1. Introduction
The majority of behavior change interventions are successful in changing behavior in the short term, but they often fail to result in the maintenance of newly adopted behaviors over time [
1]. To design effective interventions, the active ingredients of interventions, or behavior change techniques, must be identified [
2]. However, as behavior change techniques were developed to target specific mechanisms of action in behaviors [
3], understanding the salient characteristics of behavior may be important to facilitate the identification of common mechanisms of action and, therefore, the relevant behavior change techniques.
Researchers have attempted to categorize behaviors in various ways to understand the underlying mechanisms of action. For example, meta-analyses may choose to categorize behaviors by functionality (e.g., health promotive, health risk, and illness detecting behaviors [
4]) or further sub-divide categories into descriptive groups (e.g., risk, detection, physical activity, dietary, safer sex, and abstinence [
5]). However, human behavior is complex, and to sufficiently understand the drivers of behavior to effectively change them, the complexity of behaviors needs to be considered to design effective interventions [
6].
One of the most commonly noted consequences of behavioral complexity in behavior change research is that some behaviors are more complex to habitually implement or maintain than others [
7]. This is important to consider when habits can encourage behaviors in a more automatic, rapid, and unconscious manner [
8]. Therefore, due to habits being the default responses in specific contexts, they may trigger behaviors automatically, with minimal drain on self-regulatory resources and during lapses in motivation or situations with competing demands [
7]. Hence, habits can facilitate behavioral engagement and maintenance with greater consistency [
1] and understanding the characteristics of behavioral complexity may help interventions foster habitual behavior to improve its effectiveness.
Although researchers agree that some behaviors are more complex than others (e.g., active commuting involves multiple sub-actions and is more complex than taking a pill), the proposed characteristics that define behavioral complexity are subject to debate. Phillips and Mullan [
7] argue that complexity is an inherent characteristic of behaviors themselves, involving the number of meaningful, separable (e.g., in instigation and execution) and substitutable sub-actions or steps, which may require greater time to prepare and execute intrinsic motivation to form and maintain habits. Contrastingly, Rebar, Rhodes [
9] argue that rewards and behavioral complexity are related but distinct constructs, and complexity is not solely a characteristic of the behavior itself but an interaction between the behavior, context (e.g., barriers and facilitators), and actor (e.g., skill level, knowledge acquisition, and cognitive load). Similarly, Kaushal and Rhodes [
10] consider behavioral complexity as the perceived difficulty of a behavior, influenced by an individual’s self-efficacy, with McCloskey and Johnson [
11] agreeing but also considering the time, attention, and planning required to prepare and execute a behavior. Although McCloskey and Johnson [
11] have identified their proposals of behavioral complexity characteristics via data-driven methods, other proposals have not yet been tested, and no studies to date have tested all proposals together to synthesize the findings.
With varying and debated conceptualizations of behavioral complexity within the field, research is needed to assess lay individual’s opinions of the importance of the proposed characteristics of behavioral complexity. Data-driven approaches are also needed to investigate the validity of the proposed characteristics. Then, a synthesized and evidence-based definition can be developed to inform the methods required for tailored public health and environmental interventions. Understanding the salient characteristics of behavior can inform researchers, practitioners, individuals, and policy-makers of the common barriers and facilitators of various behavior types to target when aiming to change behavior. The findings can inform the suitability of tailored intervention methods and education programs to change behavior.
The Current Study
To work towards a unified definition of behavioral complexity, the aim of this study was to explore people’s perceptions of the characteristics of complexity previously proposed by researchers. A secondary aim was to identify typical behaviors that exemplify the varying degrees of complexity across the characteristics of behavioral complexity. Due to the exploratory nature of this study, there were no specific hypotheses.
2. Materials and Methods
2.1. Participants
Participants were recruited from February to June 2023 via a university undergraduate participant pool. Participants received course credit in return for completing the study. Two hundred and fifty-two participants completed the study, with 27 cases removed due to not providing consent, being under 18 years of age, missing over 50% of data, or being a duplicate response (e.g., determined by their provided identifiers). Two hundred and twenty-five participants were retained in the final sample.
Participants were aged between 18 and 51 years (M = 23.10, SD = 7.14). Most identified as women (75.1%), 20.4% as men, 2.7% as non-binary, and 1.8% did not indicate their gender. Most resided in Western Australia (95.6%), with the remaining 4.4% residing in Victoria, Queensland, and New South Wales, or did not indicate their state of residence. Most received a high school education (69.5%), 20.4% received a university degree or above, 8.4% received a TAFE certificate (vocational education training), and 1.8% did not indicate.
2.2. Measures
For complexity ratings of behaviors, the participants were asked to indicate if 24 behaviors were simple or complex for the average person to do on a regular basis (1 = simple, 2 = complex). Specifically, they were instructed, “Please indicate whether the following behaviors are simple or complex for the average person to do on a regular basis.” Behaviors were chosen based on previous meta-analyses of behavior change theories, which categorized behaviors by type, e.g., [
5,
12]. Upon participants indicating that a behavior was complex, they were then probed to indicate how complex they think that behavior is for the average person to do on a regular basis on a 7-point Likert scale (1 = not at all complex, 7 = extremely complex). Specifically, they were instructed, “Please indicate how complex (rather than simple) that behavior is for the average person to do on a regular basis.”
For factors of behavioral complexity, the participants were asked to indicate whether 12 factors of behavioral complexity, as proposed by the literature [
7,
9,
10,
11], were influential or not influential to how complex a behavior is deemed to be (1 = not influential, 2 = influential). Specifically, they were instructed, “Please indicate whether the following factors are influential to how complex a behavior is for the average person to do on a regular basis.” Proposed factors of behavioral complexity were the novelty, difficulty, number of steps, planning, the immediacy of reward, time and attention associated with a behavior, an individual’s skill level, mental resources, self-efficacy, and motivation, and the supportiveness of a behavior’s general context. Upon participants indicating a factor of behavioral complexity to be influential, they were then probed to indicate how influential they thought that factor was on a 7-point Likert scale (1 = not at all influential, 7 = extremely influential). Specifically, they were instructed, “Please rate that factor’s level of influence to how complex a behavior is for the average person to do a behavior on a regular basis.”
2.3. Procedure
Participants completed a 30-minute online survey on Qualtrics by clicking on a link within study advertisements. Participants were first displayed a participant information sheet and asked to provide consent by clicking a box to continue. Only after providing informed consent were participants permitted to continue with the survey. At the end of the survey, participants were thanked for their time and requested to enter their contact details. The University Human Research Ethics Committee approved this study.
2.4. Data Analysis
SPSS Version 28 [
13] was used for data cleaning (e.g., case deletion for high proportions of missing data, missing values analysis, renaming variable labels, and assumptions testing) and assessing for frequency counts (e.g., proportions) and descriptives statistics of participant demographics and variables (e.g., range, means, and standard deviations). To integrate findings and identify behaviors that exemplify the characteristics of behavioral complexity, post-hoc analyses were conducted by two members of the research team. Researchers used Microsoft Excel to plot participants’ mean ratings of each behavior’s complexity into individual figures, each representing a characteristic of behavioral complexity. Based on the spread of normality of ratings, behaviors low in complexity were those rated < 4, and those high in complexity were rated > 4. Behaviors were then categorized as true (1) or untrue (−1) to each characteristic based on researchers’ knowledge of the scientific literature surrounding each behavior and characteristic of complexity. The final figures consisted of four quadrants, which represented four categories of behavior: (1) low complexity behaviors that were untrue to characteristics, (2) low complexity behaviors that were true to characteristics, (3) high complexity behaviors that were true to characteristics, and (4) high complexity behaviors that were untrue to characteristics. The frequency of which behaviors were plotted in each quadrant across all figures determined the best-fitting category for that behavior. Raincloud plots to visually display the spread of distributions in responses were produced using R Statistical Software Version 4.4.1 [
14].
4. Discussion
To work toward a unified definition, the aim of this study was to explore lay individuals’ perceptions of the characteristics of behavioral complexity that were previously proposed by researchers. A secondary aim was to identify typical behaviors that exemplify the varying degrees of complexity across these characteristics. Participants rated all characteristics to be influential to the complexity of behaviors, with little variability in the ratings of influence between characteristics. Typical behaviors that reflect the varying degrees of complexity across the characteristics of behavioral complexity were also identified. Highly complex behaviors, which were true to characteristics, include abstaining from smoking cigarettes, taking insulin shots, limiting alcohol consumption, cycling for commuting, and composting. Low complexity behaviors, which were untrue to characteristics, include eating fruit and stretching. Finally, low complexity behaviors, which were true to characteristics, were exercising at the gym and taking a bus or train for commuting. Few behaviors were highly complex and untrue to characteristics. Findings support the idea that complex behaviors, compared to simple behaviors, are more likely to be true to the characteristics.
4.1. Characteristics of Behavioral Complexity
Participants perceived all proposed characteristics as influential to the complexity of a behavior. Researchers have previously speculated that characteristics include the number of steps and motivation required for a behavior [
7], supportiveness of the context, an individual’s skill, knowledge acquisition and cognitive load [
9], perceived difficulty of a behavior, self-efficacy [
10], and the time, attention, and planning involved in a behavior [
11]. However, this study expands on previous research using data-driven methods to synthesize and test these ideas to reveal that all characteristics are influential. Although findings support all individually proposed perspectives of behavioral complexity, individual perspectives alone may not be sufficient to define them. Based on the current results, which found that complex behaviors are more likely to be true to characteristics than simple behaviors, we propose that a synthesized definition of behavioral complexity may consider the additive function of all proposed characteristics collectively. Specifically, although behaviors may not always be true to all characteristics (e.g., due to potential variability in individual and situational factors), behaviors possessing a greater number of characteristics at any given time may contribute to greater complexity. This notion aligns with complex systems approaches to behavior change, considering the interconnectedness of factors of behavior, heterogeneity across individuals, and non-linear dynamics [
15]. From this perspective, complexity may be influenced by various levels of systems or scales [
16] ranging from individuals and communities to broader policies. Future research may test the validity of this definition by assessing within and between-person fluctuations (e.g., via ecological momentary assessment methods) in individuals’ perceived complexity of the identified exemplar behaviors and how true each behavior is to each characteristic of complexity. This is of importance to improve theoretical understandings regarding the mechanisms of automatic factors of behavior. Some of the participant ratings were unexpected (e.g., physical activity behaviors as low complexity) in that they do not align with definitions of complexity used in previous research to establish the influence of behavioral complexity as a moderator of the habit–behavior relationship [
17] and implicit attitude–behavior relationship [
18,
19].
Current findings indicate the importance of considering behavioral, individual, and contextual factors when aiming to tailor intervention methods to foster habit formation across behaviors of varying complexity. As behavior change techniques are designed to target specific mechanisms of action [
3], relevant behavior change techniques may be identified and implemented in public health and environmental interventions aiming to change behaviors with certain characteristics. For example, one common characteristic of complex behaviors, such as abstaining from smoking, identified in this study, is that the behavior commonly occurs in an unsupportive context. Relevant behavior change techniques that can be implemented to foster supportive environmental contexts include restructuring the physical and social environment or encouraging social support [
2]. Similarly, complex behaviors that require developed skills and strong self-efficacy, such as administering insulin shots, may be relevant to behavior change techniques such as problem-solving or behavioral practice and rehearsal [
2]. Habit-based public health and environmental intervention methods may be tailored to non-changeable characteristics of complex behaviors, such as the environmental barriers or inherent characteristics of a behavior itself (e.g., by applying implementation intentions or action and coping planning strategies to prepare for and overcome difficult situations when intentions are likely to fail [
20]). However, researchers may amend the changeable characteristics of habit formation in complex behaviors by fostering individuals’ intrinsic motivation, availability of mental resources, skill development, and self-efficacy while reducing their perceived difficulty of a behavior.
There was little variability in the ratings of influence between characteristics (
M = 4.68 − 5.40, SD = 1.12 − 1.34), indicating the importance of all characteristics for most individuals, at least in certain situations. However, future research may use qualitative methodology to contextualize and gain further insight into when and how characteristics are most influential, including the prominence of characteristics depending on the individual or context. This is important to note due to the heterogeneity in the complexity ratings of behaviors visually displayed in raincloud plots. Understanding this heterogeneity may contribute to more nuanced and relevant intervention designs [
21]. One area of research may be to explore how characteristics interact to influence habits and habitual behavior, as there are mixed proposals on how this may be influenced by behavioral complexity. For example, Rebar and Rhodes [
9] proposed that the ease in behavioral engagement may depend on dynamic interactions between behavioral, individual, and contextual factors, while Gardner and Lally [
22] propose that habitual processes for behaviors are equal regardless of complexity, and Phillips and Mullan [
7] propose that complex behaviors require greater intrinsic motivation and take longer to become habitual, therefore producing differences in the outcomes of habit impulses and intervention techniques required [
23]. Alternatively, other factors not considered in the current research, which may be more characteristic of behavioral complexity, should be explored, as the current research only assessed participants’ perceptions guided by researchers’ proposals. For example, previously proposed behavioral characteristics include the likelihood of a behavior’s consequence, severity, and tangibility of a behavior’s impact, and the emotionality and nature of a behavior (e.g., as approach or avoidance) [
24].
4.2. Typical Behaviors of Complexity Characteristics
Behaviors were most often and consistently categorized as highly complex and true to characteristics (e.g., abstaining from smoking cigarettes, taking insulin shots, limiting alcohol consumption, cycling for commuting, and composting). There was also consistency in low-complexity behaviors that were untrue to characteristics (e.g., eating fruit and stretching). The identified typical behaviors may be implemented in future research for theory testing and research with the aim of optimizing habit-based interventions across behaviors of varying complexity. It is important to have consistency in the behaviors implemented in habit research to facilitate the synthesis of findings and the accumulation of knowledge required to inform a broader understanding of habit formation and disruption for long-term behavior change [
25]. Additionally, only a few behaviors were categorized as highly complex and untrue to characteristics, with inconsistencies in behaviors included in this category. Therefore, findings are consistent with researchers’ conceptualizations, where complex behaviors, compared to simple behaviors, are more likely to possess or be influenced by these factors [
7,
9,
10,
11].
However, there was still good consistency in behaviors that were low in complexity and true to characteristics. Behaviors always included in this category were exercising at the gym and taking a bus or train for commuting. Findings are inconsistent with the literature as it is expected that behaviors that are characteristic or expected to be influenced by the proposed factors would be more complex [
7,
9,
10,
11]. However, frequently performed behaviors engaged in a more automatic manner are typically perceived as less complex [
11]. Therefore, it is possible that these behaviors may not be perceived as complex, although still true to characteristics, as individuals may already engage in the behavior habitually or perceive others to. Perceived behavioral control (i.e., self-efficacy as an identified characteristic in this study [
26]), may be related to the challenge level in non-habitually or frequently performed tasks [
27] and should be strengthened to facilitate these behaviors (e.g., by providing the resources and opportunities required).
Alternatively, findings may be attributed to behaviors being categorized post-hoc by researchers as being true to the characteristics or not. Future research should investigate participants’ perceptions of how similar or influenced by characteristics these behaviors may be, as behaviors may only be somewhat similar or influenced despite still being similar or influenced to some degree.
4.3. Strengths and Limitations
This study was the first that we are aware of to explore lay individuals’ perceptions of the characteristics of behavioral complexity based on researchers’ proposals. Findings may help resolve the currently mixed proposals in the literature to work towards a synthesized definition of behavioral complexity. This definition may consider the additive function of the identified characteristics of behavioral complexity, whereby behaviors may not always be true to all characteristics, but greater complexity is contributed to by possessing a greater number of characteristics at any given time. Therefore, a framework may be developed to inform tailored habit-based interventions and assessment methods for habit across behaviors of varying complexity [
7]. This framework can guide public health and environmental intervention designs, the methods implemented by practitioners to encourage individual behavior change, and future research aiming to optimize interventions by tailoring behavior change techniques to the behavior types of interest and the relevant characteristics of behavior identified in this study. However, it should be noted that further research is required to test this definition before applying it in future research. For example, future research may explore ways of objectively assessing behavioral complexity to counteract self-perceptions, as it is known that individuals who frequently perform behaviors may perceive them as less complex [
11].
Additionally, this study has identified typical behaviors that reflect the varying types and degrees of complexity across the proposed characteristics. The identified behaviors can be employed in future research to assess the influence of behavioral complexity in habitual processes more consistently. Therefore, findings from this research may help address one priority raised by expert panels in the behavior change field to further theory and evidence for intervention development by understanding the influence of behavioral complexity on habits [
25]. Similarly, the identified behaviors can be used to assess the utility of theoretical frameworks to behaviors of varying complexity, as there is evidence that the importance of constructs in behavioral models, such as temporal self-regulation theory, may differ depending on a behavior’s complexity [
12].
However, categorizations were conducted post-hoc by researchers to determine if each behavior was similar or expected to be influenced by the characteristics of behavioral complexity. Therefore, it is not known if or to what extent lay individuals perceive these behaviors to be true to the characteristics. Future research is required to confirm how similar behaviors are and how influenced each behavior is by the characteristics so as to determine, more robustly, the typical examples of behaviors of varying complexity. One approach may be to apply a latent profile analysis or similar methods to identify subgroups of behaviors based on the similarity of latent characteristics [
28], as the current study was limited due to the structure of the data and could not apply the methods.
Additionally, the population average for the complexity of behaviors and the importance of the characteristics to behavioral complexity may be difficult to interpret due to the questioning style, where participants were first asked to choose a category via a binary response option and then prompted by a Likert scale format question to avoid central tendency bias [
29]. It is commonly observed that binary response options are equally reliable and valid to multi-category formats, but multi-choice formats may be more suitable to assess individuals’ attitudes [
30]. Although a mid-point in the follow-up question was included in this study to allow participants to reassess their response, future research may replicate the current study to more robustly assess the true average of behavioral complexity ratings and the importance of complexity characteristics.
Further, only the importance of existing conceptualizations in the literature was assessed so that other possible characteristics of behavioral complexity may not be captured in the current study. Future research may use qualitative methodology to identify any influential factors not considered. However, the current study has included the most pertinent proposals of behavioral complexity in the literature, presented from extensive evidence surrounding habit formation across various behavior types. Additionally, further contextualization of findings or moderators of behavioral complexity may be explored as the included behaviors in the current study are not comparable based on frequency, duration, and context. Differences in the propositions for each behavior need to be considered, as for behaviors to effectively achieve the desired outcome, they must follow the recommended guidelines [
31].