Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Apr 1.
Published in final edited form as: J Behav Med. 2023 Aug 29;47(2):197–206. doi: 10.1007/s10865-023-00444-4

Wearable Device Adherence among Insufficiently-Active Young Adults is Independent of Identity and Motivation for Physical Activity

Jingchuan Wu 1, Jenny L Olson 2, Deborah Brunke-Reese 3, Constantino M Lagoa 4, David E Conroy 5,*
PMCID: PMC10902189  NIHMSID: NIHMS1927322  PMID: 37642938

Abstract

Wearable devices are increasingly being integrated to improve prevention, chronic disease management and rehabilitation. Inferences about individual differences in device-measured physical activity depends on devices being worn long enough to obtain representative samples of behavior. Little is known about how psychological factors are associated with device wear time adherence. This study evaluated associations between identity, behavioral regulations, and device wear adherence during an ambulatory monitoring period. Young adults who reported insufficient physical activity (N=271) were recruited for two studies before and after the SARS-COVID-19 pandemic declaration. Participants completed a baseline assessment and wore an Actigraph GT3X+ accelerometer on their waist for seven consecutive days. Multiple linear regression indicated that wear time was positively associated with age, negatively associated with integrated regulation for physical activity, and greater after (versus before) the pandemic declaration. Overall, the model accounted for limited variance in device wear time. Exercise identity and exercise motivation were not associated with young adults’ adherence to wearing the physical activity monitors. Researchers and clinicians can use wearable devices with young adults with minimal concern about systematic motivational biases impacting adherence to device wear.

Keywords: Data Quality, Behavior and Behavior Mechanisms, Exercise, Digital Technology

Wearable devices for ambulatory monitoring are increasingly used to support clinical care and increase the rigor of physical activity research. In clinical contexts, remote monitoring via wearable devices has potential to improve risk identification and prevention, chronic disease management, and rehabilitation (Lu et al. 2020). Device-based measures of physical activity sense momentary acceleration at the point of attachment whereas self-reports are often influenced by additional factors such as context, purpose, and perceptions (Troiano et al. 2014). In research contexts, wearable devices address a key limitation of self-report measures because they are not vulnerable to recall and response bias (Kaewkannate and Kim 2016). Yet wearable devices introduce unique threats to the validity of physical activity assessments that can limit their value in clinical or research contexts. For example, non-compliance with device wear instructions is a fundamental adherence challenge that can reduce data quality and introduce systematic sampling bias. Wear time is a particular challenge for waist-worn devices (Troiano et al. 2014). Research on the factors associated with adherence to device wear has spanned the lifespan and focused on demographic and health correlates of device wear. Less is known about psychological correlates of device wear in young adults, a high-priority group for physical activity promotion. Understanding factors that impact adherence to wearable device wear can help to improve clinical applications of remote monitoring and increase the efficiency and statistical power of future research device-based physical activity monitoring by identifying individuals who may need additional support to provide high-quality data. This study used secondary data analyses to evaluate whether young adults’ adherence to device wear was associated with motivational characteristics after adjusting for demographic and anthropometric characteristics.

Device Wear Time and Data Quality

Conventions for quality control with accelerometers require that a device be worn for at least 10 hours for the day to be counted, and for at least four days for reliable assessment of individual differences in physical activity (Aadland and Ylvisåker 2015; Donaldson et al. 2016; Migueles et al. 2017; Trost et al. 2005). Based on this standard, data loss in many cross-sectional studies is approximately 20% (Belton et al. 2013; Howie and Straker 2016). Data loss may be compounded in longitudinal studies that require repeated ambulatory monitoring periods with device-based measures of physical activity (Lee et al. 2013).

Quality control failures due to device non-wear reduce the clinical value of wearable devices in health care and drive up the cost of doing rigorous research with device-based physical activity measures (Cho et al. 2021). If patients do not adhere to prescriptions to wear a device, clinicians who ask patients to use these devices will be left with incomplete data to inform their decision-making and patients may experience poor health outcomes (Huang et al. 2022). If research participants fail to reach quality control standards, sample sizes must be increased to compensate for reduced statistical power due to data loss. This data loss then requires a re-allocation of resources (and delays) to collect additional data (Huisman et al. 1998). Second, if data are not missing at random, selection factors can bias inferences (Graham, 2009). It may be possible to reduce bias via statistical adjustments if the missingness mechanism is known but that is not always possible. Even if statistical controls can be used to reduce bias, such adjustments increase model complexity, and results may not generalize to the intended population (Kang 2013). Understanding the factors that are associated with adherence to device wear can inform research design and support efficient resource allocation.

Correlates of Wear Time

Studies investigating factors that influence device wear adherence have focused on adults’ demographic and health characteristics. In the Whitehall II study, participants who opted to wear a wrist-worn activity monitor were more likely to be men, report more physical activity, and be in better health (Hassani et al. 2015). The proportion of participants who declined to wear the monitors was relatively small (8.3%). Although participants do not appear to self-select out of activity monitoring, the amount of device wear may introduce other selection biases in sampling. For example, participants in the 2003–2004 NHANES cohort who met data quality standards for device wear (i.e., four days with 10+ hours/day) were older, better educated, in better cardiometabolic health (i.e., BMI, HDL-cholesterol, C-reactive protein, plasma glucose levels), Non-Hispanic White, married, and not smokers (Loprinzi et al. 2013). Similar results were found in other analyses of adult samples from England (Roth and Mindell 2011) and Hong Kong (Lee et al. 2013). For school-age children or adolescents, device wear adherence was associated with reduced feelings of embarrassment and perceived risk of being bullied (McCann et al. 2016). Although this research spans the lifespan from adolescence to older adulthood, little is known about the correlates of adherence to device wear during emerging and young adulthood.

This developmental period plays an important role for developing healthy habits and preserving cardiovascular health (Liu et al. 2012; Nelson et al. 2008). In addition to the demographic and health factors identified in research with adolescents, adults, and older adults, two psychological factors may be particularly relevant for device wear in young adults: identity and motivation.

Exercise identity represents how strongly an individual identifies with a particular role based on prior experience and is part of a multi-dimensional self-concept (Rhodes et al. 2016). One of the most common exercise identity measures assesses both perceptions of one’s role as an exerciser and beliefs that exercise is an important part of one’s life (Wilson and Muon 2008). Although conceptually separable, role identity and exercise beliefs are strongly associated. A recent meta-analysis indicated that exercise identity has a medium-to-strong association with physical activity (Rhodes et al. 2016). People also generally strive to present themselves in ways to construct an identity-consistent image in the eyes of others which can lead identity-related motivational processes to spill over and influence other behaviors (Leary and Kowalski 1990). A person who identifies as an exerciser may experience a wearable activity monitor as identity-consistent whereas a person who does not identify as an exerciser may experience a wearable activity monitor as identity-inconsistent. The dissonance experienced from self-presenting in an identity-inconsistent manner may interfere with adherence to prescriptions to wear an activity monitor. Building on the proposal that identity can spillover to influence self-presentation, we hypothesized that young adults who identify more strongly as an exerciser (either by role or beliefs) would have greater wear time than those who identify less strongly as an exerciser.

Motivation is another psychological process that could influence device wear. Participants whose motivation for physical activity is regulated by external rewards or punishments may be especially likely to comply with instructions about device wear because they are oriented to comply with external influences on physical activity-related behavior (Cialdini and Goldstein 2004). In self-determination theory, this motivation is referred to as external regulation, and it is the most extreme form of extrinsic motivation (Ryan and Deci 2019). External regulation for physical activity has mixed associations with physical activity, typically null but sometimes associated with reduced activity (Teixeira et al. 2012) but its associations with device wear adherence have not been investigated previously. Although self-determination theory would predict stronger associations between more autonomous behavioral regulations and physical activity, the phenomenon of interest in this study is device wear time rather than physical activity. Wear time may be influenced by sensitivity to and compliance with external incentives such as the researcher’s instructions. We hypothesized that young adults with strong external regulation for physical activity would have greater wear time than those with weaker external regulation for physical activity. Based on our specific proposal that compliance may motivate wear time, we limit our hypothesis to compliance-driven wear time produced by external regulation and will treat associations between other behavioral regulations and wear time as exploratory analyses.

The Present Study

This study aimed to evaluate exercise identity and externally regulated physical activity motivation as correlates of device wear adherence in young adults. We hypothesized that adherence would be greatest in participants with stronger identities as exercisers and stronger external regulation. This study is a secondary data analysis of screening data from two physical activity intervention development projects targeting healthy but insufficiently-active young adults (Conroy et al. 2023; Hojjatinia et al. 2021). Screening for one study was completed prior to the COVID-19 pandemic declaration and screening for the other study was completed following the pandemic declaration.

Methods

Participants

Insufficiently-active young adults were recruited for two studies (pre- pandemic and post-pandemic) using print flyers in the community and internet advertisements. Eligibility criteria and baseline protocols were similar in the two studies, so data were combined to create the analytic sample. Participants were eligible if they were aged 18–29 years, capable of reading, speaking, and understanding English, and giving informed consent, and free of visual impairment that would interfere with using a smartphone. Participants were excluded if they reported regular physical activity, participated in organized programs with mandated physical activity (e.g., varsity sports, Reserve Officers Training Corps), had contraindications to physical activity on the Physical Activity Readiness Questionnaire (Thomas et al. 1992), required an assistive device for mobility, or had any other condition that could limit participation in moderate-intensity physical activity, were pregnant or planning to become pregnant within the next four months, or had a prior diagnosis of cancer, cardiovascular disease, diabetes or metabolic syndrome. Regular physical activity was defined as engaging in 90 minutes/week of moderate-to-vigorous intensity physical activity (pre-pandemic declaration study) or 150 minutes/week of moderate-to-vigorous intensity physical activity (post-pandemic declaration study).

As a secondary analysis, the sample size was not planned based on the research question driving this study. Post hoc analysis with G*Power v.3.19.6. revealed that a sample of 271 participants afforded sufficient power to detect multiple regression model effects as small as R2 = .03 (with 16 predictors) (Faul et al. 2009). Models accounting for less than 3% of the variance in device wear time would indicate trivial bias in the data so this sample was sufficient to draw meaningful conclusions about the research question.

Measures

Ambulatory physical activity was assessed with the ActiGraph wGT3X-BT accelerometer (ActiGraph, Pensacola, FL). Participants were asked to wear the device on their waist over the midline of the dominant thigh for at least 10 waking hours/day over a 7-day monitoring period. Participants removed the device while bathing, swimming, or sleeping. The device measured acceleration in three axes at a sampling rate of 30 Hz. Devices were initialized and data were downloaded using ActiLife 6 software (ActiGraph, Pensacola, FL). The proprietary Troiano 2007 algorithm was used to identify device wear time. Participants maintained a wear log to record device removals; these logs were entered into ActiLife to filter data outside of waking hours. Three wear time scores were calculated to quantify data quality for each participant: (a) number of days with 10+ hours of wear time, (b) average daily wear time across all possible days (min/day), and (c) average daily wear time on days with 10+ hours of wear time (min/day).

Exercise identity was assessed with the 9-item Exercise Identity Scale (EIS; Anderson and Cychosz 1994). Participants responded to each item on a scale ranging from 1 (strongly disagree) to 7 (strongly agree). Scores were calculated for exercise beliefs (6 items) and exercise role identity (3 items) based on prior studies of the structure of responses (Anderson et al. 1998; Anderson and Cychosz 1994; Vlachopoulos et al. 2011; Wilson and Muon 2008). In prior work, responses to the Exercise Identity Scale (EIS) have demonstrated a two dimensional structure internal consistency estimates exceeding .80, and positive associations with exercise participation (Anderson and Cychosz 1994; Wilson and Muon 2008).

Contextual motivation for exercise was assessed using the 23-item Behavioral Regulation in Exercise Questionnaire-2 (BREQ-2) and a supplementary four-item integrated motivation scale (Markland & Tobin, 2004; Wilson et al., 2006). Participants rated each item on a scale ranging from 0 (Not true for me) to 4 (Very true for me). Scores were calculated for amotivation (lack of motivation or intention to act; 4 items), four forms of extrinsic motivation (external regulation [performing an activity to achieve an outcome separate from the activity itself; 4 items], introjected regulation [actions taken to avoid guilt or boost ego; 3 items], identified regulation [behaviors carried out due to their recognized personal value; 4 items], integrated regulation [actions fully assimilated with one’s self-identity and values; 4 items]), and intrinsic motivation (engaging in an activity for its own sake and the enjoyment it provides; 4 items). If participants did not respond to two or more questions on a scale, the scale score was set to missing; otherwise, available responses were averaged to produce each scale score. Psychometric properties of these scores have been established in prior work evaluating structural validity, discriminant validity, internal consistency reliability, and test-retest reliability (D’Abundo et al. 2014; Frederix et al. 2015; Liu et al. 2020; Markland and Tobin 2004).

Procedure

Pre-pandemic declaration study.

Screening data for the study that preceded the SARS-COVID-19 pandemic declaration were collected between April 4, 2019 and January 22, 2020 (Hojjatinia et al. 2021). Interested participants completed a telephone screening to assess self-reported physical activity. Provisionally-eligibility participants were scheduled for a face-to-face meeting with the researcher. Participants provided written informed consent and the researcher measured the participant’s height and weight (in duplicate) using a wall-mounted stadiometer and a digital scale. Participants then completed a battery of questionnaires on a web-based survey platform (Qualtrics). Next, the researcher trained the participant how to wear the Actigraph device to assess eligibility for the parent study, as well as how to complete a daily paper-and-pencil log to record device removals. Participants were informed that they would be eligible for the intervention component of the study if they accrued less than 90 minutes of moderate-to-vigorous intensity physical activity (45% qualified). The researcher scheduled a follow-up visit for the participant to return the device after the one-week ambulatory monitoring period. Participants received $25 for completing this initial part of the study.

Post-pandemic declaration study.

Screening data for the study that followed the SARS-COVID-19 pandemic declaration were collected between July 8, 2020 and April 30, 2021 (Conroy et al. 2023). Interested participants completed a telephone screening to assess self-reported physical activity and, if provisionally-eligible, provided contact information. Interactions with participants took place over telephone or video due to pandemic-related prohibitions on face-to-face research interactions. The researcher mailed the participant an Actigraph device and a pre-paid, addressed envelope to return the device at the end of the monitoring period. The video meeting was used to share a link to the web-based surveys on Qualtrics. Following questionnaire completion during the meeting, the researcher trained the participant to wear the Actigraph device to assess eligibility for the parent study as well as how to complete a daily paper-and-pencil log to record device removals. Participants were informed that they would be eligible for the intervention component of the study if they accrued less than 150 minutes of moderate-to-vigorous intensity physical activity. At the end of the one-week ambulatory monitoring period, participants returned the activity monitor they wore in the pre-paid, addressed enveloped provided in the initial mailing. Participants received $10 for completing this initial part of the study.

Data Analysis

R software was used for data analysis (R version 4.1.1) (R Core Team 2021). Descriptive statistics were calculated for the individual study samples and the combined analytic sample. Normality was evaluated based on skew (< 2) and kurtosis (< 4) (Fjeldsoe et al. 2016). Data that were not normally distributed were transformed using the BestNormalize package (Peterson and Cavanaugh 2020). The package compared square root, power law (Box-Cox), and log transformations to determine the optimal transformation for the three outcome variables used to characterize adherence to device wear instructions: days with 10+ hours/day of wear time (Qualified Wear Days), average wear time across all days (min//day), and average wear time on valid wear days (min/day).

Differences between the samples were evaluated using independent samples t-tests and chi-square tests. Internal consistency of responses to multi-item scales was estimated using Cronbach’s (1951) alpha coefficient. Linear regression models were estimated to regress each of the three adherence variables on the hypothesized psychological predictors. Given evidence that the pandemic influenced physical activity (Pépin et al. 2020; Tison et al. 2020), the timing of data collection was included as a statistical control when testing hypotheses about wear time. Analyses adjusted for demographic (i.e., sex, age, education, work status) and anthropometric (i.e., body mass index) factors as well as other intrinsic and extrinsic motivational factors.

Results

Table 1 summarizes the demographic characteristics of the pre-pandemic (n = 174) and post-pandemic (n = 97) samples as well as the combined sample (N = 271). Overall, the mean (±SD) age and body mass index of the sample were 23.0 (±3.2) years and 27.27 (±13.65) kg/m2, respectively. The samples did not differ in age (t[174.75] = 1.71, p = .09, Cohen’s d = 0.23) or body mass index (t[105.80] = 0.78, p = .44, d = 0.13). The sample recruited after the pandemic declaration was more educated (χ2[2] = 64, p = .036) and more likely to be employed (χ2[1] = 9.47, p = .002), compared to the sample recruited before the pandemic.

Table 1.

Demographic characteristics of the sample

Pre-Pandemic Declaration Post-Pandemic Declaration Overall Sample
Characteristic n (%) n (%) χ2 n (%)
Sex
  Female 119 (68.4) 67 (69.1) 0.00 186 (68.6)
  Male 55 (31.6) 30 (30.9) 85 (31.4)
Education
  Completed college or above 83 (47.7) 57 (58.8) 6.64* 140 (51.7)
  Some college or Associate’s degree 68 (39.1) 23 (23.7) 91 (33.6)
  High school or less 23 (13.2) 17 (17.5) 40 (14.8)
Race
  Minority 66 (38.2) 26 (26.8) 3.07 92 (34.1)
  White 107 (61.8) 71 (73.2) 178 (65.9)
Work Status
  Employed 35 (20.1) 37 (38.1) 9.47** 72 (26.6)
  Unemployed 139 (79.9) 60 (61.9) 199 (73.4)
Body mass classification
  Normal weight or underweight 85 (48.9) 50 (52.6) 0.22 135 (50.2)
  Overweight or obese 89 (51.1) 45 (47.4) 134 (49.8)
Open in a new tab

Note.

*

p < .05

**

p < .01.

Table 2 summarizes descriptive statistics for adherence to device wear, identity, and motivation variables in each sample and the combined sample. Participants wore the devices on almost all days during the monitoring period (95%) and met the 10-hour minimum wear time requirement for data quality on most days (81%). The distribution of the number of days that met data quality standards was skewed so it was transformed using a square root transformation. For participants who met the wear time requirements, the average MVPA duration in pre-pandemic sample was 25.85 (n=140, SD=17.48) minutes; and the average MVPA duration in post-pandemic sample was 19.83 (n=90, SD=16.43) minutes. The average MVPA in the combined sample was 23.49 (n=230, SD=17.36) minutes.

Table 2.

Behavioral & motivational descriptive statistic (raw values, original scale)

Pre-Pandemic Sample Post-Pandemic Sample t-test d Complete Analytic Sample
Variables n M (SD) α n M (SD) α N M (SD) α
Device Wear Compliance
 Days with 10+ hours of wear time 174 5.37 (1.73) -- 97 6.15 (1.36) -- 4.09** 0.49 271 5.65 (1.65) --
 Average wear time across 7 days (min/day) 174 615.75 (85.99) -- 97 604.32 (73.38) -- −1.15 −0.14 271 611.65 (81.75) --
 Average wear time on valid wear days (min/day) 173 668.30 (144.53) -- 96 626.89 (106.83) -- −2.68* −0.31 269 653.52 (133.59) --
Exercise Role 174 3.53 (1.52) .82 97 3.53 (1.42) .90 0.01 < 0.01 271 3.53 (1.48) .85
Exercise Belief 174 2.40 (1.33) .85 97 2.85 (1.42) .90 2.54* 0.33 271 2.56 (1.38) .86
Amotivation 174 0.38 (0.57) .70 94 0.36 (0.60) .88 −0.29 −0.04 268 0.37 (0.58) .76
External Regulation 173 0.93 (0.84) .78 96 1.03 (0.93) .80 0.84 0.11 269 0.96 (0.87) .79
Introjected Regulation 174 1.70 (1.08) .77 95 2.05 (1.17) .82 2.46* 0.32 269 1.82 (1.12) .79
Identified Regulation 174 2.17 (0.75) .67 97 2.54 (0.83) .72 3.72** 0.49 271 2.30 (0.80) .70
Integrated Regulation 173 0.91 (0.79) .79 94 1.48 (0.98) .84 4.80** 0.66 267 1.11 (0.90) .83
Intrinsic Regulation 172 1.88 (0.85) .86 97 2.28 (0.99) .91 3.36** 0.44 269 2.02 (0.92) .89
Open in a new tab

Note.

*

p < .05

**

p < .01. The possible range for exercise role and belief scores was 1 to 7. The possible range of behavioral regulation scores from amotivation to intrinsic motivation was 0 to 4.

Table 2 also shows that all scale scores exhibited acceptable internal consistency. Compared to the pre-pandemic sample, the sample collected after the pandemic declaration accumulated significantly more valid days of device data (d = 0.49) but wore the device for slightly less time on those days (d = −0.31). The two samples did not differ significantly in average wear time across the seven days of data collection, nor in the number of days with any device wear time (p > .05). Participants in the sample collected after the pandemic declaration reported significantly stronger exercise beliefs (d = 0.33) than participants in the pre-pandemic sample but did not differ in exercise roles. Participants in the sample collected after the pandemic declaration reported significantly stronger introjected regulation (d = 0.32), identified regulation (d = 0.49), integrated regulation (d = 0.66), and intrinsic motivation (d = 0.33) than the pre-pandemic sample, but did not differ in external regulation or amotivation.

Table 3 summarizes correlations between the three data quality measures from the activity monitors and the identity and motivation variables. None of the data quality variables had statistically significant bivariate associations with any of the identity or motivation variables. The two identity variables had a strong positive association (but not strong enough to warrant combining the scores). The motivation scores exhibited the expected pattern with the strongest correlations between scores adjacent to each other on a continuum of self-determined motivation. The two identity scores exhibited strong positive associations with more self-determined behavioral regulations (beginning with introjected regulations).

Table 3.

Correlations between behavioral & motivational variables (combined data)

Variable 1 2 3 4 5 6 7 8 9 10
1. Qualified Wear Days
2. Wear time (7 days) .57**
3. Wear time (valid wear days) .24** .53**
4.Exercise Role −.08 −.05 −.05
5.Exercise Belief −.04 −.02 −.10 .61**
6. Amotivation −.12 −.03 −.00 −.23** −.11
7. External Regulation −.10 −.00 .08 .05 .08 .28**
8. Introjected Regulation −.03 .01 .02 .43** .22** −.09 .27**
9. Identified Regulation .04 .04 −.04 .55** .45** −.27** .04 .53**
10. Integrated Regulation −.09 −.04 −.09 .57** .56** −.13* .14* .39** .68**
11. Intrinsic Motivation .09 .01 −.07 .47** .49** −.25** −.12 .25** .61** .55**
Open in a new tab

Note.

*

p < .05

**

p < .01.

Table 4 presents coefficients from the model regressing the number of days that met data quality standards (i.e., 10+ hours/day wear time; square-root transformed) on the identity, motivation, and control variables. The model was statistically significant (F[15, 249] = 2.50, p < .001, R2 = .13). Contrary to hypotheses, neither identity nor external regulation for physical activity was associated with the number of days meeting data quality standards. Participants who accumulated more days at or above the wear time criterion reported less integrated regulation, were older, and were more likely to be in the post-pandemic dataset.

Table 4.

Regression coefficients from the model of the number of valid wear days

Predictor variable b SE t(248) p
Intercept −2.18* 0.71 −3.06 0.00
Amotivation −0.12 0.11 −1.05 0.29
External Regulation 0.01 0.08 0.08 0.94
Introject Regulation −0.06 0.07 −0.89 0.37
Identified Regulation 0.16 0.12 1.27 0.20
Intrinsic Regulation 0.16 0.09 1.83 0.07
Integrated Regulation −0.30** 0.11 −2.84 0.00
Exercise Role −0.05 0.06 −0.88 0.38
Exercise Belief 0.03 0.06 0.49 0.63
BMI 0.01 0.01 1.17 0.24
Female (vs Male) 0.06 0.13 0.49 0.63
Age 0.07* 0.03 2.22 0.03
Education
 Some college vs High school or less −0.01 0.19 −0.03 0.97
 College graduate vs High school or less −0.10 0.23 −0.43 0.67
Unemployed (vs part- or full-time employed) 0.16 0.14 1.13 0.26
Post-pandemic declaration (vs pre-pandemic declaration) 0.47*** 0.14 3.38 0.00
Open in a new tab
*

p < .05

**

p < .01

***

p < .001

Neither model of daily wear time was statistically significant (average wear time across all days: F[15,248] = 1.20, p = .26, R2 = .07; average wear time on valid wear days: F[15,246] = 1.15, p = .31 R2 = .01).

Based on the mean differences between samples in identity and motivation, sensitivity analyses were performed to evaluate whether coefficients from each model varied between the pre- and post-pandemic samples. Only one moderation term was statistically significant. It implied a disordinal interaction such that exercise role and average wear time on valid days had a positive association for participants in the pre-pandemic declaration sample but a negative association for participants in the post-pandemic declaration sample. Overall, conclusions generally appear to be robust to mean differences between samples in identity and motivation.

Discussion

This study investigated whether demographic or motivational variables were associated with adherence to wearable device wear in young adults. Results indicated very limited and weak relations between individual differences in demographic or motivation characteristics and device wear time. Systematic biases linked with demographic characteristics and motivation for physical activity are likely to have minimal influence on adherence during ambulatory assessments using accelerometers with young adults. This study makes three primary contributions to the literature.

First, neither identity nor motivation appear to increase device wear adherence. Although human behavior is often the product of interactions between a person and situation, the situational influences in a research context (e.g., researcher instructions) appear to overpower the influence of individual differences that might activate identity or motivation influences (Mischel 1973). The only significant association between motivation and device wear time indicated that people with stronger integrated regulation for physical activity accumulated fewer valid days. Integrated regulation reflects a behavior being aligned with a person’s identity but the observed association was opposite from what the identity hypothesis predicted (Leary and Kowalski 1990; Wierts et al. 2022). Participants with stronger integrated regulation for physical activity may have valued the monitoring goals less than those with weaker integrated regulation, leading to fewer days with 10+ hours of wear time. Integrated regulation tends to be weak among insufficiently-active people so, in the context of physical activity promotion with this population, this threat to device wear adherence may be minimal (Teixeira et al. 2012). This association was not hypothesized and should be tested in an independent sample before drawing strong conclusions.

A second contribution of this study was the finding that young adults’ demographic characteristics were generally not associated with adherence to device wear. For the most part, neither biological (i.e., sex, BMI) nor social (i.e., education, employment) characteristics were associated with any wear time variables. This finding contrasted with prior work across the adult lifespan and in older adults, suggesting that young adults may be less vulnerable to these biological and social threats to data quality compared to other age groups (Lee et al. 2013; Loprinzi et al. 2013; Roth and Mindell 2011). An exception to this conclusion may be age: older participants accumulated more days with 10+ hours of wear time than younger participants. Prior findings across the adult lifespan and in older adults have linked age with higher levels of weekly data quality (Loprinzi et al. 2013; Roth and Mindell 2011), so this study extends knowledge of age-related association to young adulthood and to daily data quality. The consistency of this effect across the lifespan is noteworthy given that restricted age range should attenuate associations. Regardless, given the high level of device wear adherence overall (i.e., 95% of participants recording at least 4 days with 10+ hours/day), there is minimal concern about age-related bias in accelerometer data in young adults despite the association between age and the number of days meeting data quality standards.

Finally, participants wore the devices for at least 10 hours on more days following the SARS-COVID-19 pandemic declaration than before. The adverse effects of the stay-at-home orders and other pandemic countermeasures on physical activity are well-documented (Wilke et al. 2021). These findings suggest device wear adherence may be superior following the pandemic declaration compared to before that declaration. These contrasting effects may be driven by a subtle difference in screening criteria (i.e., shifting from 90 to 150 min/day of self-reported MVPA). The increased tolerance for self-reported activity in the post-pandemic sample may correspond with unmeasured motivational differences between the samples that also support device wear time. Alternately, participants may have been exposed to less contextual variation (e.g., due to stay-at-home orders, remote work/school arrangements) so there were fewer opportunities to force device removal. Wearable devices have also been increasingly adopted by consumers over time (Vogels 2020) so increased familiarity and acceptability may contribute to increased adherence.

Limitations of this study warrant attention. First, the cross-sectional nature of the research design prevents causal inferences. This limitation was mitigated by the lack of associations with device wear time, a necessary condition for causal inferences. Second, the study sample was recruited from overtly healthy young adults in Pennsylvania who expressed interest in a physical activity trial. Although more diverse than the US population, conclusions may not generalize to young adults with chronic disease(s), adults living in other regions or those not interested in volunteering for a physical activity trial. The sample was also screened for insufficient physical activity so the ranges of identity and autonomous motivation scores were restricted. This restriction may have attenuated power for detecting associations with wear time and caution is warranted before generalizing conclusions to more active populations with greater variability in identity and autonomous motivation. Third, this study used waist-worn Actigraph devices and findings may not generalize to other research wearable devices (e.g., activPAL), consumer wearable devices (e.g., Fitbit, Apple Watch) or wear locations (e.g., thigh, wrist). This study focused on motivation and identity involving physical activity; conclusions may differ for motivation or identity involving sedentary behavior. Wear time was high in these data and findings may not generalize to samples with less consistent wear time. Additionally, other unmeasured motivational mechanisms may be associated with wear time variability (e.g., motivation for research participation). Finally, analyses did not adjust for variability in sleep, mental health, or work factors.

Conclusions

Wearable devices for physical activity monitoring can be used to improve health outcomes and complement self-report measures of physical activity but the validity of inferences drawn from device-based measures is dependent on participants’ adherence to wearing the device. This study indicated that device wear adherence was largely independent of young adults’ demographic characteristics, identity, and motivation for physical activity. Researchers and clinicians can have confidence that identity and motivation motives are unlikely to influence the validity of inferences about physical activity from wearable devices.

Funding:

This project is funded by the National Heart, Lung, and Blood Institute of the National Institutes of Health (NIH) through Grant R01 HL142732, by the National Center for Advancing Translational Sciences of the NIH through Grants UL1 TR000127 and UL1 TR002014, and by the National Institute on Aging of the NIH through Grants T32 AG049676. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Footnotes

Competing interests: The authors declare no competing interests with the published work.

Ethics approval: All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional review board (Study#9455, Study#10218) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Consent to participate: Informed consent was obtained from all individual participants included in the study.

Consent for publication: Not applicable

Code availability: Analytic code is available from the corresponding author upon request.

Contributor Information

Jingchuan Wu, Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA.

Jenny L. Olson, Department of Public Health Sciences, The Pennsylvania State University, Hershey, PA, USA

Deborah Brunke-Reese, Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA.

Constantino M. Lagoa, School of Electrical Engineering & Computer Science, The Pennsylvania State University, University Park, PA, USA

David E. Conroy, Department of Kinesiology, The Pennsylvania State University, University Park, PA, USA.

Availability of data and material:

Data and study materials are available from the corresponding author upon request.

Reference

  1. Aadland E, & Ylvisåker E (2015). Reliability of objectively measured sedentary time and physical activity in adults. PLOS ONE, 10(7), e0133296. 10.1371/journal.pone.0133296 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Anderson DF, & Cychosz CM (1994). Development of an exercise identity scale. Perceptual and Motor Skills, 78(3), 747–751. 10.1177/003151259407800313 [DOI] [PubMed] [Google Scholar]
  3. Anderson DF, Cychosz CM, & Franke WD (1998). Association of exercise identity with measures of exercise commitment and physiological indictors of fitness in a law enforcement cohort. Journal of Sport Behavior, 21(3), 233–241. [Google Scholar]
  4. Cho S, Ensari I, Weng C, Kahn MG, & Natarajan K (2021). Factors affecting the quality of person-generated wearable device data and associated challenges: rapid systematic review. JMIR mHealth and uHealth, 9(3), e20738. 10.2196/20738 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Cialdini RB, & Goldstein NJ (2004). Social influence: Compliance and conformity. Annual Review of Psychology, 55(1), 591–621. 10.1146/annurev.psych.55.090902.142015 [DOI] [PubMed] [Google Scholar]
  6. Conroy DE, Wu J, Lee AM, Brunke-Reese D, & Lagoa CM (2023). Dose-response relations between the frequency of two types of momentary feedback prompts and daily physical activity. Health Psychology, 42(3), 151–160. 10.1037/hea0001271 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cronbach LJ (1951). Coefficient alpha and the internal structure of tests. psychometrika, 16(3), 297–334. [Google Scholar]
  8. D’Abundo ML, Sidman CL, Milroy J, Orsini M, & Fiala K (2014). Construct validity of college students’ responses to the behavioral regulation in exercise questionnaire (BREQ-2). Recreational Sports Journal, 38(1), 40–49. [Google Scholar]
  9. Donaldson SC, Montoye AH, Imboden MT, & Kaminsky LA (2016). Variability of objectively measured sedentary behavior. Medicine & Science in Sports & Exercise, 48(4), 755–761. [DOI] [PubMed] [Google Scholar]
  10. Faul F, Erdfelder E, Buchner A, & Lang A-G (2009). Statistical power analyses using G* Power 3.1: Tests for correlation and regression analyses. Behavior research methods, 41(4), 1149–1160. [DOI] [PubMed] [Google Scholar]
  11. Fjeldsoe BS, Goode AD, Phongsavan P, Bauman A, Maher G, Winkler E, & Eakin EG (2016). Evaluating the maintenance of lifestyle changes in a randomized controlled trial of the ‘Get Healthy, Stay Healthy’program. JMIR mHealth and uHealth, 4(2), e5280. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Frederix I, Hansen D, Coninx K, Vandervoort P, Vandijck D, Hens N, ... & Dendale P (2015). Medium-term effectiveness of a comprehensive internet-based and patient-specific telerehabilitation program with text messaging support for cardiac patients: randomized controlled trial. Journal of medical Internet research, 17(7), e4799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hassani M, Kivimaki M, Elbaz A, Shipley M, Singh-Manoux A, & Sabia S (2014). Non-consent to a wrist-worn accelerometer in older adults: the role of socio-demographic, behavioural and health factors. PLoS One, 9(10), e110816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hojjatinia S, Hojjatinia S, Lagoa CM, Brunke-Reese D, & Conroy DE (2021). Person-specific dose-finding for a digital messaging intervention to promote physical activity. Health Psychology, 40(8), 502–512. 10.1037/hea0001117 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Howie EK, & Straker LM (2016). Rates of attrition, non-compliance and missingness in randomized controlled trials of child physical activity interventions using accelerometers: A brief methodological review. Journal of Science and Medicine in Sport, 19(10), 830–836. 10.1016/j.jsams.2015.12.520 [DOI] [PubMed] [Google Scholar]
  16. Huang Y, Upadhyay U, Dhar E, Kuo L-J, & Syed-Abdul S (2022). A scoping review to assess adherence to and clinical outcomes of wearable devices in the cancer population. Cancers, 14(18), 4437. 10.3390/cancers14184437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Huisman M, Krol B, & Van Sonderen E (1998). Handling missing data by re-approaching non-respondents. Quality and Quantity, 32(1), 77–91. [Google Scholar]
  18. Kaewkannate K, & Kim S (2016). A comparison of wearable fitness devices. BMC Public Health, 16(1), 433. 10.1186/s12889-016-3059-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kang H (2013). The prevention and handling of the missing data. Korean Journal of Anesthesiology, 64(5), 402–406. 10.4097/kjae.2013.64.5.402 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Leary MR, & Kowalski RM (1990). Impression management: A literature review and two-component model. Psychological Bulletin, 107(1), 34–47. 10.1037/0033-2909.107.1.34 [DOI] [Google Scholar]
  21. Lee PH, Macfarlane DJ, & Lam TH (2013). Factors associated with participant compliance in studies using accelerometers. Gait & Posture, 38(4), 912–917. 10.1016/j.gaitpost.2013.04.018 [DOI] [PubMed] [Google Scholar]
  22. Liu K, Daviglus ML, Loria CM, Colangelo LA, Spring B, Moller AC, & Lloyd-Jones DM (2012). Healthy lifestyle through young adulthood and the presence of low cardiovascular disease risk profile in middle age. Circulation, 125(8), 996–1004. 10.1161/CIRCULATIONAHA.111.060681 [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Liu L, Xiang M, Guo H, Sun Z, Wu T, & Liu H (2020). Reliability and validity of the behavioral regulation in exercise questionnaire-2 for nursing home residents in China. Asian Nursing Research, 14(1), 11–16. [DOI] [PubMed] [Google Scholar]
  24. Loprinzi PD, Cardinal BJ, Crespo CJ, Brodowicz GR, Andersen RE, & Smit E (2013). Differences in demographic, behavioral, and biological variables between those with valid and invalid accelerometry data: implications for generalizability. Journal of Physical Activity and Health, 10(1), 79–84. [PubMed] [Google Scholar]
  25. Lu L, Zhang J, Xie Y, Gao F, Xu S, Wu X, & Ye Z (2020). Wearable health devices in health care: narrative systematic review. JMIR mHealth and uHealth, 8(11), e18907. 10.2196/18907 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. McCann DA, Knowles ZR, Fairclough SJ, & Graves LEF (2016). A protocol to encourage accelerometer wear in children and young people. Qualitative Research in Sport, Exercise and Health, 8(4), 319–331. 10.1080/2159676X.2016.1160949 [DOI] [Google Scholar]
  27. Migueles JH, Cadenas-Sanchez C, Ekelund U, Nyström CD, Mora-Gonzalez J, Löf M, et al. (2017). Accelerometer data collection and processing criteria to assess physical activity and other outcomes: A systematic review and practical considerations. Sports Medicine, 1–25. 10.1007/s40279-017-0716-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mischel W (1973). Toward a cognitive social learning reconceptualization of personality. Psychological Review, 80(4), 252–283. 10.1037/h0035002 [DOI] [PubMed] [Google Scholar]
  29. Nelson MC, Story M, Larson NI, Neumark-Sztainer D, & Lytle LA (2008). Emerging adulthood and college-aged youth: An overlooked age for weight-related behavior change. Obesity, 16(10), 2205–2211. 10.1038/oby.2008.365 [DOI] [PubMed] [Google Scholar]
  30. Pépin JL, Bruno RM, Yang R-Y, Vercamer V, Jouhaud P, Escourrou P, & Boutouyrie P (2020). Wearable activity trackers for monitoring adherence to home confinement during the COVID-19 pandemic worldwide: Data aggregation and analysis. Journal of Medical Internet Research, 22(6), e19787. 10.2196/19787 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Peterson RA, & Cavanaugh JE (2020). Ordered quantile normalization: a semiparametric transformation built for the cross-validation era. Journal of Applied Statistics, 47(13–15), 2312–2327. 10.1080/02664763.2019.1630372 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. R Core Team. (2021). R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/ [Google Scholar]
  33. Rhodes RE, Kaushal N, & Quinlan A (2016). Is physical activity a part of who I am? A review and meta-analysis of identity, schema and physical activity. Health Psychology Review, 10(2), 204–225. 10.1080/17437199.2016.1143334 [DOI] [PubMed] [Google Scholar]
  34. Roth MA, & Mindell J (2011). Who wears an accelerometer? Correlates of adherence to wearing an accelerometry motion sensor: The 2008 Health Survey for England. J Epidemiol Community Health, 65(Suppl 2), A27–A27. 10.1136/jech.2011.143586.60 [DOI] [PubMed] [Google Scholar]
  35. Ryan RM, & Deci EL (2019). Brick by brick: The origins, development, and future of self-determination theory. In Advances in motivation science (Vol. 6, pp. 111–156). Elsevier. [Google Scholar]
  36. Teixeira PJ, Carraça EV, Markland D, Silva MN, & Ryan RM (2012). Exercise, physical activity, and self-determination theory: A systematic review. International Journal of Behavioral Nutrition and Physical Activity, 9(1), 78. 10.1186/1479-5868-9-78 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Thomas S, Reading J, & Shephard RJ (1992). Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Canadian Journal of Sport Sciences, 17(4), 338–345. [PubMed] [Google Scholar]
  38. Tison GH, Avram R, Kuhar P, Abreau S, Marcus GM, Pletcher MJ, & Olgin JE (2020). Worldwide effect of COVID-19 on physical activity: A descriptive study. Annals of Internal Medicine, 173(9), 767–770. 10.7326/M20-2665 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Troiano RP, McClain JJ, Brychta RJ, & Chen KY (2014). Evolution of accelerometer methods for physical activity research. British Journal of Sports Medicine, 48(13), 1019–1023. 10.1136/bjsports-2014-093546 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Trost SG, Mciver KL, & Pate RR (2005). Conducting accelerometer-based activity assessments in field-based research. Medicine & Science in Sports & Exercise, 37(11), S531–S543. [DOI] [PubMed] [Google Scholar]
  41. Vlachopoulos SP, Kaperoni M, & Moustaka FC (2011). The relationship of self-determination theory variables to exercise identity. Psychology of Sport and Exercise, 12(3), 265–272. 10.1016/j.psychsport.2010.11.006 [DOI] [Google Scholar]
  42. Vogels EA (2020, January 9). About one-in-five Americans use a smart watch or fitness tracker. Pew Research Center. https://www.pewresearch.org/fact-tank/2020/01/09/about-one-in-five-americans-use-a-smart-watch-or-fitness-tracker/. Accessed 5 September 2021
  43. Wierts CM, Faulkner G, Rhodes RE, Zumbo BD, & Beauchamp MR (2022). Psychological needs and exercise behaviour: a comparison of two psychological needs models. Psychology & Health, 1–19. 10.1080/08870446.2022.2141484 [DOI] [PubMed] [Google Scholar]
  44. Wilke J, Mohr L, Tenforde AS, Edouard P, Fossati C, González-Gross M, et al. (2021). A Pandemic within the Pandemic? Physical Activity Levels Substantially Decreased in Countries Affected by COVID-19. International Journal of Environmental Research and Public Health, 18(5), 2235. 10.3390/ijerph18052235 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Wilson PM, & Muon S (2008). Psychometric properties of the Exercise Identity Scale in a university sample. International Journal of Sport and Exercise Psychology, 6(2), 115–131. 10.1080/1612197X.2008.9671857 [DOI] [Google Scholar]
  46. Wilson PM, Rodgers WM, Loitz CC, & Scime G (2006). “It’s Who I Am… Really!’The importance of integrated regulation in exercise contexts. Journal of Applied Biobehavioral Research, 11(2), 79–104. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data and study materials are available from the corresponding author upon request.

RESOURCES