Predicting intentions from attitudes: A reasoned action approach to religious ritual
Predicting Intentions from Attitudes i TABLE OF CONTENTS Extended Literature Review 1 The Theories of Reasoned Action and Planned Behavior 8 Family Purity 30 Current Study 36 References 39 Empirical Article: Predicting Intentions from Attitudes: A Reasoned Action Approach to Religious Ritual 43 References 69 Appendix A: Online Survey and Consent Form 78 Appendix B: Debriefing Form for Participants 89 Appendix C: Rabbis' Packet 93
Predicting Intentions from Attitudes ii ACKNOWLEDGEMENTS First, I want to thank my dissertation chairperson, Dr. Juliana Lachenmeyer, for her support and thoughtful contributions that enabled me to complete this project. I am also grateful for the other members of my committee: Dr. Gretchen Gibbs, Dr. Cynthia Radnitz, and Dr. Lana Tiersky. All of the committee members have contributed greatly, not only to this project, but also to my graduate training and my development as a clinician. To my parents, who have supported me in every way possible throughout my educational journey and my life, I will forever be grateful for everything you have done and continue to do. I have also benefited from the unwavering support of my husband, Joe, who has encouraged and pushed me to complete this project. I extend my sincere gratitude to several of my classmates: Becky, Courtney, Matt and Naomi made this journey more enjoyable and less stressful. You have provided both personal and professional support that has been invaluable. A special thank you to Bernadette Joseph, whose warmth and care allowed me to complete my degree knowing my son, Aaron, was being lovingly attended to in my absence.
Predicting Intentions from Attitudes 1 Predicting Intentions from Attitudes: A Reasoned Action Approach to Religious Ritual The Theory of Reasoned Action (TRA) and the Theory of Planned Behavior (TPB) have been used for decades as a way to understand and predict people's behaviors and intentions to perform behaviors (Ajzen & Fishbein, 2005), and is currently the most dominant model of attitude-behavior relationships (Armitage & Christian, 2003). These theories postulate that one set of factors - namely attitudes, subjective norms and perceived behavioral control - can be used to predict behaviors across multiple domains (Ajzen & Fishbein, 1980, 2005). This model has successfully been applied to predict diverse behaviors and intentions such as condom use, exercise, health behaviors and consumer behaviors (Armitage & Connor, 2001; Sheeran & Taylor, 1999; Sheppard, Hartwick & Warshaw, 1988; Sutton, 1998). Only a few published studies have used this model to predict religious and cultural behaviors (Giles & Cairns, 1996; Gorsuch & Wakeman, 1991). The objective of the current study is to apply the TPB to expand the understanding of a ritual with religious and cultural relevance in the observant Jewish community. Family Purity is a religious practice that lends both routine and meaning to the marital relationship of observant Jews. This practice calls for a physical separation between husband and wife from the time a woman begins her menses until she immerses in a mikvah (ritual bath). This ritual aims to sanctify the physical relationship between husband and wife (Ribner, 2003) and is said to strengthen the emotional bond between them, enhance the dignity and respect of the wife and teach the importance of moderation of physical pleasures (Tendler, 1988). It is a major determinant of the intimate life of the observant couple (Ribner, 2003), yet there is very little empirical research on the topic
Predicting Intentions from Attitudes 2 (Guterman, 2006). Using the TPB, this study seeks to understand what factors determine how strictly a person chooses to observe this ritual, as well as some of the factors underlying these choices. This research is only an early step to a complete understanding of how this ritual affects a person's marriage and individual religious identity, and hopefully future research can continue to build upon it. In addition, this information will add to the growing body of literature available to clinicians hoping to work in a culturally competent manner with the diverse populations now seeking therapeutic services. Brief Historical Overview of Attitude-Behavior Research Social psychologists have long assumed that human behavior is guided by social attitudes (Armitage & Christian, 2003). Based on this assumption that attitudes were the key to understanding human behavior, the field of social psychology was originally defined as the scientific study of attitudes (Ajzen & Fishbein, 2005). Despite this long held assumption, research suggests that actual behavior was not always consistent with expressed attitudes. To test this relationship, LaPiere (1934) accompanied a Chinese couple traveling across America and recorded whether or not they received service in restaurants and hotels. Upon returning home, he sent a letter to each establishment asking whether they would accommodate Chinese guests. Although the couple had received service in all but one establishment on their trip, almost all the responses to the letter were negative. LaPiere concluded that reported attitudes were merely responses to symbolic situations, and, therefore, the utility of attitude questionnaires was limited (LaPiere, 1934). In the 30 years following LaPiere's study, 45 studies were conducted attempting to predict behaviors from attitudes, most with discouraging results (Ajzen & Fishbein, 2005).
Predicting Intentions from Attitudes 3 In an early meta-analysis which examined the relationship between attitudes and overt behaviors, Wicker (1969) noted that "social psychologists have assumed that attitudes have something to do with social behavior" (pg. 41), but his review found little empirical evidence to support this assumption. After reviewing over 30 studies, which covered topics as varied as job performance to the treatment of minority groups, Wicker concluded that it is more likely that attitudes will be unrelated or only slightly related to behaviors. The correlations between attitudes and behaviors were rarely above .30, and more often near zero. Wicker noted the need for more accurate measures of attitudes, as well as further investigation into the other factors that had been postulated to impact the attitude-behavior relationship (such as personal and situational factors). Explanations for Attitude-Behavior Inconsistencies. Following Wicker's (1969) review, researchers attempted to reconcile the observed attitude-behavior inconsistencies by examining three theoretical and methodological issues (Fishbein & Ajzen, 1974). The first related to problems with measurement. Verbal answers can often be distorted or biased, and may therefore not reflect a person's true attitude. It was proposed that response biases could account for the discrepancy in reported attitudes and observed behaviors. Psychologists attempted to develop disguised procedures and physiological measures that would be less susceptible to response biases. These changes in measurement techniques were met with only limited results, as physiological measures tend to assess arousal rather than attitude and there is little empirical evidence that disguised measures have better predictive validity than undisguised measures (Ajzen & Fishbein, 2005).
Predicting Intentions from Attitudes 4 Researchers also examined how the definition of attitudes might contribute to the observed attitude-behavior inconsistency. Ajzen and Fishbein (2005) noted the prevalent view that there are three distinct components to an attitude: cognitive, affective and conative. This multidimensional definition of attitudes was rarely addressed in measurement techniques, which typically assessed an overall positive or negative evaluation of an object or behavior. Investigators began to include items to separately measure these three components of attitude on their scales. Despite the attempts to separately measure the multiple dimensions of attitudes, and thereby, more accurately measure the construct as a whole, statistical procedures revealed little discriminant validity for the separate components (Ajzen & Fishbein, 2005). Following Wicker's review psychologists also began looking for other factors that might impact the relationship between attitude and behavior (Armitage & Christian, 2003). Self- monitoring, having a vested interest in the behavior and time pressure were suggested as possible moderating factors (Ajzen & Fishbein, 2005). Those who pursued this line of research believed that in order to adequately predict behavior, one must investigate the role of other factors as moderators of the attitude-behavior relationship (Ajzen & Fishbein, 1980). Evaluative Inconsistency. Ajzen and Fishbein (2005) suggest that there are two types of inconsistencies in attitude- behavior research. The first, what they term a literal inconsistency, refers to a situation in which a person's report of how they intend to act differs from how they actually behave. The predictor and criterion are identical and relate to the same behavior, but the person does not carry out the stated intention (Ajzen & Fishbein, 2005). LaPiere's 1934 study of attitudes towards serving Chinese customers is an example of a literal inconsistency. In that study, subjects' statements of
Predicting Intentions from Attitudes 5 whether or not they intended to serve Chinese customers was inconsistent with their actual behaviors. The second type of inconsistency discussed by Ajzen and Fishbein is an evaluative inconsistency. An evaluative inconsistency occurs when a general attitude fails to predict a specific behavior. This was the case in most of the studies in Wicker's 1969 review, as most studies reviewed assessed a person's general attitude towards an object or behavior and then looked only at one specific target behavior. In their further research on the attitude-behavior relationship, Ajzen and Fishbein attempted to eliminate evaluative inconsistencies through two general principles, the principle of compatibility and the principle of aggregation. Principle of Compatibility. Fishbein and Ajzen noted that researchers often used general attitudes to predict a specific behavior and suggested that this might be responsible for the observed attitude-behavior inconsistency. For example, a study might ask researchers about their attitude towards religion, and try to predict from that attitude whether or not a person would attend religious services. However, a person's general attitude toward an object was found to be unrelated to specific behaviors (Ajzen & Fishbein, 1970). Attitudes and behaviors can be conceived of as consisting of four elements: the action, the target, the context and the time at which the behavior is performed. A behavior can be defined as a specific action directed at a target in a context in a specific point in time (Ajzen & Fishbein, 2005). These elements can be measured in either a general or a specific manner (Ajzen & Fishbein, 1977, 2005). When looking for an attitude- behavior relationship, it is important that the attitude and behavior in question be measured in a comparable manner. Ajzen and Fishbein, therefore, predict that if a person were asked "What is
Predicting Intentions from Attitudes 6 your attitude towards attending religious services this week?" there would be a greater correlation between attitude and behavior, than if the more general evaluation of "What is your attitude toward church?" is addressed. When reviewing studies that looked at the attitude- behavior relationship, Ajzen and Fishbein (1977) found that there was rarely a correspondence in all four elements of attitude and behavior. The most popular elements used were action and target, and correspondence in these elements often seemed to be sufficient to improve the observed attitude-behavior correlation. In their review of over 140 studies, they divided studies into three groups; the studies in one group had little or no correspondence in measurement, in the second group there was partial correspondence and in the last group there was high correspondence in measurement. Measurements in the studies categorized as having low correspondence were mainly non-significant. The studies categorized as having partial correspondence largely had inconsistent results: Many had no significant results or a low attitude-behavior relationship and only very few studies showed a high correlation between attitudes and behaviors. Of the studies that Ajzen and Fishbein categorized as having high correspondence, most found a high relationship between attitude and behavior, while a minority had inconsistent results, or results of a low magnitude. While these findings support the predictive values of attitudes, compatibility may sometimes be difficult to achieve. Ajzen and Fishbein themselves noted that it might often be difficult to identify correspondence, particularly if it is difficult to identify the exact target of an act. Principle of Aggregation. Early in the history of attitude research Thurstone (as discussed by Ajzen & Fishbein, 2005, Fishbein & Ajzen, 1974) recognized the possibility that two people might hold the same
Predicting Intentions from Attitudes 7 attitudes and express them differently. For example, if two people have similar, positive attitudes towards higher education, one might act on the attitude by becoming a professor, while the other might become a donor to a university. An attitude toward an object may influence the overall pattern of a person's responses, but not a specific behavior (Ajzen & Fishbein, 1977). Fishbein and Ajzen (1974) noted inconsistencies in the behaviors used as dependent variables in studies. The behavioral criterion in a study could be a single act or multiple acts. When looking at the previous research they commented that there seemed to be a higher attitude-behavior relationship when multiple acts were used as the behavioral criterion, rather than a single act. To test this, Fishbein and Ajzen (1974) gave participants a set of 100 behaviors dealing with religious matters, as well as five traditional scales measuring attitudes toward religion. Half the participants were asked to report which of the 100 acts they had performed (behaviors) and half were asked which acts they would perform (behavioral intentions). Multiple act criteria were obtained by summing the 100 behaviors or the 100 behavioral intentions. As predicted, all attitude scales were found to be highly correlated with the multiple act criteria, while prediction of the single act criterion from attitudes tended to be low or non-significant. The principle of aggregation assumes that behaviors are influenced not only by general attitude, but also by other factors unique to that situation and action observed. These other influences are cancelled out when multiple actions are aggregated together (Ajzen, 1991). Multiple acts as a general measure of behavior correspond more closely to the measurements of general attitudes.
Predicting Intentions from Attitudes 8 The Theories of Reasoned Action (TRA) and Planned Behavior (TPB) Uses and Assumptions Before the development of the theory of Reasoned Action (TRA) there was an assumption that there are different causes for different types of behavior. Researchers looked at factors that would be relevant in specific behavioral domains, rather than factors that could predict all types of social behaviors. The development of the TRA made it possible to account for all types of behaviors by referring to only a small number of concepts that are applicable to all behavioral domains (Ajzen & Fishbein, 1980). Simply stated, the theory suggests that a person's behaviors are a function of intentions, which in turn are determined by attitude toward the behavior and social norms (what a person perceives to be others' expectations; Ajzen &Fishbein, 1970). Attitudes and subjective norms are determined by underlying salient beliefs (Ajzen & Fishbein, 1980). The development and empirical research on the TRA were based on the assumption that the behaviors studied were under the volitional control of the actor (Madden, Ellen & Ajzen, 1992). However, oftentimes behaviors are not under the full control of the actor. The TPB (Ajzen, 1991) extends this theory to behaviors not under volitional control by adding the factor of perceived behavioral control, determined by control beliefs, to the reasoned action model (Armitage & Connor, 2001). Meta-analyses have demonstrated perceived behavioral control is a useful addendum to the TRA, and it is also able to better predict behavior, as well as behavioral intentions (Armitage & Connor, 2001; Madden, Ellen & Ajzen, 1992). There are four main assumptions underlying the TRA and the TPB. First, behavioral intention is the immediate antecedent of specific behavior. Intentions are a summary of the
Predicting Intentions from Attitudes 9 motivation to perform a behavior, which reflect a decision to follow a course of action (Armitage & Christian, 2003). Second, intention is determined by a person's attitude toward the behavior (evaluation of positive or negative consequences), the person's subjective norms (perceived approval or disapproval of others) and perceived behavioral control (factors that facilitate or impede the performance of the behavior). Third, these determinants are a function of a person's underlying behavioral, normative and control beliefs. Lastly, behavioral, normative and control beliefs vary as a function of a person's background and personality factors (Ajzen & Fishbein, 2005). Therefore, factors external to these theories, such as demographic variables of age or education, will impact attitudes, subjective norms, intentions and behaviors only to the extent that they impact a person's underlying beliefs (Ajzen & Fishbein, 1980). Background factors Individual Personality Mood, emotion Intelligence Values, stereotypes General attitudes Experience Soc inl Education Age. gender Income Religion Race, ethnicitv Culture Infiu malum Knowledge Media Intervention I J ti / / '- + \ \ \ \ • • • >, Behavioral beliefs Normative beliefs Control beliefs w b- Attitude toward the behavior Subjective norm f > Perceived beha\ loral t control Figure 1: The theories of reasoned action and planned behavior (taken from Ajzen & Fishbein, 2005, p. 194)
Predicting Intentions from Attitudes 10 Assumption that Behavior is Reasoned. The TRA is based on the assumption that human beings use the information available to them in a reasonable manner to make decisions. This does not mean that behavior is always objectively reasonable, because people often have incomplete or biased information. Beliefs - the cognitive foundation of attitudes and subjective norms according to this theory - are often irrational or inaccurate (Ajzen & Fishbein, 2005). Rather, the term "reasoned action" means that information is synthesized and evaluated in a systematic manner (Ajzen & Fishbein, 1980). A person's actions will rest ultimately on the information that is available to him or her about that behavior (Ajzen & Fishbein, 2005). Defining and Measuring Behavior. Throughout the early history of attitude-behavior research, much attention was paid to the construct of attitude and its measurement. However, less attention was given to how behaviors were defined and measured for the purpose of empirical studies (Fishbein & Ajzen, 1974). The principles of compatibility and aggregation discussed above were two conceptual advances that led to more precise measurement of behavior and a consequential increase in the observed relationship between attitudes and behaviors. When defining and measuring behavior, multiple acts should be taken into account, consistent with the principle of aggregation. Furthermore, the measured attitude and behaviors should be consistent in action, target, context and time, as proposed by the principle of compatibility.
Predicting Intentions from Attitudes 11 The TRA. Intentions as predictors of behavior. Despite findings that attitudes best predict behavior in general, rather than specific acts, Ajzen and Fishbein persisted in their attempt to predict specific behaviors. Based on the assumption that most social actions are under total volitional control, it can be logically assumed that the immediate antecedents - and the best predictor - of behavior is the intent (Ajzen & Fishbein, 1980). In fact, empirical research has found that when one is able to choose whether or not to perform a behavior, intentions are good predictors of behavior (Ajzen & Fishbein, 2005). Ajzen and Fishbein (1970) assumed that the prediction of intention was both necessary, as well as sufficient, for predicting overt behaviors. Like behaviors and attitudes, intentions can also be viewed as consisting of four elements: action, target, context and time period. In order for intention to adequately predict behavior, there must be correspondence between the intention and the behavior measured in the elements of action, target, context and time (Ajzen & Fishbein, 1980). The size of the intention-behavior relationship will impact the ability that attitudes and subjective norms - the determinants of intention - have in predicting behavior, since intention is the mediator between overt behavior and the constructs attitudes and subjective norms in this theory. Determinants of behavioral intentions. According to the TRA, behavioral intentions are determined by two components: a personal or attitudinal factor and a social or normative factor. The personal factor consists of a positive or negative evaluation of performing the behavior. Attitudes, as defined by Ajzen and
Predicting Intentions from Attitudes 12 Fishbein for the purposes of this theory, are evaluations of any psychological object and are distinct from beliefs, intentions and behaviors (Ajzen & Fishbein, 1980). The simplest and most frequently used method for evaluating attitudes in TRA research is the semantic differential technique (Osgood, Suci & Tannenbaum, 1957). This technique consists of items in which adjective pairs anchor the ends of a seven point bipolar scale, and participants select a point on each continuum which represents their evaluation. Although this technique continues to be used today, it is rarely done with Osgood's original formula and instructions (Krosnick, Judd & Wittenbrink, 2005). To compute an attitude score using the semantic differential, several antonym pairs are presented and the person's attitude score is computed by summing their score on each individual item (Ajzen & Fishbein, 1980). When presenting the semantic differential, it is crucial that the phrasing take into account two factors. The question must evaluate a person's attitudes towards the behavior or behavioral intention, rather than towards a general object, as attitudes towards a behavior have been shown to have greater predictive validity than attitudes towards an object (Ajzen & Fishbein, 2005). Additionally, the attitude question must match the behavior in target, action, context and time (Ajzen & Fishbein, 1980). For example, rather than asking someone to evaluate his or her attitude towards elections (good bad), the question should be phrased as "what is your attitude toward voting in the next election at your local polling station on Tuesday?" It assumed that the more positively a person evaluates a behavior, the more likely he or she is to intend to perform that behavior (Ajzen & Fishbein, 1980). The second determinant of intentions in the TRA model is the social aspect, which takes into account the perception of social pressure to perform the behavior. This is referred to as the
Predicting Intentions from Attitudes 13 subjective norm. It is an evaluation of whether or not a person thinks that those who are important to him or her believe a particular behavior should be performed. Hypothetically, the more a person perceives that others think he or she should perform a behavior, the more likely it is that there will be intention to do so (Ajzen & Fishbein, 1980). Subjective norms are also generally assessed on a seven point scale and can be phrased in the following manner: "Most people important to me think I should I shouldn't perform behavior x" (Ajzen & Fishbein, 1980, p. 57) Subjective norm should also correspond to behavior and behavioral intention in action, context, target and time (Ajzen & Fishbein, 1980). Later empirical reviews found subjective norm to be the weakest predictor in the TRA and the TPB models (i.e. Armitage & Conner, 2001). Perhaps that is due to a theoretical question about the construct raised by Ajzen and Fishbein early on in their research. As early as their 1970 article, Ajzen and Fishbein questioned whether a researcher should evaluate subjective norms in relation to the general other or a specific other (Ajzen & Fishbein, 1970). The example quoted above asks a participant about "most people", a question which generalizes the other. Alternatively, several items could ask a participant about specific people who are important (i.e. a parent, spouse or friends). The exact role of subjective norms as a predictor continues to be investigated. Relative weights of attitude and subjective norms. If individuals hold positive attitudes toward a behavior, and believe that those who are important to them think they should perform the behavior, prediction of behavioral intention is straightforward: it can be assumed that the individuals intend to perform that behavior.
Predicting Intentions from Attitudes 14 However, if attitude and subjective norm are not in agreement (for example, if one does not like attending church but one's family thinks it is important) prediction becomes more difficult, and the relative weight a person gives to the attitudes and subjective norms in making decisions becomes more important. The relative weights of attitudinal and normative factors may vary from one person to another and for each specific behavior. Weights for each component of the theory can be computed for groups of people and for specific behaviors (Ajzen & Fishbein, 1980). The difference in relative weights for each factor was seen in one of the first empirical tests of the TRA. Ajzen & Fishbein (1970) presented pairs of participants with the Prisoner's Dilemma, a problem in game theory used to analyze how people react in situations of cooperation and competition. Attitudes, subjective norms, behavioral intentions and behaviors were all measured. In half the pairs, a cooperative motivation was induced and in half the pairs a competitive motivation was induced. Results showed that subjective norms were more important in determining intentions for the cooperative group than they were for the competitive group (Ajzen & Fishbein, 1970). The role of beliefs. The reasoned action approach attempts to explain, as well as predict behavior. Consistent with this approach, the model also looks at the antecedents of the determinants of intention (Ajzen, 1991). Underlying a person's subjective norms and attitudes towards a behavior are sets of beliefs. Usually people have many beliefs about an object or a behavior, but there are a relatively small number that can be attended to at any given time. These are referred to as salient beliefs, and are thought to be the immediate determinants of attitudes and subjective norms (Ajzen, 1991; Ajzen & Fishbein, 1980). The best way to determine a person's beliefs is to ask
Predicting Intentions from Attitudes 15 open ended questions. When evaluating beliefs and how they relate to attitudes and subjective norms, researchers developed lists of modal beliefs to be tested in a pilot study. The beliefs determined to be the most salient for participants would then be applied in a more comprehensive study with a larger sample (Ajzen & Fishbein, 1980). Since a belief about something can change based on a context or time period, the principle of compatibility applies to beliefs as well. Behavioral beliefs refer to the set of beliefs that influence a person's attitude toward performing a behavior and relate to the person's understanding of the outcome of the behavior (Ajzen & Fishbein, 2005; Madden, Ellen & Ajzen, 1992). In order to understand why a person holds a particular attitude, it is important to look at their underlying beliefs. To predict attitudes from beliefs, the strength of the belief is evaluated and is multiplied by an outcome evaluation (whether that outcome is good or bad). Both factors are typically measured on a seven point scale (Ajzen & Fishbein, 1980). For example in evaluating beliefs about exercise, a subject might be asked how likely they are to lose weight from exercise (measured on a seven point scaled from very unlikely to very likely), as well as how they evaluate the outcome of losing weight (measured on a seven point scale rating from good to bad). The assigned values of the responses for each question would then be multiplied to quantify the behavioral belief. Normative beliefs influence an individual's subjective norms about performing a behavior (Madden, Ellen & Ajzen, 1992). Normative beliefs differ from subjective norms in that they refer to a specific individual or group, as opposed to subjective norms, which are typically measured by asking about others, in general. Subjective norms can be determined from normative beliefs by multiplying the strength of the normative belief by the motivation to