Appendix E

Research on Behavioral Change from Other Domains

The committee examined a significant body of research on consumer behavior from six domains: energy conservation, water conservation, waste prevention and management, recycling, diet change, and weight management. Key findings from this work are discussed in the body of the report; this appendix presents a more detailed discussion of the work reviewed. It begins with a brief historical overview of the literature and then summarizes findings from the research, including general themes about consumer behaviors and themes related to both drivers of consumer behavior and interventions. It closes with an overview of suggestions for further research made by scholars in these fields.

The committee gathered 406 peer-reviewed and published articles that were classified as systematic reviews, narrative reviews, and meta-analyses about behavior change in these six domains and reviewed their abstracts for applicability to the committee’s task. From this set, 46 articles were selected to be reviewed in full; details of the literature search can be found in Appendix B.

An important caveat of this appendix is the focus on meta-analyses, systematic reviews, and narrative reviews. While these types of analyses are helpful in presenting distilled information and can present the weight of the evidence on a topic, they can bias evidence (Sorrell, 2007). This is because these types of analyses and reviews often focus on a narrow set of questions, such as whether certain types of interventions “work” (Sorrell, 2007). But they often do not capture the important implementation and qualitative studies that answer questions about how something works or

the cost of an intervention (Sorrell, 2007). This neglects the complexities about how something works and often overlooks important information.

HISTORY OF BEHAVIOR CHANGE RESEARCH

Awareness of behavior change science and how it has transformed over time is an important backdrop to the scientific findings from current work. Numerous behavior change theories and frameworks have been used to understand and predict individual behavior. One systematic review of research on diet change mentioned that a total of 83 theories of behavior change had been identified (Samdal et al., 2017). There has been no consensus on which theories and frameworks are most useful, and some have been more commonly applied than others. Importantly, the dominant theories in use have evolved over time. Early theories of behavior change, such as social cognitive theory, theory of planned behavior, and the transtheoretical model, were most often used to explain why people adopt a behavior.

For much of the 20th century, these theories characterized human behavior as being predominately conscious and reason driven; these theories and behaviors are sometimes referred to as System 2 (Koop et al., 2019; Marteau, 2017) Because of this focus, the drivers that were captured and interventions that were designed were often centered around knowledge transfer and ways to improve self-efficacy (Koop et al., 2019). By the end of the 20th century, it was apparent that these theories were limited in their ability to explain behavioral outcomes and that there was another set of automatic and emotion-driven factors that needed to be captured (Marteau, 2017).

At this point, models of behavior emerged that depicted two or three sets of parallel processes that characterized human behavior as being more automatic and emotion-driven. These processes were described as reflective, semireflective, and automatic processes (Koop et al., 2019), with reflective being aligned with System 2, and automatic sometimes being referred to as System 1 (Marteau, 2017). Because of widespread use of these processes across domains, the drivers that were captured and interventions that were designed in the studies included in this appendix were often centered on social norms, framing, priming, nudging, and emotions (Koop et al., 2019). In addition, the evolution in these fields over time has meant that the meta-analyses reviewed in this appendix often contain more studies on System 2 interventions and more recent, but sparse, studies on System 1 interventions. Most recently, it has been recognized that individual behaviors are responsive to both System 1 and 2 processes and that a combination of the two can be more effective. Thus, researchers have begun to measure drivers and create study designs that combine System 1 and 2 into more complex

and multicomponent interventions; these studies, because they are more recent, are rarely included in the meta-analyses discussed in this appendix.

GENERAL THEMES IN CURRENT RESEARCH

Literature from the six domains uses many terminologies to categorize and measure behavioral predictors or drivers, behavioral interventions, and other related factors. The multiplicity of terms and measures used within domains can be a barrier to identifying commonalities and differences, to designing and evaluating programs and interventions, and to aligning current and future research. However, there are some common threads in the terms used for processes (e.g., reflective, semireflective, and automatic or System 2 and System 1) and for intervention types (e.g., social comparison, feedback, information, appeals, engagement, choice architecture/nudges). This section summarizes two general themes in the research from the six domains: the use of theory to drive terminology and guide programs, and the fact that intention does not always drive behavior.

Use of Theory to Drive the Use of Common Terminology and Guide Program Design and Evaluation

The terminology used to categorize behavioral drivers and interventions is inconsistent within fields and across fields even though terms often described the same phenomena. This inconsistency makes it challenging to compare within and across literatures. For example, drivers of behavior were categorized within studies in the following ways:

• individual and contextual;
• individual, behavior-specific, and general;
• sociopsychological, technical-organizational, individual, sociodemographic, and study-specific;
• sociopsychological, sociodemographic, contextual (situational);
• perceived and objective;
• demographic, institutional, economic, social/cultural; and
• environmental, situational, psychological.

The definitions for these categories often differed or only overlapped partially between studies, making it challenging to know in which category a driver belonged. Drivers were also often interchangeably termed as determinants, determining factors, motivations, predictors, or moderators. Because these terms can take different levels of meanings (e.g., statistical meanings of prediction and moderation) or serve as a determining construct within a behavioral model (e.g., motivation), it can be challenging to parse

meaning. Similarly, the terms behaviors, actions, and outcomes were often used synonymously. An analysis of household waste prevention interventions by Sharp et al. (2010) found that many studies that describe behavior change are ultimately measuring outcome change but not necessarily the behaviors leading up to the change in outcome.

Behaviors were also often subdivided into categories differentially (Koop et al., 2019; Scott et al., 2015). For example, in the waste management literature, some studies used the well-known reduction, reuse, and recycling categories while other studies created a suite of composite behaviors, including:

• basic environmental, decision-making environmental, interpersonal environmental, and civic environmental behavior (Li et al., 2019) and
• citizenship, financial, persuasion, and ecological management behaviors (Li et al., 2019).

Behaviors were also categorized as one-off, continuous or repeated, or dynamic (i.e., a mix of one-off and repeated) acts (e.g., procuring a recycling bin from the county, turning off lights, recycling) or purchases (e.g., buying an energy efficient appliance or organic food). Some studies recommended that it was more useful to conceptualize actions or behaviors by activity type (i.e., one-off, repeated) than by sector (e.g., waste management behaviors, energy conservation behaviors) because of the similarities and differences between one-off actions and habits or routines, even across domains. Consistent categorizations and terminology would allow the literatures to share common findings more easily.

Similarly, the meta-analyses and systematic reviews included in this appendix categorized interventions in many different ways (e.g., by construct, by strategy, and by process, often relating it to more reasoned behaviors or more automatic behaviors, or both), and, in many cases, described interventions as bundled strategies. For example, one meta-analysis on validated field interventions to promote household recycling appealed to psychological constructs and categorized interventions by type as information, feedback, incentives, commitment, behavior modeling, and environmental alterations (Varotto and Spagnolli, 2017). Another meta-analysis testing behavioral interventions to promote household action on climate change categorized interventions by type as information, appeals, engagement, social comparison, and choice architecture (Nisa et al., 2019). Both studies commented on the frequency with which interventions incorporated bundled strategies. Another waste prevention meta-analysis found that bundled strategy interventions often do not disaggregate which behaviors relate to which strategies and this could be improved by the use of theory

as a guide (Sharp et al., 2010). In addition, other studies categorized interventions by information processing routes. For example, Koop et al. (2019) used reflective (i.e., conscious, reason driven), semireflective (i.e., heuristics, simple cues), and automatic processing routes to categorize interventions on water conservation behavior within households. Other studies often compared only reflective and automatic processing routes and used various terms, such as System 2 or System 1, hot or cold, and reflective or impulsive (Marteau, 2017).

The use of theories of change and conceptual frameworks can help resolve these inconsistencies in terminology to an extent. In addition, theories and frameworks can guide the design of behavioral interventions, including identifying behavioral constructs and mechanisms and various levels of variables and outcomes that will need to be measured in order to disaggregate effects (Thomson and Ravia, 2011). Despite this, not enough studies use theoretical frameworks to guide design (Sweet and Fortier, 2010; Thomson and Ravia, 2011; Varotto and Spagnolli, 2017). One meta-analysis of health behavior interventions estimated that only about 30 percent of studies used theoretical frameworks to guide their interventions (Sweet and Fortier, 2010).

A significant body of research has demonstrated the efficacy of theory-driven interventions targeting modifiable behaviors (Haggar and Weed, 2019). One review of behavior change related to diet and physical activity found only sparse and inconsistent evidence that theory-based interventions are effective or lead to better outcomes (Samdal et al., 2017), while another review found that interventions structured on behavioral theory techniques are more effective (Belogianni and Baldwin, 2019). In addition, behavioral theories are often poor at explaining how the initiation and the maintenance of behavior might differ (Samdal et al., 2017). Despite this, studies in these domains discuss how theories and frameworks can help to standardize monitoring and evaluation practices and reporting of outcomes (Cox et al., 2010). By standardizing common elements, the next generation of studies could develop new methods for easier interpretability and comparison that investigate change across multiple behaviors and bundled strategies. For example, one meta-analysis suggested exploring the following new methods to capture the complexity of actions underlying behavior change: combining change scores, creating an index score, expanding the impact formula, and using an overarching measure of change (Sweet and Fortier, 2010).

Intention Does Not Always Lead to Behavior

Based on behavioral theories that are widely applied across these domains, there is an assumption that self-reported behavioral intentions lead

to implementation of a behavior. In other words, intended or self-reported behaviors are often considered synonymous with actual behavior change. However, numerous studies have documented that self-reported intentions and their actual behavior frequently do not match (Li et al., 2019; Varotto and Spagnolli, 2017). As such, many authors have suggested future studies move away from or be aware of the use of intentions or clearly distinguish between intention and actual behavior when collecting and interpreting data (Li et al., 2019; Nisa et al., 2019; Varotto and Spagnolli, 2017). For example, one meta-analysis on household recycling behaviors found that individual and contextual factors often predicted intention to recycle, but they did not observe recycling behavior (Geiger et al., 2019). Another review of health communication campaigns found that an increase in knowledge, awareness, or beliefs did not necessarily change targeted behaviors, and it recommended that campaigns should aim to target specific behavior change goals rather than only awareness and should plan to evaluate both (Snyder, 2007). The value-action gap, the awareness-behavior gap, communications gap, and the knowledge gap are also terms that are used to refer to this phenomenon (Li et al., 2019; Sharp et al., 2010; Snyder, 2007).

DRIVERS OF CONSUMER BEHAVIOR

The committee’s review found several themes in the research on drivers of consumer behavior summarized in this section.

Not One but Many Behaviors

Across the domains, the literature suggests that not one but many behaviors determine the outcomes of interest to policy makers and practitioners (Cox et al., 2010). There is no standard set of behaviors that is widely accepted as the set that determines outcomes. In the recycling domain, which has one of the more extensive and well-developed literatures, Li et al. (2019) explained that the behaviors shaping this domain are so complex that a single model would be unable to encompass all the relevant factors. In the waste prevention literature, Cox and colleagues (2010) reported that the vast majority (~70–85 percent) of behavior cannot be explained in current studies due to the multiplicity of behaviors.

A majority of interventions in the six domains were designed as packages of strategies to target several behaviors aimed at an outcome. This approach made it challenging for the meta-analyses and systematic reviews to measure, depict, and disaggregate which strategies influence which behaviors. However, Sharp et al. (2010) and Sweet and Fortier (2010) suggest this may not matter because often an individual strategy might be more influential on a single behavior and less influential on the targeted outcome,

while a package of strategies can be less influential on a single behavior but, additively, more influential on the broader outcome. For example, a review of meta-analyses comparing single and multiple health behavior interventions found that multiple health behavior interventions were more effective at reducing body weight than single behavior interventions (Sweet and Fortier, 2010). The authors explained that multiple behavioral improvements in individual behaviors (e.g., diet, physical activity) added up to greater effects on weight loss when combined (Sweet and Fortier, 2010). Notably, combining such strategies may be more cost-effective, as the additional costs of adding another mode of intervention is likely smaller than the base costs of undertaking an intervention.

Predictive Power of Sociodemographic Variables

Sociodemographic variables were assessed across all domains in terms of their ability to predict proenvironmental behavior. People who were higher income, more educated, older, living in detached properties, and women were more likely to engage in water conservation and recycling behaviors (Addo et al., 2018; Cox et al., 2010; Whitmarsh et al., 2018). People who were younger, more educated, and lower income were more likely to engage in waste reduction behaviors (i.e., buying less, avoiding buying new things) (Whitmarsh et al., 2018). People who were higher income, older, and owned a home were more likely to conserve energy (Karlin et al., 2015). Other studies reported that sociodemographic variables have no significant influence on proenvironmental behavior (Li et al., 2019), that only income predicts recycling behavior (Miafodzyeva and Brandt, 2013); or, that while well-educated people are generally more committed to resource conservation, they actually consume more (Koop et al., 2019). Studies of waste prevention find that while both genders contribute equally to municipal solid waste production, females are more willing to recycle or participate in reduction behaviors, and men are more willing to pay for waste reduction (Ma and Hipel, 2016). Similarly, young people have been found to be more willing to pay for waste prevention and reduction than older people (Ma and Hipel, 2016).

While there are trends by domain in how sociodemographic variables are associated with behaviors, many studies find that these variables do not contribute much to understanding of proenvironmental behavior and that psychological factors are more successful in predicting behavior and behavior change (Li et al., 2019). One meta-analysis suggested that there was no need to tailor recycling interventions to different target groups, such as households, students, or employees, because similar factors seemed to underlie their behavior and the relationship between individual and contextual factors did not differ by group (Geiger et al., 2019). Other studies

have illustrated that as a behavior becomes well established (e.g., recycling), external social pressure no longer predict or significantly influence behavior (Miafodzyeva and Brandt, 2013; Soderhorn, 2010).

Predictive Power of Motivational Factors

It is somewhat tempting to think that simply having enough information about a given behavior or its effects will change individuals’ choices. However, knowledge or information alone was found to be insufficient as a predictor of ability (i.e., knowledge for action) to change and maintain behavior (Abrahamse et al., 2005). By contrast, motivational factors, which take a wide range of forms, seem to be more effective. For example, motivational tools—such as altered attitudes toward outcomes, personal norms, agency or perceived control, and social norms—have turned out to be the most successful predictors and influencers of proenvironmental and weight management behaviors (Li et al., 2019; Miafodzyeva and Brandt, 2013; Samdal et al., 2017). Literature in the weight management domain indicates that people who have self-efficacy and are motivated by their own needs and desires can more readily sustain a behavior (Samdal et al., 2017; Thomson and Ravia, 2011).

Not all motivational factors are egocentric: several meta-analyses illustrate that proenvironmental behavior is more motivated by normative and sometimes environmental concerns than by individual costs and benefits (Geiger et al., 2019; Miafodzyeva and Brandt, 2013). Similarly, environmental attitudes and beliefs, concerns for the future, and an individual’s sense of responsibility—all of which can shape motivation—contribute more to understanding proenvironmental behavior than sociodemographic variables (Li et al., 2019). Norms play a particularly important role in behavior change. Moral norms (i.e., when people feel that doing something aligns with an abstract right or wrong), injunctive social norms (i.e., what one ought to do), and descriptive social norms (i.e., perceptions of what most people are doing) have increased in many societies and are strongly correlated with behavior (Miafodzyeva and Brandt, 2013; Whitmarsh et al., 2018).

These findings suggest that behaviors that are presented as useful, pleasant, important, and widely accepted activities are more likely to be adopted and sustained (Miafodzyeva and Brandt, 2013); conversely, behaviors that are viewed as someone else’s responsibility, inconvenient, or that require a high bar of self-efficacy or locus of control are less likely to be adopted and sustained (Cox et al., 2010). One caveat to this finding is that it may not apply to prevention behaviors that are unseen (e.g., changing acquisition behaviors in order to purchase less in the first place). In instances where

an action is not visible—frequently those categorized as prevention—social norms are unlikely to develop (Cox et al., 2010).

Contextual Factors Affect Opportunities to Initiate and Sustain Behaviors

Several meta-analyses of household recycling interventions found that contextual factors were seldom considered (Geiger et al., 2019; Varotto and Spagnolli, 2017). Studies that included contextual factors, such as having curbside or convenient recycling, a bin at home, or other (e.g., space at home to store recycling), found them to be very strong predictors of waste reduction and recycling behavior (Geiger et al., 2019; Whitmarsh et al., 2018). In one study by Guagnano et al. (1995), the explanatory power of personal norm beliefs decreased when curbside pickup was included. A review of the literature on water conservation behavior found that water pricing was the most important variable explaining differences in domestic consumption in 10 Organisation for Economic Co-operation and Development countries (Koop et al., 2019). Moreover, studies have found that psychosocial factors, such as attitudes and norms, are insufficient for overriding structural barriers to behavior (Karlin et al., 2015).

Interactions among Psychosocial and Contextual Factors

Contextual and psychosocial factors were often found to interact to promote behavior change and maintenance. Households were more likely to adopt behaviors when they felt capable, motivated, and had the opportunity (Addo et al., 2018; Geiger et al., 2019). In a meta-analysis of the causal mechanisms of water conservation behavior, opportunity was a moderate predictor of behavior, followed by motivation and then capability; the three together explained 37 percent of the variance in household behavior (Addo et al., 2018). In this analysis, opportunity was defined as any external physical and social environment factor, such as time, resources, location, and finances, that enables an individual’s behavior; motivation was defined as intrinsic and extrinsic factors, such as attitudes, norms, values, and beliefs; and capability was defined as physical and psychological ability to enact the behavior, such as skills (Addo et al., 2018).

Proximal and Automatic Behaviors Compared with Reasoned Behaviors

More proximal and automatic behaviors have higher behavioral plasticity potential, meaning behavior can change more rapidly or with greater magnitude in response to a stimulus. Choice architecture (i.e., nudges,

removing external barriers) and social comparison interventions (i.e., comparing one’s behavior with others) have been found to be the most efficacious for behavior change when compared with traditional interventions, such as information (e.g., statistics, simple messages, energy labels), appeals (e.g., requests to change behavior for humanity), and engagement (e.g., goal setting, implementation intentions) (Nisa et al., 2019). Thus, proximal and automatic behaviors can be an effective intervention focus.

Habits

There has been inadequate attention paid to habits in comparison with infrequent or one-off behaviors; more recent literature is finding that habits are powerful drivers of behavior. Habits predict and sustain behaviors because they are automatic (Whitmarsh et al., 2018). Habits and established behaviors are powerful because they are not easily influenced by values and norms (Cox et al., 2010; Miafodzyeva and Brandt, 2013), which can be both positive and negative. For example, as waste reduction has become relatively normative in most developed countries, social norms have become insignificant influencers in any context (Whitmarsh et al., 2018). Behavioral interventions aimed at altering habits have been less effective than interventions aimed at influencing one-off behaviors (Nisa et al., 2019). At the same time, interventions that have been successful in creating a new habit find that automatized behaviors are easier to sustain (Nisa et al., 2019). To form new habits, action repetition is needed, and this finding underscores the need for interventions that frequently reinforce or give feedback on actions (Nisa et al., 2019).

How Drivers of Behavior May Differ over Time and Context

Context cues much of human behavior, and different motivations and barriers operate in different contexts, meaning that many actions are inconsistent across different times and places (Nash et al., 2017; Whitmarsh et al., 2018). Some literatures indicate that grouping drivers and tailoring interventions by different contexts is more important than by sociodemographic groups (Cox et al., 2010; Whitmarsh et al., 2018). Similarly, behavioral drivers may differ over time, both societally and individually. However, little is known about how drivers may differ at different phases in the behavior change process (Samdal et al., 2017).

Differences between Drivers that Initiate and Those that Maintain Behavior

There are important differences in how behaviors are formed and sustained and in how established behaviors are broken to form new behaviors (Miafodzyeva and Brandt, 2013). Interventions designed to help people initiate behavior may need to target different behavioral drivers than those that help people maintain behavior (Samdal et al., 2017). For example, once behaviors are established (e.g., recycling), they are less affected by such factors as social norms and expectations (Miafodzyeva and Brandt, 2013). One systematic review of 100 theories of behavioral change found five explanations of the differential roles of motives: self-regulation, psychological resources, physical resources, contextual influences, and habits from initiation to maintenance (Kwasnicka et al., 2016). This review found that people need at least one sustained motivator to maintain a behavior change and that people will often initiate a change when motivation is high and effort is low. When motivation is reduced and effort or costs increase, people will often need some way to self-monitor in order to sustain the change, which can be challenging when other things in their lives are simultaneously occurring, such as stress, tiredness, other more precarious issues (such as finances). Thus, turning a new behavior into a habit can be advantageous because external factors (e.g., changes in motivation or effort) are less likely to affect the behavior. Stable contexts can make behavior maintenance easier (Kwasnicka et al., 2016).

INTERVENTIONS

The literature on interventions was difficult to compare and contrast because of differences in terminology and how interventions were categorized, but a number of topics were addressed across the six domains. This section summarizes research on key questions about intervention effectiveness and some themes identified by researchers in these areas.

Intervention Approaches

Single or Combined Behavioral Interventions

Consistent with the broader psychological literature, the vast majority of the reviews found that behavioral interventions were only mildly to moderately effective on their own (Cox et al., 2010; Koop et al., 2019; Marteau, 2017; Thomson and Ravia, 2011; Varotto and Spagnolli, 2017).

A review by Sweet and Fortier (2010) examined whether interventions focused on a single behavior (e.g., physical activity or diet) were more or

less effective than interventions focused on multiple behaviors (e.g., physical activity and diet) by reviewing meta-analyses and reviews. They also explored the differential effects of these interventions on weight outcomes. Notably, the analysis found that while single behavior interventions were more effective at improving the targeted behavior, multiple behavior interventions produced greater weight loss even though they appeared to be less effective at changing the individual behaviors. The authors hypothesized that this occurred because changing several behaviors at once in small, nonsignificant ways might add up to a greater overall effect. This result suggests that if a single behavior is the target that a single behavior intervention might be the most effective, but if the aim is to change more behaviorally complex outcomes, such as weight, multiple behavior interventions might be more effective. Finally, while the samples were too small to draw conclusions, there may be differences in whether multiple behavior interventions introduce behavior changes simultaneously or sequentially (Sweet and Fortier, 2010).

Targeting One-Time or Single-Action Behaviors as Well as Habits

Single-action behaviors may be less resistant to change as compared with habits and may be more effectively targeted by behavioral interventions (Nisa et al., 2019). One-off actions (e.g., purchase of energy efficient appliance) have been found to have higher behavioral plasticity, meaning they are more likely to change in response to the application of effective intervention (Nisa et al., 2019). However, emerging evidence indicates that behavioral interventions can be better designed to target habits and habitual thinking by incorporating more regular and frequent delivery of information or feedback, by providing specific tips or skills, or by disrupting existing habits to embed new habits (Cox et al., 2010). For example, behavior change techniques that facilitate self-regulation of behavior (e.g., goal setting of behavior and self-monitoring of behavior) can be effective in helping people to both initiate and maintain dietary changes, while techniques that facilitate person-centered or autonomy supportive communication (e.g., problem solving, review of behavioral goals and receiving social support) are important to maintain behavior change (Samdal et al., 2017). Behavior change techniques that combine the “how to” (i.e., facilitate behavior self-regulation, such as skills) with “the why” (i.e., addresses the underlying reasons for motivation) can reinforce both an individual’s competence and need for meaning, value, and satisfaction in order to change behavior (Samdal et al., 2017). This finding is corroborated by other reviews that have indicated that successful weight management interventions are composed of an integrated mix of information, support, encouragement, progress monitoring, and feedback (Sharp et al., 2010).

Contextual or Environmental Factors as Barriers to Behavior Change

Contextual factors can support or override an individual’s desires and attempts to consume or waste less (Cox et al., 2010). Varotto and Spagnolli (2017) conducted a random-effects meta-analysis of 36 studies (1990-2015) reporting 70 psychological strategies to promote household recycling in the home environment: they found environmental alterations to be the second most effective strategy, after social modeling. Environmental alterations were described as modifying the existing physical environment, such as adding home equipment for waste sorting (Varotto and Spagnolli, 2017). The effectiveness of this strategy was attributed to the presence of environmental cues that prompted behavior and minimized the effort required to implement the behavior. For example, the addition of bins can initiate behavior but the presence of bins in a neighborhood can increase awareness of a program and reinforce social norms. In a review of health communication campaigns, Snyder (2007) acknowledged the necessity to change other people or contexts in order to change the target population and explained that campaigns can vary in their use of communication strategies “to try to change the behavior of the target population, including strategies that attempt to change the political and economic context in which people are making decisions, those aimed directly at the populations, and those aimed at people who may have influence with the target population” (Snyder, 2007, p. S35). Often, environmental changes are needed first, and campaigns can serve the role of publicizing them or attempting to stimulate demand for a new option.

A study by Whitmarsh et al. (2018) looking at behavioral consistency across contexts examined waste reduction behaviors at home, at work, and on vacation to determine whether consistency was a function of proenvironmental identity. The study found that the proportion of waste recycled at home was greater than that in the workplace or on holiday (67 percent, compared with 39 percent and 38 percent, respectively) and that repair and reuse behaviors were more common at home than at work. The prevalence of behaviors by context was consistent with the literature that describes work and vacation contexts as places when people are less motivated to act proenvironmentally or experience less control over barriers to behavior (Whitmarsh et al., 2018). Contextual factors and perceived behavioral control were found to be as important for predicting recycling behavior as motivational and normative factors. Recycling knowledge and personal norms predicted behavior. The authors concluded that proenvironmental identity was not a significant predictor of cross-contextual consistency (Whitmarsh et al., 2018, p. 10):

Preventing Consumption in the First Place Compared with Promoting Reuse or Recycling

Modern culture’s drive to consume and the fact that waste prevention behaviors lack the same visibility as such activities as recycling make preventing overacquisition of items more challenging than reuse or recycling (Cox et al., 2010). In the recycling literature, one of the biggest barriers to household waste prevention was the fact that people often mistake recycling as waste prevention (Cox et al., 2010). Households are more likely to participate in reuse than reduce behaviors (e.g., donation vs. avoidance) (Cox et al., 2010). Moreover, because waste prevention behaviors are often not visible, there are no descriptive or injunctive social norms to support this identity (Cox et al., 2010). Thus, the notion of tapping into an individual’s intrinsic identity around “ethic of care” for products, the environment, or wider society was highlighted as a generally successful way to raise participation in the hidden behaviors of waste prevention (Cox et al., 2010).

Behavioral Boomerang or Rebound Effects

The tendency when given feedback (e.g., social comparison) for individuals that are performing better than average to increase their consumption is known as a “boomerang” or “rebound” effect (Andor and Fels, 2018). Andor and Fels (2018) performed a systematic review of causal studies and compared four behavioral economic intervention types on energy conservation: social comparison, commitment devices, goal setting, and labeling. Only 1 study of 24 found a “boomerang effect.” However, this individual study also noted that this boomerang effect could be eliminated by adding an injunctive message (Schultz et al., 2007). Similarly, Gillingham et al. (2013) looked at the rebound effect of energy efficiency policies: “studies and simulations indicate that behavioral responses shave 5-30 percent off intended energy savings, reaching no more than 60 percent when combined with macroeconomic effects” (Gillingham et al., 2013, p. 476). Even when taking rebound effects into account, interventions can result in substantial change.

Interventions Using Financial Strategies and Their Possible Moderation by Intrinsic Motivation

Financial interventions have been found to be more influential for behavior change than psychosocial behavioral interventions (Nisa et al., 2019). Within the diet change literature, both taxation and subsidization were consistently found to influence dietary behaviors in the directions in which they were designed to work (i.e., subsidies increase consumption of healthier foods, taxes reduce purchases of less healthy foods) and to work well in tandem (Niebylski et al., 2015). In the research on residential solid waste management, studies have examined the effectiveness of fees to reduce residential solid waste disposal. Facing high costs for solid waste disposal and difficulties in locating new landfill and incineration sites, about one-quarter of U.S. communities charge a fee for residential solid waste collection (Skumatz, 2008). These programs, which are also known as pay-as-you-throw or unit-based pricing programs, shift the costs faced by the community to individual households and are intended to reduce total household disposal amounts. Given estimates that about 20 percent of landfill content is wasted food and food scraps (U.S. Environmental Protection Agency, 2015), such policies have immediate implications for community efforts to reduce wasted food within households.

Bel and Gradus (2016) conducted a meta-analysis of 25 studies (1970–2013) that estimated the responsiveness (i.e., elasticity) of household disposal levels to the imposition of such fees. Across all studies, they found an average elasticity estimate of -0.34, that is, that a 10 percent increase in the price charged for solid waste collection led to a 3.4 percent reduction in the amount of waste collected. However, the responsiveness was significantly greater when fees were applied separately to compostable waste or when the price charged was based on the weight of the solid waste rather than on the number of bins or bags of waste from a household. This led Bel and Gradus (2016, p. 178) to summarize that “. . . a fee for compostable waste is . . . therefore highly effective,” while questioning the efficacy of imposing fees that did not vary with the weight of the material to be discarded. A key issue with imposing such fees is that residents may respond by disposing of materials outside the fee-based system. Fullerton and Kinnaman (1996) found that about 28 percent of the reduction in waste from a fee-based disposal program in Charlottesville, Virginia, was actually being disposed of illegally through other outlets. However, Allers and Hoeben (2010) report the most municipalities that have imposed fee-based systems are generally satisfied with the system, suggesting illegal dumping is not a large enough issue to disillusion adopters. One important caveat to financial strategies may be that they can negatively affect intrinsic motivations. Prior studies have found that while household demand responds to price, price elasticity

can be low in the short term or counterproductive because it crowds out other more altruistic or prosocial motivations (Delmas et al., 2013).

Reflective (i.e., System 2) Interventions

Reflective interventions aim at giving people information or appealing to their self-efficacy and rational decision making. Such interventions, designed to increase a person’s knowledge about reasons for performing a behavior or appealing to their self-efficacy, are insufficient to promote behavior change (Koop et al., 2019; Sharp et al., 2010; Thomson and Ravia, 2011; Varotto and Spagnolli, 2017). However, reflective or information campaign interventions may promote behavior when people are motivated but do not know exactly how to implement a behavior (Varotto and Spagnolli, 2017). Samdal et al. (2017) corroborated these findings by concluding that behavior change interventions that combine motivation with opportunity and ability can be effective in initiating and sustaining behavior change. Reflective interventions can reinforce an individual’s competence, as well as their need for meaning, value, and satisfaction in order to change behavior. Further, Ma and Hipel (2016) pointed out that while public education interventions are insufficient to change societal behavior around municipal solid waste, they can represent a long-term path to societal consciousness. In order to shift societal norms, interventions might best be designed to change beliefs, motivations, and attitudes toward policies and programs rather than having a sole focus on behavior change (Ma and Hipel, 2016).

Semireflective Interventions for Long-Term Behavior Change

Social norms, framing, and tailoring are categorized as semireflective interventions because they represent an individual’s attempt to use simple cues or rules about which choices should be made (Koop et al., 2019). In particular, normative messages are effective and repeating these messages can support long-term behavior change (Koop et al., 2019). The framing of messages appears to be important, and messages framed as suggestive, emphasizing direct impacts or real-time information, or that appeal to intrinsic motivation (e.g., conserve for the future) as opposed to extrinsic motivation (e.g., save water and reduce costs) are the most persuasive (Koop et al., 2019). In addition, messages of competitive peer ranks (i.e., social comparison) are more effective with low-consuming households than neutral rank (e.g., average neighbor household consumption), which are more effective with high-consuming households (Koop et al., 2019). Personalized messages or those that reveal attitude behavior discrepancies also invoke behavior change (Koop et al., 2019).

Automatic Interventions (i.e., System 1)

Automatic interventions are those that use emotional cues, primes, and nudges to change behavior. Using emotional shortcuts, priming, and nudging are categorized as automatic because they represent automatic responses by people (Koop et al., 2019). Koop et al. (2019) found that while the use of emotional cues, primes, and nudges to stimulate domestic water consumption has only been explored in small samples or short-duration studies, they show promise due to the amount of water savings they produced. Similarly, Nisa et al. (2019) conducted a meta-analysis of 83 randomized controlled trials (1976-2017) to explore the most effective mechanisms (i.e., choice architecture, social comparison, information, appeals, and engagement) for promoting household action on climate change. While fewer in number, the strategies that had the highest effect sizes and showed the most promise were choice architecture (i.e., nudge) approaches.

Social Influence Approaches

Varotto and Spagnolli (2017) conducted a random-effects meta-analysis of 36 studies (1990-2015) reporting 70 psychological strategies to promote household recycling in the home environment. They found social modeling to be the most effective strategy, compared with environmental alterations, combined strategies, prompts and information, incentives, commitment, and feedback. The analysis found that social modeling, which was described as the passing of information by people (e.g., block leaders, children to their parents) who also personally engage in the behavior, was effective because it engendered social norms.

Abrahamse and Steg (2013) conducted a random-effects meta-analysis of 29 studies that used social influence approaches to improve resource conservation (e.g., energy savings and use, gas and electricity savings and use, showering time, water use, recycling). The results of the analysis found that, compared with control groups, social influence approaches were effective and that greater effect sizes were found with the block leader, public commitment, and modeling approaches and smaller effect sizes with group and socially comparative feedback and social norms in information and feedback provision approaches. The authors suggested that the approaches that were found to be more effective might be due to their face-to-face delivery mode, and they questioned whether this was cost-effective. The magnitude of the effect depended on the target group but not the type of proenvironmental behavior. Specifically, employees appeared to be the most affected by social influence approaches, followed by students, households, farmers, and hotel guests.

Andor and Fels (2018) performed a systematic review of 44 causal-effect studies and compared 4 behavioral economic intervention types on energy conservation: social comparison, commitment devices, goal setting, labeling. The authors found that social comparison interventions were the most effective (ranging from 1.2 to 30 percent reduced energy consumption) as well as the most researched, both in terms of quantity and quality. Social comparison interventions appeared to differ in effect on the basis of the mode of delivery, with online or in-home displays being more effective than letters. Nisa et al. (2019), in a meta-analysis of 83 randomized controlled trials to explore the most effective mechanisms for promoting household action on climate change, found social comparisons to be the second most effective approach for behavior change after choice architecture (i.e., nudges).

Despite these supportive findings of social influence approaches, there may be implementation challenges, such as whether such a strategy can be consistently scaled in areas with low social connectedness or where block leaders are unavailable (Varotto and Spagnolli, 2017).

Feedback Approaches

Giving people information about their behaviors that they can use to modify future actions could be effective and engaging ways to alter behavior. Delmas et al. (2013) performed a meta-analysis of 156 information-based energy conservation trials in residential settings (1975-2012) and found that nonmonetary, information-based approaches can be effective for reducing energy usage. Information strategies included in the analysis were savings tips, energy audits, different forms of energy use feedback, and monetary feedback. They found that, on average, individuals in the trials reduced their electricity consumption by 7.4 percent. In general, individuals receiving real-time feedback or experiencing high involvement interventions, such as home energy audits, reduced their electricity use, and individuals receiving lower-level information or less intensive feedback, such as energy saving tips or individual usage feedback and comparative feedback, did not.

Karlin et al. (2015) performed a meta-analysis on the effectiveness of feedback intervention studies in residential settings for conserving energy, as well as how they vary by the treatment moderators of frequency, medium, measurement (e.g., cost or carbon), combination with other interventions, comparison message, granularity, and duration. Feedback resulted in an average energy savings of 12 percent across studies, which was consistent with prior research that found a range of 8 to 12 percent (Karlin et al., 2015). Variables that moderated this effect included medium, comparison message, duration, and combination with other interventions (e.g., goal,

incentive), while feedback frequency, granularity, and medium did not. More engaging mediums (e.g., computer) appeared more effective than less engaging mediums (e.g., a utility bill). Studies using goal-based comparisons showed significant effects compared with controls, while social and historical comparisons did not. The authors underscored the relevance of this finding because the use of social comparisons is the most commonly used type of feedback by industries, such as public utilities. Users’ attention to feedback can vary over the duration the feedback is provided, with users generally engaging more initially and then less over time. At the same time, longer durations of feedback may be necessary to allow habits to be created and maintained. Finally, feedback was most effective when it was combined with goal-setting or external incentive interventions.

Promoting Healthy Behaviors Compared with Reducing Unhealthy Behaviors

Carrero et al. (2019) conducted a meta-analysis of 70 interventions to assess the efficacy of implementation intention interventions for promoting healthy eating behaviors. Implementation intention interventions are defined as “volitional planning interventions that support the realization of goal intentions by delegating the control of goal-directed responses to anticipated situation cues that elicit these responses automatically” (Carrero et al., 2019, p. 239). For example, using if-then plans to detail where, when, and how one intends to behave in a future situation. These interventions are less effective at reducing unhealthy behaviors (e.g., eating less fat) and more effective when promoting healthy eating behaviors (e.g., eating more fruit), possibly because of the challenge of breaking a habit compared with initiating a new behavior (Carrero et al., 2019). Moderators for unhealthy and healthy eating goals differed. For unhealthy eating, plan formulation was the only significant moderator variable and implementation intention interventions had low efficacy regardless of intervention design. The variable plan formulation indicated that when these plans were designed only to avoid the unhealthy food, they were less effective than when they were planned with an alternate positive action in mind. For healthy eating, moderator variables explained 53 percent of the variance; effect size was negatively predicted by age, with younger people having more favorable outcomes than older people. It was also affected by an implementation intention check, meaning that an instructor checking the plans reduce the intervention’s efficacy. Effect sizes were positively predicted by initial training, off-line delivered interventions, and specific if-then plans and action plans, in comparison with more complex plans.

Communication Campaigns Aimed at One-Time or Infrequent Behaviors

In a narrative review on how health communications campaigns affect behavior, Snyder (2007) described the overall impact of communication campaigns and some of the most important lessons learned from prior health campaigns in terms of campaign planning (i.e., goals and strategies of the campaigns). The review found that, on average, health campaigns can positively affect outcomes in interventions communities by about 5 percent and have an average reach of 40 percent of their target populations. Short-term and intense campaigns with more frequent exposures resulted in greater short-term effects. In general, campaigns that promoted the adoption of new or replacing an old behavior with a new behavior or a change in an infrequent or one-time behavior were more successful than campaigns aimed at a habit, such as stopping an unhealthy behavior already in practice, or preventing initiation of risky behaviors.

Gamification Interventions

In an interesting systematic review of the use of gamification and serious games on domestic energy consumption, Johnson et al. (2017) systematically reviewed 26 studies to assess the potential of using well-designed digital games to change energy consumption behavior. Serious games were defined as “fully fledged games (e.g., a digital role-playing game in which the player completes challenges or quests designed to educate them about nutrition), while gamification refers to the application of parts of games in a non-game setting (e.g., a mobile phone app designed to track and encourage exercise that uses levels, points, and badges” (Johnson et al., 2017, p. 249). While differing widely in methodology, intervention design and framework, and disciplinary focus, the studies found that applied games had a positive effect on behavior or behavioral antecedents. Two high-quality studies in the sample compared different gaming elements, such as feedback, challenges, social sharing, rewards, leaderboards, and points, and found that competition and social sharing showed effectiveness for encouraging participants to adopt specific behaviors. Only two high-quality studies looked at cognitive outcomes; they both found positive changes in attitudes toward and awareness of energy consumption. Several studies in the sample reported improvements in general but not specific energy consumption and conservation knowledge. Interestingly, the games appeared to have led to improvements in self-reported and actual energy conservation behavior in the short term. The authors concluded that while these initial studies were far from conclusive, the use of applied games holds promise for positively impacting energy consumption.

Applying Research Findings to Intervention Design

Researchers across the six domains have begun to identify ways to apply their findings about the nature and operation of interventions to provide broader guidance to intervention designers. This section summarizes the support for some key ideas.

Targeting Multiple Behaviors Using Multiple Approaches

Based on their analyses, several authors concluded that the best approach to behavior change was a comprehensive approach that combined behavioral interventions with other approaches such as partnerships with influential organizations, social marketing programs, economic incentives, regulations, or technology (Cox et al., 2010; Koop et al., 2019; Niebylski et al., 2015; Nisa et al., 2019; Sharp et al., 2010; Thomson and Ravia, 2011). Nisa et al. (2019) underscored that behavioral interventions would not be enough because of the low behavioral plasticity of most behaviors and recommended that behavioral interventions might be more effective when used in combination with other strategies, such as financial incentives or policy regulations. For example, financial incentives might initiate behaviors but then be reinforced by behavioral strategies. Or, interventions could be sequenced to initiate with motivating, eye-catching strategies (e.g., financial incentives, social marketing) and move to or add on more information-based strategies to reinforce change (Nisa et al., 2019). Cox et al. (2010) emphasized that interventions are a part of wider social, institutional, and political conditions. Ma and Hipel (2016) explained that successful interventions should also involve all stakeholders (e.g., government, private sector, nongovernmental organizations, the informal economic sector), all factors (e.g., economic, environmental, and social), and incorporate public participation. An integrated range of intervention tools and partnerships can effectively make collective and cumulative impacts (Koop et al., 2019; Sharp et al., 2010).

Comprehensive Interventions

Many studies recommended the development of more comprehensive and conjunctive approaches that address intrinsic and extrinsic motivation, opportunity, and ability; appeal to both rational and emotional processes; and use a systems approach. The goal is to address the complexities of influences on targeted outcomes. As Geiger et al. (2019) explained, the application of several theories of behavior change are needed simultaneously in order to account for the variety of individual costs and benefits and normative and environmental concerns that play a role in explaining behavior,

illustrating the need for an integrated approach. Cox et al. (2010, p. 211) concurred: “[N]o single approach is sufficient on its own, rather a ‘hybrid’ method using a suite of monitoring approaches” and recommended that behavior change interventions be composed of a suite of interventions and measures that are needed simultaneously to facilitate and evaluate change.

Koop et al. (2019) recommended conjunctive use of reflective, semireflective and automatic tactics (i.e., reasoned, rational, and emotional processes) to influence behavior, such as persuasive technologies. In particular, the authors recommended interventions that consisted of repetitive messages, primes, and nudges that reinforce previously introduced normative messages, tailored feedback and knowledge. They found that knowledge transfer is only meaningful when people know they can change their behavior and consider it feasible and when tailored feedback is reinforced by repetition, social norms, and message framing (Koop et al., 2019). Miafodzyeva and Brandt (2013) proposed a framework for effective recycling interventions that combined the moral reasons and environmental concerns of the household with the awareness and knowledge of recycling programs and the removal of any major convenience barriers. Snyder (2007, p. S38) concluded that “a comprehensive strategy that addresses policy and environmental constraints, individual factors in behavior change, and social influences on the target population should be considered.” In a meta-analysis by Maki et al. (2019) on proenvironmental behavior spillover, the authors found that positive spillover was most likely when interventions target intrinsic motivation.

Interventions that Are Tailored by Context, Phase, and Segment

One study recommended segmenting audiences by context or behavior rather than by demographic group in order to target messages and recommendations, such as “targeting by behavior, actual and perceived risk, misinformation and beliefs, environmental barriers, and communication patterns” (Snyder, 2007, p. S35) Another study suggested that large-scale strategies can be implemented without need for tailoring as long as context was at the forefront (Geiger et al., 2019). One author underscored results showing that implementation intentions (i.e. planning interventions that support the realization of goal intentions) interventions are more effective in young adults—a time when there is a marked increase in initiation and maintenance of habits (Carrero et al., 2019).

Measuring both Isolated and Combined Effects of Different Behavioral Strategies

There is value in trying to understand both isolated and combined effects of different behavioral strategies (Nisa et al., 2019). In addition, interventions that are well designed to account for intermediate and outcome variables can best assess how a combination of variables adds up to reach an impact (Miafodzyeva and Brandt, 2013). For example, how improving recycling facilities and giving bins to homes interacts to strengthen attitudes and perceived behavioral control (Geiger et al., 2019). Belogianni and Baldwin (2019) emphasized the need to measure actual behaviors over intentions and that changes in intermediate variables, such as knowledge, self-efficacy, and attitude, were important for understanding the mechanisms of behavior change.

Increased Study Duration to Track Maintenance

More understanding is needed of how to prolong and reinforce newly formed habits (Koop et al., 2019). Future studies should be of longer duration in order to maintain and monitor behavior change (Fjeldsoe et al., 2011). Fjeldsoe et al. (2011) conducted an interesting systematic review examining the effect of physical activity and dietary intervention trials on behavior maintenance (n = 29), with maintenance defined as “a physical activity, dietary or combined intervention trial that was considered to demonstrate maintenance of behavior change if a statistically significant between-groups difference in favor of the intervention group was reported at end-of-intervention and at follow-up for at least one behavioral outcome” (Fjeldsoe et al., 2011, p. 102). After a minimum of 3 months postintervention, Fjeldsoe et al. (2011) noted several interesting findings. First, of the 157 trials initially examined, only 35 percent included behavior maintenance outcomes. Second, of the 29 trials that included maintenance outcomes, participants in 72 percent of studies achieved maintenance of at least one outcome, and 38 percent achieved maintenance on all outcomes. In addition, trials with retention rates of greater than 70 percent were less likely to achieve maintenance than those with lower retention rates. Longer duration trials (more than 24 weeks) were more likely to achieve maintenance, as were trials that included face-to-face contact, used more than six intervention strategies, and included follow-up prompts after the main part of the intervention to reinforce intervention content.

Readiness for Scaling Up

A few studies cautioned against scaling up before understanding more about which strategies affected which behaviors and outcomes because of the costs of large-scale interventions, particularly face-to-face interventions (Abrahamse and Steg, 2013; Andor and Fels, 2018). Nisa et al. (2019) illustrated the reduced effects on behavior that occur when an intervention is scaled up to the general population. Andor and Fels (2018) described the interventions in their analysis as potentially combining too many strategies, which made it difficult to discern the “pure effects” that should be scaled up. These authors recommended the practice of performing impact evaluations prior to rolling out policy or large-scale interventions. Sweet and Fortier (2010) recommended that it would be useful to understand whether strategies should be deployed simultaneously or sequentially.

Positive or Negative Messaging

Carrero et al. (2019) recommended that policy makers avoid negatively framed policies. Cox et al. (2010) and Sharp et al. (2010) discussed how tapping into a culture or ethic of care was more important than aligning with “green” behavior. And Li et al. (2019) described that focusing on the positive benefits of a particular behavior could bring higher place attachment and improve quality of life.

SUGGESTIONS FOR FURTHER STUDY

Researchers in the six domains have made suggestions for further study on behavioral change, covering a wide range of topics: habits, interventions on contextual factors, understanding why interventions work, equity, the generalizability of interventions, the persistence of intervention effects over time, the dearth of effectiveness studies in comparison with efficacy studies, better study design to track pathways, the need for more cross-context understanding, evaluation studies, and cost-effectiveness studies.

Not enough is known about habits. More understanding is needed about habits, such as how habits differ from more one-off and infrequent behaviors, how to undo old and create new habits, how to prolong and reinforce newly formed habits, and how interventions may differ between those that target one-off and infrequent behaviors and those that target habits (Koop et al., 2019).

Interventions targeting contextual factors are underrepresented. Linking the drivers-based evidence with the intervention-based research is

challenging as some determinants and interventions (e.g., contextual) are systematically underrepresented and some are widely covered (e.g., psychological: motivation, information and knowledge, beliefs/perception, social influence) (Koop et al., 2019; Nisa et al., 2019; Varotto and Spagnolli, 2017).

The vast majority of existing interventions illuminate whether specific interventions work but not why. Many meta-analyses and systematic reviews found that relatively few studies included measures of behavioral antecedents, such as social norms, attitudes, or knowledge, and thus could not explain why an intervention worked or what it changed, only whether it worked (Abrahamse and Steg, 2013; Abrahamse et al., 2005). Both the “how to” and the “why” are important in learning how to design an effective intervention that includes techniques to both initiate and maintain behavior change (Samdal et al., 2017).

Few studies addressed equity or equity components. Few studies included measures of outcomes or discussions related to equity. In one of the only studies to do so, Ma and Hipel (2016) conducted a systematic literature review on municipal solid waste management to understand the social dimensions of that management. The review highlighted that the negative effects of solid waste were inequitably distributed among populations and that more vulnerable populations often bear the negative consequences of being near or able to see waste sites. This inequity means that more vulnerable populations often have to advocate for waste management and often do not gain traction because it is not a problem equally experienced by all. For example, in many cases more vulnerable populations were more exposed to the environmental contamination of solid waste disposal and while this affected their awareness and attitudes, it did not affect other societal strata in the same way. In addition, these populations lacked the agency to change.

Little is known about the extent to which interventions are generalizable to large-scale populations. Most studies mention the challenges of generalizability in terms of the extent to which findings from a behavioral intervention implemented in a specific (geographical, cultural, and behavioral) context can be transferred to a different population. In particular, there is a question about whether the small-scale experiments that often show bigger effects can be effectively scaled up and at what cost (Sharp et al., 2010). In a meta-analysis by Nisa et al. (2019), when interventions were restricted to more generalizable studies (i.e., those with large samples and naïve subjects), the expected probability was reduced to 2-3 percent, a reminder that experimental intervention effects will be more tempered when

applied to a general population. These authors recommended conducting trials in large samples with naïve populations or restricting subanalyses within systematic reviews to large, naïve samples to understand how effect sizes might be lowered in more general populations.

Little is known about how intervention effects persist over time. Little is known about the long-lasting effects of interventions over time (Abrahamse and Steg, 2013; Koop et al., 2019; Nisa et al., 2019; Snyder, 2007; Varotto and Spagnolli, 2017). Future research is needed to understand which behaviors can be sustained and which interventions stand the test of time (Belogianni and Baldwin, 2019; Niebylski et al., 2015).

The literature has efficacy studies but is remiss in effectiveness studies. Literature in all the domains was largely focused on the efficacy of behavior change interventions but not effectiveness (Bowen et al., 2015). There is an over-emphasis in these studies on whether an intervention is successful, but not why (Abrahamse and Steg, 2013).

Better study designs are needed for parsing impact pathways. More sophisticated study designs are needed that allow for the parsing of study variables (e.g., behaviors, outcomes) so one can learn what can be expected from different approaches and different strategies within these approaches. The majority of current research does not evaluate behavior constructs or how they influence intervention efficacy and therefore best practices cannot be identified (Sweet and Fortier, 2010; Thomson and Ravia, 2011). Most meta-analyses and systematic reviews mentioned the need for a better understanding of the particular pathways of change. For example, how behavior change pathways differ for initiation versus maintenance, for one-off or infrequent behaviors versus habits, and for forming new habits versus breaking old habits, as well as for impact patterns, change over time, and how variables interact with one another.

There is not enough cross-context understanding. There needs to be a better understanding of behaviors and outcomes across contexts. There are different motivations and barriers operating in different contexts, and no single model will transfer across contexts (Whitmarsh et al., 2018). These differences are nuanced. For example, recycling is more common and more transferable across contexts than behaviors aimed at reducing, repairing, and reusing, and there are more barriers to waste reduction (i.e., recycling and reuse) outside the domestic context than within it (Whitmarsh et al., 2018).

Evaluation studies of intervention implementation are needed. There is a need for formative research, monitoring research, and evaluative research to design, monitor implementation, and evaluate how implementation affects impacts (Snyder, 2007).

Cost-effectiveness studies are needed. There was a common call among researchers for better understanding of the costs of interventions at scale (Snyder, 2007).

REFERENCES

Guagnano, G.A., P.C. Stern, and T. Dietz. 1995. Influences on attitude-behavior relationships: A natural experiment with curbside recycling. Environment and Behavior 27(5):699-718.

Hagger, M.S., and M. Weed. 2019. Debate: Do interventions based on behavioral theory work in the real world? International Journal of Behavioral Nutrition and Physical Activity 16(1):36.

Johnson, D., E. Horton, R. Mulcahy, and M. Foth. 2017. Gamification and serious games within the domain of domestic energy consumption: A systematic review. Renewable and Sustainable Energy Reviews 73:249-264.

Karlin, B., J.F. Zinger, and R. Ford. 2015. The effects of feedback on energy conservation: A meta-analysis. Psychological Bulletin 141(6):1205-1227.

Koop, S.H.A., A.J. Van Dorssen, and S. Brouwer. 2019. Enhancing domestic water conservation behaviour: A review of empirical studies on influencing tactics. Journal of Environmental Management 247:867-876.

Kwasnicka, D., S.U. Dombrowski, M. White, and F. Sniehotta. 2016. Theoretical explanations for maintenance of behaviour change: A systematic review of behaviour theories. Health Psychology Review 10(3):277-296.

Li, D., L. Zhao, S. Ma, S. Shao, and L. Zhang. 2019. What influences an individual’s proenvironmental behavior? A literature review. Resources, Conservation and Recycling 146:28-34.

Ma, J., and K.W. Hipel. 2016. Exploring social dimensions of municipal solid waste management around the globe—A systematic literature review. Waste Management 56:3-12.

Maki, A., A.R. Carrico, K.T. Raimi, H.B. Truelove, B. Araujo, and K.L. Yeung. 2019. Meta-analysis of pro-environmental behaviour spillover. Nature Sustainability 2(4):307-315.

Marteau, T.M. 2017. Towards environmentally sustainable human behaviour: Targeting nonconscious and conscious processes for effective and acceptable policies. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 375(2095).

Miafodzyeva, S., and N. Brandt. 2013. Recycling behaviour among householders: Synthesizing determinants via a meta-analysis. Waste and Biomass Valorization 4(2):221-235.

Nash, N., L. Whitmarsh, S. Capstick, T. Hargreaves, W. Poortinga, G. Thomas, E. Sautkina, and D. Xenias. 2017. Climate relevant behavioral spillover and the potential contribution of social practice theory. Wiley Interdisciplinary Reviews: Climate Change 8(6):e481.

Niebylski, M.L., K.A. Redburn, T. Duhaney, and N.R. Campbell. 2015. Healthy food subsidies and unhealthy food taxation: A systematic review of the evidence. Nutrition 31(6):787-795.

Nisa, C.F., J.J. Belanger, B.M. Schumpe, and D.G. Faller. 2019. Meta-analysis of randomised controlled trials testing behavioural interventions to promote household action on climate change. Nature Communications 10(1):4545.

Samdal, G.B., G.E. Eide, T. Barth, G. Williams, and E. Meland. 2017. Effective behaviour change techniques for physical activity and healthy eating in overweight and obese adults; systematic review and meta-regression analyses. International Journal of Behavioral Nutrition and Physical Activity 14(1).

Schultz, P.W., J.M. Nolan, R.B. Cialdini, N.J. Goldstein, and V. Griskevicius. 2007. The constructive, destructive, and reconstructive power of social norms. Psychological Science 18(5):429-434.

Scott, A., C. Oates, and W. Young. 2015. A conceptual framework of the adoption and practice of environmental actions in households. Sustainability (Switzerland) 7(5):5793-5818.

Sharp, V., S. Giorgi, and D.C. Wilson. 2010. Delivery and impact of household waste prevention intervention campaigns (at the local level). Waste Management & Research 28(3):256-268.

Skumatz, L.A. 2008. Pay as you throw in the US: Implementation, impacts, and experience. Waste Management 28(12):2778-2785.

Snyder, L.B. 2007. Health communication campaigns and their impact on behavior. Journal of Nutrition Education and Behavior 39(2 SUPPL.):S32-S40.

Soderhorn, P. 2010. Environmental Policy and Household Behavior: Sustainability and Everyday Life. Abingdon, UK: Earthscan.

Sorrell, S. 2007. Improving the evidence base for energy policy: The role of systematic reviews. Energy Policy 35(3):1858-1871.

Sweet, S.N., and M.S. Fortier. 2010. Improving physical activity and dietary behaviours with single or multiple health behaviour interventions? A synthesis of meta-analyses and reviews. International Journal of Environmental Research and Public Health 7(4):1720-1743.

Thomson, C.A., and J. Ravia. 2011. A systematic review of behavioral interventions to promote intake of fruit and vegetables. Journal of the American Dietetic Association 111(10):1523-1535.

U.S. Environmental Protection Agency. 2015. Advancing sustainable materials management: 2013 fact sheet. Available: https://www.epa.gov/sites/production/files/2015-09/documents/2013_advncng_smm_fs.pdf.

Varotto, A., and A. Spagnolli. 2017. Psychological strategies to promote household recycling. A systematic review with meta-analysis of validated field interventions. Journal of Environmental Psychology 51:168-188.

Whitmarsh, L.E., P. Haggar, and M. Thomas. 2018. Waste reduction behaviors at home, at work, and on holiday: What influences behavioral consistency across contexts? Frontiers in Psychology 9:2447.

This page intentionally left blank.