) poses significant health risks. Still, older adults frequently engage in SB, presumably driven by habit. As stable contexts and rewarding experiences reinforce habits, understanding these mechanisms is essential for designing effective interventions. This study used sensor‐triggered ecological momentary assessments (EMA) in 90 older adults (age 70.2 +/− 6.4, 64.4% female) for 14 consecutive days. Participants wore an ActivPAL accelerometer, a Fitbit activity tracker, and used a smartphone EMA app. Up to six times a day, they were prompted after 30 minutes of SB to report on their activity, context, habit strength, and reward value. Descriptive statistics explored habit strength, context stability, and reward value. Generalized linear mixed models examined the associations between these variables. Results showed that sedentary activities are both habit‐driven and often rewarding, performed for pleasure, self‐care, or personal values. Context stability depends on the sedentary activity. Activities performed in more stable social contexts and more rewarding activities were associated with higher habit strength. Interventions should adopt habit substitution strategies that disrupt the automatic response of sitting, while offering a substitute that is equally rewarding, in a different posture or with more frequent interruptions.

INTRODUCTION

Sedentary behavior (SB), defined as any waking behavior characterized by a low energy expenditure (≤1.5 Metabolic Equivalents of Task) while in a sitting, lying, or reclining position, has been associated with a range of negative health outcomes such as cardiovascular diseases, cancer, type 2 diabetes, and increased mortality (de Rezende et al., ²⁰¹⁴; Matthews et al., ²⁰¹²; Tremblay et al., ²⁰¹⁷). Although these health risks apply to all age groups, they are particularly concerning for older adults, who spend approximately 80% of their awake time sedentary, making them the most sedentary segment of the population (Giné‐Garriga et al., 2020). Aging itself is already accompanied by a higher likelihood of chronic conditions and functional decline (Salive, 2013; World Health Organisation, ²⁰²⁴, October 4), and excessive SB can further exacerbate these negative health effects (Wirth et al., 2017). Reducing total sedentary time or frequently breaking up long periods of uninterrupted SB can be promising for promoting healthy aging (Benatti & Ried‐Larsen, 2015; Chastin, Egerton, et al., ²⁰¹⁵; Chastin, McGregor, et al., ²⁰²¹; Ekelund et al., ²⁰¹⁹).

A variety of SB interventions have been developed and evaluated. However, the effectiveness of these interventions remains mainly limited (Chase et al., 2020; Chastin, Gardiner, et al., ²⁰²¹). Most of these interventions are based on social‐cognitive models of behavior (e.g., the theory of planned behavior), primarily relying on reflective processes to establish new behavioral patterns or change behavior (e.g., conscious decision‐making) (Chastin, Buck, et al., 2015; Van de Velde et al., ²⁰²⁵). Research increasingly suggests that SB often occurs with minimal cognitive forethought, largely driven by automatic processes or habit (Conroy et al., 2013; Maher et al., ²⁰²¹; Maher & Conroy, ²⁰¹⁶; Maher & Dunton, ²⁰²⁰). According to a dual‐process theory of sedentary behavior in older adults, both reflective and automatic processes contribute to its regulation (Maher & Conroy, 2016). Still, habit strength was shown to be the strongest predictor of device‐measured sitting time, contributing more to sedentary behavior than any controlled‐process variable (Maher & Conroy, 2016). As a result, interventions that solely focus on increasing intentions or self‐efficacy may fail to reduce sedentary time (Chastin, Buck, et al., 2015; Van de Velde et al., ²⁰²⁵), as habitual responses can overrule reflective intentions to be more active (Gardner et al., 2020). To effectively target these automatic processes in future interventions, understanding how SB is driven by habit is crucial.

Habit can be defined as a mental association between a cue and a behavior, learned through consistent repetition in a stable context (Phillips & Mullan, 2023). As this association strengthens, an impulse to enact the behavior is automatically triggered when encountering the cue, resulting in habitual behavior (Gardner, 2015; Lally & Gardner, ²⁰¹³; Wood & Rünger, ²⁰¹⁶). Unless particularly strong counter‐habitual motivation overrides this impulse, people will automatically react to the triggering cue with their habitual behavior (Gardner et al., 2020; Hagger et al., ²⁰²³). Consequently, stronger habits lead to more frequent engagement in the behavior (Gardner et al., 2020; Hagger et al., ²⁰²³; Rebar et al., ²⁰¹⁶). Context stability is inherently linked to the strength of habit. A behavior that is repeatedly performed in a consistent physical, social, or temporal context results in higher habit strength (Lally et al., 2010; Lally & Gardner, ²⁰¹³; Mazar & Wood, ²⁰¹⁸; Verplanken, ²⁰¹⁰). For instance, a person who regularly watches TV on the couch with their partner after dinner may develop a strong SB habit in this context. The repeated association with the physical (the couch), temporal (after dinner), and social (being with the partner) context reinforces this behavior, making it automatic. Over time, this person may find themselves watching TV each evening without making a deliberate decision to do so. To date, only one study has assessed the link between context stability and habit strength of SB. Although context stability is a key factor in habitual behavior, Maher et al. (2021) did not find any association between the degree of context stability and habit strength for SB in older adults. However, this lack of association may be due to methodological limitations. For example, context stability was measured at the individual level, resulting in one aggregated score per person. Although this might provide insight into which contexts are more likely to be associated with automatically performed sedentary behavior, older adults engage in a wide range of activities while being sedentary, occurring in a variety of contexts (Leask et al., 2015; Palmer et al., ²⁰¹⁹), which could suggest that it may be more informative to examine context stability concerning specific activities rather than SB as a whole across the entire study period. A more fine‐grained understanding of the degree of stability of specific contexts could inform which contexts should be targeted in future interventions.

Next to context stability, habit theory suggests that habits are reinforced by the reward associated with the behavior. Intrinsic rewards are thought to lead to stronger habits by reinforcing the relationship between behavioral repetition and habit strength (de Wit & Dickinson, ²⁰⁰⁹; Wiedemann et al., ²⁰¹⁴). Although this has been shown in several health behaviors, such as physical activity or fruit and vegetable consumption (Fremling et al., 2025; Kilb & Labudek, ²⁰²²), little is known about the reward value of SB (Cheval et al., 2018). In older adults, sitting is deeply embedded in daily routines and is therefore instrumental to meeting everyday goals (Palmer et al., 2019; Ten Broeke et al., ²⁰²²). From a hierarchical perspective, behavior is typically guided by higher‐order goals (e.g., socializing, reading), while the lower‐level actions that serve these goals, such as sitting, may be enacted with little thought or deliberation (Ten Broeke et al., 2022). When people are asked what they are doing, they rarely perceive or report sitting as a distinct activity, reflecting its automatic and incidental nature within broader goal pursuits (Gardner et al., 2019). In this way, sitting is both goal‐directed and automatic: individuals sit to achieve a meaningful goal, but the act of sitting itself may be triggered by contextual cues such as time of day, social setting, or location (Ten Broeke et al., 2022). Consequently, the habitual nature of sitting may be reinforced by the reward of the activity older adults engage in to pursue their higher‐order goals. Reward is a broad concept, but Eakman and Eklund (2011) argue that activities that are experienced as meaningful typically provide rewarding outcomes such as enjoyment, satisfaction, and a sense of belonging or competence. Currently, no validated instruments exist to assess the full reward value of sedentary behavior in daily life, but assessing the meaningfulness of sedentary activities might already offer insight into the reward value of these activities. At the same time, sedentary behavior might not only be rewarding because of the activities performed, but also because of its low level of physical effort (Maltagliati, ten Broeke, et al., 2024). According to the theory of effort minimization, SB may emerge from an automatic tendency to favor actions that require minimal physical effort (Cheval & Boisgontier, 2021). This likely has evolutionary roots, as conserving energy would have provided a survival advantage in times of resource scarcity. For older adults, this tendency might be amplified due to age‐related declines in physical strength, endurance, or the presence of chronic conditions that make standing or being physically active more demanding. Experiences of pain or fatigue may therefore further reinforce the preference for low‐effort behaviors. Sitting, therefore, may become a rewarding activity because it aligns with the evolutionary drive to minimize unnecessary effort, making it an attractive and default choice (Cheval et al., 2018).

Despite increasing recognition of the role of habit in driving SB, research on how context stability and perceived rewards contribute to SB in older adults remains limited. Habits are highly personal. They are linked to different cues, occur in distinct contexts, and may differ in their reward value (Kwasnicka et al., 2018). Identifying these within‐person habitual patterns can reveal which contexts or activities are most likely to elicit automatic sitting and, in turn, help inform personalized intervention strategies aimed at disrupting or substituting these habits. Because these habitual processes are context‐dependent and unconscious, retrospective questionnaires are limited in their ability to capture them accurately (Wood & Rünger, 2016). Ecological Momentary Assessment (EMA) is particularly well‐suited for studying sedentary habits, as it allows the real‐time collection of data on the contexts, rewards, and automaticity of sedentary activities as they naturally occur (Kwasnicka et al., 2018; Lally et al., ²⁰¹⁰). Moreover, since sedentary behavior is linked to a variety of activities, it seems more meaningful to ask participants to reflect on those specific sedentary activities rather than on SB in general (Gardner et al., 2019). To develop the dual‐process theory of older adults' SB, Maher and Conroy (2016) also used EMA, but measured habit strength only once at baseline, and for sedentary behavior as a general behavior. As a result, it could not account for variation in automaticity across different habits related to specific sedentary activities. However, a fine‐grained, activity‐level approach could offer the opportunity to understand how distinct sedentary habits accumulate to form overall sedentary patterns throughout the day.

To address these gaps in the literature, the present study aims to: (1) explore habit strength and reward value of sedentary activities, and context stability of prolonged SB and sedentary activities in older adults, (2) identify which types of context stability (physical, social, temporal, behavioral) are most related to habit strength, and (3) determine the relationship between reward value of sedentary activities and habit strength. By making use of EMA, this study aims to better understand sedentary behavior by providing insight into how uninterrupted SB is driven by habit, which is a crucial first step towards developing SB habit disruption interventions.

METHODS

Study design and participants

Between February and October 2024, an intensive longitudinal sensor‐triggered event‐based EMA study was conducted in a study sample of older adults aged 60 years or older. Sensor‐triggered event‐based EMA involves a device (e.g., Fitbit) automatically prompting a survey (sensor‐triggered) when a predefined event of interest, such as 30 minutes of SB, is detected (event‐based) (Dejonckheere & Erbas, 2021). To ensure heterogeneity in age and sex, a combination of purposive convenience and snowball sampling was used. Older adults were recruited via word of mouth, social media, and local service centers, which are neighborhood‐based facilities that provide accessible information and activities to support social cohesion and facilitate self‐reliance. Participants had to be Dutch‐speaking, community‐dwelling, and able to walk at least 100 m with or without assistance. Those with significant visual or hearing impairments, making it difficult to read or hear the EMA notifications, respectively, were not eligible for inclusion. Furthermore, participants diagnosed with neurological disorders (e.g., dementia or Parkinson's disease) were also excluded. This paper was written according to the CREMAS guidelines for EMA studies (Supplementary file S1) (Liao et al., 2016).

Procedures

Older adults who expressed interest in participating were contacted by phone to arrange a first home visit. During this visit, participants were informed about the study goals and procedure. All participants signed the informed consent form and completed a baseline questionnaire that assessed sociodemographic characteristics, health‐related factors, social isolation, and sleep patterns. Participants received a Fitbit Inspire 3 and an Android smartphone (Motorola G20/E40 or a Samsung Galaxy A35) with the pre‐installed EMA application HealthReact (version 1.1.12). A recent validation study has demonstrated the validity of the Fitbit Inspire 3 in detecting bouts of prolonged sedentary behavior (i.e., 30 minutes) in the same population, based on step count (Delobelle et al., 2024). Additionally, an ActivPAL4 accelerometer (attached to their right upper thigh) was used to describe participants' SB patterns. To familiarize participants with the EMA application and the types of questions they could expect throughout the study period, a test survey was sent to their smartphones. The researcher and participant reviewed this survey together. Furthermore, a brief training session was given on the correct use of the wearable devices. Participants were instructed to wear the Fitbit activity tracker continuously on the non‐dominant wrist from waking until bedtime. For the ActivPAL4 accelerometer, participants were advised to wear it continuously, including during sleep time, and were shown how to re‐attach the device if it needed to be removed during water‐based activities.

The day after the first home visit, the EMA protocol started for 14 consecutive days. This duration was deliberately chosen to capture participants' weekly and weekend routines regarding SB, while also taking feasibility into account. Activity data collected by the Fitbit device were automatically transmitted to the Fitbit application on the participant's smartphone. These data were then synchronized with the Fitbit cloud server and subsequently transferred to the HealthReact server. On this server, incoming data were monitored for the occurrence of an “event of interest,” which was defined as a period of 30 minutes of uninterrupted SB, consistent with common definitions of prolonged SB in literature (Mickute et al., 2021). These events were detected if, in a time window of 30 minutes, two steps or fewer were captured by Fitbit, in combination with continuous heart rate detection to confirm the device was worn (Delobelle et al., 2025). This approach prevented triggering in case of non‐wear. A recent validation study has shown that the Fitbit Inspire 3 can accurately capture these periods of uninterrupted SB. However, the synchronization lag between Fitbit and the HealthReact server could delay the appearance of the survey on the smartphone by up to sixteen minutes (Delobelle et al., 2024). When an event of interest occurred, participants were prompted by an auditory signal to fill out a survey on their smartphones. They had fifteen minutes to respond, after which the survey disappeared until a new event occurred. More details about the sensor‐triggered event‐based EMA surveys can be found below, under “measures.” At the end of the two weeks, a second home visit was conducted to collect the ActivPAL, Fitbit, and smartphone.

Measures

Baseline measurements

Socio‐demographic characteristics were collected using a paper‐based survey. The questionnaire included items on age, sex, height, weight, marital status, educational level, and current occupation. Furthermore, health‐related quality of life was assessed using the SF‐36 (Brazier et al., 1996), and the Patient‐Reported Outcomes Measurement Information System (PROMIS) was used to measure aspects of social isolation and sleep quality (Cella et al., 2019). Finally, participants performed the short physical performance battery (SPPB) to assess their physical functioning (Guralnik et al., 1994; Kameniar et al., ²⁰²⁴).

Sensor‐triggered EMA surveys

To balance data collection and participant burden, and based upon previous compliance studies (Compernolle et al., 2024; Maher et al., ²⁰¹⁸), participants received a maximum of six surveys across the day: two prompts in the morning (7–12h30), two in the afternoon (12h30–18h), and two in the evening (18–22h), each separated by a minimum interval of 90 minutes. Although participants were instructed to wear the Fitbit only during the daytime, an additional rule was implemented to avoid surveys being sent while participants might still be in bed if they had worn the device overnight. Specifically, the first morning survey was sent only after they had walked at least 10 steps within two consecutive minutes.

The EMA survey comprised 16 items, taking approximately one to two minutes to complete. The full EMA survey can be found in the supplementary file S2. Before data collection, the survey was pilot tested with older adults from the target population to ensure that all questions were clearly understood and applicable to a wide range of sedentary situations.

Behavioral context

Participants were asked to indicate the main activity they were doing while sitting or lying down. A list of activities was provided in which only one option could be indicated: watching TV, using the computer/tablet/smartphone, reading, eating/drinking, talking to friends/family/ …, doing a task for my job, household task (e.g., cooking, ironing), taking care of myself, transport, listening to music, doing a creative activity (e.g., knitting, sewing), doing a cognitive activity (e.g., crossword puzzles, a card game), sleeping, or other which allowed providing more details (Hevel et al., 2021).

Habit strength

One item from the Self‐Reported Behavioral Automaticity Index (SRBAI) examined the habit strength of the indicated sedentary activity: “Without thinking, I perform this activity while sitting.” This approach follows the recommendations of Gardner, Lally, and Rebar (2024), who demonstrated that each SRBAI item shows strong construct and predictive validity and concluded that the single‐item version is an appropriate substitute when very brief measures are required, such as in intensive repeated‐measures designs (e.g., EMA studies). Responses were rated on a seven‐point Likert Scale from “totally agree” to “totally disagree,” which is identical to that of the original SRBAI instrument, ensuring conceptual and psychometric consistency with prior applications.

Reward value

Reward value was also assessed at the level of the sedentary activity. Participants were asked to indicate their level of agreement on six specific statements derived from the Engagement in Meaningful Activities Scale (EMAS). This validated instrument is designed to measure the extent of engagement in meaningful activities (Cruyt et al., 2023), and Eakman and Eklund (2011) argue that meaningful activities typically provide rewarding outcomes such as enjoyment, satisfaction, and a sense of belonging or competence. An activity is considered meaningful if it aligns with one of the following clusters: selfhood, social, or pleasure (Eakman et al., 2018). The twelve original EMAS items were clustered into five items to lower participant burden. Each item followed a stem: “I do this activity because …” (e.g., “… it gives me pleasure and/or satisfaction”). Additionally, recognizing that SB may also occur specifically to rest or relieve pain, independent from the activity they are engaging in, an extra item was included: “… I want to rest or relieve pain”. Each item had four response options to capture the degree of agreement and minimize central tendency bias by omitting a neutral midpoint: disagree, somewhat disagree, agree, and completely agree. Total reward value was calculated as a mean score on all six items, with a higher score indicating a higher reward value.

Physical and social context

To capture the contexts in which sedentary behavior occurred, two closed questions asked about where and with whom participants were at the moment of the trigger (Giurgiu et al., 2020; Liao et al., ²⁰¹⁵). While all other questions in the survey were static, the question on physical context was adaptive: participants could first indicate whether they were indoors or outdoors, and depending on this response, they were presented with a tailored list of location‐specific options (e.g., at home, in my garden, a bar). Based on these items, context stability was calculated at two levels: (1) the individual level (per person), and (2) the behavioral level (per activity). One score was calculated for each of the four contexts of SB: the physical, social, and temporal context (calculated at both levels), and the behavioral context (only measured at the individual level).

Pain and fatigue

At the end of the survey, two additional questions were included to assess participants' levels of pain and fatigue before sitting down. For pain, participants responded to the question: “How much pain did you experience before sitting down?” using the following options: none, very mild, mild, moderate, severe, very severe. Fatigue was assessed with the question “How fatigued did you feel before sitting down?” with the options: not fatigued at all, slightly fatigued, neither fatigued nor rested, fatigued, very fatigued. These measures aimed to provide insight into how physical discomfort and low energy levels might influence the association between the perceived reward associated with SB and habit strength.

Device‐measured sedentary behavior

The ActivPAL 4 accelerometer (PAL technologies, Glasgow, Scotland), attached to the midline of the right anterior thigh and worn during both day and night, was used to measure SB during the entire study period. Data were summarized in epochs of fifteen seconds and were processed using the ActivPAL CREA algorithm (PALanalysis v8.11.8.75). Three measures of SB were calculated (Boerema et al., 2020): (1) total sedentary time, expressed as the average sedentary time on valid days (i.e. at least 10 hours of waking wear time) (Blackwood et al., 2022; Winkler et al., ²⁰¹⁶), (2) minutes spent in prolonged bouts of SB, defined as uninterrupted sedentary bouts of 30 minutes or more (Diaz et al., 2019; Niemelä et al., ²⁰¹⁹), and (3) usual bout duration, which represents the bout duration at which half of all sedentary time is accumulated (Chastin, Winkler, et al., 2015). In case of outliers or impossible values in sedentary time, the sleep data of the corresponding participants were manually checked and adjusted based on the data of other study days in the tab “time in bed adjustments” in the PALanalysis software (Winkler et al., 2016).

Data management

Baseline data were collected and managed using REDCap (Research Electronic Data Capture, version 14.0.21) hosted at Ghent University, Belgium (Harris et al., 2019, 2009). EMA data were stored on the HealthReact server (v20240830–1). All study data, including baseline questionnaires, ActivPAL data, and EMA responses, were merged into one dataset in RStudio (version 2023.6.2.561).

Data analysis

Compliance

Currently, there are no clear guidelines on compliance thresholds. Therefore, various thresholds between 50% and 80% have been explored. To find a balance between maximizing data inclusion and minimizing the risk of bias due to excessive missing data, participants who responded to fewer than 60% of the EMA prompts were excluded from further analysis (Hartson et al., 2023; Perski et al., ²⁰²⁴). More detailed information about the compliance thresholds can be found in Supplementary File S3.

Descriptive statistics

Descriptive statistics were conducted to describe the study sample, reward value, context stability, and habit strength of prolonged SB. To avoid bias due to differences in prompt counts across participants, all individual‐level variables were first averaged within each participant. These participant‐level means were then used in the main analyses to ensure that participants who completed more prompts did not disproportionately influence group‐level results (Siepe et al., 2025).

Calculating context stability

Context stability was examined at two levels: (1) at the individual level, capturing the stability of contexts in which an individual was sedentary, and (2) at the behavioral level, reflecting the stability of these contexts for each specific sedentary activity. The stability of the behavioral, physical, and social context was assessed based on the EMA items regarding what activity participants were doing (behavioral context), where (physical context), and with whom (social context) they were at the moment of the trigger. Temporal context was derived from the timestamp of the EMA response, extracted from the Fitbit data. For each participant, four context stability variables were computed at the individual level (physical, social, behavioral, and temporal stability) to capture the overall stability of each contextual domain. Additionally, context stability was calculated at the behavioral level, reflecting how stable the physical, social, and temporal context was for each specific sedentary activity. All context stability scores were computed using the shrinkage entropy estimator from the R entropy package, applying a robust maximum likelihood estimator with James‐Stein‐type shrinkage intensity. Entropy scores quantify the uncertainty or instability in the data, with higher scores indicating greater instability, thus indicating less stable contexts (Hausser & Strimmer, 2009). Context stability scores were calculated by inverting entropy scores. As a result, all context stability scores are negative, with a score closer to zero representing more stable contexts.

Generalized linear mixed‐effects models

An empty model was first run to estimate the intraclass correlation coefficient (ICC), quantifying the proportion of variance in habit strength attributable to between‐person versus within‐person differences. To avoid overfitting, the random effect for study day was removed when it did not improve model fit. Subsequently, a series of gamma generalized linear mixed‐effects models with a log link were applied, given the hierarchical structure of the EMA data (observations nested within days within individuals) and because the assumption of normality of the dependent variable (i.e., habit strength) was violated. Benjamini‐Hochberg corrections were applied to correct for multiple model testing (Benjamini & Yekutieli, 2001).

A first set of seven models examined the association between the reward value of SB and habit strength, with a separate model for each cluster of reward (pleasure, social connection, social help, self‐care, personal values, and pain relief/relaxation) and total reward value. A second set of models assessed the association between context stability and habit strength, with separate models for each of the contextual dimensions at both levels (individual and behavioral level): physical, social, temporal, and behavioral context stability.

In all models, habit strength served as the dependent variable, while the corresponding variable on reward value or context stability served as the independent variable. Each model included age, pain, fatigue, and activity type as control variables. Random intercepts were included for participant ID. Before the analysis, model assumptions were checked, including multicollinearity, linearity of continuous predictors on the log scale, and the distribution of residuals. P‐values below 0.05 were considered statistically significant.

RESULTS

Sample description

A total of 107 community‐dwelling older adults were initially recruited for the study. Two of them dropped out after a few days due to a loss of motivation. Considering a compliance rate of at least 60%, the total sample size consisted of 90 participants, with a mean age of 70.2 years (SD = 6.4), ranging from 60 to 87, of whom 64.4% were female. The mean BMI of the study sample was 26.0 kg/m² (SD = 3.1). Participants spent on average 555.0 (SD = 109.4) minutes per day in SB, of which 288.9 minutes (SD = 102.6) were spent in prolonged bouts, with a usual bout duration of 27.5 (SD = 12.6) minutes. Throughout the entire study period, participants received on average 34.7 (SD = 16.2) prompts, of which 28.7 (SD = 14.0) were answered, resulting in an average compliance rate of 83.3% (SD = 11.6%). All descriptive statistics on the sample characteristics can be found in Table 1. Patterns of nonresponse were explored in a previous publication using the same dataset, in which sociodemographic characteristics were compared between the included sample and participants excluded due to low compliance (<60%). The group of noncompliers entailed significantly more men compared to the group that answered at least 60% of the prompts. Detailed results can be found in Compernolle et al. (2025).

TABLE 1.

Descriptive statistics on sample characteristics and EMA prompts.

Participant characteristics	Study sample (n = 90)
Sociodemographic data
Age
Mean (SD)	70.2 (6.4)
Range	60–87
Body mass index
Mean (SD)	26.0 (3.1)
Range	19.1–33.8
% female	64.4
Educational level (%)
Primary education	24.4
Secondary education	21.1
Bachelor's degree	41.1
Master's degree	13.3
Occupational status (%)
Retired, volunteering work or flexi‐job	37.8
Retired, no volunteering work or flexi‐job	51.1
Not yet retired	11.1
% familiar with smartphone use	91.1
Sedentary behavior, mean (SD)
Total sedentary time (min/day)	555.0 (109.4)
Prolonged bouts (min/day)	288.9 (102.6)
Usual bout duration (min/day)	27.51 (12.6)

Describing context stability, reward value, and habit strength of SB

Participants spent most of the prolonged SB events inside (92.4%), at home (83.1%), mostly accompanied by their partner (63.2%), or alone (34.3%). Regarding the temporal context, 45.7% of the events were triggered in the evening, 34.0% in the afternoon, and 20.3% in the morning. The most frequently indicated activity was watching TV (39.3%), followed by using electronic devices (15.9%), eating or drinking (12.9%), and reading (12.8%). More detailed information about the contexts in which the participants were sedentary can be found in Compernolle et al. (2025). Table 2 shows the stability at both the individual and the behavioral level, and the frequencies of the behavioral context. The more negative the value, the more variation and less stable the context; values closer to zero indicate higher context stability. At the individual level, the social, temporal, and behavioral context in which SB occurred varied considerably, suggesting rather low context stability for prolonged SB. In contrast, the physical context showed more stability.

TABLE 2.

Context stability at the individual and behavioral context, and frequencies of the behavioral context.

Context stability at the individual level
	Mean	SD		Range
Physical context	−0.82	0.61		−3.58‐0
Social context	−1.13	0.48		−2‐0
Temporal context	−1.47	0.23		−1.58‐0
Behavioral context	−2.29	0.56		−3.7‐0

Context stability at the behavioral level
	Participants N	Observations N (%)	Physical context (range: −3.58; 0)	Social context (range: −2; 0)	Temporal context (range: −1.58; 0)
Watching TV	83	869 (36.7)	−0.14	−1.10	−0.88
Using electronic devices	74	353 (14.9)	−0.44	−1.26	−1.55
Reading	66	284 (12.0)	−0.71	−1.22	−1.55
Eating/drinking	69	287 (12.1)	−1.29	−1.43	−1.57
Creative activity	9	27 (1.1)	−0.88	−1.41	−1.39
Cognitive activity	21	63 (2.7)	−1.55	−1.80	−1.58
Talking to friends/family	56	148 (6.3)	−2.35	−1.54	−1.45
Household task	34	63 (2.7)	−1.79	−1.47	−1.58
Job‐related task	19	35 (1.5)	−2.19	−1.48	−1.58
Listening to music	10	18 (0.8)	−0.34	−1.17	−0.83
Self‐care	20	32 (1.4)	−1.61	−1.82	−1.24
Transport	24	38 (1.6)	−2.10	−1.77	−1.51
Sleeping	48	148 (6.3)	−0.34	−1.24	−1.28

At the behavioral level, context stability varied by sedentary activity, particularly concerning the physical context. Some activities were consistently associated with a uniform physical context, while others occurred in more diverse physical contexts. Activities such as watching TV (−0.14), listening to music (−0.34), using electronic devices (−0.44), and creative activities (−0.88) were performed in relatively stable physical contexts. In contrast, activities such as job‐related tasks (−2.19), transport (−2.10), and talking to friends/family (−2.35) showed lower physical context stability, suggesting they occurred across a broader range of physical settings. Compared to the physical context, the social context showed less stability across all activities. Temporal context stability was low for most activities, except for watching TV (−0.88) and listening to music (−0.83). Watching TV showed the highest context stability across all contexts, meaning that this activity is frequently performed in the same physical, social, and temporal context.

Of all prompted sedentary activities, the mean score for reward value was 9.55 (SD = 1.6), on a scale of 0 to 18 (n = 2,356). Social interactions had the highest total reward value (10.8, SD = 1.1), followed by listening to music (10.6, SD = 1.0) and job‐related tasks (10.3, SD = 1.9). Self‐care activities had the lowest total reward value (8.5, SD = 1.5), followed by sleeping (8.7, SD = 1.2). Figure 1 shows the relative frequencies of the answers per activity for each cluster of reward. Job‐related tasks scored the most positively on all clusters of reward. Most sedentary activities were performed for pleasure or to express creativity (65.3%), followed by pursuing personal values (58.3%) and self‐care (51.4%). Additionally, 46.1% of the activities were performed to foster social connections, and 35.6% helped others. In 21.7% of the prompts, participants indicated they were sedentary to relieve pain or relax.

FIGURE 1.

Relative frequencies (x‐axis) of four answering options (represented by the colors), depending on the activity (y‐axis) for each cluster of reward.

On a scale of one to seven, the mean habit strength (n = 2,251) of performing the activities in a sitting or reclining position was 5.6 (SD = 0.7). As shown in Figure 2, most of the triggered activities were reported to be performed without conscious thought about doing it in a sitting or reclining position, indicating that habit strength is high in all activities. Compared to the other activities, self‐care and transport were more frequently reported as performed with conscious thought.

FIGURE 2.

Relative frequencies (x‐axis) of each answer on a seven‐point Likert scale for habit strength (colors represent the answering options) depending on the activity (y‐axis).

Associations between context stability, reward value, and habit strength

Of the total variance in habit strength, 39.9% is attributable to between‐person variance (ICC = 0.39), and 2.4% to day‐to‐day fluctuation within individuals. The remaining 57.6% reflected within‐person, within‐day variation. To avoid overfitting, the random effect of study day was left out of the generalized linear mixed models. All outcomes of the empty model and the generalized linear mixed models can be found in Table 3. At the individual level, no associations were found between context stability and habit strength. At the behavioral level, a positive association was found between social context stability and habit strength. Activities performed in more stable social contexts are more automatically performed in a seated position (t = 39.60; p < 0.001). In contrast, no association was found between physical context stability (t = 1.28, p = 0.20) or temporal context stability (t = 1.53, p = 0.13) at the behavioral level and habit strength when controlling for age, pain, and fatigue.

TABLE 3.

Overview of the outcomes of the empty model, and generalized linear mixed models on the effects of context stability and reward value on habit strength (dependent variable), with and without covariates.

EMPTY MODEL Habit strength
Random effect	Variance					Standard deviation
Person‐level	0.98					0.99
Residual	1.42					1.19

GENERALIZED LINEAR MIXED MODELS Habit strength
PREDICTOR	Model without covariates					Model controlled for age, pain, fatigue, and activity type (except for the behavioral context)
Context stability at the individual level	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC
Physical context	0.01	0.05	0.31	0.88 (0.95)	8817.6 8840.5	0.01	0.05	0.17	0.96 (0.96)	8325.4 8438.6
Social context	0.03	0.06	0.43	0.49 (0.81)	8817.18840.0	0.05	0.07	0.71	0.32 (0.64)	8324.3 8437.6
Temporal context	0.02	0.11	0.17	0.80 (0.95)	8817.5 8840.4	0.06	0.12	0.49	0.52 (0.81)	8324.9 8438.2
Behavioral context	0.01	0.05	0.15	0.88 (0.95)	8817.6 8840.5	0.01	0.05	0.22	0.83 (0.95)	8362.9 8402.5

Context stability at the behavioral level	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC
Physical context	0.01	0.01	1.45	0.15 (0.42)	8812.9 8835.7	0.01	0.01	1.28	0.20 (0.46)	8358.7 8398.3
Social context	0.11	0.02	2.29	<0.001** (0.01*)	8806.2 8829.1	0.08	<0.01	39.60	<0.001** (<0.001**)	8352.7 8392.4
Temporal context	0.03	0.02	1.57	0.12 (0.42)	8812.9 8835.8	0.03	0.02	1.53	0.13 (0.42)	8358.4 8398.0

PREDICTOR	Model without covariates					Model controlled for age, pain, fatigue, and activity
	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC	B	SE	t‐value	p‐value (adjusted p‐value)	AIC/BIC
Pleasure	0.06	0.01	4.63	<0.001** (<0.001**)	8758.5 8781.3	0.06	0.01	4.15	<0.001** (<0.001**)	8274.2 8387.4
Connection	0.02	0.01	1.56	0.08 (0.16)	8758.2 8781.1	0.02	0.01	1.44	0.09 (0.16)	8280.6 8393.8
Help	−0.003	0.01	−0.64	0.79 (0.79)	8765.3 8788.1	0.005	0.01	0.01	0.69 (0.74)	8279.7 8392.9
Values	0.02	0.01	1.91	0.08 (0.16)	8728.0 8750.8	0.02	0.01	2.08	0.06 (0.16)	8248.1 8361.2
Self‐care	0.02	0.01	1.58	0.14 (0.18)	8768.2 8791.1	0.02	0.01	1.30	0.14 (0.18)	8320.8 8360.4
Pain & Relaxation	−0.02	0.01	−1.51	0.13 (0.18)	8758.2 8781.1	−0.01	0.01	−1.71	0.15 (0.18)	8290.6 8403.7
Total reward value	0.01	0.004	2.26	0.02* (0.07)	8623.5 8646.3	0.01	0.004	2.40	0.01* (0.05*)	8177.1 8290.0

^*p‐value <0.05;

^**p‐value<0.001.

Regarding reward value, a positive significant association was found between total reward value and habit strength for performing activities in a seated position, controlled for age, pain, fatigue, and activity type (t = 2.40, p = 0.01). Looking at the clusters of reward individually, only activities performed for pleasure were positively associated with habit strength (t = 4.15, p < 0.001) when controlling for age, pain, fatigue, and activity type. None of the other clusters of reward was significantly associated with habit strength for performing activities in a seated position.

DISCUSSION

This study explored context stability, reward value, and habit strength of prolonged, uninterrupted SB in older adults, using sensor‐triggered event‐based EMA. Furthermore, it examined which specific types of context stability are associated with habit strength and whether more rewarding sedentary activities are more automatically performed in a seated position. Context stability was examined at the level of SB in general and for specific sedentary activities, whereas reward value and habit strength were only assessed at the level of sedentary activities. The findings show that SB is most often performed in stable physical contexts. In contrast, the social, temporal, and behavioral contexts in which SB is performed showed less stability. However, context stability varies depending on the activity, with some activities performed in more stable physical or temporal contexts, while others occur in more variable contexts. Importantly, activities performed in more stable social contexts are associated with higher habit strength. Furthermore, it was shown that sedentary activities are rewarding for older adults, often performed for pleasure, self‐care, or to pursue personal values. More rewarding activities are more likely to be automatically performed in a seated position.

Most of the variation in habit strength was observed within‐person, within‐days, rather than between‐person, showing that the automaticity of SB fluctuates within individuals and throughout the day. This finding is not surprising, as habit strength is based upon personally made cue‐behavior associations (Phillips & Mullan, 2023). Furthermore, the mean score for habit strength in this study was close to the upper limit of the response scale, indicating that many sedentary activities performed by older adults are highly automatically performed in a seated position. The same results were found in the study of Maher et al. (2021), where habit strength for SB was measured with the full SRBAI. These findings align with the evidence that SB is integrated into the daily routines of older adults. It often occurs with minimal effort or conscious reasoning (Conroy et al., 2013; Maher & Conroy, ²⁰¹⁶; Maher & Dunton, ²⁰²⁰).

Habitual behavior takes place in stable contexts. However, the social, temporal, and behavioral context showed rather low stability in general, suggesting that older adults engage in prolonged SB in a variety of social settings, spread throughout the day, and in a wide range of activities. In contrast, the physical context demonstrated greater stability. This is what would be expected, as the events were most often triggered at home, where older adults are most often sedentary, as suggested by previous research (Palmer et al., 2019). Furthermore, the study of Compernolle et al. (2025), based on the same dataset, found that older adults were most willing to interrupt their sedentary behavior when they were at home. This suggests that the home may be an optimal context for implementing interventions that prompt the interruption of prolonged sedentary behavior. Looking at the behavioral level, across almost all sedentary activities, the social and temporal contexts showed low stability, suggesting that most of the activities occurred in a variety of social settings and at different times throughout the day. However, certain activities, such as watching TV or listening to music, showed more stable physical and temporal contexts compared to others. This echoes previous research suggesting that watching TV is an activity that is commonly performed by older adults in the evening and at home (Palmer et al., 2019; Van Cauwenberg et al., ²⁰¹⁵). As a result, it can be concluded that not all sedentary activities are equally tied to stable physical contexts or a specific part of the day.

Building on these findings, the second research question explored which types of context stability (i.e., physical, social, temporal, or behavioral) are related to habit strength. Although the literature is clear on the role of stable contexts on habit strength, no associations were found between the degree of context stability and habit strength at the individual level. This aligns with previous findings of Maher et al. (2021) and may be explained by the fact that SB is spread out across the entire day, as SB is linked to specific activities (Palmer et al., 2019; Ten Broeke et al., ²⁰²²), and is therefore linked to several contexts (Leask et al., 2015). For example, the habit of flossing may be cued only once a day when a person has brushed his teeth, while sedentary behavior is automatically cued in several contexts (e.g., in the morning when a person gets the newspaper or in the afternoon when a friend visits) (McCloskey & Johnson, 2019). When examining context stability at the behavioral level, which makes it more specific, a positive association was found between social context stability and habit strength. This suggests that activities performed in more stable social contexts are more automatically performed in a seated position. Over time, social expectations or norms may become internalized as contextual cues that trigger this automatic seated behavior (Prapavessis et al., 2015; Rollo et al., ²⁰¹⁶). In contrast, no associations were found between the stability of the physical and temporal context and habit strength. This, again, could be because sedentary activities can be triggered in a variety of physical and temporal contexts. Furthermore, it could be that the response options for the item on physical context may not have captured enough variation across physical settings, which could have limited the ability to detect meaningful associations. For example, the response option “at home” includes a range of distinct physical environments (e.g., in the kitchen, in the living room). Additionally, only the link between the individual types of context stability and habit strength was examined, while contextual cues can interact with each other (Meghani et al., 2023). For instance, it could be that the physical context acts as a habit cue in a specific social context.

Next to context stability, the reward value of sedentary activities was examined. Most episodes of prolonged SB were performed for pleasure, satisfaction, or to express creativity. The other clusters of reward (social connection, helping others, taking care of themselves, and pursuing personal values) were also frequently reported. These results indicate that the sedentary activities older adults perform are often meaningful to them and can therefore be considered rewarding (Eakman & Eklund, 2011). This speaks to previous research, where participants mentioned enjoyment and social contact as one of the most important benefits of their sedentary activities (Compernolle et al., 2020; Greenwood‐Hickman et al., ²⁰¹⁶; Mcewan et al., ²⁰¹⁷). This again highlights the goal hierarchy. Being sedentary is lower on the goal hierarchy as people try to achieve a higher‐order goal (e.g., socializing), and being sedentary is instrumental in serving this goal. Lower‐order goals are more often automatically triggered, while the higher‐order goals are more reflectively triggered (Marien et al., 2018; Ten Broeke et al., ²⁰²²). To better understand how these rewards relate to the automatic nature of SB, the second part of the analysis focused on the relationship between the activity‐related reward value and habit strength. A positive association was found between total reward value and habit strength, suggesting that activities that score high on all clusters, and are therefore more rewarding, are associated with higher habit strength. Interestingly, pleasure, as an individual cluster, also showed a significant positive association with habit strength. Although it was initially hypothesized that only activities scoring high across all clusters would be associated with stronger habits, the findings indicate that the more enjoyable an activity is, the stronger the habit of performing it while sitting. This may reflect that pleasure acts as an immediate reinforcer that strengthens cue‐behavior associations, thereby facilitating the development of automatic responding. Pleasure might relate more closely to intrinsic motivation (i.e., engaging in an activity for its inherent enjoyment), whereas clusters such as helping others or acting according to personal values represent autonomous, but not purely intrinsic, forms of motivation (Judah et al., 2018; Lally & Gardner, ²⁰¹³). Different forms of motivation may contribute in distinct ways to habitual performance of SB (Fremling et al., 2025). Future research should therefore examine how different types of motivation relate to habit strength. This may help to disentangle how goal‐directed and habitual processes interact in SB.

Furthermore, the item on rest and pain relief was not associated with habit strength, although such an association could have been expected (Cheval & Boisgontier, 2021; Maltagliati, ten Broeke, et al., ²⁰²⁴). The study of Maltagliati, ten Broeke, et al. (2024) indicated that the attractiveness of SB is partly due to the low effort required for sitting, suggesting that low effort is rewarding and could potentially impact habit strength. Moreover, research indicates that pain is often a reason for sitting down (Chastin et al., 2014). One possible explanation for the absence of an association could lie in the measure of habit strength, which referred specifically to the activity participants engaged in at the time of the prompt. It may be that these activities were performed while seated due to momentary needs like rest or pain relief, but are not consistently carried out in a sedentary posture under other circumstances. In that sense, pain or fatigue may act as situational triggers for SB without contributing to habit strength (Chastin et al., 2014). Future research should explore this distinction further, ideally using more specific measures of both reward and habit strength.

The findings from this study contribute to the understanding of SB in the field of health psychology. Beyond confirming that older adults spend time sitting while engaging in rewarding activities, the findings provide valuable insights into how SB may be so resistant to change. The observed association between the rewarding nature of sedentary activities and their automaticity suggests that pleasurable or rewarding activities might contribute to reinforcing sedentary patterns over time. This adds to current dual‐process models of SB by providing evidence of how reward value and context stability reinforce habit strength in real‐world contexts. From a health psychology perspective, these insights help to explain why intentions to sit less may often fail to translate into behavior change: automatic responses shaped by rewarding sedentary activities may override reflective intentions. Understanding these potential mechanisms represents an important step towards designing interventions that address both reflective and automatic processes underlying SB. Most importantly, the findings suggest that SB is habit‐driven, reflecting the combined effect of several sedentary activities that are automatically performed in a seated position across different contexts. While each activity may have distinct cues and motivations, their recurrence throughout the day contributes to the overall pattern of habit‐driven SB observed in older adults. Interventions should be particularly mindful of activities that are pleasurable and rewarding, and that typically occur in stable social contexts, as these are most strongly associated with habit strength. Once such behaviors are identified, habit substitution strategies could be used to disrupt the automatic seated position (Gardner, Rebar, et al., 2024, ²⁰²¹). To date, only one study has used habit substitution strategies to reduce SB in older adults. A booklet listing 16 tips for substituting SB with light physical activity was found to be feasible and acceptable, and participants reported that the recommended behavior became more automatic, but negligible changes were found in SB and SB habit strength (Matei et al., 2015; White et al., ²⁰¹⁷). However, this study did not take reward value into account. The challenge for future interventions is to identify substitutes that are at least as pleasurable and rewarding as the sedentary activities they replace, while also addressing the perceived effort associated with changing posture (Maltagliati, Sarrazin, et al., 2024; Maltagliati, ten Broeke, et al., ²⁰²⁴). Crucially, what is experienced as pleasurable or rewarding is likely to vary between individuals. As such, qualitative research may play an important role in identifying which non‐sedentary activities older adults personally experience as pleasurable, rewarding, and feasible in daily life. These insights could inform more tailored habit substitution strategies, such as maintaining the same activity and goals but promoting a different posture (e.g., reading the newspaper while standing), pursuing the same goal through an alternative, less sedentary activity (e.g., socializing while walking rather than sitting), or retaining the activity itself while introducing more frequent interruptions of sitting time. In this way, interventions may address both the instigation and execution of prolonged sedentary activities, while acknowledging the rewarding value they serve in older adults' daily lives.

Strengths and limitations

The study has several strengths. A first key strength lies in the use of sensor‐based EMA, which enables the triggering of surveys in real time during prolonged sedentary episodes, allowing for the assessment of context stability, reward value, and habit strength of prolonged SB in older adults. The 14‐day monitoring period enabled the capture of a wide range of sedentary activities and contextual variations, thereby enhancing the ecological validity of the findings. Furthermore, context stability was analyzed at both the individual level and at the behavioral level, which provides a more nuanced view of context stability compared to previous research (Maher et al., 2021). Finally, the study specifically focused on the types of activities participants engage in while being sedentary, acknowledging that SB is often instrumental to higher‐order goals (Gardner et al., 2019; Ten Broeke et al., ²⁰²²). In doing so, it addresses a critical gap in previous research, which has often overlooked the functional and purposeful nature of sedentary activities (Maher et al., 2021; Maltagliati, ten Broeke, et al., ²⁰²⁴).

Nonetheless, several limitations should be considered. First, the sample was recruited via convenience sampling, which means that more vulnerable older adults may not have been reached or included in the study, limiting the generalizability of the results. Second, the study did not fully account for the role of internal states such as momentary mood, which could influence habit strength for prolonged SB. Moreover, objects in the physical context might also facilitate prolonged SB, regardless of the physical location (e.g., chair or sofa). Third, concerns could be raised regarding the validity of self‐reporting habits (Hagger et al., 2023). However, the SRBAI does not require insight into habit, but requires individuals to report on the experience of “symptoms” of habit, such as automaticity (Orbell & Verplanken, 2015). Furthermore, while a single‐item assessment is not ideal for capturing a complex concept, Gardner, Lally, and Rebar (2024) have shown that single‐item measurements on habit strength produce highly similar results to a multi‐item version and are particularly suited to intensive longitudinal designs such as EMA. Nevertheless, using self‐reports in EMA introduces the risk of response biases, which can potentially limit the accuracy of the data. One specific consideration in this study concerns the interpretation of the single‐item measure of habit strength. Although the item was adapted from the SRBAI and intended to capture automaticity, the item did not explicitly distinguish between habit instigation or execution (Gardner et al., 2016). Given the use of the verb “perform” and the fact that prompts were delivered after at least 30 minutes of prolonged SB, the item was intended to primarily reflect the habitual execution of activities performed while sitting rather than their habitual instigation. However, the wording remains somewhat ambiguous. During the first home visit, researchers clarified the intended meaning of the item with participants, but variation in interpretation over the two‐week study period cannot be ruled out. Finally, the reward value of SB was assessed using a modified version of the EMAS (Cruyt et al., 2023; Eakman et al., ²⁰¹⁰). Although the EMAS was developed to capture engagement in meaningful activities, it provides insight into activity‐related reward value, as activities experienced as meaningful are typically accompanied by subjectively rewarding experiences such as enjoyment, satisfaction, or relaxation (Eakman & Eklund, 2011). At the same time, this adapted measure captures only a limited part of the broader reward construct and does not assess all potential rewards associated with SB, such as effort minimization (Maltagliati, ten Broeke, et al., 2024). Consequently, the present findings reflect a partial and activity‐focused capture of reward value. Future research should therefore develop and validate instruments that more comprehensively assess the reward value of sedentary behavior, including both rewards related to the sedentary posture itself and rewards derived from higher‐order goal pursuit during sedentary activities. Nevertheless, this study provides a first exploratory step toward understanding how activity‐related reward value may contribute to the automatic performance of prolonged sedentary activities in daily life. Future research should build on these findings and look further into how SB is driven by habit.

CONCLUSION

This study offers important insights into the role of context stability, reward value, and habit strength of prolonged, uninterrupted sedentary activities in older adults. The findings highlight that SB is habit‐driven, reflecting the accumulation of multiple activity‐specific habits that are automatically performed in a seated position, and that sedentary activities performed by older adults are often rewarding to them. Sedentary activities that are more rewarding or pleasurable, and those performed in stable social contexts, are more likely to be performed automatically in a seated position. Therefore, to effectively disrupt habit‐driven SB, it is essential to identify these specific activities and explore the most feasible substitutes that are at least equally rewarding and pleasurable but involve a different posture or include more frequent interruptions in SB.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflict of interest.

ETHICS STATEMENT

The study was approved by the Ethical Committee of Ghent University Hospital (ONZ‐2023‐0531).

Supporting information

File S1 CREMAS checklist.

File S2 EMA questionnaire.

File S3 Descriptive statistics of EMA compliance across three minimum response thresholds (50%, 60%, and 80%).

Van de Velde, L. , Gardner, B. , Van Dyck, D. , Větrovský, T. , & Compernolle, S. (2026). Exploring habit strength, reward value, and context stability of sedentary activities in older adults: An ecological momentary assessment study. Applied Psychology: Health and Well‐Being, 18(2), e70144. 10.1111/aphw.70144

DATA AVAILABILITY STATEMENT

Data are publicly available at OSF (https://osf.io/tp9hz/).