Intraindividual Variability of Sleep Opportunity Attitudes, Perceived Norms, Perceived Behavioral Control, and Intentions

Abstract

Background

Recent research has used the theory of planned behavior (TPB) to predict sleep. However, this research has focused on between-person effects and has failed to capture the intraindividual variability of sleep-related attitudes, subjective norms, perceived behavioral control, and intentions.

Purpose

The current study sought to characterize the between- and within-day patterns of these constructs.

Methods

Participants (N = 79) completed a 1 week ecological momentary assessment protocol in which they reported their attitudes, subjective norms, perceived behavioral control, and intentions toward nightly sleep opportunity four times per day.

Results

Analyses revealed both between- and within-day variability of these constructs, with perceived behavioral control and intentions demonstrating greater variability than attitudes and subjective norms. Mixed linear models revealed that attitudes and subjective norms significantly increased throughout the week, while perceived behavioral control and intentions significantly decreased throughout the day.

Conclusions

The between- and within-day patterns of the TPB constructs highlight important methodological considerations and provide insight into the potential refinement of sleep promotion efforts.

In healthy young adults, thoughts about sleep may change both throughout the week and across a day.

Introduction

The theory of planned behavior (TPB) [1] states that volitional behavior results from intentions to perform that behavior, and intentions are influenced by one’s attitudes, subjective norms, and perceived behavioral control toward that behavior. Recent research has applied the TPB to predict sleep duration (for a review, see [2]), and this work suggests that its constructs explain a significant amount of variance in sleep-related intentions and behavior. However, this research is limited by its emphasis on between-person designs, which fail to capture the intraindividual variability of these constructs. While focusing on between-person differences can help identify individuals or groups vulnerable to poor sleep health, characterizing the intraindividual variability of such constructs can identify vulnerable days and contexts for all individuals. Furthermore, longitudinal designs can identify the stability of these constructs over time, elucidating whether these constructs are stable or fluctuate due to variations in states or contexts throughout the week or day. Indeed, the temporal stability of these constructs may be related to the intention–behavior gap [3], which is when intentions do not translate to behavior.

To date, no study has characterized the intraindividual variability of the TPB’s constructs as they relate to sleep health behaviors [2], and, therefore, it is presently unknown how these constructs vary in a given week or day. The current study sought to use ecological momentary assessment (EMA) to characterize the intraindividual variability of sleep opportunity (i.e., the amount of time an individual allows themselves to sleep at night) attitudes, subjective norms, perceived behavioral control, and intentions. EMA is a measurement tool that allows for momentary assessment of psychological constructs and behaviors, which reduces recall bias and improves ecological validity [4]. As these patterns and trajectories have not yet been explored, there were no specific hypotheses.

Methods

Participants

Participants (N = 79) were recruited from a Midwestern university undergraduate research pool. Students had to be between ages 18 and 25 and have a mobile phone with the capability of receiving text messages and connecting to websites to participate. Individuals were ineligible for participation if they had an atypical sleep schedule (e.g., shift work), were currently being treated for a sleep disorder, were binge drinkers, smoked cigarettes or e-cigarettes, used sleep aids or medications that can disrupt sleep, or had a medical condition that can influence sleep.

Procedure

Data for analyses were from a larger parent study that examined the associations between constructs of the TPB and sleep opportunity (for preregistration, see osf.io/2q6s8). All study materials and procedures were approved by the institution’s institutional review board. After completing an online questionnaire to screen for inclusion/exclusion criteria, eligible individuals received an email that briefly described the study’s purpose, procedures, and compensation, and participants provided their consent electronically. To reduce reactivity (i.e., participant responses influenced by participation in study protocol), participants completed bedtime and morning sleep diaries for 3 days prior to their Time 1 session. Upon arrival at the lab for their Time 1 session, participants met with a research assistant in a private room to complete a demographic questionnaire. An EMA protocol, sampling of participant’s experiences in real time, was used to assess sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions four times (morning, afternoon, evening, and bedtime) throughout each of the study days. The bedtime and wake time assessments were event contingent and were initiated by participants before they went to bed and when they woke up in the morning. The afternoon and evening EMA signals were prompted by an automated text message containing a link to their assessment. All afternoon signals were sent between 12:00 pm and 2:00 pm, and all evening signals were sent between 5:00 pm and 7:00 pm. During the initial session, participants indicated a fixed time during each of these time periods to receive their afternoon and evening signals to accommodate individual school and work schedules and maximize adherence to the EMA protocol. As an additional incentive for protocol adherence, participants who completed at least 95% of these assessments during the study were eligible for two $50 raffles. Adherence checks were conducted daily, and participants who missed any signal were contacted via email with encouragement to adhere to the study protocol. Following the week-long in-home assessment, participants received credit toward their psychology course.

Measures

Attitudes, subjective norms, perceived behavioral control, and behavioral intentions to allow adequate sleep opportunity were assessed with four items collected four times daily using EMA. These questions were adapted from Robbins and Niederdeppe [5], which consisted of 14 items and were assessed on a 1–7 scale. For the current study, responses for each question ranged from 1 to 11 to reduce the possible restriction of range in repeated assessment. Adding more response choices to the TPB measures does not diminish their reliability [6]. To minimize participant burden, one item for each construct was included in the EMA signals. Constructs were measured with the following items. Attitudes: “Overall, I think allowing myself the time to sleep 8–9 hr tonight is” 1 (good) to 11 (bad). Subjective norms: “People who are important to me think that I should allow myself the time to sleep 8–9 hr tonight” 1 (strongly disagree) to 11 (strongly agree). Perceived behavioral control: “For me to allow myself the time to sleep for between 8 and 9 hr tonight is” 1 (easy) to 11 (difficult). Intentions: “I intend to allow myself the time to sleep for between 8 and 9 hr tonight” 1 (definitely do not) to 11 (definitely do). Attitudes and perceived behavioral control were reversed scored.

Data Analysis

Descriptive statistics were used to estimate the between- and within-day variability for attitudes, subjective norms, perceived behavioral control, and intentions toward nightly sleep opportunity. To characterize the between-day variability of these constructs, observations from all four daily time points were averaged into a daily score for each construct and participant. The between-day variability was assessed by calculating the standard deviation (SD) across the 7 days of the study period for each individual. The within-day variability of these constructs was assessed by calculating the SD within each of the seven study days for each individual, then averaging across the 7 days into a single measure of within-day variability for each participant.

In addition to the preregistered descriptive statistics, mixed linear models were used to characterize the between- and within-day slopes [7, 8]. The between-day slopes were calculated by first averaging observations from all four daily measures of each construct into a daily score for each participant. Four separate models were conducted with each construct as the dependent variable, study day as the predictor, and a random intercept. The within-day slopes for each construct were tested by entering each of these variables (each was tested in a separate model) as the dependent variable, both signal and study day as predictor variables, and a random intercept. Heterogenous autoregressive covariance structures were used in each model to allow for the variances at each timepoint to covary. Study day was added as a random effect to account for the nesting of the signals. Age, gender, and race were tested as covariates but were not significant and were removed from the final model.

Results

Protocol Adherence

Overall protocol adherence was 92% of all signals, and missing data were not related to any study variables or participant characteristics (all ps >.05). Twenty-five signals were removed due to being taken at incorrect times (e.g., taking the morning assessment in the evening), and two study days were removed due to having less than 50% of signals. One participant was removed from analyses due to protocol nonadherence. Out of 548 total valid study days, 29 days had two signals and 92 days had three signals.

Descriptives

Participants were majority female (58.2%) and white (83.5%), with an average age of 19.01 (1.16) years (see Table 1). On average, participants had positive attitudes (M = 9.45, SD = 2.24) and subjective norms (M = 9.01, SD = 2.10) and moderate perceived behavioral control (M = 7.16, SD = 3.19) and intentions (M = 7.43, SD = 3.04).

Table 1.

Demographic and theory of planned behavior characteristics (N = 79)

Gender, n (%)
Male	33 (41.8%)
Female	46 (58.2%)
Race, n (%)
White	66 (83.5%)
Black	4 (5.1%)
Asian	9 (11.4%)
Age, mean (SD), years	19.01 (1.16)
Socioeconomic status, n (%)
10,000–49,999	9 (11.4%)
50,000–99,999	22 (27.5%)
100,000 and over	29 (36.7%)
Decline to answer	18 (22.8%)
Mean level, mean (SD)
Attitudes	9.45 (2.24)
Subjective norms	9.01 (2.10)
Perceived behavioral control	7.16 (3.19)
Intentions	7.43 (3.04)
Between-day SD, mean (SD)a
Attitudes	0.71 (0.70)
Subjective norms	0.64 (0.75)
Perceived behavioral control	1.15 (0.76)
Intentions	1.18 (0.64)
Within-day SD, mean (SD)b
Attitudes	0.71 (0.68)
Subjective norms	0.60 (0.52)
Perceived behavioral control	1.28 (0.70)
Intentions	1.45 (0.68)
Between-day slope, b (SE)c
Attitudes	0.10 (0.03)*
Subjective norms	0.11 (0.04)*
Perceived behavioral control	0.03 (0.04)
Intentions	0.06 (0.04)
Within-day slope, b (SE)d
Attitudes	−0.04 (0.03)
Subjective norms	−0.04 (0.02)
Perceived behavioral control	−0.19 (0.04)*
Intentions	−0.33 (0.04)*

*p < .05.

^aDegree to which variables varied day to day.

^bDegree to which variables varied within each day.

^cLinear slope of variables across the seven study days.

^dLinear slope of variables within each day.

Between- and Within-Day Variability

Across the sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions (each measured on an 11-point scale), there were low to moderate levels of between- and within-day variability (see Table 1). Subjective norms had the lowest between-day (M = 0.65, SD = 0.75) and within-day variability (M = 0.60, SD = 0.52). Attitudes also had somewhat low between-day (M = 0.71, SD = 0.71) and within-day (M = 0.71, SD = 0.68) variability. Thus, both between and within days, participants generally did not fluctuate more than 1 point from their typical trend. Of the three theoretical predictors of sleep opportunity intentions, perceived behavioral control had the greatest between-day (M = 1.15, SD = 0.76) and within-day (M = 1.30, SD = 0.70) variability. Lastly, intentions had both between-day (M = 1.18, SD = 0.64) and within-day (M = 1.45, SD = 0.68) variability.

Between- and Within-Day Slopes

Attitudes, subjective norms, perceived behavioral control, and intentions also showed patterns of between- and within-day slopes across the seven study days and four within-day signals, respectively (see Figs. 1 and 2). Attitudes had a nonsignificant decrease during the day (b = −.04, p = .09) and significantly increased throughout the week (b = .10, p = .003). Subjective norms had a negative, but not significant, slope during the day (b = −.04, p = .054) and significantly increased during the week (b = .11, p = .005). Perceived behavioral control significantly decreased during the day (b = −.19, p < .001) and had a nonsignificant positive slope throughout the week (b = .03, p = .484). Intentions significantly decreased during the day (b = −.33, p < .001) and had a nonsignificant positive slope throughout the week (b = .06, p = .085).

Fig. 1.

Weeklong slopes of sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions.

Fig. 2.

Within-day slopes of sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions.

Discussion

The current study sought to characterize the between- and within-day variability of attitudes, subjective norms, perceived behavioral control, and intentions toward obtaining a sleep opportunity of 8–9 hr each night. Attitudes and subjective norms had high mean levels in this sample, and perceived behavioral control and intentions were somewhat lower. Attitudes and subjective norms had the lowest intraindividual variability, with their between- and within-day variability deviating 6%–7% from an individual’s typical trend. This demonstrates that attitudes and subjective norms relating to sleep opportunity may be both positive and relatively stable. In contrast, participants deviated between 11% and 13% both between and within days in their typical perceived behavioral control and intentions. This indicates that one’s sleep opportunity intentions and perceived behavioral control are more fluid, perhaps due to states or experiences that also vary throughout the day. Sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions also showed between- and within-day trajectories. All four constructs showed increases throughout the week, but only attitudes and norms were significant. It is possible that these increases throughout the week are explained by participant reactivity. Attitudes, subjective norms, perceived behavioral control, and intentions also showed within-day patterns and decreased as the day went on. However, only perceived behavioral control and intentions showed significant decreases. Taken together, these results show that sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions show distinct patterns of intraindividual variability and between- and within-day trajectories.

These within-day patterns highlight methodological considerations when using the TPB to study sleep opportunity. Specifically, the time of the day in which these constructs are measured may impact the construct validity of this measurement. For example, assessing intentions in the morning, as compared to later in the day, may show higher than average intentions. These results also have implications for interventions. Both perceived behavioral control and intentions significantly decreased throughout the day, which could help guide the timing of sleep health promotion efforts. Just-in-time adaptive interventions adapt to an individual’s needs and identify specific times and contexts in which intervention efforts would be most effective [9]. These results would suggest that intentions to obtain 8–9 hr of sleep opportunity may be highest in the morning, and, thus, adaptive intervention efforts at different times of the day may require different approaches. Efforts in the morning may aim to maintain high levels of intentions and perceived behavioral control throughout the day. For example, time management interventions may be effective [10] by teaching skills to effectively allocate time throughout the day, thus reducing the need to stay up late completing daily tasks and preserving perceived behavioral control. In addition, the implementation of intention interventions [11] could help translate these intentions into achieving a healthy sleep opportunity, while interventions later in the day may focus on increasing intentions toward sleep opportunity.

This is the first study to use EMA to characterize the intraindividual variability and trajectories of sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions. While this is a great strength of the study, the study is not without its limitations. There was likely participant reactivity in response to the EMA protocol. In fact, self-monitoring is a strategy for sleep health improvement [12]. A 3 day “practice” period was implemented to reduce reactivity, but future studies could consider a longer assessment period (e.g., 14 days) to further reduce reactivity and provide more reliable estimates of variability [13]. Future research could add to this work by characterizing the intraindividual variability of these constructs in other sleep-related behaviors, such as sleep timing and sleep hygiene. In addition, future research might characterize the stability of other facets of the TPB, such as descriptive norms [1]. Some research has already demonstrated that physical activity intentions fluctuate both between [14] and within [15] days. Lastly, this work is limited by its predominately white sample of college students, and future research should characterize the intraindividual variability of these constructs in more diverse populations.

The current study is the first to characterize the intraindividual variability of sleep opportunity attitudes, subjective norms, perceived behavioral control, and intentions. These data demonstrate that the constructs of the TPB fluctuate both between and within days, which have important implications for the refinement of both future research methodologies and refinement of sleep health promotion efforts.

Compliance With Ethical Standards

Authors’ Statement of Conflict of Interest and Adherence to Ethical Standards The authors declare that they have no conflicts of interest.

Authors’ Contributions All participants provided study consent before participating in the study protocol.

Ethical Approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed Consent Michael P. Mead is the primary author. He developed the topic, performed analyses, and wrote the paper. Leah A. Irish also contributed to the development of this paper and was very active in providing feedback during the writing process.