The theory of reasoned action (TRA) is a special case of the theory of planned behavior (TPB). The only difference between the two theories is that the TPB includes perceived behavioral control as an additional determinant of intentions and behavior. In the development of the TRA it was assumed that people have volitional control over the behavior of interest (and that they realize that they are capable of performing the behavior if they so desire). Under these conditions, perceived behavioral control becomes irrelevant and the theory of planned behavior reduces to the theory of reasoned action.

There is no official TPB questionnaire. In most published research papers, you can find detailed information about the questionnaire used. A sample questionnaire is availabe online (Sample Questionnaire). However, the items employed in previous research and the items shown in the sample questionnaire may not be appropriate for your behavior, population, or time period. Formative research is therefore required to construct a new questionnaire suitable for the behavior and population of interest. If beliefs are to be assessed, they must be elicited anew from a representative sample of the research population. Similarly, items designed to directly assess the theory's constructs must be validated prior to construction of the final questionnaire. (See Constructing a TPB Questionnaire.)

First search the list of TPB references on my website, at: http://www.people.umass.edu/aizen/tpbrefs.html If you cannot find anything of relevance, you could try one of the social science or online databases, such as PsycINFO or Google Scholar.

We are often interested in predicting, explaining, or changing categories of behavior, such as exercising, studying, or conserving energy -- not any single action. It is possible to deal with such a criterion by assessing attitudes, subjective norms, perceptions of control, intentions, and actual behavior with respect to each of a representative set of actions that comprise the category of interest. These measures can then be aggregated to arrive at indices representing the behavioral category. However, if the investigator has no particular interest in the individual actions that comprise the category, a simpler strategy can be adopted. Participants can be given a description of the behaviorl category, and all TPB measures are obtained in relation to the category as a whole. For example, the participants could be told that we are interested in vigorous exercise defined as 30 minutes or more of physical activity that produces a sweat, at least three times a week. The TPB constructs are then assessed with respect to this criterion.

Strictly speaking, every behavior involves a choice, even if the choice is between performing or not performing the behavior. In some cases, beliefs about not performing a behavior may largely be mirror images of beliefs about performing it. In this case, we can simply assess the TPB constructs in relation to performing the behavior. However, sometimes different beliefs are readily accessible in relation to the behavioral alternatives. For example, if we are interested in travel-mode choice, the beliefs that are readily accessible with respect to using public transportion may not simply be mirror images of using one's car. If practical, therefore, it is recommended that in this case the TPB constructs be assessed in relation to both options under consideration.

Not necessarily. Theoretically, attitudes are based on behavioral beliefs, subjective norms on normative beliefs, and perceived behavioral control on control beliefs. However, these propositions are subject to empirical test. The validity of the belief composite measures is typically tested by correlating each composite with the direct measure of the corresponding construct.

The answer depends on the purpose of your research. If you just want to predict intentions and behavior, the direct measures of attitude, subjective norm, and perceived behavioral control are sufficient. However, if you would like to understand the basis for these factors (perhaps to design an effective behavior change intervention) you also must elicit and assess the accessible behavioral, normative, and control beliefs.

Standard attitude scaling procedures are used to assess attitude, subjective norm, and perceived behavioral control in a relatively direct manner. According to the theory, the summative indices of behavioral, normative, and control beliefs capture, respectively, the underlying determinents of these constructs. The direct measures should therefore be used to predict intentions.

It is possible to measure intention and behavior at the same time. For example, we can measure intention to exercise in the next three months and, at the same time, obtain a measure of exercise behavior during the previous three months. Of course, the correlation between these two measures only provides an indication of the extent to which current intentions are consistent with previous behavior. The correlation can be taken as an indication of the intention's predictive validity only if there is independent evidence to suggest that the behavior in question has a high degree of temporal stability. Thus, if it is know that later exercise behavior correlates strongly with prior exercise behavior in the population of interest, then the measure of past behavior can serve as a proxy for a measure of future behavior. However, this correlation may be inflated if participants misremember, misconstrue, or misrepresent their reported past behavior such that it is more consistent with current intentions than it actually was.

The reliability and validity of direct TPB measures are estimated in formative research. First, a TPB questionnaire is constructed in accordance with established guidelines (See Constructing a TPB Questionnaire). The direct items designed to assess a given TPB construct (attitude, subjective norms, perceived behavioral control, intention, or behavior) are then submitted to an internal consistency analysis to establish reliability. Cronbach's alpha is the most commonly used coefficient. However, internal consistency is not a requirement of the behavioral, normative, and control belief composities because different accessible beliefs may well be inconsistent with each other. If reliability in the sense of temporal stability is also considered important, the questionnaire must be administered a second time and test- retest correlations are computed. Estimation of validity is, as always, more difficult. Usually, all we can do in pilot work is to establish the convergent and discriminant validities of the TPB measures. Confirmatory factor analyses are employed to show that the items measuring a given construct can be considered indicators of the same latent variable; and that a model in which the attitude, subjective norm, perceived control, and intention items are treated as assessing separate constructs is superior to a model in which all items are considered to measure the same underlying construct.

Direct TPB measures should be developed in pilot work to make sure that they have satisfactory psychometric properties. If you have multple items for each construct, you may be able to drop one or two and thereby increase internal consistency. When this is not possibble, the correct procedure is to start over again, develop adequate measures, and then collect your data.

No. All readily accessible behavioral beliefs (usually elicited in a free-response format) are considered to be important determinants of attitude, all readily accessible normative beliefs are considered to be important determinants of subjective norm, and all readily accessible control beliefs are considered to be important determinants of perceived behavioral control. No regression analysis is needed to estimate a belief's relative importance or weight. All beliefs are given equal weight in the expectancy-value model.

Most investigators working with the TPB use 7-point bipolar adjective scales (e.g., harmful - beneficial), not "Likert" scales. The format of these scales is based on work with the semantic differential which found 7 points to be optimal. However, there is nothing sacred about 7 points and it is at the investigator's discretion to use fewer or more scale points.

In the case of modal accessible beliefs (i.e., the same set of beliefs for all respondents), it is immaterial whether the beliefs are aggregated by summation or averaging. The mean is a simple linear transformation of the sum. Such a transformation has no effect on the correlation of the aggregate measure with any other variable. Note that different considerations apply to personal accessible beliefs, i.e., beliefs emitted by the participants themselves. Here, it is assumed that the beliefs listed represent a person's total set of accessible beliefs. Unlike modal accessible beliefs which are the same in number for all participants, different individuals may well hold different numbers of personal behavioral beliefs, different numbers of personal normative beliefs, and different numbers of personal control beliefs. According to the theory, the greater the number of favorable beliefs, the more favorable will be the attitude, subjective norm, or perceived behavioral control. Personal accessible beliefs should therefore be summed, not averaged, because computing a mean would elminate these theoretically meaningful differences between individuals.

Some outcomes associated with a behavior receive uniformly positive or negative evaluations. For example, one of the behavioral beliefs about eating a low-fat diet may be that this behavior "reduces the risk of cardio-vascular disease." When asked to evaluate this outcome on, e.g., a 7-point good - bad scale, virtually all participants will give it a positive evaluation, and having the negative side of the scale may be irritating. Nevertheless, some people may value the outcome more favorably than others. To capture this variance, and avoid annoying the partcipants, we can use a unipolar evaluative scale, such as a 7-point scale that ranges from slightly good to extremely good. Alternatively, we could ask participants to rate how important it is for them to reduce the risk of cardio-vascular disease on, e.g., a 7-point not at all important to extremely important scale. Note, however, that importance ratings can be used only when the valence of the outcome (positive or negative) is clearly established because positive as well as negative oucomes can be judged important.

Attitude (A), subjective norm (SN), and perceived behavioral control (PBC) are conceptually independent predictors of intentions. However, empirically they are usually found to be intercorrelated because the same information can influence behavioral, normative, and/or control beliefs, the theoretical antecedents of A, SN, and PBC, respectively. For example, the information that medical research has established the superiority of an easy to prepare low-fat diet for lowering serum cholestorol may lead to the formation of the behavioral belief that "my eating a low-fat diet will reduce my serum cholestorel level," the normative belief that "my doctor thinks I should eat a low-fat diet," and the control belief that "I have the ability to prepare a low-fat diet."

The relative importance of the three predictors is not measured but, instead, is estimated by means of multiple regression or structural equation analyses. The standardized regression or path coefficients serve as estimates of the relative importance (weights) of the predictors.

Conceptually, there is no difference between perceived behavioral control (PBC) and self-efficacy (SE). Both refer to people's beliefs that they are capable of performing a given behavior. Operationally, however, PBC and SE are usually assessed in different ways. Research with Bandura's concept of self-efficacy usually defines a graded series of potential obstacles to performance of the behavior, and participants are asked to indicate how likely it is that they could overcome each obstacle. To assess PBC, participants are usually asked to rate the extent to which they have the ability to perform the behavior, how much the behavior is under their control, etc. (see Constructing a TPB Questionnaire). It would be a mistake to assume that some of these items (e.g., rated ability to perform the behavior) assess SE whereas other items (e.g., the extent to which the behavior is under my control) assess PBC.

Perceived behavioral control (PBC) refers to people's beliefs that they are capable of performing a given behavior. Locus of control (LC) has to do with the generalized belief that events in one's life are caused by internal factors (e.g., ability or motivation) as opposed to external factors (e.g., other people, nature, or luck). Even domain-specific LC (e.g., health locus of control) has to do with a broad range of behaviors and events, not control over performance of a particular behavior. Moreover, internal control should not be equated with high PBC, nor should external control be equated with low PBC. Perceived lack of ability is an internal causal factor, but it would produce a sense of low PBC. Similarly, performance of a behavior may depend on the cooperation of others (an external factor), but people may believe that they nevertheless have high behavioral control because they are confident of obtaining the needed cooperation.

In the TPB, actual behavioral control (ABC) moderates the effect of intentions on behavior. An essential prerequisite for assessing a person's ABC is a good understanding of the various internal factors (skills, knowledge, physical stamina, intelligence, etc.) and external factors (legal barriers, money, equipment, cooperation by others, etc.) that are needed to perform the behavior or that can interfere with its performance; as well as a way to assess the extent to which the person has or can obtain the requisite resources and overcome potential barriers. Because it is usually much more difficult to measure actual behavioral control than perceived behavioral control (PBC), most studies rely on PBC as a proxy for ABC.

Several factors may be responsible for a relatively low proportion of explained variance in observed intentions or behavior. There may be little meaningful variance in the criterion if, for example, most participants in the study intend to, and/or actually engage in the behavior under investigation. Events intervening between questionnaire administration and observation of behavior may have produced changes in beliefs, attitudes, and intentions. As a result, the original measures are no longer predictive of the behavior. The measures of the theory's constructs, including intention and behavior, may be unreliable, containing a large proportion of error variance. The predictors and criteria are incompatible, i.e., they are measured at different levels of specificity or generality. The factors included in the TPB are not sufficient. Other factors, such as moral norms or strong habits, may also influence the particular intention or behavior under investigation. It may be possible to measure some of these factors and add them to the theory in order to improve prediction. It is, of course, also possible that the TPB, even with the inclusion of additional factors, is unable to provide accurate prediction of the intention or behavior under investigation.

The TPB is, in principle, open to the inclusion of additional predictors. However, for the sake of parsimony, additional predictors should be proposed and included in the theory with caution, and only after careful deliberation and empirical exploration. Among the criteria that should be met by any proposed addition are the following. Like the theory's existing predictors (attitude toward the behavior, subjective norm, perceived behavioral control, and intention), the proposed variable should be behavior-specific, conforming to the principle of compatibility. That is, it should be possible to define and measure the proposed factor in terms of the target, action, context, and time elements that describe the behavioral criterion. It should be possible to conceive of the proposed variable as a causal factor determining intention or action. Proposed additions should be conceptually independent of the theory's existing predictors, rather than be redundant with them. The factor considered should potentially be applicable to a wide range of behaviors studied by social scientists.

Personality traits, intelligence, demographic variables, values, and other variables of this kind are considered "background factors" in the TPB. They are not neglected but assumed to influence intentions and behavior indirectly by affecting behavioral, normative, and/or control beliefs. That is, the components of the TPB are assumed to mediate the effects of background factors on intentions and actions. The theory acknowledges that background factors can provide valuable information about possible precursers of behavioral, normative, and control beliefs, information not provided by the theory itself. Conversely, with the aid of the TPB it becomes possible to examine why a given background factor influences behavior by tracing its effects via the more proximal antecedents of the behavior.

No, there is nothing in the theory to suggest that attitude, subjective norm, and perceived behavioral control will each make a significant contribution to the prediction of intention. The relative importance of these three factors is likely to vary from one behavior to another and from one population to another. In some cases, one or another of the three factors will be found to have no significant effect on intention. Assuming that the factors were measured with equal reliability, lack of predictive validity merely indicates that for this particular behavior and population, the factor in question is not an important consideration in the formation of intention.

No, investigators define their behavioral criteria at a level of specificity or generality suitable for their purposes, typically at an intermediate level (e.g., donating blood or exercising on a regular basis). The details (mostly context and time) specified in an implementation intention are usually of no interest to the investigator. When forming an implementation intention, participants are asked to indicate when, where, and how they plan to implement their (goal) intentions to increase the likelihood that the intention will be carried out, not because of any inherent interest in these details. The behavioral criterion remains the more general goal intention. Furthermore, even when the goal intention of interest to the investigator is relatively specific (e.g., exercising every weekday in the gym during lunch hour), implementation intentions -- although also very specifc as to place, time, and modality -- can be formulated only for positive goal intentions. Thus, it makes little sense to ask where, when, and how you plan to exercise if you have not formed an intention to engage in this behavior.

Knowledge, or correct factual information, plays no direct role in the TPB. Although information in the form of behavior-relevant beliefs is a central component of the theory, whether that information is correct or incorrect is immaterial. What matters is whether the information works for or against performance of the behavior. Consider, for example, a woman's belief that breast self-examination allows her to distinguish between benign lumps and malignant tumors. Although factually wrong, this item of "knowledge" would tend to support performing BSE. Thus, we cannot expect a direct correlation between amount of correct factual information on one hand and attitudes, intentions, and behavior on the other.

The TPB was developed to predict and explain human social behavior, and to serve as a framework for behavior change interventions. Like other theories of this kind, it can be used as a heuristic framework to guide questions to be raised in qualitative research. However, the standard methods developed over the years for use with the theory are largely quantitative in nature. The only part of these methods that requires qualitative research is the elicitation and coding of readily accessible behavioral, normative, and control beliefs (See Constructing a TPB Questionnaire).

The theory of planned behavior is in the public domain. No permission is needed to use the theory in research, to construct a TPB questionnaire, or to include an original drawing of the model in a thesis, dissertation, presentation, poster, article, or book. However, if you would like to reproduce a published drawing of the model, you need to get permission from the publisher who holds the copyright. You may use the drawing on this website for non-commercial purposes so long as you retain the copyright notice.

I receive many requests for assistance and it is simply impossible for me to devote time to each. If you have a budget for consultation, I will be glad to discuss the terms with you. Otherwise, I suggest you carefully examine this website. On it you will find guidance on how to apply the theory of planned behavior, a sample TPB questionnaire, as well as answers to many other questions you may have. If, after perusing the site, you still have one or two quick questions, email me and I'll try to answer them.