Determinants of Exercise Behavior After a Myocardial Infraction : Beliefs , Intention , Behavior

Exercise is a critical component of cardiac rehabilitation associated with physical and psychological benefits; however, patients are often reluctant to participate in cardiac rehabilitation programs. Their reticence often grows out of their own attitudes towards exercise and the attitudes of people they value. Ajzen and Fishbein's (1980) theory of reasoned action suggest that personal attitudes and subjective norm influence behavioral intention which in tum predicts the likelihood of performing a specific behavior. Both attitude and subjective norm can be measured directly and indirectly. However, Ajzen and Fishbein report the direct measures of attitude and subjective norm result in better prediction of behavioral intention than the indirect measures. Therefore, this study tested the direct and indirect measures of Ajzen and Fishbein's theory of reasoned action in predicting the behavioral intention to enroll in and subsequent attendance behavior in a 36-session exercise cardiac rehabilitation program for post myocardial infarction patients. This study was conducted in three phases in eight participating hospitals. Phase I was conducted on 50 post myocardial infarction patients who.identified a belief inventory for cardiac patients contemplating exercise. Phases II and III were conducted on a second sample of 194 post myocardial infarction patients that resulted in a 6-scale measure that tested the study hypotheses. The regression analyses of Phase III revealed that the attitude and subjective norm measures taken directly and indirectly were found to be significant predictors for the behavioral intention to enroll in a cardiac rehabilitation exercise program. Limitations of the study, recommendations for practice and implication for further research are discussed.


Introduction IbeProblem
Myocardial infarction is the leading cause of death in America. The 1987 United States Statistics (estimates) revealed the incidence of myocardial infarction at 1,500,000 victims per year, with a mortality rate of 513,700 (American Heart Association, 1990).
Data from the Framingham Heart Study revealed 45% of all myocardial infarction victims are under the age of 65, and 5% are under age 40 (American Heart Association).
Almost 80% of the people who have a heart attack will survive, but many will have another one (Kannel, Hjortland, McNamara, & Gordon, 1976). "No matter how uncomplicated the initial clinical course the survivors are always faced with the inherent risk of another myocardial infarction or sudden death" (Naughton, 1985, p. 21). The incidence of potential reoccurrence of a myocardial infarction necessitates preventive measures be incorporated into the prescribed medical regimen to ultimately change behavior.

Theory of Reasoned Action
The term "reasoned action".represents the assumption that people consider the implications of their behavior and make conscious decisions to engage or not engage in actions based on the use of information available to them (Ajzen & Fishbein, 1980).

1
The theory first introduced by Fishbein in 1967 has been further developed, refined and tested (Fishbein & Ajzen, 1975;Ajzen & Fishbein, 1980). The theory is based on the assumption that human beings are quite rational and make systematic use of information available to them. Ajzen and Fishbein's goal is to predict and understand human behavior. The simplicity of the theory is a result of its use of relatively few constructs to explain the attitude-behavior relationship. In its current form, the theory is best represented in summary by the equation: B -I= (AB) W1 +(SN) W2. In this equation, B is a specific overt behavior under the volutional control of the individual, I is the intention to perform that behavior, AB is the attitude toward the behavior, and SN is the subjective norm or the individuals' perception of what most people who are important to the individual think he or she should or should not perform regarding the targeted behavior. W1 and W2 are empirically determined standardized regression coefficients.
Both attitude and subjective norm can be measured directly and indirectly. The direct measure of attitude is derived by having subjects rate the targeted behavior on bipolar scales with adjective endpoints such as good-bad and wise-foolish. The direct measure of subjective norm is derived by having subjects rate the likelihood that significant referents think the subject should engage in the targeted behavior. Although Ajzen and Fishbein's theory states that the direct measures of attitude and subjective norm should predict behavioral intention better than the indirect measures, the indirect measures are more useful in that their subcomponents provide the researcher with more information that can be used to develop strategies for behavioral change (Riddle, 1980). The indirect measure of attitude is derived by two subcomponents: beliefs about the outcomes (bi) of a behavior and the evaluation of those outcomes (ei) where i represents the number of salient beliefs elicited. A sum of the product of the beliefs (bi) weighted by their respective evaluations (ei) produces the indirect measure of attitude toward the behavior.
The indirect measure of subjective norm also is derived by two subcomponents: normative beliefs (nbj) and motivation to comply (mcj). where j represents the number of salient referents elicited. A sum of the product of nonnative beliefs (nbj) weighted by their respective motivation to comply (mcj) produces the indirect measure of subjective norm toward the behavior (Appendix A).

.statement of the Problem
This study utilized the model and methods of Ajzen and Fishbein's theory of reasoned action ( 1980) to identify beliefs that predict the behavioral intention to enroll and participate in an cardiac rehabilitation exercise program. To facilitate this examination, this study was conducted in three phases. The purpose of Phase I was to elicit beliefs regarding exercise in a cardiac rehabilitation exercise program from a sample of post myocardial infarction patients. Phase II consisted of developing a preliminary inventory and administering it to 194 post myocardial infarction patients.
The purpose of Phase II was to perform an item analysis and to refine the instrument.
The purpose of Phase III was to test whether pertinent responses to the scales developed in Phase II predicted exercise intentions and/or participation in an exercise cardiac rehabilitation program.
The specific hypotheses tested in Phase III were: Hypotheses Hypothesis I: Attitude and subjective norm measured by the direct and indirect indices will significantly predict the behavioral intention to enroll in a cardiac rehabilitation exercise program.
Hypothesis II: Attitude and subjective norm measured by the direct and indirect indices will significantly predict participation behavior in a cardiac rehabilitation exercise program.

Lone Term Goal
The ultimate goal of this study was to develop a conceptual foundation for producing and promoting a behavioral change (i.e. modify lifestyle post myocardial infarction in relation to exercise). In order to change behavior, it becomes necessary to determine those components that influence behavior. According to the theory of reasoned action, behavioral change is ultimately a result of a change in beliefs.
4 Therefore, determining a person's salient beliefs and salient referents provides a viable foundation upon which to construct nursing strategies for inducing behavioral change.
The theory of reasoned action provides a scientific basis for uncovering the determinants of intention to enroll and participation behavior in an cardiac rehabilitation exercise program.

Back~round of the Problem
Most people who live through a myocardial infarction will be required to change their behavior and lifestyle to survive. Modifiable habits which put patients at-risk would include cigarette smoking and failure to treat hypertension and/or elevated blood cholesterol. Other factors have been associated with increased risk of cardiovascular disease, although their significance and prevalence have not been precisely determined.
Physical inactivity has not been clearly established as a risk factor for heart disease, but when lack of exercise is associated with excess weight, it unquestionably becomes a factor contributing to heart disease. Cardiac rehabilitation, in order to be successful, must include modifications of behaviors that put patients at-risk.
The goals of cardiac rehabilitation programs for myocardial infarction patients are designed to return the patient to an optimum lifestyle and enhance psychologic recovery, while encouraging a wide variety of behavioral changes. Supervised exercise is the major focus of cardiac rehabilitation programs. Exercise has been associated with improvements in cardiovascular efficiency (Blackburn, 1983), as well as a sense of psychological well-being (Frolicher & Brown, 1981).
The survivor of a myocardial infarction receives numerous directives and prescriptions from physicians, nurses and family members concerning adherence to specific behaviors. Many patients find it difficult to change in all of the ways suggested.
They initially comply with the prescribed medical regimen but their adherence declines after they begin to recover (Carmody, Senner, Malinow, & Matarazzo, 1980;Mark, Kersbkowitz, Mark, & Coleman, 1975;Taglizacozzo & Ima, 1970;Tirrell & Hart, 1980). This decline may be attributed to the resumption of usual activities dominated by modifications (McMahon, Miller, Wikoff, Garrett, & Ringel, 1986). Hence, it is important for the health professional to understand the determinants of exercise behavior and their relationship to one another as they influence this behavior post infarction.

Ihe Cardiac Patient and Exercise
Of the promising available measures, regular physical activity has come to be considered a valuable prescription post infarction. Although research is lacking regarding the incidence of reoccurrence of infarction with and without participation in a cardiac rehabilitation program, numerous research studies detail the positive results of an exercise regimen post infarction. Parchert and Simon (1988) suggest that there are physiological benefits from the adaptive changes that occur to meet the increased oxygen demands of aerobic exercises (p. 11). Other studies have revealed significant reductions in resting heart rate and exercise heart rate for a given exercise level (McArdle, Karch & Katch, 1986;Rovario, Holmes & Holmsten, 1984;Naughton, 1974) and a reduction of both systolic and diastolic blood pressure both at rest and during submaximal work (Hedback, Perk & Perski, 1985;Rovario et al., 1984;Stem & Cleary, 1981). Radionuclide angiography has revealed that exercise therapy improves stroke volume and cardiac output, thus enhancing the ejection fraction, that is, ratio of blood ejected from the left ventricle to end diastolic volume (Fontana, Kerns, Rosenberg, Marcus & Colonese, 1986;McArdle et al., 1986). Folkins, 1976 andMcArdle et al., 1986 found that exercise therapy increases maximal oxygen consumption, which determines the aerobic exercise capacity for an individual. This allows the individual to exercise at a higher workload with a lower oxygen demand (Parchert & Simon, 1988). A growing body of research thus affirms that exercise of an aerobic nature improves cardiovascular efficiency.
The psychological benefits of exercise have also been well-documented. After participating in an exercise program, patients felt better about themselves and exhibited greater confidence (Erdman, Duivenvoorden, Verhage, Kazemier & Hugenholtz 1983;Finnegan & Suler, 1985;Giese & Schomer, 1986;Rovario et al., 1984). Positive effects of exercise participation on mood and emotional stability have also been reported (Folkins, 1976;Giese & Schomer, 1986;Shepard, Kavanagh & Klavora, 1985). In addition, exercise has been associated with decreased depression and anxiety after myocardial infarction (Folkins, 1976;Schomer & Noakes, 1983;Shepard et al., 1985;Stern & Cleary, 1981). It has been documented that exercise therapy reduces stress and enables patients to cope better with a chronic illness (Folkins & Amsterdam, 1977;Sutterley, 1979). In general, exercise participation does much to improve the psychological status of the individual, thereby enhancing the quality of life.
Participation in an exercise program can produce social benefits as well.
Participation in group exercise fosters a sense of camaraderie among post-coronary 6 patients. For example, Erdman et al., 1983Hedback et al., 1985and Rovario et al., 1984 noted that individuals participating in group exercise activities were able to return to work and experienced less work-related stress.
While the various benefits of exercise post myocardial infarction have been welldocumented, exercise rehabilitation trials have not demonstrated a lower morbidity in exercising coronary-heart-disease patients (Rechnitzer et al, 1983;Wilhelmson, Sanne, Elmfeldt, Grimby, Tibblin & Wedel, 1978;Kallio, Hamalainen, Hakkila & Luurila, 1979;Shaw, 1981). Exercise rehabilitation has not been positively associated with a reduction in the risk of recurrence of myocardial infarction (Wilhelmsen, Sanne, Elmf eldt, Grimby, Tibblin & Wedel, 1978;Shaw, 1981;Rechnitzer et al., 1983). Oldridge, Guyatt, Fischer and Rimm (1988), however, carried out a meta-analysis on the combined results of ten randomized clinical trails. Their results suggested that for selected post myocardial infarction patients, cardiac rehabilitation has a beneficial effect on mortality, but not on non-fatal recurrent myocardial infarction. Similarily, Naughton (1985) reported that in three of five clinical trials of myocardial infarction patients, exercise therapy tended to reduce mortality rates; however, exercise did not affect reoccurrence rates of nonfatal infarctions. Cardiac rehabilitation programs, on the other hand, are designed to lessen the immediate adverse physiologic and psychologic impact of the infarction and enhance the long-term physical, emotional and vocational status of the individual. Schroeder (1985) articulated the goal of cardiac rehabilitation as, "not only add years to life but add life to years" (p. 21 ).

Cardjac Rehabilitatjoo
The concept of cardiac rehabilitation connotates three major components: education, counseling, and exercise therapy, the latter being a crucial component of cardiac rehabilitation. Cardiac rehabilitation programs are standardized across the country. Based on an exercise-stress-test, the cardiac patient's physical status is assessed and exercise prescription is then formulated. Cardiac patients exercise at 60% of their maximal heart rate to ensure safe, yet beneficial, exercise effects. The participants exercise three times per week for half an hour to one hour. Such sessions include warmup exercises, peak exercise and cool-down exercises. ·Educational classes on various topics related to risk-factor-modification and individual counseling are provided. This The challenge for the nurse caring for the swvivor of a myocardial infarction lies in helping the patient integrate the physical and psychosocial changes into an acceptable lifestyle (Owen, 1987). More specifically, the nurse's role in the rehabilitation process encompasses: (1) educating the patient about risk factors and lifestyle modification, (2) providing emotional support to patients and their families, (3) assessing exercise tolerance and improvement therein, (4) motivating patients, (5) improving patients attitudes about health risks, and (6) decreasing fear and self-doubt about the future for patients with coronary artery disease (Parchert & Simon, 1988). Within the past five years nurses have focused more on increasing and promoting participation in the exercise regimen of the rehabilitation process (Parchert & Simon).
It has not been shown definitively what factors motivate patients to enter and participate in exercise programs (Holm, Fink, Christman, Reitz & Ashley, 1985). Few studies account for the belief factors that may influence an individual's participation in a rehabilitation program. One theoretical formulation that addresses the beliefs, attitudes, social norms and intentions of specific behaviors is Azjen and Fishbein's (1980) theory of reasoned action. The patient's attitudes and perceptions of what significant others believe about exercise in a cardiac rehabilitation program may determine the behavioral intention to enroll in a cardiac rehabilitative exercise program and therefore participation in the program. Beliefs are amenable to change. The nurse caring for the swvivor of a myocardial infarction is in a critical position to influence this patient's attitude through education, role-modeling and instillation of values related to health and exercise. It was thus necessary to examine those components that have been suggested to influence and determine behavior. One underlying key variable to understanding patient behavior is attitude; the patient's learned predisposition to respond to a situation in a consistently favorable or unfavorable way.

Jlistorical Perspective on Attitude
The concept of attitude has played a major role throughout the history of social psychology (Ajzen & Fishbein, 1980). The simple concept of attitude, first alluded to as a state of preparedness or a set to make a particular overt response, has grown over time into a complex, multi-dimensional concept consisting of affective, cognitive and conative components (Fishbein, 1967). As early as 1862, Herbert Spencer, a psychologist, stated that "the concept of attitude has been recognized as a factor contributing to judgmental processes, where much depends on the subject's attitude of mind while he or she is coping with a controversy" (Ajzen & Fishbein, 1980, p. 13). Znaniecki and Thomas (1918) were the first to use the attitude concept to explain social behavior. They viewed attitudes as individual mental processes that determine a person's actual and potential responses. This notion of attitudes as behavioral dispositions went unchallenged until the late 1960s.
Conceptual and methodological issues associated with the attitude construct complicated one another in early attitude research. However, L. L. Thurstone (1931) and Thurstone and Chave (1929) made significant contributions by applying psychometric methods to the measurement problem. Thurstone developed a scale from which an attitude score represented a person's position on a bipolar affective continuum in relation to the attitude object. "In Thurstone scaling, the attitude score represents a person's evaluation of an object implied by a set of his beliefs, intentions, or actions" (Ajzen & Fishbein, 1980, p. 15 ). Very early on, Thurstone recognized the attitude-behavior inconsistency by noting similar attitudes among individuals, yet arrived at by entirely different routes. Rensis Likert (1932) then proposed a similar, yet simpler, scaling procedure in which summated attitude scores reflect patterns of beliefs, intentions or actions. Like Thurstone scaling, the Likert scale resulted in a single score that represented the degree to which a person is favorable or unfavorable with respect to the attitude object. A given attitude score could again reflect different patterns of beliefs, intentions or actions. These early scaling techniques provided the groundwork for further psychometric developments.
Conceptually, early attitude research of the 1930s was predominantly descriptive in nature and used attitude surveys. G. W. Allport (1935) addressed the complexity of the attitude concept, inferring that attitudes are comprised of affect, cognition and conation. However, early research utilizing unidimensional attitude scales, which captured affect only, revealed a close link between attitude and behavior. R. LaPiere's 1934 study of racial prejudice against Chinese people impacted attitude research with a classic example of the failure of attitudes to predict behavior using affect only as an indicator of attitude. As additional studies supported this finding, L. W. Doob (1947) proposed a rationale similar to that of Thurstone where similar attitudes can be expressed in different actions. As time passed, further research expanded the definition of attitude to include affect and cognition; this return to an idea, first proposed by Allport, was an effort to explain the low attitude-behavior relationship. During the 1950s, additional attitude measurement techniques were developed to capture the cognitive and conative components of attitude as well. Of particular interest, are two scaling techniques, the Guttman scale (1944) and Osgood, Tannenbaum and Socci semantic differential scale (1957). Both scales result in a single score representing a person's evaluation of an attitude or affect These additional · attitude measurements techniques did little to verify the attitude-behavior relation. However, Thurstone and Likert's early work significantly impacted future directions for the attitude-behavior dilemma. Both Thurstone and Liken scales rely on beliefs or intentions to infer an individual's attitude. This implies that in providing a measure of affect, the standard scaling procedures already take into account cognitions, conations or both (Ajzen & Fishbein, 1980). In light of this revelation, measurement as well as conceptualization problems of the attitude-behavior relations prevailed. Ajzen and Fishbein (1980) revealed that most early attitude investigators worked on the assumption that attitudes explain and predict behavior and, thus, dealt with studies designed to examine attitude formation and change. Actual investigations dealing with the attitude-behavior relationship were lacking.

IJle Cardiac fatient and Reasoned Action
Consideration of attitudes and subjective norms constitute a beginning toward understanding why patients behave the way they do. Ajzen and Fishbein (1980) assert that possession of the factors that influence behavior requires knowledge of the detenninants of the attitudinal and normative components of the patient's environment Ajzen and Fishbein view beliefs as underlying a person's attitudes and subjective norm, ultimately determining intentions and behavior. Only salient beliefs (those beliefs attended to regarding a given object or situation) are the immediate determinants of attitude while salient referents (those individuals that will influence behavior) determine the subjective norm measure. In the final analysis, a patient's behavior is best explained by reference to beliefs. Beliefs, therefore, influence attitude and subjective norm, which influence intention, which, in turn, influence behavior. For the nurse caring for the survivor of a myocardial infarction, insight into the beliefs of the post infarction patient regarding a cardiac rehabilitation program provide an understanding of factors influencing a behavioral change.
According to the theory of reasoned action, behavioral change is ultimately the result of a change in beliefs. Ajzen  This study was an attempt to look systematically at the underlying beliefs post myocardial infarction patients have about exercise in a cardiac rehabilitation program.
The specific task of this investigation was to link the beliefs about exercise behavior of t my ocardial infarction survivor to the intention to enroll in a cardiac rehabilitation the pos gram and subsequent participation behavior via Ajzen and Fishbein's strategies. Phase pro 1 provided the belief item pool for development of the Phase II questionnaire. Phase II encompassed item analysis and instrument refinement of the questionnaire developed from the belief item pool. Phase ID of this study tested the validity of the developed instrument.

Si2Jlificance of the Study
Field work with a representative sample of post myocardial infarction patients revealed that they use the following activities to comply with the exercise prescription of the physician: self-directed activities such as walking; exercise directed by non-medical personnel, such as attending aquatic programs or aerobic classes; and medicallysupervised activity, such as participation in a cardiac rehabilitation program. There are some patients, however, who do not comply with their physician's exercise prescriptions at all, these patients presumably do not believe that exercise is a critical component of rehabilitation. Few researchers have examined the beliefs of post myocardial infarction patients regarding exercise in a cardiac rehabilitation program.
Health professionals concern themselves with various strategies to alter patient behaviors, but do not always base these strategies on a professional understanding of the motivators of patient behavior. Through application of the theory of reasoned action, this study was undertaken to examine whether such strategies should be based on an understanding of the determinants of behavioral intention; specifically, the determinants of attitude and subjective norm that undergrid behavioral change. Post myocardial infarction patients were selected for study since they comprise a large population that require substantial lifestyle change. Of particular interest in this study was the patient's willingness to engage in recommended exercise regimens following myocardial infarction. This study, therefore, was designed to provide a better understanding of the l . contributions of the direct and indirect determinants of behavioral intention and re anve behavior. These assessments were made at hospital discharge, and at the completion of 13 36 sessions in a cardiac rehabilitation exercise program, re~pectively. The information obtained from the first phase of this study reflected the pattern of beliefs common to survivors of a myocardial infarction toward exercise. In addition, the findings of this study have the potential to contribute to the ongoing research of the theory of reasoned action by complimenting the predictability of the theory, assessing the role of behavioral intention, and acknowledging the usefulness of the theory's constructs and their determinants to influence behavior, specifically patient behaviors.

Definition of Terms
For the purpose of this study, the following definitions of terms were used: Behavior: Observable acts that are studied in their own right. Behavior in this study is operationally defined as: attendance in a cardiac rehabilitation program generated from the number of sessions attended in a 36 session cardiac rehabilitation program.

Behavioral intention:
The individual's intention to perform a specific behavior.
Behavioral intention is operationally defined as a score derived from a semantic differential type scale designed to reflect the subjects intent to enroll in a cardiac rehabilitation exercise program.
Attitude: An individual's predisposition to respond in a consistently favorable or unfavorable manner with respect to the behavior. Attitude is operationally defined by a direct measure of attitude (AB) and an indirect measure of attitude (L bi ei).
Direct measures of attitude (AB) are generated from a 7 point semantic differential scale.
Indirect measures of attitude (L bi ei) are comprised of two measures: Beliefs about the outcome of a behavior (bi) are generated from a 7 point bipolar scale.
Evaluations of those outcomes of a behavior ( ei) are generated from a 7 point semantic differential scale. 14 Subjective Norm: One's perception that most people, who are important to the individual, think one should or should not perform the behavior in question. Subjective norm is operationally defined by a direct measure of subjective norm (SN) and an indirect measure of subjective norm (L nbi mci).
Direct measures of subjective norm (SN) are generated from a 7 point bipolar scale.
Indirect measures of subjective norm (L nbi mci) are comprised of two measures: Strength of normative beliefs concerning the referents (nbi) generated from a 7 point bipolar scale.
Motivation to comply with the referents (mci) are generated from a 7 point unipolar scale.

Referent:
A person whose opinion about the behavior is valued by the subject. Referents in this study are those individuals cited in the Phase 1 questionnaire (e.g. Which individuals or groups of people would approve/disapprove of your exercising in a cardiac rehabilitation program?).

Assumptions of the Study
Assumptions of the study were: patient gender, culture, occupation, or personality would not affect the behavorial intention, or plan to enter a cardiac rehabilitation program and/or subsequent exercise behavior at 4 to 6 weeks post-infarct and 36-weeks respectively; all patients would be able to financially afford a cardiac rehabilitation program; all patients would be provided similar explanations regarding cardiac rehabilitation at discharge in all participating hospitals ;and all patients would have access to a cardiac rehabilitation program. A final assumption is that all programs are equally good and equally attractive.

IJmjtatjons of the Study
The sample of subjects in this study were first time myocardial infarction patients or post myocardial infarction patients who have not previously attended a cardiac rehabilitation program. Generalization of the findings to other patient populations (coronary-artery bypass, coronary angioplasty, arthrectomy patients, etc.) or settings (general physical activity programs, etc.) is not appropriate.
Predischarge anxiety and/or pressures generated by transition from the sick role to what Barie (1969) has termed the "at-risk" role may alter the study's findings since the final questionnaire was administered just prior to discharge. It is possible that the nursing recruitment efforts during hospitalization accounted for a positive behavioral intention to enter a cardiac rehabilitation program at hospital discharge.
Other intervening variables that could have affected the study results were: attempts to acquire medical knowledge after myocardial infarction by, for example, reading medical literature, etc. Personal contact with others who have had a myocardial infarction, or who know an individual who has experienced a myocardial infarction, may also have altered the study results.

Conceptual Framework
Jl.ulany's Theory of Proposjtjonal Control Amidst the attitude-behavior controversy of the 1950s, Don Dulany (1961Dulany ( , 1964 proposed a theory of propositional control. Developed initially in the framework of learning theory, Dulany's proposal was essentially a theory that led to the prediction of overt behavior. This theory was developed within the context of studies examining and testing verbal conditioning and concept attainment. Dulany's concern was with predicting the probability with which an individual will make a particular verbal response or class of verbal responses in a particular situation.
Critical to Dulany's theory is behavioral intention, as the immediate antecedent of overt behavior. Dulany was concerned with a precise and specific type of behavioral intention or one's intention to perform a given action in a given situation. Hence, Dulany focused on measuring the subject's intention to perform the specific behavior he was interested in predicting.
Dulany's theory takes the form of a linear multiple regression equation, where behavioral intention (BO is composed of two distinct components, each of which has two parts. The first component of this equation comprises a quantification of the subject's hypothesis that the occurrence of the particular response will lead to a certain event or class of events. This is termed RHd. RHd is multiplied by a numerical expression of the subject's evaluation of (or attitudes toward) those events labeled A. Dulany's theory (Fishbein, 1967) has been developed within the context of studies of verbal conditioning and concept attainment. Moreover, Dulany has been concerned with predicing the probability with which an individual will make a particular verbal response or class of responses ( p. 487).

fjshbein's Adaptation of Dulaoy's Theory
Martin Fishbein ( 1967) proposed a theoretical model of behavioral prediction as an adaptation of Dulany's theory . Fishbein's formulation provides a comprehensive understanding of the attitude-behavior relationship by identifying other variables that interact with attitudes as determinants of overt behavior. Fishbein (1967) conceptualized beliefs as determinants of attitude and behavioral intention as a consequence of attitude.
Later research revealed that behavioral intention could be predicted from attitude, and intentions were related to behaviors . Fishbein asserted (1967a) that behavioral prediction entails knowledge of beliefs, attitudes, behavioral intention and behavior. Fishbein (1967) Fishbein (1967) made minor changes in the meaning of the constructs, as well as in the way in which they are measured. Fishbein ( 1967) also attempted to emphasize the importance of beliefs as building blocks of attitude and subjective norm, with behavioral intention as a link in the causal chain to behavior.
It can be seen that Fishbein's (1967) adaptation of Dulany's theory leads to the prediction that an individual's behavioral intention and performance of the behavior are a function of attitudes and subjective norms; and attitudes and subjective norms are themselves functions of beliefs. Fishbein's theory is algebraically expressed as follows: The two major components of this model are weighted for their importance in the prediction of behavioral intention (W 0 W1). For some individuals, normative considerations may be more important in determining behavioral intention than are attitudinal considerations. "The assignment of the relative weights to the determinants of behavioral intention greatly increases the explanatory value of the theory " (Ajzen & Fishbein, 1980, p. 6).
Research during the early 1970s revealed a low empirical relation between knowledge of a subject's attitude and prediction of behavior.  proposed that the attitude-behavior relationship is contingent upon its relationship to other determinants, specifically behavioral intention. These authors concluded that a person's attitude toward an object may not be related to any single behavior which is performed with respect to the object as postulated by the traditional attitude approach.
However, attitude should be related to the overall pattern of behavior. This notion takes into account beliefs, attitude, and behavioral intention as determinants of behavior. This hypothesis was proposed to account for the low attitude-behavior relation.

Du:ory of Reasoned Actjon
The model holds that two types of constructs underlie any intention under the individual's volitional control (Miniard & Cohen, 1981). Both constructs, attitude and subjective norm, are functions of beliefs.
Underlying assumptions of the theory of reasoned action are that: (1) behavior is directly caused by behavioral intentions (conation), which are caused by attitudes (affective evaluations), which in tum reflect beliefs about the consequences of behavior influenced by the subjective evaluation of the consequences, and (2) subjective norm functions as an independent variable which is caused by normative beliefs and motivation to comply, which indirectly effects behavior through its effect on intentions.
Subjective norms are thought to reflect beliefs about the behavioral expectations of significant others, influenced by the motivation to conform to them, summed over all significant others. All other variables are assumed to effect behavior via their effects on the cognitions or beliefs which underlie attitudes I subjective norms and on the model weights. Ajzen and Fishbein's model (1980) provides a deeper understanding of human behavior by observing the effects of variables external. to the basic theory . Factors such as personality traits, demographic characteristics, habits, social norms, situational factors and traditional measures of attitudes toward persons or situations have been found by Ajzen and Fishbein to exert no influence on any specific behavior, since they have no consistent effects on the beliefs underlying the behavior. Thus, the relation between external variables and behavior, is at most, indirect. Ajzen and Fishbein posit that variables external to the theory cannot provide explanatory value of behavior. The theory of reasoned action proposes a single set of constructs that account for the observed relations between external variables and behavior. This model distinguishes the three components of the traditional attitude concept (affect, cognition and conation). Ajzen and Fishbein's model represents a prominent and influential attitude-behavior model. Liska (1984), Ajzen and Madden (1986) have proposed various revisions to the present model in an ongoing effort to improve the attitude-behavior relationship. The theory of reasoned action appears to be a potentially useful model to predict, explain, and understand the determinants of human behavior that influence the actions of post myocardial infarction patients. Ajzen and Fishbein (1980) posit the behavioral criterion or the behavior of interest entails four elements: action, target, context and time. There must be correspondence between each of these elements in relation to the behavior of interest.
The action element reflects the behavior of interest; the target element is that at which a behavior is directed; the context and time elements reflect the context in which the behavior occurs and the time at which it is performed. Each of these elements can be very specific or more general. These behavioral elements are important to consider when defining the behavior of interest. Once the behavior is chosen, it must be measured.
Clearly, the way observations are made influences the data obtained. Ajzen and Fishbein suggest that clearly defining the behavioral elements are preliminary measures when applying the theory of reasoned action. For use in this study, the behavioral elements of the theory were specified as: the behavioral criterion (action) was exercise participation with respect to a specific target, the post myocardial infarction patient, in the context of a cardiac rehabilitation exercise program, at four to six weeks after hospitalization.
In this study, the theory of reasoned action provided the guiding principles to understand and predict the behavioral intention to enroll in a cardiac rehabilitation exercise program and subsequent participation in the program (number of sessions attended) for the post myocardial infarction patient.

Review of Literature
This chapter provides a review of the literature related to the theory of reasoned action proposed by Fishbein (1967), and modified by Fishbein and Ajzen (1975) and Ajzen and Fishbein (1980). Empirical research associated with the theory of reasoned action is reviewed, with a focus on research related to the beliefs of cardiac patients in regard to exercise behavior. Application of the theory of reasoned action to nursing research is also reviewed. The final section of this chapter offers a justification for applying the theory of reasoned action to investigate the intention to enroll in a cardiac rehabilitation exercise program and subsequent exercise participation behavior.

21
Attitude research has long plagued american social psychology. Social psychologists have used attitude to explain overt behavior. However, an accumulation of contrary evidence has frequently refuted this relationship in both the laboratory (Berg, 1966;Bray, 1950,;Kutner, Wilkins & Yarrow, 1952;LaPiere, 1934) and field settings. Ajzen and Fishbein (1970) designed an experiment to test the validity of a theoretical model of behavioral prediction presented by Fishbein ( 1967) in the context of a Prisoner's Dilemma (PD) game situation. Over the total sample of subjects, the correlations between behavioral intentions and game behavior were .897 and .841 (p < .001) for the two games. The high correlations between behavioral intention and behavior were deemed to be attributable to the fact that the measures of behavioral intention were: (a) behavior specific (b) taken immediately prior to the behavior and (c) were taken after eight warm-up trials allowing the subject to form fairly accurate beliefs about consequences of cooperating and about the partner's expectations. This hypothetical situation provided early insight into the complexity of understanding or predicting human behavior while raising many theoretical and methodological problems associated with Fishbein's (1967) model.

Laboratory Studies
In an attempt to provide answers to the problems they discovered in the model, Ajzen and Fishbein (1972) examined four hypothetical decision situations: investment, operation, renting and transplant--each involving a different degree of risk in making a behavioral choice. Each situation included information about whether the subject's own estimate of probability of success was high or low and whether the probability of success as estimated by close family and friends was high or low. Consistent with Fishbein's model, attitude and subjective norm were found to predict behavioral intentions in all four hypothetical situations with correlations ranging from .299 to .814. Attitude carried more weight than subjective norm in determining intentions.
The utility of the Fishbein model in the family planning area was demonstrated by Jaccard and Davidson (1972) in a test of two hypotheses: (1) behavioral intention to use birth control pills is a function of (a) one's attitude toward using birth control pills and/or (b) one's normative beliefs weighted by one's motivation to comply with those perceived norms and (2) attitude toward using birth control pills is a function of (or is highly correlated with) the sum of one's beliefs about the consequences of using birth control pills times the evaluation of those beliefs. The findings which indicated substantial correlations among attitude, subjective norm and behavioral intention (r = .84 p < .01) and indirect attitude measures (r = .79 p < .01) partially supported Fishbein's model.
According to Fishbein's model, variables not included in the theory can affect behavioral intentions and behavior indirectly, by influencing attitude, subjective norm and/or the relative weights of these components. Ajzen and Fishbein (1974) investigated two such variables: (1) situational variations that influence interdependence of group members and (2) subject's beliefs about and attitudes toward their co-workers. A series of three studies was conducted in which the task of three-person groups was to balance a board in the shape of an equilateral triangle by moving their respective corners of the d T 0 behaviors were assessed: the number of instructions the subject sent to a coboar · w k ( Omm unicative behavior) and the proportion of instructions from each co-worker wor er c with which the subject complied (compliance behavior). Subjects were 144 male undergraduates. Results supported the model when subjects' intentions to perform communicative and compliance behavior correlated highly with their attitudes toward the behaviors and with their normative beliefs about the behaviors. Behavioral intentions correlated significantly with communicative behavior (r = .690, p < .01) and compliance behavior (r = .211, p < .01), although intervening events such as percent of reenforcement and incompatibility were found to attenuate the intention-behavior relation . Newman (1974) tested Fishbein's model (1967) in work organization to assess its usefulness in organizational situations. Newman's study examined the relative efficacy of Fishbein's model and traditional job attitude measures as predictors of absenteeism and job turnover in 108 nursing home employees. The data of this study suggested that traditional job attitude measures were more efficacious in predicting absenteeism while Fishbein's model was more effective in predicting voluntary resignations. Fishbein and Coombs (1974) tested the validity of Fishbein's model in the context of a political election. Application of the theory to voting behavior allowed them to investigate the locus of effect of a political campaign on the beliefs of the voting public.

Fjeld Studies
It was found that the basic assumptions of Fishbein's model fit well with the data collected in the 1964 presidential election between Goldwater and Johnson. Specifically, the correlations between the direct measure of attitude toward a candidate and the specific measure of attitude ranged from + .69 to + .87. The direct measure of attitude was consistently a better predictor of voting behavior. It was found that the specific measure of attitude toward a candidate is more highly related to voting intention than is the direct Voting intentions correlate even more highly with the actual vote while both measure· 24 the direct and indirect measures of attitude correlated well with actual voting behaviors .
. hbe. and Coombs' data suggested additional validity for Fishbein's model within the flS ID political arena.
In response to continuing skepticism about Fishbein's (1967) model andAjzen's (1975) refinements of that model, which was at variance with the traditional measure of attitude, Ajzen and Fishbein (1977) re-examined the attitude-behavior relationship with a review of pertinent empirical research. The extensive review supported the contention that strong relationships between attitude and behavior were obtained only when there was high correspondence between at least the behavioral elements (action, target, context, time) which define the behavior.
Helping behavior too, has been assessed using the theory of reasoned action. Pomazal and Jaccard (1976) found that donating blood during a college blood drive was significantly related (r = .59) to the intention to donate blood. Behavioral intention was found to relate to attitudes and subjective norms (multiple R = .60) in this study. Also, Pomazal and Jaccard found that both donating behavior and the corresponding intention were significantly correlated with perceived moral obligation to donate blood (r = .43 and r = .50, respectively). Zuckerman and Reis (1978)  u· on behavior relationship decays more rapidly than the attitude-behavior the 111ten -. shi 'p Thus when the behavior is immediate, attitudes do not add to intentions in relanon · • predicting behavior, however, when the behavior is deferred, the relationship between intention and behavior is weaker, and attitudes help significantly in accounting for the behavior. Saltzer (1980) applied Fishbein and Ajzen (1975) theory of reasoned action to investigate the determinants of behavioral intentions toward losing weight and actual weight loss. The questionnaire measured the predictor variables of the model: personal attitudes, subjective norms and behavioral intent. Only the number of pounds intended to be lost in 6 weeks was used in the final analysis since all subjects indicated a very strong positive intention to lose weight. The behavioral criterion was actual weight loss. In the 79 subjects who completed the weight reduction program, the two predictive variables of the model when taken together were significantly correlated (R = .38, p < .01) with behavioral intention. The subjective norm measure was the significant predictor of behavioral intention with a beta weight of .28, where attitude = .16. Behavioral intention was significantly associated with absolute weight loss (r = .44, p < .001). Saltzer's study supported Fishbein and Ajzen's model and demonstrated the importance of social pressures and personal beliefs in the prediction of a health-related behavior (weight loss). Smetana and Adler (1980)  survey instrument was mailedto 296 respondents and returned. All subjects who returned a questionnaire were telephoned two weeks later to assess jogging behavior. The major test ofFishbein's model was whether or not the behavioral intention (to jog regularly) can be predicted from a linear combination of attitude and· subjective norm measures.
Attitude and subjective norm when· combined provided high prediction for the intent to jog (R = .742). The attitude measure proved to be a better predictor of intention to jog, r = .643; than the normative measure r = .157, both significant (p < .05). The intercorrelation between behavioral intent and behavior was .820. Riddle states that this model should be tested using other health-related behaviors inclusive of other types of exercise, to augment current understanding of human health behavior. Bentler and Speckart (1981) assessed the causal structure of Fishbein's model to evaluate the causal predominance of attitudes over behavior. They assessed attitudes, subjective norms, intentions related to studying, exercising and dating in a sample of 158 tu dents Structural equation analysis revealed that attitudes have causal priority colleges · over behaviors. Attitudes exerted a significant direct effect on all three behaviors while nl e of the three behaviors had a significant direct effect on attitude. Intentions to 0 yon study and date accurately predicted the behaviors while the intention to study most strongly correlated with subjective norm. A direct path was found between attitude and behavior. Bentler and Speckart concluded that attitudes cause behavior.
Bagozzi ( 1981 a) employed a causal modeling methodology to examine key hypotheses within the context of two leading theories of the relationship between attitude and behavior, Fishbein and Ajzen expectancy-value model and Trandis, semantic differential model. Following Fishbein strategies, two independent samples of 50 subjects identified salient beliefs related to the behavioral criterion of donating blood. In one sample, respondents were asked to list their beliefs about the consequences that giving blood would have for them personally while the second sample listed their beliefs about the costs that giving blood would have for them personally. Using Liseral the results revealed that attitude influenced behavior only through its impact on behavioral intent. The relationship between attitude and intention was stronger when Fishbein's expectancy-value attitude measures were used as predictors than when Trandis' semantic differential measures were employed. Bagozzi (1981b) also performed a quasi-experiment utilizing the same sample previously discussed (n = 284). Three groups of respondents were examined: those individuals who had never given blood; those individuals who had given before but in the distant past (2 months or greater); and those individuals who had given blood within the immediate past (20 minutes or less). The study determined the construct validity; the reliability; and the convergent, concurrent, discriminate, predictive and nomonological validities within the context of blood donation for expectancy-value and semantic differential measures of attitude. Both the Ajzen-Fishbein and Trandis models achieved convergent, concurrent, discriminate and predictive validity. The nomonological validity . d remained questionable. The extent of past behavior and the elapsed time since of atntu e ce of the behavior appeared to attenuate both the attitude-intention relationship perrorman and the intention-behavior relationship. Personal and social normative beliefs as copredictors of intention were found to be insignificant. Fishbein and Ajzen's expectancy -value attitudes were found to occur as multidimensional responses indicating the impact of consequences of the act and the evaluation those consequences have on attitude while Trandis semantic differential attitudes were found to exist as unidimensional responses due to the global nature of the consequences of the act.
Two concepts related to the theory of reasoned action are the passage of time and the effect of past experiences. Vinokur and Ajzen ( 1982) examined the notion of these concepts as influencing factors in human judgement Vinokur and Ajzen (1982) conducted two experiments which tested the hypothesis that earlier causes in a chain of events are credited with greater relative importance than later, more immediate causes. In the first experiment, 170 undergraduates judged the relative contributions to success or failure made by members of a team who initiated a problem-solution process versus team members who terminated it, representing prior and immediate causes. In the second experiment, 206 undergraduates rated the importance of prior and immediate causes of four varied situations, each of which involved two events that could have produced an important outcome. In both experiments, prior events in a causal chain were perceived to be more important than immediate events. In addition, the second experiment revealed that when one event is considered to make a greater causal contribution to an outcome than another event, the temporal or causal relationship does not change the perception of the relative importance. This experiment supported the fact that the causal primacy effect is not just a matter of the order in which prior and immediate events occur. Instead, it appears to be related to the perceived freedom of prior events that control and thereby appear to restrict the freedom of subsequent events. A . n Timko and White (1982) examined the role of individual differences as JZe , .
ton people's tendencies to act in accordance with their attitudes. "Selfthey unpac · . ·ng" is a term frequently utilized to define the extent to which behavior is moniton ta ble to situational or interpersonal cues. The behavior of high self-monitoring suscep . individuals is guided by situational factors with little correspondence between attitude 29 and behavior, while the behavior of low self-monitoring individuals is guided by interpersonal cues with substantial correspondence between attitude and behavior. The effect of self-monitoring on the attitude-behavior relationship led to the following hypotheses. The researchers hypothetized that: (1) attitude towards a behavior predicted intention of both high and low self-monitoring individuals, (2) the relation between intention and behavior is stronger for low than high self-monitors, and (3) perceived behavioral relevance may mediate the relation between general attitude and specific behavior where low self-monitoring indivduals are more likely to perceive the action implications of general attitudes than are high self-monitors.
A total of 155 undergraduates responded to three questionnaires over a six week pericx:l before and after the 1980 presidential election. The first questionnaire assessed self-monitoring tendency, the second questionnaire assessed attitudes and intention with respect to two behavioral criteria; voting in the election and smoking marijuana, and included self-reports of actual behavior. Two weeks after the presidential election participants were telephoned to assess behavior. The third questionnaire assessed attitudes toward smoking marijuana. Pearson-product moment correlations were computed between attitudes of high, moderate and low generality and the behavioral criteria. The sample of respondents were dichotomized at the median score forming high and low self-monitoring subsamples.
The findings supported the first two hypotheses. Low self-monitors tended to exhibit stronger attitude-behavior correlations than high self-monitors. There was, however, no difference in the prediction of intentions from attitudes, a finding that d th e perceived relevance of the effect of self-monitoring. Instead, the difference refute 30 1 te d in the relation between intentions and behavior; low self-monitors exhibited was oca significantly stronger intention-behavior correlations than did high self-monitors. The findings support the interpretation that the two types of individuals are equally aware of the implications of their attitudes, but that high self-monitors, being sensitive to situational demands, are less likely to carry out their previously formed intentions than low self-monitors, who are more attentive to internal cues. This study illustrated the value of measuring behavioral intentions as a variable that mediates between attitudes and behavior. Kantola, Syme and Campbell (1982) applied Fishbein's model to household water conservation. Variables external to the model were included in the questionnaire to test the sufficiency of the model in explaining water behavior intentions to conserve water.
The external variables were also used to examine the role of individual differences. The study was done on the western coast of Australia during a three-year draught. Thirty-nine subjects participated in a pilot study which determined consequences of conserving water during the ensuing summer. The six salient beliefs composed a section of the final questionnaire while the three normative beliefs questions were researcher determined.
The behavioral intention measure was a Likert scale response to the statement "I intend to conserve more water this year than iast year." The final questionnaire was given to 125 subjects attending the Perth Royal Show (State Fair).
In support of Fishbein's model, attitude and subjective norm were found to predict the intent to save water. Examination of correlations showed that subjective norm had the largest correlation with behavioral intention. Results also indicate that individual differences in age and sex played a significant role in intentions to conserve water. It must be noted that the subjective norm measure was not derived by Fishbein strategies yet achieved the largest correlation (r2 = .18) with behavior intention. aumkrant and Page (1982) examined the convergent, discriminant and predictive validity of Fishein's model in relation to blood donation. Two exploratory research di .,ovided the salient beliefs, salient referents and the 20 most discriminant items stu es P• for the final questionnaire. The final questionnaire was administered over a 2 week period preceding the blood drive. Results from 124 respondents supported each component of the model as a determinant of behavorial intention. Predictive validity of the constructs was supported although behavioral intention to donate blood was found to be primarily under attitudinal control. Convergent and discriminant validity were supported where attitude and subjective norm were found to be separate, unidimensional but correlated constructs. In general, the results support a simpler model than the current Fishbein (1975) model, in which a single attitude construct and a single normative construct are antecedents of intention. Ryan (1982)  tal effects of cognitive and nonnative information on intentions than did the experunen attitude measure. Manstead, Proffitt and Smart (1983) examined the applicability of Ajzen and Fishbein's theory of reasoned action to the prediction and understanding of infant feeding. This study proposed two distinct theoretical advantages: (1) in the context of adult behaviors it is unusual for primiparous mothers to choose and use a method of infant feeding and (2) prenatal intentions cannot reflect past behavioral experience.
Salient belief statements and social referents were not derived according to Fishbein and Ajzen but were derived from a study conducted by the British government in 1975.
Twelve salient beliefs and four salient referents were used in the study.
One hundred twenty-three primparous and 127 multiparous mothers were given the predictive questionnaire during prenatal clinic visits. Each group was sent a second questionnaire 6 weeks after the birth of the baby. Results of the study were based on 215 completed questionnaires of which 106 were the primiparous test group. This study utilized a control group in which 85 primiparous mothers completed only the follow-up questionnaire. Behavioral intention was assessed using a seven point scale. The followup questionnaire elicted prenatal intentions and post partum infant feeding behavior used in seven specific stages (in hospital through weeks 1 through 6) of the baby's life Infant feeding behavior was measured dichotomously. A score ofO was given the mothers who never breast-fed their babies while a score of 1 was given to mothers who had breast-fed their babies at some point in time during the 6-week postpartum period.
Behavioral intention and behavior were strongly related (r = .82). Using multiple regression, the attitude and subjective norm measures were found to exert equal influence on behavioral intention in primiparous mothers. The attitudinal component of behavioral intention was weighted more heavily for multiprimiparous mothers than for primiparous mothers. Chi-square analysis revealed that the control group of mothers did not differ significantly from that of the primiparous test group. As a secondary aim of the study, k nn an and Reis (1978) "time interval" hypothesis was tested. This hypothesis the Zuc e sed the attitude-behavior relationships of mothers with differing intervals between asses the measurement of predictor variables and the measurement of behavior. The 33 zuckennan-Reis "time interval" hypothesis predicts that mothers with long intervals between the assessment of predictor variables and the assessment of behavior will be more likely to display a direct relationship between attitudes and behavior (Manstead et al., 1983, p. 659). Results for this hypothesis indicated for the short delay group (less than 25 day interval), the attitude-behavior correlation was .64 while the intentionbehavior correlation was .83. The corresponding correlations in the long delay group (25 day or longer interval) were .71 and .81. The length of delay did not significantly alter the correlations. Understanding why some mothers breast-fed and yet others bottle-fed required analysis at the behavioral belief level. The two groups were found to differ significantly on all six behavioral beliefs about breast feeding and differed on three of the six beliefs about bottle feeding. This study suggested Fishbein's model worked well within the behavioral domain of inf ant feeding intentions and behaviors for British mothers.
Wittenbraker, Gibbs and Kahle (1983) conducted a longitudinal study of seat belt usage utilizing Fishbein and Ajzen theory and guidelines. The construct of habit was also tested within this context to assess behavior that is not wholly volitional but typically enacted. One hundred thirty-four subjects enrolled in an introductory psychology course responded to two identical questionnaires separated by a 1 month time interval. The questionnaire consisted of essentially the relevant Fishbein and Ajzen measures with the addition of a frequency measure regarding the specific behavior and a social desirability scale. The dependent variables consisted of the frequency of riding in or driving in a car.
The researchers analyzed the data using multiple regression and cross-legged panel correlation. The regression results supported the theory of reasoned action. Attitudes and subjective norms predicted intentions, and intentions predicted behaviors; habits, too, predicted behavior. The cross-legged results affirmed the cycle of influence from . . norm to intention to attitude and back to subjective norm, implyed that each of subJecnve an . ables plays a legitimate, important role in the use of seat belts. these v Originally, the Fishbein model (1967) included perceived moral values as a third measure of behavioral intention. This measure was eventually removed from the model by Ajzen and Fishbein (1970) because moral obligation was discovered to serve as an alternate measure of behavioral intention. Gorsuch and Ortberg (1983) hypothesized that a measure of perceived moral obligation would add significantly to the predictive power of attitude and social norms in moral situations but not in non-moral situations. Four hypothetical situations, two in which morality was relevant and two in which it was not, were used in the study. The four situations were: an erroneous tax refund; working on Sundays (unable to attend church); attending a party (unable to attend early worship next day); and, second church service (too tired to attend early worship). One hundred thirteen subjects from an adult Sunday school class of a Baptist church responded to statements developed to assess attitude, subjective norm, perceived moral obligation and intention. The moral obligation component significantly predicted behavioral intentions in the morally relevant situations while attitude was significant for predicting just one of the non-moral relevant situations.
The findings of this study suggested that a measure of personal moral obligation is a necessary component of the Fishbein Ajzen (1980)  Adhering to Fishbein and Ajzen (1980) format, Fredericks and Dosset (1983) 35 conducted a pilot study on 123 summer school students in psychology and determined modal beliefs and significant others in relation to the consequences of the specific behaviors: class attendance/absence. Modal beliefs are salient beliefs for a representative population. Eleven modal categories, seven modal consequences of attending class and four modal consequences of being absent were identified for the primary study questionnaire. The primary study used a different sample of nine, 5-week summer school classes. Two hundred thirty-six subjects completed the final questionnaire. The behavior, class attendance or absence was assessed by signature on class attendance sheets. Prior behavior was indexed by attendance/absence data collected for 2-weeks prior to the midpoint of the summer session. A semantic-differential questionnaire assessed attitudes, subjective norms and behavioral intention between weeks 3 and 4 of the summer session. Behavioral intention was assessed by two questions regarding intention to either attend or be absent from class during the final two weeks of the summer session. Data analysis was done using Liseral IV. The results supported Bentler-Speckan's (1983) hypothesis that there is a direct path from prior behavior to both intention and target behavior but Bentler and Speckan's hypothesis that there is a direct path from attitude to target behavior was not supported. This study is significant because its conclusions can be applied more broadly. The basic interrelationships between variables in the Fishbein and Ajzen formulation apply to industrial, governmental and service organizations where attendance behavior is of interest. Pagel and Davidson (1984)  . d r behavioral plan) the models were compared in terms of the criterion variables. atntu e o ear chers also tested the specificity hypothesis, which states that a predictor and Theres . · n are defined by action, target, context, time and there is correspondence between cnteno each pair of elements. To evaluate the predictive utility of attitudes, within-versus across-subjects prediction of contraceptive behavior was used. Each model was put into operation according to each theorist's specifications. Seventy female psychology students completed a questionnaire which assessed attitudes toward, beliefs about, and intentions to use either oral contraceptives, a diaphragm or natural methods. The questionnaire operationally defined the components of each model through scales that reflected the particular model's dimensions. Thus, scales were developed to assess the instrumentality and value components of the Rosenberg model, the Fishbein's model and the Beach model. The dependent variables were self-reported measures of attitudes and behavioral plans regarding the designated contraceptive behaviors. Pagel and Davidson analyzed their data using stepwise regression. All three models exhibited significant predictive utility for behavioral intention. Personal normative beliefs emerged as a strong independent predictor of behavioral plan. As hypothesized, the "within-subject" procedure yielded greater predictive accuracy than the '.'across-subject" procedure for the Fishbein and Beach models with no clear pattern of predictive superiority between them.
The results of the specificity hypothesis revealed that each model made its best linear predictions of attitude toward the behavior when correspondence between elements was demonstrated, however, as attitudes became less specific for behavior predictability decreased. Schifter and Ajzen (1985) examined the use of attitudinal and personality variables as predictors of success in attempted weight reduction. The study was based on the theory of planned behavior (Ajzen, 1985;Ajzen & Timko, 1986) which extends the theory of reasoned action to include non-motivational determinants of behavior (requisite 'ties and resources). As in the original model, the prime concern is the opportUill individual's intention to lose weight. This intention is said to be a function of three independent variables: attitude, subjective norm and perceived control over one's body . ht This theory through perceived control, takes into account some of the realistic we1g · ' constraints that exist regarding weight reduction. 37 Schifter and Ajzen (1985) hypothesized that: (1) intentions to lose weight can be predicted from attitudes, subjective norms and perceived control; (2) intentions and perceived control should permit prediction of actual weight loss, and (3) individuals who score high on variables related to actual control (perceived competence, ego strength, action control, health locus of control) are expected by others to be better able to carry out their intentions to lose weight. The major focus of this study was on the extent to which individuals made detailed weight reduction plans and examined a number of individual difference variables. Eighty-three college students expressed their attitudes, subjective norms, perceived control, and intentions with respect to losing weight. The study was conducted during two stages of the spring semester. Stage 1 assessed attitudes, subjective norm, intentions, and perceived control. Six weeks later, 76 of the same subjects responded to a Stage 2 questionnaire which assessed individual differences. In support of the theory of planned behavior, intentions to lose weight were accurately predicted on the basis of attitudes, subjective norms and perceived control; perceived control and intentions when taken together, predicted fairly successfully the amount of actual weight loss over the 6-week period and perceived control was the best of five predictor variables of actual weight loss. The individual difference variables: self knowledge, planning, and ego strength, revealed a significant correlation with weight loss. Hierarchical multiple regression analysis was performed to examine the total amount of variance in weight loss that could be explained by considering all factors that were found to have signficant relations with weight loss. Intention and perceived control (Variable I) and self-knowledge, planning and ego strength (Variable 2) alone did not ·gnificant contribution, yet their simultaneous effect was highly significant: F Jl)akeast (6,67) == 3.73, p < .01. Ajzen and Madden ( 1986) tested the theory of planned behavior in two 38 experiments with college students. Perceived behavioral control is specific to the theory of planned behavior. Underlying perceived behavioral control is a set of beliefs that deal with the presence or absence of requisite resources and opportunities. These beliefs are also presumed to determine intention and behavior.
Similar to Fredricks and Dossett's (1983) study on class attendance, Ajzen and Madden (1986) assessed attitudes, subjective norms, perceived behavioral control, and intentions in two experiments on class attendance and the goal of obtaining an "A" in a course. In the first experiment 169 undergraduates' attendance at class lectures was recorded over a 6-week timeframe. Attendance data was collected at 16 regular class sessions. The questionnaire, designed to obtain a measure of the constructs within the theory of planned behavior, was administered eight sessions into the semester. The number of sessions attended during the first and second eight-week periods served as measures of prior and later behavior. The perceived behavioral control measure was developed in a pilot study. Twenty-four college students listed factors which could prevent them from attending class, frequency of occurrence, and the degree to which they felt in control of attending all class sessions. Consistent with the theory of planned behavior, perceived behavioral control greatly improved the models predictive power (F = 46.16, p < .01) in comparison to attitude and subjective norm only.
In the second experiment, 90 undergraduates participated in an experiment where the behavioral goal was getting an A in a course. As with the first experiment, perceived behavioral control added significantly to the prediction of intentions, independent of attitude and subjective norm. Bagozzi (1986) presented an alternative expectancy-value model, derived from Fishbein and Ajzen ( 197 5) formulation, and compared the alternative model with the . 1 c nnulation found in the theory of reasoned action. Bagozzi focused upon uamnona 10 c ation specifically upon evaluations. He argues that neither a moral nor an attitude 1onn • .
valuation is sufficient to capture the motivational component of the original af{ecnvee m00el. Bagozzi then proposed a re-conceptualization based on subjective conditional h/ avoidance reactions where an individual appraises the consequences of his or approac 39 her actions. These appraisals are hypothesized to represent approach/avoidance reactions.
These approach/avoidance reactions are hypothesized to be the results of affective evaluations, moral evaluations, and other psychological processes.
Following Fishbein and Ajzen (1980) strategies, Bagozzi (1986) elicited 12 salient beliefs about the consequences of the act of donating blood from 40 respondents.
The final questionnaire assessed attitude, beliefs, evaluations, and subjective approach/avoidance responses. Two independent samples of 110 respondents were randomly selected and separated into an experimental group and a control group. The subjects were male students at the University of Saarland in Germany. The experimental manipulation for those donating blood consisted of viewing a series of slides of a person giving blood. Hierarchical regression was the method of data analysis chosen to test the hypothesis concerning the multiplicative combination of beliefs with either evaluations or approach/avoidance responses. It appeared that neither moral evaluations nor affective evaluations combine multiplicatively with beliefs to predict attitudes. Bagozzi's reconceptualization with approach/avoidance evaluations were found to combine with beliefs to account uniquely for variation in the attitude criteria (Bagozzi, 1986). Fishbein and Ajzen (1975) inferred that external variables such as personality traits influence behavior indirectly by affecting attitudes, social norms, or the weights placed upon these components. Miller and Grush (1986) examined certain dispositional variables that might moderate the relationship between attitudes-behavior and subjective norm-behavior. The study predicted that individuals who are both aware of their own attitudes and unconcerned with the opinions of others would display high attitude-. rrespondence. Also, it was hypothesized that individuals with other behavtorco . u· ns of these personality traits (high and low) were expected to display high combma o norm-behavior correspondence. The subjects studied were 226 students enrolled in an 40 introductory psychology course. The subjects completed a self-consciousness scale and a self-monitoring scale to assess individual differences in relation to the attitude-behavior, norm-behavior relationships. Following Ajzen and Fishbein (1980) guidelines, Miller and Grush distributed questionnaires to assess social norms, attitudes and behaviors with respect to spending time on school work. To create an overall behavior index, they developed a summed scale of eight academic activities in which students might have engaged in that semester. Hierarchical regression analysis supported both hypotheses.
The study clearly revealed the joint effects of dispositional variables to provide a more adequate account of the attitude-behavior relationship. Ajzen and Timko (1986) examined the relationship between health attitudes and behavior. These authors articulated the principle of correspondence, which posits that specific health behaviors are likely to correlate only with equally specific attitudes towards those behaviors. This same principle can also be applied to perceived control, where measures of perceived control over specific preventive health behaviors are expected by researchers to facilitate relatively accurate·prediction of corresponding actions. This study also considered other factors related to matters of health: perceived vulnerability to illness, perceived harmfulness of illness, concern about illness, and perceived effectiveness of recommended health practices.
In a pilot study, 40 undergraduate students responded to how frequently they performed specific health behaviors. Item analysis revealed 24 of the 53 common behaviors correlated .40 or greater with the total score; these behaviors were selected for funher examination. In the subsequent expanded study, 113 undergraduates completed a questionnaire that assessed attitudes and general beliefs about health and illness, and specific beliefs regarding the 24 health-related behaviors. The frequencies regarding how h behavior was perf onned generated single-act criteria and were used to often eac an aggregate measure of health behavior. The results demonstrate the constn1ct 41 importance of correspondence between measures of independent and dependent variables.
Attitudes and perceived control with respect to specific health-related actions correlated highly with corresponding behaviors, yet global attitudes toward generally recommended health practices and general beliefs regarding health locus-of-control were found to account for little variance in specific health behaviors. Perceived control over specific health related actions exhibited strong relations to self-reports of corresponding behaviors. This study also found that health behavior was predicted with greater accuracy from an affective than from an evaluative measure of attitude.
The construct, behavioral intention, of Ajzen and Fishbein's model (1980) continues to be a questionable variable. Behavioral intention is defined as a person's subjective probability that he or she will engage in the behavior of interest. Azjen and Fishbein (1980) posit that the correlation between behavior and behavioral intention should be high. This high correlation has been found by Pomazal and Jaccard's (1976) study of blood donation, Jaccard and Davidson's (1972) family planning work, Fishbein and Coombs (1974) study of voting, De Vries and Ajzen's (1971) cheating in college work, and Holman's (1956) study of attending football games.
However, Fishbein (1972) inferred several factors may weaken the obtained intention-behavior relationships (e.g., the specificity of the intentional measure, the length of time between the measurement of intention and the observed behavior, the degree to which carrying out the intention depends on others or the occurrence of Particular events, lack of or partial correspondence between the elements [action, target, context and time] of the theory's constructs). There appears to be evidence that the mediating role of behavioral intention may be artifactual. In a study of blood donation, Zuckerman and Reis (1978) found a direct contribution of attitudes to predicting blood donation behavior without the mediating role of behavior intention. Social norm's effect in predicting blood donating behavior was derived from the mediating role of intentions, . w·wdes retained a direct, non-mediated influence on behavior. Bentler and while a speckart ( 1981) in a study of three behaviors of college students found that attitude ed a significant direct effect on the behavior of studying. The other two behaviors exert · revealed intention predicted dating but not exercise. Manstead, Proffitt and Smart (1983) studied infant feeding behavior. More specifically, this study examined the Zuckerman -Reis hypothesis that the longer the interval between the expression of behavioral intent and behavior, the weaker the relationship between intent and behavior, and the stronger the relationship between attitude and behavior. No statistical difference was noted, yet the pattern was consistent with the hypothesis. Fredricks and Dossett (1983) compared Ajzen and Fishbein's model (1975) to Bentler and Speckart's (1981) expanded version of that model to include prior behavior and a direct path from attitude to behavior. The results did not suppon a direct path from attitude to the behavior of class attendance.
In summary, few studies suggest that the effect of attitudes on behavior is not completely mediated by behavioral intentions. These studies cast doubt on this construct of Ajzen and Fishbein's theory (1980). As Liska (1984) noted the independent effect of attitude is often substantial; sometimes stronger than the indirect effect mediated by behavioral intentions, and then sometimes even stronger than the independent effect of intentions. This study intends to address the predictive efficiency of the direct and indirect measures of attitude and subjective norm to behavorial intention and subsequent behavior. The mediating role of behavior intention will be assessed.
This section summarized the review of the literature for Ajzen and Fishbein's (1980)  Thirty coronary-bypass patients participated in a teaching program to evaluate its effect on long-term compliance to exercise. The teaching program consisted of an individualized exercise regimen, how to obtain target heart rate and to maintain an activity log. Six to eighteen months after surgery patients participated in a interview survey. The results of this study revealed a low compliance rate with the specifics of the heart-walk exercise regimen. Perception of barriers was the variable that showed the strongest relationship with walking compliance. The greater the number of perceived barriers the lower the compliance level. The perception of susceptibility with the walkregimen revealed an inverse relationship, those who perceived themselves as most susceptible were least compliant Specific barriers to compliance were weather and short bouts of acute illness. This study did not report statistical results, only global findings.
The use of the health belief model in Tirrell and Hart's (1980) study provided minimal information regarding the health beliefs of post-coronary -artery -bypass graft patients. Muench (1987) also used the health belief model to examine health beliefs regarding a cardiac rehabilitation program with bypass patients and post myocardial infarction patient's overtime. Seventy-two patients completed a questionnaire regarding health beliefs, perceived self-efficacy of exercise and other risk factors, and health . .
Patients were also asked to list, in an open format, the benefits and barriers 1110 uvanon.
di ac exercise program. The results of this study indicated subjects who perceived 0 fa car benefits of the exercise program had higher levels of general health motivation and selfefficacy (r = .55, p < .001) while subjects who perceived more benefits of exercise noted fewer barriers to attendance (r = .22, P = 0.31). Three benefits were reported regarding a cardiac rehabilitation program: improved stamina, medical supervision and regular participation in a scheduled exercise program. Three barriers were noted: early morning schedules, conflict with other activities, and transportation. The health belief model as a conceptual framework is useful but limited by its constructs to elicit health beliefs of a specific situation.
Other studies involving cardiac patients have focused on factors associated with participation and adherence to exercise. Oldridge and Jones (1983) examined the effect of a written agreement on compliance in a cardiac rehabilitation program. Patients were asked to sign an agreement to comply with the program for 6-months and to keep a diary of self-monitored heart rate, daily physical activity and weight changes and smoking habits. One hundred twenty subjects with coronary heart disease participated in the study. Of the 120 subjects, 51 % were classified as dropouts within 6-months. A 42% compliance rate resulted for the control group while a 54% compliance resulted for the experimental group. These compliance rates were not statistically significant (P > .10 < .20). However, 15 of 63 experimental subjectes refused to sign an agreement. The compliance rate for those who agreed to participate for the 6-months was 65% higher than it was in those experimental group subjects who did not sign the agreement (20%) (P < .005) and higher than the compliance rate in the control group (42%) (p < .01). Factors significantly associated with dropout in both groups were blue collar work, smoking at entry, inactive leisure habits and a younger age. The compliance rate for self-monitoring was 77% of 31 experimental subjects and 52% for daily physical logs. Mirotznik, Speed.ling, Stein and Bronz (1985) examined specific characteristics . al fitness beliefs and attitudes) of those who join and adhere to a cardiovascular (physic ' P .. 0 gram and those who do not. These authors conducted their study on 215 fimess • people who came to a Coronary Detection and Intervention Center for a coronary heart disease (CHO) risk assessment with the opportunity to enroll in an exercise fitness program· The non-joiners were those individuals (N = 154) who underwent the risk assessment and exercise electrocardiogram (EKG) only. The joiners (N = 61) were those individuals who enrolled in the fitness program. Both groups were compared in terms of: sociodemographic and physical characteristics, fitness, CHD risk factors, self-assessed health, and general health attitudes and behavior. Using both chi-square and t-test the following were noted. Joiners were older, more likely to be retired or working part-time and 41 % had graduate degrees. Significant differences were noted with regard to fitness variables, CHD risk assessment and health attitudes.
In the final analysis, three variables were able to differentiate joiners from nonjoiners. In order of explanatory importance, the three variables were: number of minutes on the exercise EKG protocol (.366, p .002), worry or concern about health (1.318, p .035), and the belief that improved health would lead to increased activity (1.475, p .045).
More specifically, joiners and non-joiners had different health attitudes. Joiners worried more about their health and viewed improved health as important in other areas of life.
Health beliefs were not found to affect adherence significantly.
Other investigators have focused attention on the relationship between attitude and exercise behavior. Godin and Shephard (1986b) compared Ajzen and Fishbein's (1980) concept of attitude with Kenyon's attitude inventory toward intentions to exercise.
Simple Pearson coefficients were computed between pairs of the measured variables.
Stepwise multiple regressions were performed, predicting the intention to exercise from, Its indicated intentions to exercise correlated greater with Ajzen and Fishbein's Tberesu attitude than with any of the other measurements. The cumulative R2 for attitude and . · e norm was 0.308. Attitude being the only variable to carry a significant beta subJecnv 47 weight (0.448, p < 0.001 ). Kenyon's inventory yielded a cumulative R 2 of 0.109, where the health subdomain carried the only significant beta weight (0.226, p < 0.05). The third multiple regression, combining both Ajzen and Fishbein and Kenyon's subdomains yielded a cumulative R2 of 0.308 with the attitude construct to carry the only significant beta weight (0.448, p < .001). The results of this study support the notion that attitudes defined specifically according to Ajzen and Fishbein are more strongly related to intentions than general attitudes toward an object as defined by Kenyon.
In yet another study, Godin, Desharnais, Joben and Cook (1987) investigated the effect of two persuasive techniques to modify: (1) intention to exercise with and without knowledge of the results of a fitness test and health appraisal, and (2) intention and behavior to exercise over3-months. Ajzen and Fishbein's methods (1980) were used to determine intention to exercise. Subjects were randomly assigned to one of four experimental conditions either with or without knowledge of test results: (1) physical fitness test, (2) appraisal of health, (3) fitness testing and health age combined, and (4) no treatment-control.
Of the 250 subjects, the final sample consisted of 140 subjects. An analysis of variance revealed the following: to complete one or both tests had no significant immediate effect (without knowledge of results) on the intentions to exercise during the next 3-months (F = 4.01, P < .01). With knowledge of results, the intention of the fitness test and fitness test plus health appraisal groups differed significantly from the control group. However, 3-months later, no significant difference in intentions remained between the groups (F = 0.70).
Investigators continue to search for factors relevant to an individual's decision to initiate and maintain regular physical exercise. Dzewaltowski (1989) compared . , ocial cognitive theory and Ajzen and Fishbein's theory of reasoned action sanduras s 0) redict exercise behavior. Both theories conceptualize determinants of (198 to P . but differ with respect to the variables and their causal relationship with each behaVlOr other. A questionnaire consisting of both theories constructs and an exercise log was 48 completed by 328 undergraduate students in a physical education skills class. The results of this study supported the theory of reasoned action where direct attitude and subjective norm correlated significantly with indirect attitude and subjective norm (.584 and .677) respectively. Contrary to the theory, a weak but significant path was noted between indirect attitude and direct subjective norm (.123). Only direct attitude was a statistically significant predictor of behavioral intention with a path coefficient of .508. The theory of reasoned action explain 5% of the exercise behavior variance. Two social cognitive theory variables, self-efficacy and self-evaluated dissatisfaction, significantly predicted exercise behavior (23.8, 7.91) respective beta weights. Social cognitive theory variables accounted for 10.8% of the variance in exercise behavior. Also a multiplicative function of self-evaluated dissatisfaction and outcome expectations increased the amount of predicted exercise behavior variance to 16%. Finally, hierarchical regression analyses determined that the theory of reasoned action did not account for any unique variance in exercise behavior that the social cognitive theory variables did not explain, R 2 change = .006, F change < 1.
Many studies have applied attitude-behavior models to examine exercise behavior. Godin and Shepard (1990) provide an extensive overview of the main attitudebehavior models that have been used to analyze exercise behavior. Among the models discussed, the health belief model (HBM) was reviewed in terms of existing empirical data. More specifically, since the HBM is concerned with perceptions of illness, it has been applied to study exercise behavior in patients with ischemic heart disease. Of three such studies (Holm et al., 1985;Muench, 1987;Tirrell & Hart, 1980) the results were .
and contradictory (Godin & Shepard, 1990). This existing data provided no arob1guous . dication that the HBM is appropriate for the study of exercise behavior. c1ear10 The protection motivation theory of Rogers is similar to the HBM with the focal point as motivation to protect oneself. The model has limited usefulness when studying exercise behavior (Godin et al., 1983. However, the addition of the self-efficacy concept to this model and the HBM has revealed positive findings. Bandura's social cognitive theory has been successfully applied to exercise behavior (Sallis, Hovell,,Hofstetter,,Faucher, & Elder 1989). The perceived ability to participate and to exercise regularly in a supervised exercise program appears to be the variable of prime importance (Godin & Shephard, 1990). Ajzen and Fishbein's (1980) theory of reasoned action has also proved to be a very successful model toward the understanding of the decision-making process that underlies exercise behavior. Of 22 studies using Ajzen and Fishbein's model reviewed by Godin & Shephard (1990), 13 reported a substantial (30%) amount of variance in intentions to exercise explained by the persons attitudes. The subjective norm construct is less consistently associated with intention toward exercise. The effect of external variables and post behavior may influence intention independent of the theory's constructs (Valois, Desharnais, Godin, 1988).
Triandis's theory of interpersonal behavior has been successfully applied to some health-related behaviors. This model supplements the theory of reasoned action with the concepts of role beliefs and person normative beliefs. Studies using this model have found the effect of past behavior and the affective attitude dimension to be important determinants of intention and exercise behavior.
Finally, Ajzen and Madden (1986) added the construct of perceived behavioral control to the theory of reasoned action and extended the theory which is entitled theory of planned behavior. Few studies have investigated the theory of planned behavior toward health-related behaviors.
The HBM and protection motivation theory say health-related behaviors are ood in terms of their specific potential to protect against disease or to optimize underst health, while all other attitude-behavior theories analyze behavior in terms of social ·ons The variables, expectations of self-efficacy, attitude toward exercising, dimeDSl • · ed barriers and past behavior all exert strong influences upon intention and perce1v • behavior (Godin & Shephard, 1990).
Dzwaltowski, Noble and Shaw (1990)  In summary, social cognitive theory and the theories of reasoned action and planned behavior, are among the most useful frameworks to study exercise behavior.
Attitude-behavior models serve to understand the decision-making process that underlies and precedes an action (Godin & Shephard, 1990). Of these theories, an individual's Predispositions are of central concern. In an attempt to uncover the post myocardial infarction patients' predispositions toward a cardiac rehabilitation exercise program, the theory of reasoned action was applied because of its ability to qualitatively uncover . bel"efs regarding a cardiac rehabilitation program and to link these beliefs to panent 1 exercise behavior.

Nursine Research: General
The theory of reasoned action has been applied or cited in nursing research since appro:ximately 1981. The model's utility to patient and nurse situations has been demonstrated. This section of the literature review will first review general nursing research employing and/or testing Ajzen and Fishbein's model (1980). The second section will focus on specific cardiac nursing research which utilized this model over a period of twenty years. Also, two studies that address entrance into a cardiac rehabilitation program will be reviewed.
The theory of reasoned action was tested in an examination of nurses' charting behavior. Schmidt's (1981) study tested the utility of the Fishbein model in nursing. The study also examined a specific intervention designed to change nurses' attitudes toward charting behavior. The quality assurance framework of the American Nurses Association specifies that designated criteria need to be identified for each patient population to be audited. The patient population for this study was myocardial infarction patients. The study's treatment which was designed to create a positive attitude consisted of nurses participation in the development and revision of patient outcome criteria. The subjects were 16 nurses employed on a post coronary care unit of a private hospital. A pre-post experimental design with a single treatment effect was used to assess attitudes, personal and social norms and motivation to comply. Schmidt conducted a retrospective chart audit, during the pre-post treatment Phase, to determine nurses' charting behavior. Three semantic differential scales were used to measure attitudes toward charting as well as personal and social norms with regard to charting. Two Likert-type scales measured motivation to comply with personal and social norms. The study's negative findings may be attributed to a variety of factors with emphasis on the lack of Fishbein and Azjen . to first determine beliefs, attitudes, subjective norms and intentions toward strategies charting behavior. Specifically, instrument validity for post-test data had a Cronbach's 52 alpha reliability of 0.00 which may have accounted for the lack of support of Fishbein's model. The study's results failed to report a strong relationship between subjects' attitudes and their behavior. The first hypothesis which stated the behavior prediction coefficient for pretest charting of outcome criteria will not differ significantly from zero was not rejected. Motivation-social and nursing service guidelines correlated significantly with the audit score for pretest data (r = .62, p <.01). In the pretest audit scores, 45% of the variance was accounted for by ideal nurses' notes (personal norm}, nursing service guidelines (social norm), and nursing notes (attitude). The second hypothesis dealt with the behavior prediction coefficient for posttest charting of outcome criteria and revealed no significant correlation among the three variables and the audit score. The third hypothesis resulted in a one-way chi-square (x2 = 2.22, df = 1, p = .13) which indicated no significant difference using the total number of criteria charted for all subjects prior to and following treatment. No significant differences were found in attitude scores of the pre and posttest Phases. Davidhizar (1982) utilized Fishbein's expectancy-value model (Fishbein, 1963, Fishbein & Ajzen, 1975 to develop a tool for profiling -the attitude of schizophrenic clients toward their medication. Fislibein's expectancy-value model modifies the theory of reasoned action by considering only three major concepts: belief, evaluation and attitude. Davidhizar carefully considered the mechanisms of the model to elicit the desired data from the schizophrenic patient population. Ajzen and Fishbein (1980)  A major finding in this study was that the attitude-adjustment relationship was positive (R = .52, p < .001) and the relationship was much stronger for the mothers (R = .67, p < .001) than for the fathers (R = .31, p = .49). Parental attitude scores were not found to be influenced by seizure control, perception of seizure control nor the length of time the child had epilepsy (r = .16 to .18). A one-way ANOV A was used to examine gender of parent, gender of child, and seizure control differences. Parental attitudes were found to be significantly lower (F = 7.56, p < .01) when the child was female (X = -.82) rather than male (X = 3.28). This study also compared the open-ended belief format with the fixed-belief format. Two attitude scores were obtained from each parent. Ajzen and Fishbein (1980) suggest the fixed-belief format for identification of belief strength can be a substitute for comparing a wide range of beliefs to constancy of beliefs by all subjects elicited in the open-ended format. During structured interviews parents were asked to h th ey believe to be true about epilepsy in their child. To assess belief strength state w at po nded with how sure they were (0% to 100%) that the belief was associated parents res with the attitude object. A positive correlation between attitude scores was found (r = 60 < .001). The range for the open-ended attitude score ranged from -18.0to+12.2, . ,p while the fixed-belief attitude score ranged from -14.8 to +23.4. This finding suggests measuring belief strength on a 0% -100% scale is an alternative to the open-ended approach. Shamansky, Schilling and Holbrook (1985) utilized the theory of reasoned action to detennine factors associated with the behavioral intention to use nurse practitioner services among health care consumers. Six hundred names were drawn from the New Haven and Guilford telephone directories. A letter was mailed to each potential subject one week prior to the telephone interview to explain the study. An "innovation subscale", measuring respondent's reactions to an innovative service, nurse practitioner care, was chosen from a 71-item instrument, adopted from Smith and Schamansky (1983).
Spearmen-Brown internal consistency for this innovation subscale yielded a .54 reliability coefficient. The subscale measured: five items related to consumer behavior on a Liken scale 1 (definitely would not) to 10 (definitely would buy); five items related to the value of selected nurse practitioner functions; 5 items assessed cost of nurse practitioner services; and other items assessed demographic and psychographic characteristics of respondents. Three hundred thirty-one respondents provided the following results: intent to use nurse practitioner services was not statistically associated with a summary measure of innovative products (OR= 1.13; p = .74); the cost factor of nurse practitioner services revealed: if nurse practitioner was covered by insurance, ?9.3% said the would seek their care, if not covered by insurance, 63.8% would decline nurse practitioner care, if nurse practitioner care cost the same as a physician, 41.1 % would seek their care and 44.9% would decline it, if nurse practitioner care cost more than a physician, 74.8% would decline care, and if nurse practitioner care cost less than a . . ?0 2% would seek care. Demographic and psychographic factors that physician. · ed intent to use nurse practitioner services when covered by insurance which influenc 56 · u·cally significant: age (OR= 1.8, p = .2), satisfaction with present health care were stans (OR= 2 , p = .04) and home size (OR= 11.0, .01 < p < .001). This study used the theory of reasoned action to a limited degree.
Chang, Uman, Linn, Ware and Kane (1985) examined selected components of nurse practitioner's care to determine which contributed most to the intent to adhere to a health care plan. To accomplish this end, 268 women aged 56 to 89 years at 26 senior citizen nutrition sites in a metropolitan area were asked to participate in this study.
Respondents were asked to view a videotape which portrayed a nurse-patient encounter.
Respondents were asked to answer a set of questions as if they were the patient in the videotape. Each videotape consisted of a hypothetical situation in which a patient with stable chronic angina was cared for by a nurse practitioner and came to a return office visit Eight videotapes with varying levels of technical, psychosocial and patient participation were produced. Among a five-member panel, 100% agreement was achieved for the high low level of each component depicted in the eight tapes.
Intent to adhere to health care plan was defined as the extent to which the respondents felt they would perform various behaviors agreed upon in the videotape. A five-item intent to adhere scale consisted of items such as the following: "If you were the patient in the videotape, would you cut down on your potatoes and bread as discussed with the nurse practitioner?" Responses ranged from definitely would to definitely would not on a five-option scale. Four items modified from the Patient Satisfaction questionnaire (Ware, Snyder & Wright, 1976) were used to determine the subject's attitudes regarding health care. A five-point scale ranging from strongly agree to strongly disagree was used.
Analysis of variance indicated that high psychosocial care (tapes #1,#2,#3) resulted in significantly greater intent to adhere. An analysis of covariance found the .
h cteristic variables (covariates) to be significantly correlated with the subJect c ara .
adhere to a hypothetical health care plan (F = 6.22, p < .001). The covariates intennon to .
ed ( widowed marital status, Jewish religion, high pre-existing satisfaction with select health care, high importance of a physical exam and high social network) were selected baSed on significant correlations with one or more of the dependent variables.
The researchers did enumerate various limitations of the study (volunteer sample hypothetical situation, previous experience with angina, etc.). However, the study did demonstrate that personal characteristics are influential in stimulating intent to adhere to the care plan. Fishbein and Ajzen (1975) model was used to study contraceptive behavior of college-age males. Ewald and Roberts (1985) study was a partial replication of a study done by Fisher (1978). Fisher developed a measure derived from Fishbein's model  Pender and Pender ( 1986)  .233, P < .01) accounting for 5.5% of the variance in intention. The correlation coefficient for subjective norm with intention to exercise was, (r = .263, p < .01) while the attitude measure and intention was, (r = .127, p < .01). When the variables of weight of the individual was regressed on subjective norm and attitude, the variance explained in intention to exercise more than doubled (13%) (R = .364, p < .01). Only attitude was associated with both intention to eat a diet conducive to attaining/maintaining ded weight (r = .127, p < .01) and intention to manage stress by avoiding recoIDIIlen , ful S ituations r = .271, p < .01. "The multiple regression coefficient for bighlY stress ' . d ight and perceived health status was .428, p < .001 with 18% of the variance atutu e, we n ·on to control weight explained" (Pender & Pender,p. 17). The results of this in inten study suggest attitudes exert an effect on intentions to engage in the three health promoting behaviors examined. However, it was found that subjective norm did exert more influence on intention to regularly exercise than did the attitude measure. Hence, the determinants of behavior are crucial to understand and to promote health behavior. Bowles (1986) used Fishbein and Ajzen's (1975) theory of reasoned action to construct and validate a semantic differential measure to determine womens' attitudes toward menopause. Conceptually, the theory of reasoned action provided a sound rationale to construct a measure of attitude toward menopause. Bowles indicated, "A woman learns or forms a number of beliefs about the menopausal experience. The woman's attitude toward this experience is determined by her beliefs that menopause is associated with other events, feelings, and symptoms, as well as her evaluation of them" three groups and each group completed one of the following measures to divided mto . h srruct validity: Attitudes Toward Menopause Scale to establish convergent establis con validity (Neugarten, Wood, Kraines & Loomis, 1963;Kogan, 1961), Attitudes Toward Old People Scale to establish discriminant validity, and the Attitude Toward Women scale (Spence, Helmreich & Stapp, 1973) and the MAS as a second measure of discriminant validity. When the scores for each scale were correlated with the MAS, pearson product moment correlation coefficient revealed the following correlations respectively:( r = .63; r = .42 ) and (r = -.04) . Test-retest reliability for the MAS was .87 after a 6-week time interval. Multiple regression analysis revealed age and menopausal status as two explanatory variables for the variance of the MAS scores. The use of the MAS for attitude assessment provides an understanding of "other variables" related to menopause. It also provides an understanding of menopause and its relationship to other aspects of a women's mid-life experiences. Prestholdt, Lane and Mathews (1987) used the theory of reasoned action (1980) to build a mcxiel of nurse turnover. The theory provided these researchers with an understanding of complex decision-making processes. "Specifically, the process is revealed as a hierarchical sequence leading from beliefs, through attitudes and social norms, to intention, and finally, to behavior" (Prestholdt et al.' 1987, p. 221).
Specifically, the researchers tested the theory of reasoned action and compared its predictive effectiveness with two modifications to the theory . The first modification involved the addition of moral obligation as a predictor variable for intention. The second modificaton involved the use of a differential measure (the difference between remaining and resigning) as opposed to the usual single behavioral alternative. In a pilot study, 109 nurses provided salient beliefs and referents. The final questionnaire administered to 885 nurses in 21 Louisiana hospitals consisted of a measure of attitude, subjective norm, moral obligation and behavioral intention. A differential measure was constructed for each of the above components studied. Behavioral intention was measured for two behaviors: "remaining on the staff of this hospital" and "resigning from ·ta1" To assess behavior, 6 months later the hospital provided the employment this bosp1 · f a ch nurse. Thirteen predictor variables (five components of the theory , moral status o e obligations and seven demographic variables) were entered into a hierarchical multiple regression analysis with employment status as the dependent variable. Differential intention was the only significant predictor for employment status. Differential attitude (B = .72, < .001), differential subjective norm (B = .10, p < .01), and differential moral obligation (B = .10, p < .01) were the only significant predictors of differential intention.
The regression analyses were repeated using model measures that related only to resigning behavior to determine the second modification to the theory . The expanded model consistently demonstrated the advantage of using measures that relate to both behavioral alternatives. This research suggests the utility of Ajzen and Fishbein's (1980) theory of reasoned action to nurse turnover. Knowledge of the underlying beliefs, attitudes, intentions of nurse turnover will aid those concerned with recruitment/retention efforts. Savage, Cullen, Kirchhoff, Pugh and Foreman (1987) utilized the theory of reasoned action to examine the extent to which attitude, subjective nomis and behavioral intentions influence a nurse's decision to comply with a "do not resuscitate order." The theory applies to this situation since nurses involved in the care of such a baby must come to tenns with their beliefs about the situation and explore their personal attitudes about the do not resuscitate decision. Three registered nurses in 10 perinatal centers in one midwestern state participated in the study. Subjects were asked the prevalence of a "do not resuscitate" policy in their neonatal intensive care unit and to complete a demographic data form. Also, using Ajzen and Fishbein (1980) format, nurses' attitudes, subjective norms, and behavioral intentions toward complying with a do not resuscitate order in four hypothetical situations were elicited. The four hypothetical situations described inf ants .
xpected outcomes of recovery and potential capabilities. Test-retest with vanous e . m· ver a two-week time interval was (r = .765).
reliab lY 0 The results of this study indicated that nurses' intention to resuscitate an inf ant . "do not resuscitate" order, depended on a description of the infant's physical despite a condition and a perceived unfavorable prognosis. Attitudes and subjective norm were significantly correlated with intention to resuscitate, p < .001 with correlation coefficient ranging from .58 to .93. Multiple regression analysis revealed subjective norm (B = .41 to .82) exerted more powerful influence on nurses' decisions not to resuscitate than attitudes (B = .17 to .39).

This study also depicted beliefs about caring for infants with various disabilities.
This data indicates attitudes and subjective norms, as determinants of intention, impact professional behaviors of the neonatal intensive care nurse. Lierman, Young, Kasprzyk and Benoliel (1990) cited several advantages of the theory of reasoned action: (a) the relationships of the model components are clearly specified by mathematical formulations, (b) there are specific guidelines for operationalizing the major constructs, ( c) instrument development is grounded in the target population, and (d) a social normative component is included. These researchers used the theory of reasoned action to predict intention to perform breast self-examination and breast self-examination behavior~ It is common nursing practice to encourage and teach self-breast examination to patients. Effective teaching must be based upon an understanding of the factors which influence this behavior hence, the theory of reasoned action provides a methodology to determine these factors.
The pilot study was done on 29 women from women's organizations of 12 local churches. Ajzen and Fishbein (1980) strategies were utilized for the pilot measure. Also included were current breast self-exam practices; beliefs about cancer, and cancer treatment; and people with whom they had discussed breast self-exam.
The revised instrument, referred to as, the Beliefs and Attitudes Questionnaire, investigations. Behaviors ranged from male contraception to nurse turnover. Subjects for two of the studies were nurses while all others were either patients or lay public. All of the studies either applied one or more constructs of the theory of reasoned action or utilized the theory as a conceptual framework. One study specifically used the theory of reasoned action as a conceptual framework to develop and validate a semantic differential instrument to measure attitude toward menopause. Five studies utilized some degree of Ajzen and Fishbein's strategies for elicitation of beliefs and construction of the final questionnaire. In general, the theory of reasoned action appears to possess utility for nursing practice which is designed to understand and in some instances alter human behavior.
In conclusion, Ajzen and Fishbein's (1980) model appears to be a useful model for understanding human behavior. This model provides the researcher with a theoretical fralilework to predict human behavior when operationalized in a research study.
The theory of reasoned action was used in the general nursing research literature for the following purposes: to determine its usefulness to nursing; to derive attitudes toward selected health behaviors; to elicit parental attitude and adjustment to a chronic illness; to determine behavioral intention to use selected health services; to determine intent to adhere to a health regimen; to identify attitude, subjective norm and behavioral intention toward selected health behavior; conceptual considerations of the theory of reasoned action to develop an instrument to measure attitude toward a life event; to test and expand the theory of reasoned action; to determine how the constructs of the theory of reasoned action influence nurse behavior; for prediction of selected health behaviors.
The focus of the general nursing research literature involving the theory of reasoned action centers upon the ability of the theory to impart an understanding of the determinants of human behavior, specifically health-related behaviors. Application of the theory of reasoned action in this study enabled the researcher to discover whether the direct or indirect measures of attitude and subjective norm contributed more weight in predicting behavioral intention and subsequent adherence behavior. Critical underlying beliefs were derived and may prove useful in attempting to change this health-related behavior.

Nursim: Research: Cardiac
The nursing research literature involving specifically the cardiac patient population, is composed largely of a series of investigations by the same researchers. Spearmen's rank order correlation was used to assess attitude and adherence behavior relationships indicative of predictive validity. Attitudes and behavior were found to correlate significantly with subscales of diet (r = .31, p <.05); activity (r = .46, p < .01) and smoking (r = .62, p < .001), but not for medication (r = .12) or stress (r = .15).
To provide a measure of accuracy of patient responses, a "significant other" also completed a similar version of the Health Behavior Scale. Patient-family correlations revealed significant relationships for diet, activity, medication and smoking. Factor analysis of the data revealed that eight factors accounted for the majority of the variance.
Each of the five behaviors of the medical regimen defined a factor using a varimax rotation. The sixth factor was termed the "burden factor" which loaded significantly on items: difficult versus easy and aggravating versus soothing. Subjects rated the overall regimen with these terms rather than specific behaviors. A seventh factor termed "fairness" loaded significantly on fairness versus unfair adjective pair for overall regimen prescriptions. The eighth factor was termed "desirability" based on the paired adjectives harmful versus helpful, unnecessary vs helpful, again for overall regimen prescriptions.
The study established the reliability of the Miller Attitude Scale for assessing attitudes of cardiac patients toward five behaviors of the medical regimen. Alpha reliabilities for both groups indicated a high degree of internal inconsistency of scale items for all five subscales. This investigation indicated that assessment of attitudes may provide valuable infonnation for patient compliance. Miller, Wikoff, McMahon, Garrett and Ringel (1985) utilized Fishbein's model 0980 ) to examine the relationships between attitudes, perceived beliefs of others, and . s toward medical regimen adherence and actual behavior for myocardial intennon infarction patients. These variables were assessed at three points in time: during hospitalization, and at 6 and 9 months after hospitalization. The subjects were 112 myocardial infarction patients from five institutions that provided cardiac rehabilitation programs. Prior to discharge from the hospital, patients completed the Miller Attitude Scale and the Health Intention Scale. Medical and demographic data forms were also completed to assess relationships.

68
Six to nine months after hospitalization patients completed the Miller Attitude Scale, the Health Behavior Scale and medical and demographic data forms. The Health Behavior Scale-Family was completed by a family member to verify the patient's adherence to the medical regimen. The Health Behavior Scale is an adaptation of the Health Intention Scale in which intention statements are changed to behavior statements.
Multiple regression analysis was used to determine prediction of patients ' adherence behaviors. The results indicated that intentions during hospitalization are related to attitudes and perceptions of the beliefs of others concerning one's intentions to obey the five prescriptions. The perception of significant other's beliefs contributed most to prediction of intentions for each of the prescriptions. Adherence to the medical regimen 6 to 9 months after hospitalization as related to one's attitudes and intentions during hospitalization was significant for diet and activities (r -.33 with diet and r = .26 with activities). Attitudes and intentions related to smoking, medications and stress during hospitalization were not significantly related to adherence after hospitalization.
Both attitudes and perceived beliefs of significant others after hospitalization were significantly related to patients' statements of adherence for all prescriptions except for medications at 6 to 9 months after hospitalization. Intentions during hospitalization and a t 6 to 9 months after hospitalization were not statistically related to adherence P hic variables as hypothesized. demogra The findings of this study indicate that the point in time to provide information 69 and discuss lifestyle adjustments to promote medical regimen adherence is after hospitalization. Based on the findings attitudes and perceived belief of others at 6 -9 months post hospitalization were indicators of regimen adherence. Therefore, home visit was compared to hospital discharge data. Second, the 30 day data was used for problem identification related to medical regimen adherence and societal adjustment.
Third, the nurse developed a health plan which included patient goals that addressed f atn .tudes medical regimen adherence, coping methods, and societal problems o ' .
t in different life situations. The sixty days after hospitalization visit provided adJustmen rtu nity to measure the effect of the nursing intervention on medical regimen an oppo adherence and societal adjustments.
Analysis of variance revealed no significant differences for the first hypothesis: medical regimen adherence and societal adjustment will increase more in patients in the nursing intervention group than in the control group at 30 or 60 days post hospitalization.
The second hypothesis: attitudes, intentions, perceived beliefs of others and coping methods as predictors of societal adjustment and regimen adherence was tested using stepwise multiple regression. Perceived beliefs of others were found to be a predictor of medical regimen adherence for smoking, activity, stress, and medications at the 30 day visit. The addition of diet to this list occurred at the 60 day visit. Attitudes were predictive of medical regimen adherence for diet, smoking, and medication at the 30 day visit and diet, smoking, stress, and medication at the 60 day visit.
Intentions during hospitalization were predictive of medical regimen adherence to smoking, and activity at the 30 day visit and diet and stress only at the 60 day visit.
Helpfulness of coping methods was predictive of medical regimen adherence for stress modification prescription (R2 .64) (F 18.82) at the 60 day visit. Attitude toward activity prescription was the only variable predictive of societal adjustment at either the 30 day or 60 day visit (R 2 .32, R 2 .50; F 16.04, F 27 .69) respectively for the 30 and 60 day visit.
The findings of this study are in direct contrast to the predictions of the Fishbein (1980) model. These researchers concluded that a lack of sensitivity of the instruments to measure medical regimen adherence and societal adjustment may have accounted for no differences between groups. Also, the high score of the control group may have resulted from the home visit by the nurse for data collection or merely by their own selfevaluation after completing the scales at 30 and 60 days. However, the study did reveal unponant relationships concerning cardiac rehabilitation: perceived beliefs of others was S t predictor of medical regimen adherence for all prescriptions; attitudes and the stronge f 1n S s of coping methods were predictive of adherence for stress modifications at help u e the 60 day visit; and for societal adjustment, only attitude toward physical activity regimen was predictive. These findings support the utility of Fishbein's model for prediction of exercise behavior in the myocardial infarction patient population.

72
In conclusion, Miller has used Fishbein's model (1980) as a theoretical basis for their studies of myocardial infarction patients and their adherence to the medical regimen over the past ten years. Miller (1988) re-emphasized the findings of their work and pointed out that demographic characteristics and personality traits have no direct effect on behavior and are related to behavior only if they influence the beliefs that shape its attitudinal or normative determinants. Miller also indicated their findings, over the ten years, suggested intention to behave in a specific manner changes over time, which emphasizes the need for concurrent measurement of these variables.
The suitability for using Ajzen and Fishbein's (1980)  . for patient elicitation of beliefs or what this patient population is thinking is oppartUJUlY afforded. Hiatt, Hoenshell-Nelson and Zimmerman (1990) similarly investigated the factors influencing patient entrance into a cardiac rehabilitation program. This study was a descriptive pilot study that utilized the Patient Entrance into Cardiac Rehabilitation Program (PECRP) questionnaire proposed by Hijeck to assess factors that distinguish subjects who participate or choose not to participate in a cardiac rehabilitation program.
Thirty-nine patients with a cardiac problem completed the questionnaire. Demographic variables, marital status and income, had significant differences on the perceived benefits/barriers construct of the health belief model. Chi-square and t-test statistics revealed no significant differences in any of the demographic variables in those who entered a cardiac rehabilitation program and those who did not.
Both of the above mentioned studies concluded a systematic method to assess patients health beliefs would be helpful in predicting patients' decisions whether to enter a cardiac rehabilitation program or not. Specific nursing strategies could then be devised and implemented to reinforce accurate beliefs and dispell misconceptions. However, their methods to determine beliefs among this heterogenous patient population remains questionable.
In summary, the nursing research performed with the cardiac patient population indicated the usefulness of Ajzen and Fishbein's (1980) Miller at al.,(1982aMiller at al.,( , 1982bMiller at al.,( , 1983Miller at al.,( , 1988 and McMahon et al.(1986) tested the relationship between the various constructs of the theory of reasoned action, yet the direct and indirect measures were not derived according to Ajzen and Fishbein's (l980) guidelines. Therefore, the validity of their results should be interpreted with caution.
More specifically, Miller and colleagues (1982a)

development of the Miller
Attitude Scale and the analysis of the data, were not consistent with Ajzen and Fishbein's fior construction of a questionnaire. Miller et al., (1982aMiller et al., ( , 1982b made reference format 10 Fishbein and Ajzen's (1975) guidelines for instrument development, however, lack of adherence to these guidelines may have resulted in the lack of support for the theory of reasoned action in both of Miller et al. ( 1982aMiller et al. ( , 1982b studies. Miller and colleagues (l983, 1985) examined the attitude-adherence relationship again with the five actions of the medical regimen. Significant findings supported the theory of reasoned action, yet revealed insufficient use of the theory of reasoned action. McMahon et al. ( 1986) examination of behavioral intention and medical regimen adherence over time for four life situations and Miller et al.(1988) study of the influence of a nursing intervention on medical regimen adherence and societal adjustment was based on Ajzen and Fishbein's theoretical model, however, neither study used Ajzen and Fishbein's methodology. Miller's (1988) findings are in direct contrast to the Ajzen and Fishbein model. The author concluded that the lack of sensitivity of the instruments to measure medical regimen adherence and societal adjustment may account for the lack of difference between the experimental and control groups. On the other hand, conclusions that can be drawn from the work of Miller et al. ( 1982aMiller et al. ( , 1982bMiller et al. ( , 1983Miller et al. ( , 1985Miller et al. ( , 1988 and McMahon et al. (1986) with regard to the theory of reasoned action were that the various studies did not adhere to Fishbein and Ajzen (1975) Ajzen and Fishbein (1980)  Nonetheless, these empirical works did reveal some significant nursing implications for example: (1) the Miller Attitude Scale may be useful planning individual rehabilitation programs, (2) that identification of unfavorable attitudes during hospitalization may be effectively managed with situational support, and (3) a measure of attitude may provide one indicator of adherence. Therefore, the current study was a unique undertaking because it was grounded in the cardiac patients beliefs. Unlike Miller et al (1982aMiller et al ( , 1982bMiller et al ( , 1983Miller et al ( , 1985Miller et al ( , 1988 and McMahon et al. (1986) this study tested the predictive relationships between the direct and indirect measures of the Ajzen and Fishbein model (1980) rather than simply employing Ajzen and Fishbein's model as a conceptual framework. Unlike the works of Hijeck (1984) and Hiatt, Hoenshell-Nelson and Zimmerman (1990)  To fulfill Ajzen and Fishbein's (1980) criteria for development of questions to . alient beliefs the behavioral elements of the theory were specified as: the eliett s ' behavioral criterion (action) was exercise participation with respect to a specific target, the post myocardial infarction patient, in the context of a cardiac rehabilitation exercise gram at 4 to 6 weeks after hospitalization. To ensure the correct belief was elicited, pro ' Ajzen and Fishbein recommended correspondence between the four behavioral elements.
Changing one of the four elements defining the behavior can elicit a completely different set of salient beliefs (Ajzen & Fishbein, 1980). For example, to elicit beliefs underlying a patient's attitude toward a cardiac rehabilitation exercise program the following questions were worded according to Ajzen and Fishbein's criteria of correspondence for the four elements of the Phase I questionnaire: 1.
What do you believe are the advantages of undertaking a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?

2.
What do you believe are the disadvantages of undertaking a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?
3. Do you have any other beliefs regarding a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?
Salient referents were also identified by responding to the following open-ended questions regarding enrollment in an exercise program post myocardial infarction.

4.
What individual or group of people would approve of your participation in a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?

5.
What individual or group of people would disapprove of your participation in a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?

6.
Which people whom you know other than the ones just listed would be interested in your taking part or not taking part in a cardiac rehabilitation exercise program 4-6 weeks after a heart attack?
The actual questionnaire for elicitation of beliefs may be found in Appendix B.

Elicitation of Beliefs
The sample (N=50) completed the Phase I open-ended questionnaire. Forty . ts were male while the other ten patients were female. The subjects age ranged pan en from 2 6 to 68 years with a mean age of 55 years. The mean educational level was 14 80 years. The employment status of the sample revealed 40% professionals, 40% white/blue collar status and 20% were not employed. Forty-nine subjects were caucasian and one subject was black. The majority of the sample was married (80%) while the other 20% were either widowed, divorced, or single. All patients ( frequency of the response noted as suggested by Ajzen and Fishbein (1980) (see Appendix C). "Once the beliefs were listed, a decision was required to ascertain the number and kind of belief to be included in the salient set. The first step was to organize the responses by grouping together the beliefs that referred to similar outcomes and counting the frequency with which each outcome in a group was elicited" ( p.68 ). Table   1 shows a list of beliefs concerning the safety of exercise after a heart attack. The beliefs were grouped by identifying the similiar outcomes each belief represented. In this case, each belief in Table 1 inferred that exercise in a cardiac rehabilitation program is safe after a heart attack. These beliefs were analyzed and a decision was made by the researcher that as a group, the beliefs represented a single belief with a similiar outcome.
For example, the belief statement, "Help me to realize that exercise is healthy, not dangerous" was then identified as a salient belief. Ajzen and Fishbein say, "Common sense is required since no clear rules can be provided" (p.69 ) to clarify this issue of grouping similiar outcomes or treating them as separate outcomes. However, one solution is to return to the original inventory and assess whether the same individual emitted similar beliefs. These eight beliefs were obtained from responses one, two, and three of the open-ended questionnaire. Table 1 also lists the belief number as found in Appendix C and the frequency of response. Table 1 depicts this grouping process.
It can be seen that five of the beliefs referred to a similar outcome, namely that exercise could potentially be dangerous after a heart attack. Despite the fact that each outcome by itself was mentioned by only a few patients, when grouped together they suggested a salient belief in the patient population concerning the safety of exercise in a cardiac rehabilitation program. The salient belief then was phrased such that the belief ded to the attitude toward the behavior. Each belief statement was reviewed and grouped by the researcher. To assess the extent of agreement among observers, inter-judge reliability was used. One reviewer was a cardiac rehabilitation nurse while the other reviewer was a lay person. No.

Table2
Identification of Salient Beliefs

Cardiac-Rehabilitation Exercise Program Frequency
Program helps to increase and strengthen heart 14 Program helps to build confidence 17

No disadvantages 27
To get in shape 14 Stress from fear that exercise can cause another heart attack 9 Program prepares you for your regular routine 17 Something could happen if your body is not ready 26 Be dangerous so soon after your heart has been damaged 13 Eight salient beliefs were retained for the final questionnaire Then each belief statement was re-phrased for conciseness and clarity of meaning. Belief statements were Phrased at a 6th grade reading level. The 8 salient beliefs retained and rephrased were: l. Help increase and strengthen my heart muscle.
2 . Increase confidence regarding physical condition.
3 . Cardiac rehabilitation program will have no bad effects.
4. Get me back in good physical shape.
6. Prepare me for my regular daily routine.
7. Help me to realize that exercise is not dangerous.
8. Be dangerous so soon after heart attack. Prospective subjects were identified by the nurse manager of the stepdown/telemetry unit or by a nurse liaison in each of the eight participating hospitals.
Again, eligibility criteria was given to each participating hospital nurse liaison and recruitment efforts were made by this individual. Each subject was informed of the nature of the study, given a brief verbal overview of the study and, if they agreed to Participate, informed consent was obtained by the nurse liaison. The Phase II questionnaire was administered by the nurse liaison to patients during their hospital stay on a stepdown or telemetry unit. Each patient was given ample time to answer the questionnaire (approximately 20 minutes to a half hour).
Prior to the start of the study, nurses acting as liaisons were given an overview of d 's purpose Directions for administering the questionnaire, specifically how to the stu Y • treat potential problems involving subject interpretation of items was presented. Nurse . . 's were instructed to have patients answer the questionnaire without the assistance liatSOn of others and have patients notify the nurse liaison or the researcher if questions arose.
Each of the eight participating hospital nurse liaison's were given twenty-five questionnaires to distribute, to total 200 post myocardial infarction patient questionnaires.
Ample time was given to each hospital nurse liaison to distribute questionnaires.
Frequent communication with each nurse manager or nurse liaison revealed some difficulty in data collection. Various problems arose which prevented questionnaires from being returned (i.e., nurse failure to return to patient to collect questionnaire; patient discharge; busy nursing units, etc.). Since the number of questionnaires was less than the desired 200, data collection was supplemented by having patients complete 50 additional questionnaires in one large hospital. The Phase II questionnaires were collected from each participating hospital at the end of a three month time period.
Two hundred twenty-five eligible patients were asked to participate. Data from the Phase II questionnaire was collected during the subjects' hospitalization. Five patients declined to participate, and gave no reason for their unwillingness to participate. Twenty patients who agreed to participate in the study were discharged from the hospital and failed to return the questionnaire. Data generated from five subjects was incomplete thus, excluded from the final sample. In addition, the data from one outlier was deleted. The final sample for this study consisted of 194 post myocardial infarction patients.
Completed demographic data were obtained from 132 patients (Appendix F). The descriptive data obtained indicated that the majority 71 % of these patients were male.
The mean age for this portion of the sample was 61 years with a range of 35 years to 80 Years of age. The majority (70%) of the patients were married. Ninety-three percent of the patients indicated that they had no problems with transportation to the respective . h bilitation program, and for 96% of the sample the time of the program agreed cardiac re a . d 31 ·1y schedule. Table 3 provides the summary statistics for the demographic with their data.  For attitude, the direct measure is given in Section III of the questionnaire in Appendix D.
Measurement of jndjrect subjective norm. As previously discussed in Phase I, three salient referents were elicited. The strength of the normative belief concerning the referent was assessed using a bipolar scale with anchor items "I should" and "I should not" These anchor items represent normative expectations of significant others. To fully understand subjective norm, the motivation to comply with the referent must also be assessed. This measure was obtained using a unipolar scale ranging from "not at all" to h" to capture the individual's general motivation to comply with the referent. ''vert muc . s are then multiplied together to comprise the indirect subjective norm score Tbese1tem be found in Sections IV and V of the questionnaire in Appendix D. and can ,Gmeral subjective norm assessment. As with attitude, Ajzen and Fishbein 89 (l 9 SO) specify a general component of subjective norm or the direct measure of subjective norm. This measure was obtained to evaluate an individuals general perception of what most important others think "should" or "should not" be done in reference to the behavior. This measure can be found in Section VI of the questionnaire in Appendix D.
Jbe a priori beliefs. Ajzen and Fishbein (1980) Table 4 and are given in Section VID of the questionnaire in dix D Thus the resultant beliefs for Phase II were derived from the elicited Appen . , bel .efs of the open-ended questionnaire and the a priori belief statements. salient 1 Table 4 A PRIORI BELIEF STATEMENTS 90 My participation for the next twelve weeks in an cardiac rehabilitation exercise program would: 1. 2. 3.

7.
Not be necessary because I am in good physical condition.
Probably bring on an early death.
Be unlikely since I don't have the energy or physical stamina to exercise.
Not be necessary because I would rather exercise by myself than participate in a group exercise program.
Be unlikely since I need to take it easy from now on.
Not be necessary because I may not be able to return to work after my heart attack.
Not be necessary since I am not the person I was before my heart attack.

Ak'sures of Indirect Attitude
The indirect measures of attitude were obtained by multiplying the strength of the beliefs about exercise participation by the evaluation of the consequences of exercise participation behavior and then the summed products provided a score for the eight beliefs. Each item score could range from -9 to 9. The potential sum could range for all eight beliefs from -72 to 72.

Direct Attitude Measure
In contrast, the direct attitude score was obtained by the sum of the responses to three 7 point semantic differential bipolar adjective pairs scored on a scale from -3 to 3.
This potential sum of the three adjective pairs could range from -9 to 9 (Sections I, II and m of the questionnaire in Appendix D depict the direct and indirect attitude measures).

Indirect Subjective Norm Measure
Scores for normative beliefs were obtained from responses on a 7 point bipolar scale. The motivation to comply was obtained from responses on a unipolar scale. To obtain the indirect measure of subjective norm, the three normative beliefs were multiplied by the corresponding motivation to comply and then summed as with the indirect attitude measure to form the indirect subjective norm score. Each item score could range from -3 to 3. This potential sum score could range from -27 to 27.

General Subjective Norm Assmment
The direct subjective norm was obtained from a single item that captured the patients perception of what people important to the patient think should be done in regard to the behavior. The score for this item could ranged from -3 to 3 (Sections IV, V and VI of the questionnaire in Appendix D depict the direct and indirect subjective norm measures).

J]IC A pcjocj BeUefs
The a priori beliefs were obtained from responses on a 7 point bipolar and 7 point . differential scale, respectively. The multiplicative products were summed as semannc with the indirect attitude and subjective norm component. Each item score could range form _ 9 to 9. This sum score could potientially range from -63 to 63 (Section vm of the questionnaire in Appendix D) .

.Bfbayjoral Intention
Behavioral intention was assessed, as previously mentioned, using a single item, "Do you plan to enroll in a cardiac rehabilitation exercise program in 4 to 6 weeks?". This measure was assessed using a bipolar scale with anchor items likely to unlikely (Section VII of the questionnaire in Appendix D). Have no bad effects. Get me back in good physical shape. Produce stress from fear that exercise can cause another heart attack. Prepare me for my regular daily routine. Help me to realize that exercise under supervision is healthy, not dangerous.

BES
Be dan~erous so soon after a heart attack.

BE9
Not be necessary because I am in good physical condition. BElO Probably bring on an early death. BEl l Be impossible since I don't have the energy or physical stamina to exercise. BE12 Not be necessary because I would rather exercise by myself than participate in a group exercise program 93 BE13 Be unlikely since I am unable to exercise due to my heart attack and need to take it easy. BE14 Not be necessary because I may not be able to return to work after my heart attack. BE15 Not be necessary since I am not the whole person I was before my heart In ~neral. most of the peQPle or wups that are important to me think.
I intend to exercise in a cardiac rehabilitation program in four to six weeks after my heart attack.

Results
JliS:riptjye Fiodjn~s of Beliefs Table 5 presents the means and standard deviations for the 15 belief statements, direCt attitude, subjective norm, general subjective norm and behavioral intention. The descriptive data was interpreted to assess if the item analyses supported the conceptual responses. In Table 5 it can be seen that beliefs, BEl, BE2, and BE7 which dealt with the physical status of the patients received positive endorsement by the subjects. Beliefs, BE3, BE4 and BE6 which dealt with resuming a healthy lifestyle also received positive endorsement by the subjects. These belief statements reflected a positive outcome of exercise in a cardiac rehabilitation program. However, beliefs, BES and BES reflected stress and danger associated with a cardiac rehabilitation exercise program. It can be seen that in general patients negatively endorsed these beliefs (i.e., they did not believe that exercise would cause another heart attack or would be dangerous coming soon after the original heart attack).
The a priori belief statements 9-15 were comprised of a group of beliefs that connoted disinclination to an exercise program. However, in this sample, patients negatively endorsed the a priori beliefs, that is they negated belief items that suggested a negative outcome (e.g .. they responded that it would be unlikely that they would not exercise because of a lack of physical stamina).
The direct attitude measures were positively endorsed by the patients (Table 5).
Patients regarded exercise in a cardiac rehabilitation program to be good, wise, and pleasant with pleasant the least favored of all responses. No subject negatively endorsed the good-bad outcome of a cardiac rehabilitation exercise program.
The subjective norm indirect measures were also positively endorsed by patients, as were the direct, general subjective norm measure. Patients regarded family, physician, and friends, respectively, as influential in deciding to exercise in a cardiac rehabilitation program. The physician subjective norm received no negative endorsements by subjects.
The behavioral intention item to enroll in a cardiac rehabilitation exercise program was also positively endorsed by patients. In general, patients appraised the positively phrased belief items favorably and denied the negatively phrased belief items.
Subjective norm items were favorably endorsed and patients indicated that they intended to enroll in a cardiac exercise program. However, subjects failed to complete all items of the a priori belief statements. The N for Table 5 was 194 with the exception of the a priori beliefs, where the N ranged from 17 4 to 194 .   (Ajzen & Fishbein, 1980 , p. 222-223). As seen in the evaluation measures, EVl, EV2 and EV7 reflect positive endorsement by the subjects, (i.e. increasing and strengthening the heart muscle, increasing confidence, and awareness that exercise is healthy not dangerous are viewed as positive outcomes). Belief evaluation, EV3, EV4 and EV6 were also positively endorsed by the subjects and viewed as positive outcomes.. On the other hand, the generally negative belief evaluations, (EV5, EV8) which reflected harm as an outcome and the a priori statements, (EVll-15), stimulated a wide range of responses. Notably, scores in five of the a priori items (EV 11-15) yielded distributions that were not unimodal suggesting less concensus. The motivation to comply component contributes to the patients willingness to eoJDPlY with the referents wishes. The data suggested patients were motivated to comply with their family (MCl), physician (MC2), friends (MC3), respectively. The N for Table 6 ranged from 121 to 181 for the a priori beliefs.
The 15 beliefs were multiplied by their respective evaluative components. The same procedure was done with normative beliefs and motivation to comply. Products represented indirect measures of attitude and subjective norm. Table 7 presents the JDeaDS, standard deviations and range of scores for the individual items that comprise the indirect attitude and subjective norm indices. 3.4 4.9 -9 9 4 5.6 3.3 -9 9 JDdirect measures of attjtude. In this study all indirect attitudes, with the exception ofJNDATfl, ranged from -9 to 9. Indirect measure of subjective norm. The indirect subjective norm index which encompassed the motivation to comply component multiplied by the referent, revealed the family, physician, and friend, respectively, to be positively endorsed. The physician referent received the highest mean score (Table 7).

Principal Component Analysis
Two sets of principal component analyses (PCA) were done to examine the factoral structure of items for, (1) belief components and (2) indirect attitude components.
Belief components. Given the Phase II data, a principal component analysis (Stevens, 1986;Dunteman, 1989) was conducted to: (1) determine if the sets of beliefs and indirect attitudes were unidimensional or not, (2) to delete items which did not contribute to any well-identified component, and (3) to identify the dimensions measured within the belief index and indirect attitude index. Item and reliability analyses assisted _,.,;ning which items best related to the identified dimensions. in dete•.u ...
An initial PCA utilizing the SPSSX program and varimax rotation was performed 00 the 15 beliefs. The decision rule for item retention included a factor loading of at least .40 with this loading being at least .10 or greater than loadings on any other component (Guadagnoli & Velicer, 1988 • threC beliefs perceived a cardiac rehabilitation exercise program to be associated with .. rdial fear or the potentially hannful effects of exercise. However, statements were myoc .... · pmased as negations of these fears, so were scored in a positive direction. Therefore, this was believed to represent a confidence component.

DISINCUNATION
Not be necessary because I am in good physical condition. Be impossible since I don't have the energy or physical stamina to exercise. Not be necessary because I would rather exercise by myself than participate in a group exercise program. Be unlikely since I am unable to exercise due to my heart attack and need to take it easy. Not be necessary because I may not be able to return to work after my heart attack. Not be necessary since I am not the whole person I was before my heart attack. BECONFI BES BES BElO = = = CONFIDENCE Produce stress from fear that exercise can cause another heart attack. Be dangerous so soon after a heart attack. Probably bring on an early death. lDdirect attitude comoopeots. Table 9

CONFIDENCE
Have no bad effects. Produce stress from fear that exercise can cause another heart attack. Be dangerous so soon after a heart attack.

Jlescriotiye Summarv of Scales
The indirect attitude set along with subjective norm measures were selected to test the study hypotheses. Indirect attitude components were chosen because: ( 1)  Cronbach's coefficient alpha, (Anastasi, 1988;Nunnally, 1978) was used to evaluate the internal consistency of the scales. It can be seen from  Range -9.00 to 9.00 -9.00 to 9.00 -9.00 to 9.00 -9.00 to 9.00 -3.00 to 3.00 -3.00 to 3.00 -3.00 to 3.00 Cronbach Alpha .92 .87 .47 .98 .67 The direct attitude scale (A TI) consisted of three items. It can be seen from Patients were entered into respective rehabilitation programs at various times based on the occurance of their infarction. Each cardiac rehabilitation nurse assessed exercise participation by attendance (number of days absent) and program completion.
The researcher, at a specified time, obtained all data from the respective cardiac rehabilitation program nurse to complete Phase ID of this study.

.De5ilD
A predictive design was used in Phase ill. Two hypotheses based on the theory of reasoned action were developed to study intention to enroll and participation behavior in a cardiac rehabilitation exercise program. The level of significance was set at p < .05 to iest the study hypotheses.
Hypathesis I: Hypathesis II: Attitude and subjective norm measured by the direct and indirect indices will significantly predict the behavioral intention to enroll in a cardiac rehabilitation exercise program.
Attitude and subjective norm measured by the direct and indirect indices will significantly predict participation behavior in a cardiac rehabilitation exercise program.  Table 11 presents the mean, standard deviation, range and number of subjects for: (1) behavioral intention to enroll in a cardiac rehabilitation program and (2) the number of sessions attended during the program. One hundred and eighty five patients completed the behavioral intention question. Of the 185 patients, 165 responded positively that they intended to enroll in a cardiac rehabilitation exercise program four to six weeks after their heart attack, five responded "neither" (0) and 15 said it was "unlikely" (-1,-2 or -3) they ll One hundred and fifty nine patients initiated the program and 157 patients would enro .  (Table 12). The three indirect attitude scales as a set significantly predicted BI (F = 5.76, p < .001) (See lower portion of Table 12). Of these three scales, INDRECUP had the largest beta of .32. Beta weights for INDISIN and JNDCONFI indicate that they contributed little to the prediction of behavioral intention. ormative beliefs as a set (SBN) improved this prediction significantly (F = The three n < ()()()). The two indirect components explained 31 % of the variance in behavioral 15.4, p .
. The first study hypothesis was accepted in the case of the direct and indirect intennon.
attitude and subjective norm indices. .000 Hierarchical regression analysis was employed also to ascertain to what extent attitude and general subjective norm ~xplained attendance in a cardiac rehabilitation program. Exercise participation was measured by the number of sessions attended in the cardiac rehabilitation program. Table 13 presents the Beta, R squared and F value for the independent variables predicting exercise participation in a cardiac rehabilitation program. Again based on prior exercise research , the first entered variable or variable set was the direct attitude variable (A TT). Direct attitude was not a significant predictor for exercise (Beta = .03). Direct attitude predicted less than 1 % of exercise participation behavior.
General subjective norm (GSN) was entered on the second step. It failed to J>l'Oduce a significant prediction (Beta = -.11 ). Taken together, the direct attitude and general subjective norm indices were able to account for just over 1 % of the variance in exercise participation.
Hierarchical regression was also employed to determine to what extent INDISIN, JNDREcuP, INDCONFI, and SBN explained exercise participation (Table 13).
The three indirect attitude scales were entered first into the regression equation as a set, but did not significantly predict attendance at exercise sessions (Beta= -.10, .00, -.09 respectively, for the three indirect scales). The three indirect attitude set of variables predicted only 2% of exercise participation behavior.
The three normative beliefs as a set (SBN) did not improve this prediction (Beta = .02 ). The indirect attitude and indirect subjective norm indices together were able to account for 2% of the variance in exercise behavior. The second study hypothesis was rejected in the case of both direct and indirect attitude and subjective norm indices  Ajzen and Fishbein (1980) patients identified eight similar beliefs in regard to physical, psychological and social concerns across the four periods of data collection in Phase I. The findings of Phase I of this study indicated that of the eight salient beliefs elicited, six reflected the positive aspects of resuming physical activity following a heart attack. One belief suggested that psychological security was a crucial element, and two of the eight beliefs reflected fear that exercise could induce a reoccurrence of a cardiac event. Data generated in regard to subjective norm showed that . across all four periods identified family, physician, and friends, respectively as panents their roost salient referents.

Ebasell
The Phase II questionnaire was developed to measure: (1)  participation. Items such as, "I don't have the energy or physical stamina to exercise" or "I am unable to exercise due to my heart attack, I need to take it easy" inferred a sense of disinclination.
The second component can be considered a recovery component. It consisted of items such as "Exercise in a cardiac rehabilitation program will help to increase and strengthen my heart muscle"; "Get me back in good physical shape". This component supports the concept of recuperation.
The essence of the third component confidence, consisted of a myocardial fear dimension. Two of the items loading on this component involved myocardial fear. Items such as "A cardiac rehabilitation program produces stress from fear that exercise can cause another heart attack"; "Be dangerous soon after a heart attack" mainly comprised this component. The third item that loaded on this component "Cardiac rehabilitation program will have no bad effects" was inversely related in meaning to the other two . This item received the highest loading and inferred a sense of confidence. There was a moderate relationship between the direct, general subjective norm variable and only the indirect attitude scale, INDRECUP (r = .26) and indirect subjective norm and only the INDRECUP scale (r = .34). These moderate coefficients suggested somewhat of an association between attitudes toward recuperation and the desire to listen to important others. The patients, as a whole, intended to enroll in exercise participation di ac rehabilitation program. Later they attended most sessions. Prior exercise in a car 115 research has investigated the theory of reasoned action to predict intentions and exercise participation in a group of participants. However, few empirical studies have tapped the · ns of a belief inventory to better understand the process post myocardial dimenStO infarction patients go through to come to a decision regarding enrollment and exercise participation. Within nursing research, only Miller et al. (1982) have developed a Health Attitude Scale designed to assess cardiac patients attitude toward performing recol1llllendated behaviors. The findings of this study provide the nurse with areas of concern (e.g. recovery, psychological security, and fear) that may be pertinent to assessment of patients following a myocardial infarction.

pbaseW
The purpose of Phase III of this study tested the ability of the developed scales to predict the behavioral intention to enroll in a cardiac exercise program , as well as actual exercise participation. The first hypothesis tested: attitude and subjective norm measured by the direct and indirect indices will significantly predict the behavioral intention to enroll in a cardiac rehabilitation exerciseprogram was supported. The acceptance of the first hypothesis highlights the fact that in a situation of crisis, both attitude and social influence contribute to the decision making process regarding intention to enroll in an exercise program after a myocardial infarction.
In a review of studies that applied the theory of reasoned action to the study of exercise behavior, Godin and Shephard (1990) reported that 13 of 22 studies found 30% of the variance in intentions to exercise explained by attitude (p. 110), and that subjective norm was less consistently associated with intention to exercise, But as the theory postulates (Dzewaltowski et al. 1990, p 339) has noted "that for some behaviors, attitudes will dominate the formation of intentions whereas with other behaviors subjective norm will be a dominant influence." Other researchers that have applied the theory of reasoned .
the prediction of exercise intentions and participation have generally found that acuon to mbination of attitude and subjective norm significantly predicted intention to a linear co . and that attitude contributed more to the prediction of intention than did exeretse subjective norm (Dzewaltowski et al., 1990;Dzewaltowski, 1989;Godin & Shepard, 198 6b;Riddle, 1980). Similarily, Godin and Shephard (1987), Godin, Cox and Shephard, 1983;Pender and Pender, (1986) and Riddle, (1980) have also found subjective norm to be statistically significant but to a lesser degree than attitude. Valois et al. (1988) attributed these findings to the fact that it is the individuals' decision to exercise or not. However, in this sample of post myocardial patients, the decision was influenced by the situation, where the patient were vunerable and compliant. They exhibited compliant behavior regarding the intention to enroll, as well as, exercise participation.
The present study supported the literature in that both the direct and indirect attitude and subjective norm measures were significant predictors of behavioral intention to enroll in a cardiac rehabilitation exercise program, however, in contrast to the findings above, subjective norm contributed more to the prediction of intention than attitude. One explanation for this difference in the intention to enroll in a cardiac rehabilitation exercise program, may have been a socially desirable response for this sample of post myocardial infarction patients, who were very positive in their intentions to enroll. Also, in this situation of crisis, patients exhibited a docile attitude that may have predisposed them to be influenced by family, physician and friends. Like, Dzewaltowski (1989), and Dz.ewaltowski, Noble & Shaw (1990) attitude was a statistically significant predictor of behavioral intention in this study. In contrast with other authors, in this study and Di.ewaltowski's work, both general subjective norm and indirect subjective norm, seemed to play a more significant role in the prediction of behavioral intention to enroll in a cardiac rehabilitation exercise program. Similarly, Daltroy and Godin (1989) found subjective norm to be a statistically significant predictor regarding spouses' intentions to e cardiac patients participation in exercise. Other studies have shown that encourag 117 . behavior is socially influenced (Carron, Widmeyer, & Brawley (1988),; Taylor, exercise Bandura. Ewart, Miller & DeBusk ( 1985) whereas a modest number of non-health related st0dies have found subjective norm to mediate behavioral intention more strongly than attitude (Bentler & Speckart, 1981;Kantala et al., 1982;Pagel & Davidson, 1984).
From a different perspective, Daltroy and Godin (1989), found only subjective norm and not attitude to significantly predict intention to encourage spouse participation in a cardiac exercise class. In Daltroy and Godin's study (1989) all other variables (attitude, etc.) added only 2.3% to the total variance explained in intention. Although, few studies have examined the spouse's influence on physical activity, Heinzelman and Bagley, (1970) Nye and Paulsen, (1974) Andrew, Oldridge, Parker, Cunningham, Rechnitzer, Jones, Buck, Kavanagh, Shepard, and Sutton (1981) and Dishman, Sallis, and Orenstein (1985) found the spouse's attitude to be a positive influencing factor for coronary patients to attend and adhere to an exercise program. The present study suggested that strategies to recruit patients into cardiac rehabilitation programs, during the hospital course, be directed toward the underlying beliefs elicited in this study, regarding exercise, as well as, the social role of the family, physician, and friends to encourage lifestyle changes regarding exercise. Consistent with past exercise research, (Dzewatowski, 1989;Godin and Shepard, 1986b;and Wurtele & Maddux, 1987) the present study supported Ajzen & Fishbein's (1980) theory of reasoned action, where direct attitude and subjective norm taken together accounted for 43% of the variance in behavioral intention and also, indirect attitude and subjective norm accounted for 30% of the variance in behavioral intention to enroll in a cardiac exercise program. Godin and Shephard (1990) state the relationship between subjective norm and intention depends upon idiosyncracies within subsegments of the population under study.
In this regard, Ajzen (1985) has maintained that attitude and subjective norm have different effects in the model, depending on the situation under study.
The second hypothesis which tested: attitude and subjective norm measured by . t and indirect indices will significantly predict exercise participation in a cardiac tbedirec rehabilitation program was rejected. This finding seemed to result from a condition in which patients, in a situation of crisis, tried to comply with a medical regimen and the wishes of others. A proposed rational for this finding centers on the group of patients stodied. In this study, the post myocardial infarction patients were a homogenous group of patients whose mean scores on behavioral intention and exercise participation displayed a highly skewed distribution. However, while there was sufficient variance in behavioral intention to test the model, that was not the case for the behavior variable, exercise participation. As a group, the patients intended to enroll and indeed, participated in the exercise program. Although the data suggests that less than 1 % of the variance was explained by behavioral intention to exercise behavior, direct examination of the frequency distribution indicated that 165 of 185 patients indicated a positive intention to enroll. These findings suggest that lack of variance in the criterion variable, exercise participation behavior, may have obscured the actual findings.
Numerous studies, within exercise research, have investigated both the antecedent-intention and intention-behavior relationship Godin & Shephard, 1985;Pender & Pender, 1986;Valois et al 1988) where the intention variance for attitude and subjective norm ranged from R2 =.25 to .54. In the present study, a low inverse relationship was found between behavioral intention and attendance in a cardiac rehabilitation program (r = -.09). In the case of this study, the lack of variance influenced the findings in regard to exercise participation behavior.
In contrast, this study found both direct and indirect measures of attitude and social influence to predict behavioral intention. Godin and Shephard's (1990) review of attitude-behavior models in exercise promotion, indicated that the findings of this study are consistent with the exercise research literature regarding the linear combination of attitude and subjective norm predicting intention. Unlike previous exercise research, . . nonn significantly improved the intention prediction. The social support subJecnve required during this specific patient situation is unquestionable as evidenced in the Moreover the physician's prescription to enroll in a cardiac rehabilitation findings. ' 119 program is more likely to occur with positive spousal support and friendly advice as the findings of this study indicated. This result illustrated the power of the social environment, as well as, patient attitude, to shape exercise behavior post myocardial infarction. When efforts are made to recommend exercise in a program, attention should be focused, not only on the patient, but also on those individuals with whom the prospective participant relates to most directly, such as family, physician and friends.
The attitudes of the three referents and the way they are perceived, are significant antecedents to the decision-making process, whether to enroll in a cardiac rehabilitation program or not and to partake in exercise or not.

Other Situational Factors
Specific situational factors related to the intention of post myocardial infarction patients participating in a cardiac rehabilitation exercise program were identified during Phase II data collection. Situational factors are defined as variables that are external to the theory that can influence intentions, hence behavior is influenced independently.
The personal opinion of the researcher is reflected in the following observations.
Post myocardial infarction patients represented a unique group of patients with various idiosyncracies. More specifically, the personality profile of a myocardial infarction patient was a classic finding. Grounded in the patient interviews in Phase I, post myocardial infarction patients characteristically possessed a sense of time urgency, were compulsive about home and work activities, were driven individuals with impatience as a compelling force and they often performed several activities simultaneously. This fastidious profile may be a determinant to behavioral intent. The stress level of these individuals was moderate to high. Stress produces physiologic responses that result from tal · li· Both these crucial factors have been well-researched in the . nmen snmu .
cnvifO Jiteratul'C and contributed to the findings that patients did what they were told. Post 120 di al infarction patients appeared to be a group of individuals who operate from an 01 yocar "all or none" principle. This was manifested in their exercise participation behavior.
This sample of patients were a captive group of individuals who experienced a cardiac event that had the potential to be life-threatening. Numerous lifestyle changes were mandated by both physicians and nurses requiring permanent adherence. This stiuation renders the post myocardial infarction patient vulnerable. Godin and Shephard (1990) indicate that a useful feature of the theory of reasoned action is its ability to identify details of the cognitive profile underlying exercise decision-making in specific populations. The situational factors external to the theory that may have influenced the decision to enroll and participate in the exercise program were: the patients response to the nurse recruiters efforts to enroll them in cardiac rehabilitation programs, the staff nurses educational series of lifestyle changes, family and friends advice and the fear of another myocardial infarction or even death. These variables may have contributed to the significant subjective norm findings. Exercise in a cardiac rehabilitation program is a health behavior under volutional control. In light of this situation and the vulnerable state of the patient, the behavioral intention to enroll in a cardiac rehabilitation program and participate in exercise is not an unexpected finding. However, the inability of attitude and subjective norm to predict exercise participation, in light of the high enrollment and attendance, was attributed to the non-normal distribution of the data. In general, the idiosyncracies and situational factors delineated may explain the patient responses, intentions, and compliant behavior which resulted in Phase ill.

MethoJ02icaI Issues
In general, patients found the Phase II questionnaire too lengthy. This factor contributed to many questionnaires left uncompleted, lost or simply not returned.
aior problem with the Phase II questionnaire was the level of difficulty in AJlotherm ;i coropleting the items. Although a simple explanation was provided to all patients, S comments were made regarding the difficulty in answering the various sections numerou of the questionnaire. In particular, patients were confused by the repetiviness of the .
specifically the evaluation items. Many patients reported they had already 1te1J1S, answered the items in the previous (belief) section. It may be that evaluation items should have been re-worded to reflect the evaluation of the outcome. For example, the belief statement, "Exercise in a cardiac rehabilitation program will increase and strengthen my heart muscle" is easily answered with the likely-unlikely bipolar scale.
The corresponding evaluation statement should have read, "Increasing and strengthening my heart muscle in a cardiac rehabilitation program would be "good" ... "bad". This clarification may have resulted in less response problems. Also, the a priori belief section posed many patient problems and a significant number of patients did not respond to the items. Of interest, in the a priori beliefs (Section VIII Appendix D) patients were asked to complete the belief and evaluation of belief items together and were not separated out as in the other sections of the questionnaire (Sections I and II). Patients could not comprehend how to answer these items and left them unanswered. In general, the patients experienced difficulty not with the belief statements but with the evaluation of the outcome (Section II Appendix D)." Young, Lierman, Powell-Cope, Kasprzyk, and Benoliel (1991) discussed a number of potential threats to reliability related to the theory of planned behavior.
Similarly, the threats apply to the theory of reasoned action. Four threats were examined.
The first reliability issue focused on the attitude scores when a specific belief is not endorsed by a participant. For example, most subjects agree with the belief, "Exercise in a cardiac rehabilitation program will help my heart to extract oxygen more efficiently." The problem arises when a respondent does not believe or disagrees with this statement and is asked to evaluate it. The respondent is unsure how to answer this question.
The second issue related again to the evaluation section of the questionnaire. The ·terns of good-bad assumes the respondent holds an absolute opinion regarding anchor 1 that belief without considering the circumstances occurring at that time. The particular life circumstances or the context in which events occur, greatly impact the way items are ered The third reliability issue pertains to the similarity of the belief and evaluation answ · items. Respondents, as previously mentioned in this study, perceived the items to be the same, potentially answering the items the same without giving any further thought to the items.
Fourth, social desirability was a threat that cannot be underestimated in the health arena. The theory of reasoned action is unable to capture this source of bias. Young et al., (1991) state, "This bias can be amplified when participants are recruited through health care providers" (p. 143). Nurses recruited patients in this study and exercise post myocardial infarction is a health behavior.
Despite the methodological issues raised, the theory of reasoned action possesses utility for predicting health behaviors. The theory of reasoned action continues to be examined to predict selected health behaviors, e.g., breast self-examination (Lierman et al., 1990) and condom use as an AIDS risk behavior among black women (Jemmott & Jemmott, 1991). The theory of planned behavior provides a new dimension to explore when predicting health behaviors, perceived behavioral control. The advantage of additional variables to the theory's predictive validity cannot be underestimated.
This study explored phenomena in the domain of client (Kim, 1983).
Specifically, the post myocardial infarction patient population. Explaining human behavior is a difficult task. The theory of reasoned action was an initial attempt and further refined by the theory of planned behavior to explain, understand and predict human behavior. If we knew why patients thought the way they did, nurses would possess a unique body of knowledge no other discipline could explain. It is through sroall efforts such as this study, that one large step is taken in the client domain toward understanding, explaining and predicting happenings in clients .caoctnsions The principle conclusions drawn from the results of this study of 194 post dial infarction patients are: (1) post myocardial infarction patients possessed JllyOCllf ble beliefs about a cardiac rehabilitation exercise program; (2) three salient favora referents were identified who influenced the post myocardial infarction patient in the 124 deeision-making process; (3) the indirect measures of attitude revealed three components labeled: disinclination, recuperation.and confidence; and (4) there exists support for applying the theory of reasoned action to predict intention to enroll and participation behavior in a cardiac rehabilitation exercise program.
The significant contribution of the subjective norm construct's prediction to the behavioral intention to enroll in a cardiac rehabilitation exercise program was demonstrated in this study. This predictive ability provides a better understanding of the beliefs and social influence that effect behavior during situations of threat in the myocardial infarction population.

Implications for Nursim: Practice
The purpose of Phase I was to elicit beliefs regarding exercise in a cardiac rehabilitation program. The significance of the Phase I findings lies in understanding the patients beliefs regarding exercise after a myocardial infarction and using these findings to facilitate recovery. The importance of the three salient referents suggested that strategies and interventions be designed to target these significant others.
Once beliefs about a particular behavior and the most important social influences are identified, nursing practice can be directed toward the determinants of the specific behavior. Although knowledge of a patient's intention to enroll in a cardiac rehabilitation exercise program is useful in the prediction of adherence to a long term activity. It is interesting to note, the behavior under study required repeated performances (36 sessions) to meet the single behavioral criteria for exercise in a cardiac rehabilitation program. The S lifestyle changes required for this patient population entail repeated actions. nUIIlerou Then urse is directly involved in these lifestyle modifications during hospitalization.
J{nowledge of the patients' responses to the attitude and subjective norm measures 125 regarding exercise in a cardiac rehabilitation program determined two points. The first point is: patients in general believed cardiac rehabilitation was beneficial, however, a nursing intervention is required for the beliefs that reflect unfavorable responses, recognizing the elicited patient fears. Secondly, who is important to the patient. It is those individuals that become part of the strategy to incur a change in behavior.
Moreover, the educational efforts of the nurse at this time should be directed toward the beliefs of patients who do not intend to enroll in a cardiac rehabilitation program. Strategies designed to alter the intention to enroll and subsequent attendance behavior should be developed from the favorable indirect attitude statements elicited in this study. The basic assumption of the theory of reasoned action is that a change in beliefs results in influence over the targeted behavior. The influence attempt or strategy must be directed at the targeted beliefs. Attitudes are thus changed by changing the existing beliefs or as this study suggested, strategies to improve patients' intention to enroll and exercise participation should be directed at the social role, as well as, the inaccurate or unfavorable attitudes toward exercise participation. Beliefs provide the basic foundation upon which strategies to incur a change in behavior are derived.
However, this is but the first step in such a complex process. The nurse possesses the given source of knowledge, is the channel by which new information is incurred, captures the patient's attention and provides the persuasive communication. Young et al. (1991) state "an understanding of the factors that influence a health-related behavior is of potential use in the development of preventive nursing approaches" (p. 143).

.Be""'"meodatjoos
There is a need to continue to study the theory of reasoned in the search for answers to health-related behaviors. The constructs of the theory of reasoned action possess utility for nurses as they assist patients to recover from illness. Specific recommendations for further research are: 126 1.
To apply and test the theory of reasoned action with another group of cardiac patients, such as the coronary-artery-bypass-graft-patient population, who are in a similar situation as the post myocardial infarction patient, regarding exercise post surgery.

2.
To apply and test the theory of reasoned action with post myocardial infarction patients, examining the differences in beliefs, attitudes, and subjective norm between those who intend to exercise after a myocardial infarction with those who do not intend to exercise.
Recommendations for theory development:  :wtiat do you believe are the adyanta~es of undertakin~ a cardiac rehabilitation ~ercjse pro~am 4 to 6 weeks after a heart attack? 28. People who don't take a heart attack seriously: 29. People who don't understand positive effects of exercise on heart attack victim:

17.
Please make a mark that most closely describes what you think or feel.
In general, I think that exercise is a cardiac rehabilitation program is: Please mark the response that most nearly describes what you think or how you feel.
In general, most of the people or groups that are important to me think I should Should not extremely quite slightly neither slightly quite extremely exercise in a cardiac rehabilitation program in four to six weeks after my heart attack.
Please respond to the foil owing questions. 142 vu.

25.
1 intend to exercise in a cardiac rehabilitation program in four to six weeks after my heart attack.

27.
Probably bring on an early death.

Likely
Unlikely extremely quite slightly neither slightly quite extremely Good Bad extremely quite slightly neither slightly quite extremely 28.
Be impossible since I don't have the energy or physical stamina to exercise.

Likely
Unlikely extremely q"Qite slightly neither slightly quite extremely Good Bad extremely quite slightly neither slightly quite extremely 29.
Not be necessary because I would rather exercise by myself than participate in a group exercise program.

Likely
Unlikely extremely quite slightly neither slightly quite extremely Good Bad extremely quite slightly neither slightly quite extremely

31.
Be unlikely since I am unable to exercise due to my heart attack and need to take it easy.

Likely
Unlikely extremely quite slightly neither slightly quite extremely Good Boo extremely quite slightly neither slightly quite extremely Not be necessary because I may not be able to return to work after my heart attack.

Likely
Unlikely extremely quite slightly neither slightly quite extremely Good Boo extremely quite slightly neither slightly quite extremely 32.
Not be necessary since I am not the whole person I was before my heart attack.

Likely
Unlikely extremely quite slightly neither slightly quite extremely Beliefs, Intention, Behavior I have been asked to take part in a research project, described below.
The researcher (or my doctor), Elaine Amato-Vealey, RN, MS, will explain the project to me in detail. I should feel free to ask questions. If I have more questions later, Elaine Amato-Vealey, the person mainly responsible for this study, (Phone: 277-4780), will discuss them with me. 7. 8.

149
There is no alternative to my participation in this study.
CONFIDENTIALITY. My part in this study is confidential. None of the information will identify me by name. The findings of this study may be used for medical publication. All records for this project will be handled according to hospital policy for medical records, Federal guidelines, and Rhode Island law on confidentiality of health-care information.
MY DECISION AND RIGHT TO QUIT AT ANY TIME. The decision whether or not to take part in this study is up to me. I do not have to participate. If I decide to take part in the study, I may quit at any time.