c06_kg.pdf

Author Query AQ1 Kindly provide explanation for ‘*”. FS...

Author Query AQ1 Kindly provide explanation for ‘*”. FS O O PR D TE EC R R O C N U 0005800230.INDD 1 03-09-2024 08:10:05 PM  CHAPTER 6 THEORY OF REASONED ACTION, THEORY OF PLANNED BEHAVIOR, INTEGRATIVE BEHAVIORAL MODEL, AND REASONED ACTION APPROACH Amy Bleakley FS Shawnika Hull O O PR Vignette KEY POINTS This chapter will: The last five years of her life, she probably did not know who I was. It’s... I got Describe assumptions and main constructs emotional about it. Because just about everyone in my mother’s family has of Theory of Reasoned Action (TRA), Theory experienced dementia and Alzheimer’s. of Planned Behavior (TPB), Integrative D —­Kay’s story (National Institute on Aging, 2019) Model of Behavioral Prediction (IM), and Reasoned Action Approach (RAA). TE Explain research methods, measurements, Alzheimer’s disease (AD) is a family disease not because of its hereditary compo- and analyses used with these reasoned nent but due to the emotional and other challenges families face watching a loved action theories. one’s cognitive decline. Having a family history of AD is one reason people may Demonstrate how reasoned action theories EC participate in AD research. Unfortunately, there are not nearly enough volunteers, were used to (1) understand intention to use pre-­exposure for prophylaxis (PrEP) especially from historically underrepresented groups who are at higher risk of to prevent HIV infections; and (2) identify developing AD. For example, Black adults are twice as likely to have AD compared salient constructs and beliefs for message to their non-­Hispanic white counterparts; Hispanic adults are 1.5 times as likely. design regarding enrollment in a brain R Research to develop improved prevention, treatment, and care for AD is advanc- health registry for Alzheimer’s-­related research. ing, but it could be accelerated by having more diverse study participants. R Present commonly cited critiques of the Diversifying prevention and clinical trial participation in medical research is a pri- reasoned action theories and future ority, with both scientific and ethical justification. Efforts to recruit participants directions. O should be based on an understanding of different beliefs and motivations that can inform recruitment efforts and promote enrollment. Researchers and practitioners C can use health behavior theories for that purpose. They offer a framework to iden- tify individual-­level determinants specific to specific groups—­such as specific N racial or ethnic groups, genders, or political identifications—­that may predict and explain participation. More importantly, they provide a structure to identify salient U group-­specific modifiable beliefs that can be used to design effective messaging about participating in AD research and build an evidence base to advance the sci- ence of recruitment. The theories described in this chapter—­Theory of Reasoned Action (TRA), Theory of Planned Behavior (TPB), Integrative Model of Behavioral Prediction (IM), and Reasoned Action Approach (RAA)—­offer frameworks that can help understand and address challenges like this. Health Behavior: Theory, Research, and Practice, Sixth Edition. Edited by Karen Glanz, Barbara K. Rimer, and K. Viswanath. © 2024 John Wiley & Sons, Inc. Published 2024 by John Wiley & Sons, Inc. 0005800230.INDD 1 03-09-2024 08:10:05 PM  2 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model Introduction and History One measure of a theory’s contribution is the degree to which subsequent theorizing and research are informed by its key constructs and applications. By that standard, Theory of Reasoned Action (Fishbein & Ajzen,1975), Theory of Planned Behavior (Ajzen, 1985, 1991), Integrative Model of Behavioral Prediction (Fishbein, 2000), and the most recent iteration, and Reasoned Action Approach (Fishbein & Ajzen, 2010) are some of the most influential psychoso- cial theories of individual behavioral prediction and explanation. Since the TRA was first i­ntroduced almost 50 years ago, the theoretical approach continued to evolve and expand into TPB, Integrative Model of Behavioral Prediction (IM), and, finally, Reasoned Action Approach (RAA). As articulated by Marco Yzer, “The d ­ ifferent formulations build FS on one another in a developmental fashion and reflect improvements—­in conceptualization and measurement of the theory’s key constructs for the purpose of improving the precision with which behavior can be explained” (Yzer, 2017, p. 2). The popularity of these reasoned action theories is tied to their theoretical robustness, parsimonious conceptu- O alization of the factors that contribute to behavior, meticulous attention to measuring those constructs, and utility for guiding intervention development and evaluation. The main proposition in all formulations is that intention is the best predictor of one’s behavior, and that intentions are formed by some combination of attitudes, norms, and control or O efficacy. We define these constructs in more detail below, but first offer a brief ­history of the evolution of this frame- work. At times, we refer to “reasoned action framework” or “the framework” to represent the set of these theories PR which share common propositions and constructs (Figure 6.1). Evolution of Reasoned Action Theories from TRA to RAA Figure 6.1 shows a graphical depiction of each theory—­TRA, TPB, IM, and RAA—­and Table 6.1 highlights the D ­differences in the four versions since the inception of TRA. The TRA was formulated to address the weakness of ­general attitudes in the prediction of behavior. In TRA, the most powerful predictor of behavior is intention to perform TE a behavior. Intentions are a function of attitudes and social norms regarding the behavior. Generally, attitudes are a disposition or tendency to evaluate a psychological object, such as a behavior, as favorable or unfavorable. Social norms, however, refer to the acceptability of a behavior by a group (or society in general). In TRA, social norms are referred EC to as subjective norms. These subjective norms are injunctive, which are perceptions concerning whether important people in an individual’s life think they should or should not perform the behavior (approval). TPB was developed to extend the utility of TRA, with addition of the perceived behavioral control (PBC) construct to account for agency, or the capacity to influence one’s own thoughts and behaviors. In TPB, PBC was added as a distinct predictor of intention R R Behavioral Attitude toward beliefs × behavior evaluations O C Normative beliefs × Subjective N Intention Behavior motivation to norm comply U Control beliefs × Perc. behavioral facilitating power control Theory of reasoned action (1975) and theory of planned behavior (1985) Shaded areas/constructs and paths were added to TPB Figure 6.1 Visual Depictions of TRA and TPB 0005800230.INDD 2 03-09-2024 08:10:06 PM  Introduction and History 3 Table 6.1 Constructs and Pathways for TRA, TPB, IM, and RAA Theory Predictors of Direct Determinants Predictors of Intention Predictors of Behavior Theory of Reasoned Action Behavioral beliefs × Evaluation ➔ Attitude ➔ Intention Normative beliefs × Motivation to comply ➔ Subjective norm ➔ Theory of Planned Behavior Behavioral beliefs × Evaluation ➔ Attitude ➔ Intention Normative beliefs × Motivation to comply ➔ Subjective norm ➔ PBC FS Control beliefs × facilitating power ➔ Perceived Behavioral Control ➔ Integrative Model of Behavioral Prediction Behavioral beliefs × Evaluation ➔ Attitude ➔ Intention O Normative beliefs × Motivation to comply ➔ Normative Pressure (Injunctive and Skills, Abilitiesa Descriptive) ➔ O Efficacy beliefs ➔ Self-­efficacy ➔ Environmental constraintsa Reasoned Action Approach PR Behavioral beliefs × Evaluation ➔ Attitude (Instrumental and Experiential) ➔ Intention Normative beliefs × Motivation to comply ➔ Normative Pressure (Injunctive and Actual controla Descriptive) ➔ Control beliefs × facilitating? Power ➔ Perceived Behavioral Control (Autonomy and Capacity) ➔ a Moderators of the intention-­behavior relationship. D TE that can exert a direct influence on behavior as well, unlike attitudes and perceived norms which are thought to operate through their effects on intention. Other behavioral theories commonly used in the context of public health and public health communication when TPB EC was developed had considerable overlap. These theories, including Health Belief Model (Janz & Becker, 1984) (Chapter Five), Social Cognitive Theory (Bandura, 1986) (Chapter Nine), and Theory of Subjective Culture and Interpersonal Relations (Triandis, 1977), share assumptions of reasoned action and also focus on outcome expectations, norms and effi- cacy. To examine this overlap, the National Institutes of Mental Health convened a small group of prominent behavioral R theorists in 1991, including Drs. Martin Fishbein, Albert Bandura, Marshall Becker, Harry Triandis, and Frederick Kanfer, for a workshop to reach consensus around which variables are the most important determinants of behavior. Though the R workshop did not result in consensus, there was general agreement on eight propositions relevant to the process of behav- ior change involving assertions about the importance of (1) positive behavioral intentions, (2) prohibitive environmental O constraints, (3) requisite skills and abilities, (4) positive attitudes toward the behavior, (5) supportive social norms, (6) perceived self-­efficacy to perform the behavior, (7) consistency with one’s self-­image, and (8) positive emotional reac- C tions to behavioral performance (Fishbein et al., 2001). Fishbein further refined these propositions when he introduced Integrative Model of Behavioral Change and Prediction, or IM (Fishbein, 2000), also referred to as the Integrated Behavioral N Model (Montano & Kasprzyk, 2015) or the Integrative Model of Behavioral Prediction (IM). IM was a step forward in the reasoned action lineage, as it began to account more comprehensively for the complex ways U perceptions of social pressure shape intention formation by including a descriptive normative component (Cialdini et al., 1990; Fishbein, 2000). In IM, descriptive norms are incorporated into the normative pressure construct along with injunctive norms. The IM also reconfigures the constructs introduced earlier, such that consistency with self-­image and emotions are considered outcome expectations and background factors, respectively. Finally, IM includes self-­efficacy rather than PBC. In 2010, Fishbein and Ajzen published Predicting and Changing Behavior: A Reasoned Action Approach, which presents the latest iteration in this set of theories—­R AA. The most significant distinction between RAA and previous versions is its emphasis on the dual aspects of attitudes, norms, and control. Attitudes are conceptualized as experiential and instrumental. Normative pressure, as in IM, includes an injunctive and descriptive component. And the PBC con- struct, which is treated synonymously with self-­efficacy, is based on autonomy and capacity. RAA also diverges from IM 0005800230.INDD 3 03-09-2024 08:10:06 PM  4 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model by conceptualizing skills and environmental constraints as a single construct: actual control. Actual control over a behavior is proposed to moderate the intention-­behavior relationship and shapes perceptions of behavioral control. As in earlier models, perceptions of behavioral control moderate the intention-­behavior relationship when actual control is not measured or accounted for in the model. Key Constructs In TRA, TPB, IM, and RAA, the most powerful determinant of behavioral performance is intention to enact the behavior. FS Performing the behavior is most likely if individuals have the necessary skills and/or abilities, believe they can perform the target behavior, and if environmental constraints are not prohibitive. Depending on the theoretical iteration, intention formation is determined by attitudes, social norms, and PBC. Generally, attitudes refer to the perceived favorability or unfavorability of a behavior. Perceived norms (also, normative pressure) are of two types. Injunctive norms are perceptions O about whether the behavior is approved by people who are important to an individual, and descriptive norms are whether these important others themselves perform the behavior. PBC refers to perceptions about one’s capability to accomplish O the behavior, even in the context of important barriers (self-­efficacy), and about having the autonomy to do so. Each of these determinants is a function of corresponding underlying beliefs, which have an expectancy-­value form PR (Ajzen & Fishbein, 2000). That is, attitudes (favorability) are determined by experiential (having to do with affective or emotional responses) and instrumental (having to do cognitive responses about outcomes) beliefs weighted by the valence of the outcome. For example, to the extent that someone feels that exercising to sweat for at least 30 minutes/ day, 5 days/week leads to primarily positive outcomes (e.g., increased energy, increased self-­confidence, better health outcomes), she would demonstrate a more positive attitude toward exercising. If she associates exercising primarily D with negative outcomes (e.g., soreness, discomfort, embarrassment), she would have a more negative attitude toward the behavior. This formulation of attitudes accounts for the fact that two individuals might hold the same belief (that TE exercising to sweat for at least 30 minutes/day, 5 days/week results in weight loss), but that belief may contribute differ- ently to attitudes, depending on the evaluation of the outcome as desirable or undesirable. Similarly, normative pressure is determined by normative beliefs; beliefs about whether specific important others (normative or social referents) approve of the behavior weighted by the motivation to comply with the referent. PBC is determined by control beliefs: EC perceptions of one’s ability to overcome specific barriers to behavioral performance. Which Theory Should be Used? R With each iteration in this set of theories, construct measurement and conceptualizations became more refined, and predictive validity improved. However, the availability of multiple versions may raise questions about which version R should be used for a new purpose. For example, although IM and RAA are more recent, TPB is more frequently referred to in the scientific literature, even in recent publications. In an informal search on Google Scholar since the last edition O of this book in 2015, approximately 408,000 results were returned from a search simply for “theory of planned behavior” compared to about 64,000 results for “reasoned action approach.” Decisions about which model to use involve several C considerations, including the empirical question(s) being asked, context of use (e.g., for intervention development versus explanation), and measures available. The process for identifying beliefs may also vary depending on available resources. N For example, the measurement development process, including elicitation research, can be labor-­intensive and costly. When resources are limited, some researchers may use alternative strategies to identify salient beliefs, including literature U review and secondary data analysis. While the latest iteration of the model may offer a more complete understanding of various behaviors and maximize explained variance (McEachan et al., 2016), earlier versions still are valuable. Common Assumptions of TRA, TPB, IM, and RAA Reasoned is Not Necessarily Rational The use of the word “reasoned” may lead to a misconception about these theories. Critics sometimes assume that the intention formation process is “rational” or “deliberative” (Ajzen, 2011; Reyna & Farley, 2006). However, here, the word “reasoned” reflects how intention formation is predicted by development of attitudes, norms, and control beliefs in a 0005800230.INDD 4 03-09-2024 08:10:06 PM  Measurement of TRA, TPB, IM, and RAA Constructs 5 consistent way, but not necessarily through a process that is unbiased or logical. In other words, these theories assume nothing about how attitudes are cognitively constructed. In addition, these theories recognize that individuals’ underly- ing beliefs may be “irrational” and may not be grounded in objective reality, yet, still be predictive of behavior. Behavioral Specificity and Compatibility According to this reasoned action framework, the target behavior must be defined at “various levels of generality or specificity” based on the four TACT components: target, action, context, and time (TACT). RAA posits that changing any aspect of the formulation results in a different behavior (Fishbein & Ajzen, 2010, p. 30). For example, suppose we FS are interested in the goal of women asking health providers about pre-­exposure prophylaxis (PrEP, an HIV prevention medication). The action is “asking about PrEP,” the target is the “health care provider,” the context might be “during your check-­up” or “at your local clinic,” and the time “in the next 30 days” or “the next time you go to your provider.” A behav- O ior defined only by action and time, for example, is the most general application of the components but is often appro- priate for some behaviors in which the target and context are not applicable. A common mistake is confusing goals, behavioral categories, and behaviors. For example, “losing weight” is not a O behavior; it is a goal. Behaviors needed to achieve this goal might be a range of actions, such as avoiding sugary drinks, walking to work, increasing exercising, or bariatric surgery. A behavior in the avoiding sugary drink category would be PR “eliminating sugary drinks at mealtimes every day for the next six weeks.” The general principle is that a target behavior should be specifically defined using TACT so there is a clear understanding of what is being predicted or explained, but not so detailed that the behavior is impractical or of little interest. Related to defining the target behavior is the concept of compatibility. Compatibility refers to measuring intention and behavior at the same level of TACT specificity (Ajzen & Fishbein, 1980; Fishbein & Ajzen, 2010). Compatibility extends to D measures for attitudes, norms, and control as well. That is, they should all refer to the same definition of the target behav- ior. A lack of compatibility attenuates the relationships among intention, its determinants, and the target behavior. TE Volitional Control EC Volitional control refers to the actual (not perceived) ability of an individual to act on their intention to perform a behavior. In TRA, behavioral intention is determined based only on attitudes and norms because it was presumed that agency was high, and the target behavior was volitional. In response to criticisms about this assumption, PBC was added to TRA, cre- ating TPB, to account for behaviors in which volitional control is low, or may be perceived as low. Studies have shown that R PBC explains more variance in behavior beyond that explained by intention when perceived control is low; its contribution to behavioral prediction is limited in other instances when perceived control is higher (Fishbein & Ajzen, 2010). In RAA, R PBC moderates the relationship between intention and behavior, as does actual control. Although some behaviors may occur involuntarily, those behaviors were not meant to be explained by the theory, as they are “outside of awareness.” O Theoretical Sufficiency C In the context of the reasoned action framework, theoretical sufficiency is the notion that performance of the target behavior is predicted directly by intention to perform the behavior and, sometimes, actual behavioral control; and N behavioral intention is predicted solely by attitude, normative pressure, and control factors. The effect of additional variables (demographics, media exposure, personality traits) on intention and behavior is mediated through these atti- U tudinal, normative pressure, and control factors. In instances where sufficiency may not occur or intention is a poor predictor of behavior, measurement error (e.g., poor reliability) or a lack of compatibility among measures are often to blame (Ajzen, 2020). Measurement of TRA, TPB, IM, and RAA Constructs In the formative stages of program or intervention development, application of reasoned action theories can help iden- tify important pathways to behavioral intention as well as the specific beliefs within those pathways that are relevant for intention formation (Cappella et al., 2001). This is a common theme across these theories and can guide identification 0005800230.INDD 5 03-09-2024 08:10:06 PM  6 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model of intervention and communication strategies that address the salient pathways and beliefs. In evaluation stages, the theories are also useful for guiding analyses to explain intervention and program effects (Yzer, 2017). This is possible, in part, because processes for measuring the theoretical constructs in RAA are well-­prescribed (Bleakley & Hennessy, 2012; Fishbein & Ajzen, 2010; Hennessy et al., 2012). Table 6.2 includes the direct and indirect measures for each model com- ponent, their definition, and examples of commonly used wording. The measures fall into two categories: direct and indirect. Direct measures are those of attitudes, perceived norms, and PBC, the three primary or proximal determinants of intention (these may also be referred to as global or universal measures). The wording of these items does not vary across different target behaviors. Indirect measures are those assessing the underlying beliefs: wording of these items is FS specific to each target behavior. Usually, direct and indirect items are measured on a 5-­point or 7-­point scale, although scoring may vary from either 1 to 7 or −3 to 3. Composite measures of the beliefs for each proximal determinant should correlate with that determinant. Intention. Measures of behavioral intention represent an individual’s readiness to perform the behavior, which O includes aspects of willingness and expectation. Various wording is used in intention items, which can be framed as statements such as “I will [perform the behavior],” “I intend to [perform the behavior],” “I plan to [perform the behavior],” O or questions such as “How likely are you to [perform the behavior]?” Responses are typically on a seven-­point scale with response options based on a Likert index (Likert, 1932). Thus, respondents indicate their level of agreement with state- PR ments ranging from “strongly disagree” to “strongly agree” on a 7-­point or 5-­point scale. Attitudes are specific to the behavior of interest and reflect the extent to which an object or behavior is evaluated favorably or unfavorably. They are determined by salient behavioral beliefs weighted by evaluations of the attribute or outcome. These beliefs focus on expectations about the likely outcomes of performing (or not performing) the target behavior. The more one believes that the behavior will lead to positive consequences and prevent negative ones, the more favorable the attitude toward the behavior. D Attitudes are measured using a series of semantic differential items that use pairs relevant to a particular behavior TE and ask respondents to rate each item using a scale with opposite adjectives at each end (such as easy-­hard, ugly-­ beautiful, safe-­dangerous). The series of items should result in a unidimensional scale that evaluates the object or behavior according to positive and negative valence. However, attitude items may be classified into two factors: experi- ential (e.g., unpleasant-­pleasant) and instrumental (e.g., harmful-­beneficial) evaluations. Some commonly used pairs for EC the semantic differential items include good-­bad, unpleasant-­pleasant, foolish-­wise, and harmful-­beneficial. Other adjectives can also be added as additional items that are relevant to the target behavior of interest. Perceived norms, or perceived social pressure, refers to perceptions about what others think and do with regard to performing the behavior, reflecting the contribution of both injunctive and descriptive normative influences. R Normative pressure is determined by underlying normative beliefs that refer (1) to whether important referents approve or disapprove of performing the behavior, weighted by their motivation to comply with the referents R (i.e., injunctive norms), and (2) beliefs about whether similar others are performing or not performing the behavior (i.e., descriptive norms). O Direct measures of norms capture the injunctive and descriptive components of normative pressure. The injunctive norm item deals with approval or disapproval from “important people in your life,” whereas the descriptive normative C item(s) refers to perceptions about whether others, and what percentage of others like the individual are performing the behavior. N PBC is based on beliefs about capacity to perform the behavior in the presence of barriers and the autonomy to do so. Control beliefs emphasize the ability to perform the behavior and the perceived ease or difficulty of doing so, U as well as the degree of control over performing the behavior. PBC is often used as a proxy for actual control when contextual factors, skills, and abilities are not measured. Self-­efficacy (Bandura, 1986) is another way to describe beliefs about capabilities and is conceptually indistinguishable from PBC (Fishbein & Ajzen, 2010), although it has distinct measures. Efficacy beliefs are based on perceptions about factors that can impede or facilitate performing the behavior. Direct measures of PBC reflect perceived capabilities to perform the behavior. Aspects of autonomy and capacity are captured in a combination of items that ask about the extent to which behavioral performance is up to an individual and/or under their control. Self-­efficacy measures are also used to measure PBC, and are often used to operationalize autonomy (McEachan et al., 2016). 0005800230.INDD 6 03-09-2024 08:10:06 PM  Table 6.2 RAA Construct Definitions, Measures, and Example Wording Construct Definition Measure Item Wording Examples Attitude 5-­or 7-­point scale Direct measures Semantic differentials Experiential attitude Overall evaluation of the positive or negative E.g., pleasant/ unpleasant; [Performing the behavior] is... experiences of engaging the behavior enjoyable/ unenjoyable Please rate whether [performing the behavior] would be... Instrumental attitude Overall evaluation of positive or negative E.g., good/bad; wise/foolish [Performing the behavior] is... consequences of performing the behavior Please rate whether [performing the behavior] would be... Indirect measures Likert Behavioral beliefs Belief that behavioral performance is Extremely unlikely to How likely do you think each of the following are? If I were to FS associated with certain attributes or extremely likely [perform the behavior], it would... outcomes If you were to [perform the behavior], it would... Evaluation Value attached to a behavioral outcome or Bad to good [Behavioral outcome mentioned in belief] would be a bad/ O attribute good result of [performing the behavior]? Perceived norm O Direct measures Likert Injunctive norm Belief about whether most people Think I should not/Think I Most people important to me think I should/should not important to individual approves or should [perform the behavior] or would approve/disapprove of me PR disapproves of the behavior Strongly disapprove/strongly [performing the behavior]. approve Most people who are important to me want me to [perform the behavior]. Descriptive norm Belief about whether most people perform “Most people like me will not” How many people like you will [perform the behavior]? the behavior to “Most people like me will” Indirect measures Injunctive belief Belief about whether each referent D Likert SD/SA [Referent] thinks that I should [perform the behavior]. TE approves or disapproves of the behavior Motivation to Motivation to do what each referent thinks SD/SA In general, I want to do what my [referent] wants me to do. comply Descriptive belief Belief about whether each referent SD/SA My [referent group] [performs the behavior]. EC performs the behavior None, a few, some, a lot, alla About how many of the following people or groups do you think would [perform the behavior]? Perceived behavioral control R Direct measures Likert Capacity Belief about ability to perform the SD/SA I am confident/I am certain that I can [perform the behavior]. R behavior and the perceived ease or If you really wanted to, how certain are you that you could difficulty of performing the behavior “Certain I could not” to [perform the behavior]? “certain I could” O Autonomy Belief about the degree of control over “Not at all up to me” to It is not at all up to me/completely up to me whether I performing the behavior “Completely up to me” [perform the behavior]. C SD/SA I have complete control over [performing the behavior]. Indirect measures Likert N Control beliefs Belief about factors that can impede or “Certain I could not” to The following are some situations in which certain people facilitate performing the behavior “Certain I could for overall might find it hard to [perform the behavior]. How certain are U behavior” (Self-­efficacy) you that you could [perform the behavior] even if [barriers or SD/SA facilitators present]? Overall behavior (Control) SD/SA I will have [the control factor]. Power of control Belief in the factor’s power to facilitate or Difficult to easy [The factor] would make [performing the behavior]... impede behavioral performance Intention Perceived readiness of performing the Extremely unlikely to How likely are you to [behavior]...? behavior extremely likely I intend to [perform the behavior]. a Not Likert scale; SD/SA = Strongly disagree to strongly agree. 0005800230.INDD 7 03-09-2024 08:10:06 PM  8 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model Constructing Indirect Measures Using Elicitation Research Direct measures in RAA generally are standardized across behaviors, but underlying beliefs for each of the correspond- ing constructs should be specific to the target behavior. Elicitation research should be conducted to identify which beliefs are important to the population of interest for a particular behavior (Erbe et al., 2020; Vézina-­Im et al., 2021). Elicitation research is crucial to determine salient beliefs that are potentially modifiable and can be incorporated into persuasive health messaging (Yzer, 2012) and intervention design. When new elicitation research is not feasible (it can be labor-­intensive and costly), scientific literature and other studies of the same behavior in the same or similar popula- tions can be used to generate a list of beliefs that will likely be relevant. FS An elicitation study can be conducted using focus groups or interviews, although interviews are preferred because they involve open-­ended discussion and center participants’ perspectives to identify relevant beliefs. To elicit underly- ing behavioral beliefs, individuals are asked about “good things” (advantages) and “bad things” (disadvantages) that would happen if they were to perform the behavior. For normative beliefs, important referents who might approve or O disapprove of performing the behavior are listed (e.g., spouse/partner, family doctor). Referents for descriptive norms are elicited by asking about groups performing the behavior and/or groups that might be relied upon by the individual O in figuring out whether s/he will perform the behavior. Barriers (“what makes it hard”) to performing the behavior and facilitators (“what makes it easy”) often are elicited to represent control beliefs. PR Behavioral beliefs are measured through extent of agreement with statements about the consequences of perform- ing a behavior. Beliefs incorporate the same aspects of the behavior mentioned in measures of the proximal determi- nant and intention (TACT). Each belief/consequence/expectation is then evaluated as being a good thing or a bad thing, also on a 5-­to 7-­point scale. Evaluation and beliefs item scores are then multiplied. Often, evaluation of a par- ticular expectation is straightforward—­some consequences are inherently negative (e.g., getting lung cancer from D smoking) or positive (e.g., feeling less stressed after exercise), and a separate evaluation item is not necessary. Others may require a more formal evaluation. For example, adolescents’ evaluation of “losing their virginity” because of hav- TE ing sex could be perceived by some as good and by others as negative. The product terms can be summed to create a composite measure of beliefs. Normative beliefs. Injunctive normative beliefs assess the extent to which each identified social referent (ideally EC elicited as “important” approves or disapproves of the individual performing the behavior). Motivation to comply with each of the referents is assessed using a unipolar scale and multiplied by the normative belief. Although motivation to comply appears in RAA, it rarely is used. Empirical evidence has not supported its contribution to the prediction of injunctive norms (Fishbein & Ajzen, 2010). Descriptive normative belief items ask about perceptions of whether specific R referents are performing the behavior. Control beliefs. These are items to assess perceived presence of barriers and facilitators and the extent to which they R impede or foster the behavior. More specifically, control beliefs comprise two measures for each control factor (barrier or facilitator): (1) assessments of the belief strength regarding the presence of specific control factors and the power of O (specific) control factors to facilitate or impede behavioral performance. C Analyzing TPB, TRA, IM, and RAA Data N Causal pathways in the RAA are the starting points for statistical analysis. The various ways the RAA frameworks are analyzed are related directly to two main uses of the model: to explain and predict behavior and to identify salient U underlying beliefs as a basis for intervention development. A multiple regression model with direct measures predicting intention will identify which factors predict a particular behavior. However, often the effects of various background or precursor variables on a behavior are the focus of empirical studies. In each version of reasoned action, effects of all precursor variables on intention and behavior are mediated through attitudes, norms, and control (Hennessy et al., 2010; Lee et al., 2018), although in Figure 6.2, they are only p­ ictured for IM and RAA. It is always an empirical question whether a particular variable will be related to any of the underlying beliefs or direct determinants, and which precursors to include in a model varies according to the population and behavior of interest. Background variables are not limited to individual-­level characteristics and may also include social or ­structural variables (such as density of health care providers in a community). In instances when precursors are 0005800230.INDD 8 03-09-2024 08:10:06 PM  Supporting Evidence 9 Integrative model of behavioral Attitude toward Background prediction (2000) Behavioral beliefs × behavior factors evaluations Reasoned action approach (2010) instrumental Shaded constructs added to RAA Demographics experiential SES variables Environmental Attitude toward target constraints FS Culture Subjective norms Normative beliefs × motivation to Intention Behavior Religion comply injunctive descriptive O Persuasive messages Skills and O abilities Personality Perceived behavioral PR Values control Efficacy beliefs × facilitating power capacity Knowledge autonomy Figure 6.2 Visual Depictions of IM and RAA D TE included in the model, path analysis, which is a regression-­based approach that allows one to assess the effects of a system of variables, is used to estimate mediated effects from the precursor to the observed determinants and then on to intention. Advanced statistical techniques, such as structural equation modeling, also can be appropriate (Hennessy et al., 2012). EC Ideally, component items for the higher-­order constructs (injunctive and descriptive norms; autonomy and ­capacity) should be combined into a single measure. However, they can be separate and entered individually into ­regression models. One important reason to keep component measures separate is that they can be used individually if there is interest in exploring the underlying beliefs, or if they are not well-­correlated. R Because underlying beliefs specific to a target behavior are often modifiable, they may be incorporated into ­messaging and other behavior-­change interventions. They can also be used to divide a population according to a R ­particular characteristic (Slater, 1996): In analyses using the reasoned action framework, the audience is segmented by their behavioral intention into intenders and nonintenders (Fishbein & Yzer, 2003). Several data points on the belief O level may be used together to inform selection of salient beliefs. These may include: correlation of the belief with inten- tion, mean difference in the beliefs between intenders and nonintenders, and the percent of agreement with the belief C within intenders and nonintenders (for examples see Bleakley et al., 2018; Hornik et al., 2019). N Supporting Evidence U The RAA and its various iterations have been used widely in developing and evaluating a diverse range of behavioral and public health communication interventions. As a result, many meta-­analyses show support for the theories applied to a range of behaviors, including: condom use (TRA, TPB; Albarracin et al., 2001), alcohol consumption (TPB; Cooke et al., 2016), exercise (TRA, TPB; Hagger et al., 2002), sun protection (TPB; Starfelt Sutton & White, 2016), smoking (TPB; Topa & Moriano, 2010), vaccine hesitancy (TRA, TPB, RAA; Xiao & Wong, 2020), and dietary choices (TPB; McDermott et al., 2015). Other meta-­analytic studies that pool research reports across behavioral domains also have shown support for ­reasoned action-­hypothesized relationships (e.g., TPB Godin & Kok, 1996; Hagger et al., 2018). McEachan et al. (2011) conducted a meta-­analysis of 237 prospective tests of the TPB. They found that the TPB constructs account for a 0005800230.INDD 9 03-09-2024 08:10:07 PM  10 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model c­onsiderable amount of variance in behavioral outcomes (19%) and intentions (44%) and the intention-­behavior ­correlation was.43, which indicates a moderate relationship. Some argue that intention does not always translate into behavior. In a later meta-­analysis, McEachan et al. (2016) demonstrated significant prospective intention-­behavior ­relationships and showed that the components of the RAA were associated with behavioral intentions. Experiential and instrumental attitudes, descriptive and injunctive norms, and capacity, but not autonomy, were significant predictors of intentions, and these variables explained more than half of the variance in intentions (59%). In turn, intention and capacity but not autonomy, significantly predicted behavior, explaining almost one-­third of the variance in behavior (31%). These analyses provide strong support for RAA but also raise important unanswered questions about the autonomy dimension FS of the PBC construct. Applications O Case 1: Analyzing Background Variables in the Context of PrEP for HIV This section offers an example to demonstrate how to identify the ways sociodemographic and other background char- O acteristics exert their influence on behavioral intentions by shaping determinants of intention. The analysis is based on data collected from women in the District of Columbia, one of the 48 US counties with the highest HIV incidence PR (HHS, 2019), to understand why women do or do not plan to use PrEP (Hull et al., 2022). PrEP is a pill that is highly effective in preventing HIV infection when taken daily and is recommended for people without HIV who are at risk of being exposed to HIV through sex or injection drug use. PrEP is underutilized by women generally and particularly by Black women relative to the number of HIV infections diagnosed in these groups (Siegler et al., 2018). Because of low rates of PrEP uptake, this study sought to understand Black women’s perspectives about PrEP and psychosocial factors D most relevant to their intentions to use PrEP to inform the development of communication efforts to increase its use. The data were used to identify how potentially relevant background variables—­age, a history of injection substance use, TE and the proportion of casual (as opposed to committed) sex partners—­affect attitudes, norms, and perceived control to ultimately shape women’s intentions to use PrEP. The study was conducted in community (rented private office space in the community) and clinic settings. Recruitment methods included approaching women in the waiting room of the women’s health clinic to invite them to EC be screened, outreach by community-­based partners, and distributing flyers and palm cards. Cisgender women (or women who identified socially as female) who met study criteria based on an eligibility screening survey were invited to participate (n = 398). Eligibility criteria included being sexually active, HIV negative, age 20–49 and potentially eligible for PrEP due to residence in high prevalence area and the presence of any HIV risk factor (having an HIV-­positive sex R partner, a recent sexually transmitted infection, participation in sex work, inconsistent condom use). The survey assessed risk perceptions, knowledge of PrEP, demographic and background characteristics, HIV risk and R prevention behaviors, and reasoned action theoretical constructs (behavioral intention, attitudes, injunctive and descrip- tive norms, PBC). The target behavior was “My using PrEP daily for HIV prevention in the next 12 months.” Attitudes O were assessed with a semantic differential item: “Overall, would you say that using PrEP daily to prevent HIV is a good or a bad thing?” In addition, 5-­point Likert scales were used to measure injunctive norms (“Thinking about the people who C are important to you, would they support or not support your using PrEP for HIV prevention in the next 12 months?”), descriptive norms (“Thinking about people who are similar to you, how likely would they be to use PrEP for HIV preven- N tion in the next 12-­months.”), and PBC (i.e., “If I really wanted to, I could use PrEP daily for HIV prevention,”). All items were coded such that higher scores represented more positive attitudes, more normative support, and higher PBC. Age U was assessed continuously; injection drug use was a dichotomous assessment of participants’ lifetime history of “inject[ing] any drug other than those prescribed to you.” Sexual partnerships were assessed by asking: “Of the men you had sex with in the past 12 months, how many were casual sex partners?” (Responses: none, a few, some, most, all). The statistical analysis tested mediation of the relationship between the background variables and intentions to use PrEP, through the proximal determinants. A path analysis was conducted with age, substance use history, and number of partners as the independent variables, behavioral intention as the dependent variable and attitudes, descriptive and injunctive norms, and PBC as parallel mediators. Injunctive and descriptive norms were kept separate in this analysis, because they were correlated at.23. As shown in Figure 6.3, attitudes, descriptive norms, and PBC were significantly associated with intentions at the p descriptive norms (b = 0.10, SE = 0.04: indirect effect b = 0.06, SE = 0.02)); Casual partners-> PBC (b = –0.09, SE = 0.04; indirect effect (b = –0.06, SE = 0.03)); PR Figure 6.3 Conceptual Model with RAA and Precursor for Intention to Use PrEP (Case Study 1) Figure 6.3 shows statistically significant paths for background variables to the determinants. Age was associated positively with descriptive norms, and the relationship between age and behavioral intentions worked (in part) through descriptive norms, as mediation was significant. That is, older women anticipated to a greater extent that their peers D would utilize PrEP to prevent HIV infection, relative to younger women, and this belief shaped intentions. In addition, those with more casual partnerships reported lower PBC for using PrEP daily. Further, having more casual partnerships TE was associated with lower intentions to use PrEP, and this relationship is explained in part by lower PBC to utilize PrEP. Evidence suggests that women perceive PrEP to be a tool primarily for those in serodiscordant relationships, in which one of the partners is known to be living with HIV (Bond & Gunn, 2016). We anticipate that those with more casual partnerships may have low certainty about their HIV risk exposure and may have less confidence in their ability to EC adhere to such a tool, which has uncomfortable side-­effects and required a daily pill at the time the research was conducted. History of injection-­drug use was not associated significantly with attitudes, norms, or self-­efficacy in this context; however, substance use may be related to other important background variables. For example, Roth and her colleagues R (Roth et al., 2015) reported that, among a sample of individuals who inject drugs, PrEP awareness was associated posi- tively with use of syringe-­exchange programs, STI testing, and drug treatment. Injection-­drug use may shape and con- R strain individuals’ abilities to enact intentions, thereby moderating the intention-­behavior relationship. Considering persistent inequities in HIV infection, and disparately low utilization of PrEP among women, raising O awareness of this HIV prevention option among women who may benefit from using it is an important step toward ending the epidemic. Findings from analyses like those reported here are being used to inform communication strate- C gies to increase the use of PrEP. For example, given that older women in this sample were more likely to perceive their peers would use PrEP, strategies that cultivate positive descriptive norms and/or that rely on important normative ref- N erents to convey information may be particularly promising strategies for communicating effectively about PrEP. These results also suggest that efforts to build actual and perceived efficacy may be an important component of communica- U tion strategies that are designed to reach women in casual sexual relationships who may be eligible for PrEP. Case 2: Using the RAA to Increase Recruitment into Alzheimer’s Disease Research Registries The advancement of medical science depends on rigorous and representative trials to test new preventive therapies and treatments for diseases. Such trials require participation from people who are affected or are at-­risk and who come from diverse groups—­including people of different race/ethnic groups. Health behavior theories can help understand and address the behaviors and intentions of individuals and groups participating in clinical trials. Black and Hispanic 0005800230.INDD 11 03-09-2024 08:10:08 PM  12 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model participation in registries and clinical trials related to AD is quite low (Langbaum et al., 2019; Manly & Glymour, 2021; Weiner et al., 2018). However, under-­representation of traditionally marginalized groups in clinical trials is not unique to AD trials. Nonwhite racial and ethnic groups report lower rates of participation in clinical trials across a range of diseases (Coakley et al., 2012). Although there are structural barriers such as access and system-­level barriers such as inclusion criteria that may reduce Black and Hispanic participant enrollment, individual-­level factors and decision-­making are also consequential. This second case demonstrates how the RAA can be used to determine whether the determinants of behavioral inten- tion vary for different subgroups based on race and ethnicity, sex, and/or any number of possible individual or structural FS characteristics. The StepUp (Study to Expand registry Participation of Underrepresented Populations) is designed to increase enrollment of cognitively healthy older adults and to advance recruitment science. StepUp is aimed at increasing par- ticipation of groups that are underrepresented in AD-­focused participant registries for AD prevention trials. Examples O of these registries include the Alzheimer’s Prevention Registry (APR), the Alzheimer’s Association’s TrialMatch, the Brain Health Registry, and APR’s GeneMatch Program. O The message-­design process began with formative research that included elicitation interviews with 60 white, Black, and Hispanic older adults ages 49–80 years old, from the Philadelphia, PA area. Phone interviews were con- PR ducted in spring and summer 2020 by two experienced interviewers and were about 30 minutes long. The goal of elicita- tion interviews was to identify attitudinal, normative, and control beliefs among the target audience that were associated with registry enrollment and could serve as a basis for subsequent quantitative research. Findings from the elicitation study (Bleakley et al., 2022) were incorporated into a nationally representative survey about one year later. RAA data on the target behavior of “signing up for a brain health research registry” were collected through an online survey using a D probability-­based sample of adults ages 49–80 years old (n = 1,501) with oversamples of Black (n = 334) and Hispanic (n = 309) adults. Direct measures and underlying beliefs were used to measure all three determinants of intention. TE Table 6.3 provides descriptive statistics on the RAA direct measures for the three racial and ethnic subgroups of women. Regression analysis of proximal determinants on intention was conducted within subgroups of interest—­white women (for comparison; n = 415), Black women (n = 194), and Hispanic women (n = 134). In these analyses, the control and efficacy items (used to measure autonomy and capacity) were correlated with one another at r = 0.44, and the EC injunctive and descriptive norms were correlated at r = 0.55, and so they were each entered separately into the models. Attitudes were related significantly to intentions for all subgroups of women at the p < 0.01 level (see Table 6.4 for all regression coefficients), and the association between intention and descriptive norms was also significant (p < 0.05) for all three groups. Capacity was related uniquely to intention for Black women. These findings emphasize the importance R of assessing specific groups within the target audience: different determinants are relevant for different groups. This has implications for interventions, such as health messaging, in which persuasive efforts are focused on specific beliefs R about the salient determinants. Here, there are clear similarities and differences in important constructs for intention formation. This type of analysis could be conducted with an intersectional lens, which attends to the ways power and O privilege are conveyed differently for people at different intersectional locations to examine the “dynamics of differences and sameness” (Cho et al., 2013, p. 787). For example, one might examine regional differences or differences between C socioeconomic categories within groups of racially similar women. Having identified salient global constructs, our next step was to conduct a belief-­level analysis that offered more N specific information on potentially modifiable underlying beliefs for each group. Although attitudes are important for all groups of women, the beliefs that underlie those attitudes may be very different. Using an audience segmentation U approach in which respondents were categorized into intenders and nonintenders, we examined relevant beliefs for each subgroup. For the purposes of an illustrative example, Table 6.4 includes only select behavioral beliefs. Beliefs are ordered by correlations of intention and each belief for Black women, from strongest to weakest. Highlighted cells represent beliefs identified as most salient for each group. Some beliefs about joining a registry, such as “help to advance science, “and “help people like you in the future” would be candidates for inclusion in a message aimed at women in general, although “helping others like me” is stronger for Black and Hispanic women, who are the actual groups targeted for enrollment. “Improve one’s community trust in medical research” also is more salient for Black and Hispanic women. Segmentation reveals numerous beliefs that, if bolstered, could lead to increased intentions for this behavior. The combination of these statistics helps researchers, practitioners, and clinicians make assessments 0005800230.INDD 12 03-09-2024 08:10:08 PM  Table 6.3 Measures and Descriptive Statistics for RAA Proximal Determinants of Enrolling in a Brain Health Registry, by Racial and Ethnic Subgroups Black Women White Women Hispanic Women Construct Items Range of Scores from −3 to 3 n = 194 n = 415 n = 134 Outcome Mean (SD) Mean (SD) Mean (SD) Intention How likely are you to sign up for a brain health registry in −0.49 (1.84) −0.32 (1.74) −0.43 (1.79) the next 30 days? Extremely unlikely/extremely likely % Intenders 36.6% 34% 31.3% Proximal determinants Attitudes Signing up for a brain health registry in the next 30 days Mean (SD) Mean (SD) Mean (SD) Alpha=0.93 would be... 0.65 (1.27) 0.57 (1.19) 0.47 (1.33) Bad/good, Unpleasant/pleasant, Harmful/beneficial b(SE) b(SE) b(SE) FS Selfish/altruistic, Unnecessary/necessary, Foolish/wise 0.63 (0.10) 0.60 (0.08) 0.42 (0.13) Perceived norms Injunctive norm Do you think most people who are important to you think Mean (SD) Mean (SD) Mean (SD) O that you should or should not sign up for a brain health −0.12 (1.38) 0.13 (1.28) −0.15 (1.50) registry in the next 30 days? Should not/should b(SE) b(SE) b(SE) 0.09 (0.09) 0.06 (0.07) 0.11 (0.11) O Descriptive norm Will most people like you sign up for a brain health Mean (SD) Mean (SD) Mean (SD) registry in the next 30 days? Will not/will −0.40 (1.63) −0.30 (1.53) −0.09 (1.58) PR b(SE) b(SE) b(SE) 0.21 (0.08) 0.24 (0.06) 0.23 (0.09) Perceived behavioral control Autonomy It is completely up to me whether I sign up for a brain Mean (SD) Mean (SD) Mean (SD) Strongly agree D health registry in the next 30 days. Strongly disagree/ 1.86 (1.74) b(SE)* 2.23 (1.28) b(SE) 1.74 (1.73) b(SE) AQ1 TE −0.00 (0.06) −0.11 (0.07) 0.002 (0.08) Capacity If I really wanted to, I am certain that I could sign up for Mean (SD) Mean (SD) Mean (SD) a brain health registry in the next 30 days. Strongly 0.84 (1.68) 1.09 (1.44) 0.63 (1.82) disagree/ EC b(SE) b(SE) b(SE) Strongly agree 0.09 (0.06) 0.23 (0.08) 0.13 (0.09) Notes: Scores were coded −3–3. Unstandardized regression coefficients (b) and standard errors (SE) in regression model on intention with all six determinants included. Bolded, italicized coefficients are statistically significant at least at the p < 0.05 level. R Table 6.4 Audience Segmentation Analysis for Women: Enrolling in Brain Health Registries for Alzheimer’s Research R Black Women (n = 194) White Women (n = 415) Hispanic Women (n = 134) O r with Mean % Likely r with Mean % Likely r with Mean % Likely intention difference difference intention difference difference intention difference difference C Behavioral beliefs Help people like you in the 0.42 1.09 27.4 0.34 0.64 14.6 0.42 1.17 30.5 futurea,c N Help to advance sciencea–c 0.35 0.79 26.7 0.35 0.62 17.5 0.35 0.94 23.7 Help others in the futureb 0.35 0.73 15.2 0.41 0.76 20.1 0.35 0.80 17.3 U Be a novel experiencea,b 0.30 0.77 29.0 0.31 0.62 15.7 0.32 0.51, ns 8.8, ns Improve your community’s 0.27 0.73 23.8 0.21 0.38 15.1 0.32 0.85 21.9 trust in medical researcha,c Make others think of you as 0.11,ns 0.38,ns 11.5,ns 0.16 0.25,ns 5.3, ns 0.17,ns 0.36,ns 11.9,ns a role model Compromise your privacyb −0.25 −0.83 19.0 −0.34 −0.76 15.0 −0.20 −0.71 20.2 Notes: Differences were calculated from intender-­nonintenders. a Black women. b white women. c Hispanic women. Mean and likely differences are statistically significant at p < 0.05 unless otherwise noted. Shaded cells represent beliefs identified as most salient for each group. 0005800230.INDD 13 03-09-2024 08:10:08 PM  14 Chapter 6: Theory of Reasoned Action, Theory of Planned Behavior, Integrative Behavioral Model about which beliefs are most salient and promising for intervention. For each belief, the larger the difference between intenders and nonintenders, the more room there is to move individuals; conversely, if the belief is not correlated with intention or attitude, it would not be a good candidate. This example highlights the versatility of the RAA, and all its predecessors, to find both similar and different points of potential intervention to encourage behavior change among various groups and demonstrates the necessity of moving beyond a one-­size-­fits-­all approach as it applies to under- standing underlying behavioral determinants. Critiques FS While the models described in this chapter have been used widely over several decades, critics (Amaro & Raj, 2000; Sniehotta et al., 2014; Trafimow, 2015; Ogden, 2003) have raised issues that warrant consideration. Some of the most common critiques include that the models are difficult to falsify, too parsimonious or simplified, and that inappropri- O ately assert theoretical sufficiency. For a review of more critiques, see Hagger (2019). Falsifiability. Ogden (2003) argued that models are not falsifiable on the grounds that when data from a study osten- O sibly provide disconfirmation, the results could be explained as conceptually consistent with theoretical assumptions or as the result of methodological shortcomings. In their response to these and other critiques, proponents have countered PR that it is indeed plausible to disconfirm the theories, as would be the case if none of the theoretical predictors was sig- nificantly associated with behavior (Ajzen, 2015; Ajzen & Fishbein, 2004; Ogden, 2003). Ajzen (2020) also points out that the models hypothesize various “mediation and moderation processes,” grounded in its sufficiency claims, that can also be disproven. Parsimony. Other critics argue that this reasoned action framework is too parsimonious because it oversimplifies D factors shaping behavior and excludes important variables (Sniehotta et al., 2014). For example, emotions, including anticipated affect (Conner et al., 2013) and behavioral consistency with self-­identity (Armitage & Conner, 1999) have TE received considerable empirical attention, leading researchers to advocate including these factors in the reasoned action framework. s As noted earlier, this framework does not preclude adding predictor variables that are not redundant with the original constructs and that consistently add additional variance across a range of behaviors (Ajzen, 2020). Further, the reasoned action theory predicts considerable variance in intentions and behaviors across a wide array of behavioral EC domains with relatively few predictors. Theoretical sufficiency. The sufficiency principle in a reasoned action context also has been challenged (Sniehotta et al., 2014), particularly with the inclusion of past behavior in the prediction of intention and/or behavior. According to the theories, the effect of past behavior on intention should be mediated through attitudes, norms, and intentions. R There is empirical evidence, however, that demonstrates that adding past behavior to the prediction of intention as well as behavior can increase the variance explained in those outcomes (Albarracin et al., 2001; Hagger et al., 2018; Rise R et al., 2010). The causal mechanism through which past behavior influences intention remains unclear. O Future Directions C As models of individual decision-­making, TRA, TPB, IM, and RAA consistently are invoked in developing individual-­ level interventions and in explaining the psychosocial mechanisms of behavioral performance. Scholars have noted that N these models rarely have been applied to understand social and structural factors that shape behavior, particularly in the context of underserved groups where a social-­structural approach could produce the most relative impact (Sniehotta U et al., 2014). However, factors like these, which are external to the individual, can be incorporated and these models accommodate such factors in two ways. First, social and structural factors could/should be treated as precursors that shape intentions indirectly through their direct effect on perceptions of behavioral control. Additionally, structural and social barriers may represent environmental constraints that reduce an individual’s actual control to perform the behav- ior. Actual control and PBC (Hagger et al., 2022) not only moderate the intention-­behavior relationship but can also precipitate a feedback loop that affects underlying beliefs of any of the proximal determinants. Though the importance of actual control is highlighted in RAA, few studies assess actual control. Methodological advances in measurement of social-­structural factors in health behaviors, such as spatial stigma (Smiley et al., 2020; Taggart et al., 2022), police-­ based discrimination (English et al., 2017) and structural racism (Doshi et al., 2020), facilitate more accurate modeling 0005800230.INDD 14 03-09-2024 08:10:08 PM  References 15 of this important set of factors in behavior. Conceptual and empirical integration of reasoned action constructs within social-­ecological models of communication (Goulbourne & Yanovitzky, 2021; Young & Bleakley, 2020) and behavior (Bronfenbrenner, 1992) is also fertile ground for future scholarship. Summary TRA (1975), TPB (1985, 1991), IM (2000), and RAA (2010), together, offer behavioral researchers and practitioners a comprehensive framework of psychosoci