Theory of reasoned action

From CEOpedia
Theory of reasoned action
See also


Theory of reasoned action (TRA) - is a model from social psychology which states that a person's behaviour is determined by his or her intention to perform it, and that the intention depends on two things: the person's attitude toward the behaviour and the perceived social pressure to perform it, called the subjective norm[1]. It was developed by Martin Fishbein and Icek Ajzen and set out in full in their 1975 book Belief, Attitude, Intention, and Behavior.

The idea sounds almost too plain. People do what they intend to do, and they intend what they like and what important others expect of them. But the theory adds something older attitude research lacked: a precise recipe for measuring each element and combining them, plus a warning that the measures must match the behaviour exactly. That made it testable. And tested it was, in hundreds of studies on voting, dieting, blood donation, buying, software use and job choice. Marketing researchers adopted it early, and so did health psychologists. Later it became the parent of two of the most cited models in management research: the theory of planned behaviour and the technology acceptance model.

Background

Through the 1950s and 1960s attitude researchers had an awkward problem. General attitudes (towards a church, an ethnic group, a political party) often predicted specific actions badly. Ajzen and Fishbein reviewed this literature in 1977 and argued that the weak correlations were largely an artefact of poor measurement: researchers kept comparing broad attitudes towards objects with narrow, single acts[2]. Their answer was to shift the focus from the attitude towards an object (say, "electric cars") to the attitude towards a behaviour ("buying an electric car in the next twelve months").

Fishbein had worked since the early 1960s on an expectancy-value account of attitudes, in which an attitude is built from beliefs about an object and the evaluation of each belief. The 1975 book combined this with a normative component and with the concept of intention as the immediate antecedent of action[1]. The 1980 book Understanding Attitudes and Predicting Social Behavior turned the theory into a practical guide, with step-by-step instructions on eliciting beliefs and building questionnaires, and with chapters applying the model to areas such as voting, family planning and consumer choice[3].

The word "reasoned" is often misread. Fishbein and Ajzen didn't claim people are rational calculators. Beliefs can be wrong, biased or emotional. The claim is only that behaviour follows reasonably and consistently from whatever beliefs a person holds.

Components of the theory

The model is a causal chain running from beliefs to behaviour:

  • Behavioural beliefs - what the person thinks will happen if he or she performs the behaviour, together with an evaluation of each outcome.
  • Attitude toward the behaviour (AB) - the overall positive or negative judgement of performing the behaviour. It is related to, but not identical with, the cognitive component of attitude described in classical three-component models.
  • Normative beliefs - what specific referents (spouse, parents, friends, boss, doctor) think the person should do, weighted by the person's motivation to comply with each of them.
  • Subjective norm (SN) - the overall perceived social pressure to perform or not perform the behaviour. It is close to what sociologists describe as group norms, though measured from the individual's point of view.
  • Behavioural intention (BI) - the person's subjective probability that he or she will perform the behaviour.
  • Behaviour (B) - the observable act.

Everything else (personality, demographics, general attitudes towards objects, social roles) is treated as an external variable. It may affect behaviour, but only indirectly, by changing beliefs or the relative weights of the two components.

Formula of the theory of reasoned action

The central equation states that intention is a weighted sum of attitude and subjective norm:

\[ B \approx BI = w_1 A_B + w_2 SN \]

The weights w1 and w2 are not fixed. They are estimated empirically, usually as standardised coefficients in a multiple regression analysis, and they change from one behaviour and population to another. For some decisions (buying a toothpaste) attitude dominates; for others (having a child, joining a strike) social pressure may weigh just as much.

Attitude is computed using the expectancy-value principle:

\[ A_B = \sum_{i=1}^{n} b_i e_i \]

where bi is the strength of the belief that the behaviour leads to outcome i and ei is the evaluation of that outcome. The logic is the same one found in Vroom's expectancy theory of work motivation: expectations multiplied by values.

The subjective norm has the same structure:

\[ SN = \sum_{j=1}^{m} nb_j mc_j \]

where nbj is the normative belief about referent j (does this person think I should do it?) and mcj is the motivation to comply with that referent.

In typical questionnaires belief strength and evaluations are measured on seven-point bipolar scales scored from -3 to +3, and motivation to comply on a unipolar scale from 1 to 7[3]. The salient beliefs are not invented by the researcher. They are elicited first in open-ended interviews with a sample of the target population ("What do you see as the advantages of...?"), and the most frequently mentioned ones go into the final instrument.

Principle of compatibility

A measure of intention predicts behaviour well only when both refer to the same four elements: the action, the target at which it is directed, the context and the time[2]. "I intend to buy an electric car" is a poor predictor of "bought a Nissan Leaf from a local dealer in March". "I intend to buy an electric car within the next year" matched against "bought any electric car within the year" fits much better. Ajzen and Fishbein showed that studies meeting this condition reported much higher attitude-behaviour correlations than those that ignored it[2]. It sounds like a technicality. In practice it's the single most common reason why survey-based forecasts fail.

Empirical support

The best-known quantitative review is the meta-analysis by Sheppard, Hartwick and Warshaw, published in the Journal of Consumer Research in 1988. They found strong overall evidence for the predictive power of the model, with an average correlation of about .66 between the combination of attitude and subjective norm and intention, and about .53 between intention and behaviour[4]. The predictive power held up even in the many studies that stretched the model beyond its original boundary conditions, for example by applying it to choices among alternatives or to goals not fully under the person's control. The authors identified moderating variables and recommended extensions, notably for situations with several alternatives and for goals rather than behaviours[4].

In management terms: roughly speaking, attitude and norm explain a little over 40% of the variance in intentions, and intentions explain a bit over a quarter of the variance in behaviour. Good for social science. Far from perfect.

Example of the theory of reasoned action

The following example is hypothetical and serves only to illustrate the calculation.

A car manufacturer surveys potential buyers about the intention to buy an electric car within the next 12 months (scale 1-7). Elicitation interviews produced five salient outcomes and four relevant referents. One respondent, Anna, gave these answers:

Outcome of buying an electric car Belief strength b (-3..+3) Evaluation e (-3..+3) b × e
Lower running costs +3 +3 9
Lower personal CO2 emissions +2 +2 4
Limited range on long trips +2 -2 -4
High purchase price +3 -3 -9
Quiet and pleasant to drive +1 +2 2
Attitude AB 2
Referent Normative belief nb (-3..+3) Motivation to comply mc (1..7) nb × mc
Spouse +3 6 18
Friends +1 3 3
Colleagues at work -1 2 -2
Parents -2 4 -8
Subjective norm SN 11

Anna's attitude is barely positive. The savings and the green image are almost cancelled by the price and range worries. Her subjective norm is clearly positive, mostly because of a spouse who wants the car and whose opinion matters to her.

Suppose a regression on the whole sample of respondents produced the following raw-score equation:

\[ BI = 3.2 + 0.08 \cdot A_B + 0.03 \cdot SN \]

For Anna: BI = 3.2 + 0.08 × 2 + 0.03 × 11 = 3.2 + 0.16 + 0.33 = 3.69 on the seven-point scale. Undecided, leaning slightly towards "no".

Raw coefficients can't be compared directly because the two indices have different ranges. If the sample standard deviations were 12 for attitude, 25 for subjective norm and 1.6 for intention, the standardised weights would be 0.08 × 12 / 1.6 = 0.60 for attitude and 0.03 × 25 / 1.6 ≈ 0.47 for the subjective norm. Attitude carries more weight, but social pressure clearly matters too.

Now the practical part. What would change Anna's mind?

  • A purchase subsidy that weakens her belief that the price is high (b drops from +3 to +1) raises the price product from -9 to -3 and the attitude to 8. BI climbs to 3.2 + 0.64 + 0.33 = 4.17.
  • A referral campaign that makes friends visibly enthusiastic (nb from +1 to +3) raises the friends' product to 9 and SN to 17. BI = 3.2 + 0.16 + 0.51 = 3.87.
  • Both together give 3.2 + 0.64 + 0.51 = 4.35.

The model doesn't just predict. It points at which belief, and which referent, the marketing research and communication budget should target.

Applications

Marketing and consumer behaviour. TRA was quickly adopted for brand choice, purchase intention and consumer buying behaviour studies, partly because the 1988 meta-analysis appeared in a consumer research journal and partly because the belief tables translate directly into advertising messages. It fits the evaluation-of-alternatives stage of the consumer decision making process and is one of the ways researchers operationalise factors influencing consumer behaviour.

Health and social marketing. Condom use, smoking cessation, screening, vaccination, exercise. Health campaigns often start with a belief elicitation study to find out which consequences and which referents matter to the target group. That's the core of many social marketing programmes.

Human resources and organisational behaviour. Intention to quit, intention to take part in training, acceptance of new procedures, whistle-blowing, safety behaviour. In organizational behavior research the model links employee beliefs about outcomes with the influence of supervisors and co-workers, and turnover intention is often used as the leading indicator for the employee turnover rate.

Information systems. Davis, Bagozzi and Warshaw compared TRA with the newly proposed technology acceptance model in a longitudinal study of users of a word-processing program; both predicted intentions and usage reasonably, with TAM doing so more parsimoniously[5].

Related approaches

Theory of planned behaviour

TRA was designed for behaviours under full volitional control, things a person can simply decide to do. Many behaviours aren't like that. Losing weight, getting a promotion or quitting smoking depend on skills, resources and other people. Ajzen extended the model by adding perceived behavioural control, the person's sense of how easy or difficult the behaviour is. In the theory of planned behaviour, perceived control affects intention and can also predict behaviour directly, to the extent that it reflects actual control[6]. The construct is close to Bandura's self-efficacy from social cognitive theory. When control is complete, the theory of planned behaviour reduces to TRA.

Technology acceptance model

Fred Davis adapted TRA to the use of information systems. TAM keeps the chain beliefs → attitude → intention → use but replaces the open-ended set of salient beliefs with two general ones: perceived usefulness and perceived ease of use. The subjective norm was left out of the original TAM[5]; later versions brought social influence back in. See technology acceptance model for details.

Reasoned action approach

In 2010 Fishbein and Ajzen published Predicting and Changing Behavior: The Reasoned Action Approach, which merged TRA and the theory of planned behaviour into one framework with attitude, perceived norm and perceived behavioural control, each with its own belief basis[7]. The perceived norm there includes both injunctive norms (what others think I should do) and descriptive norms (what others actually do).

Other relatives include the diffusion of innovations theory, which also deals with adoption decisions but from the angle of the innovation and the social system, and the broader research on attitudes and behaviour.

Advantages of the theory of reasoned action

  • Simple, parsimonious structure: two predictors, one intention, one behaviour.
  • Precise measurement rules and a standard procedure for building questionnaires, which makes studies comparable.
  • Good predictive validity for volitional behaviours, documented by meta-analysis[4].
  • Diagnostic value. The belief-level data show why people intend or don't intend to act, and which beliefs or referents an intervention should target.
  • Wide applicability: marketing, health, HR, information systems, environmental behaviour, politics.
  • The principle of compatibility is useful well beyond this theory, for any survey that tries to forecast behaviour.

Limitations of the theory of reasoned action

  • Volitional control. The model assumes the person can perform the behaviour at will. When money, skills, time or co-operation of others are needed, intentions overestimate behaviour. This was the main reason for the theory of planned behaviour[6].
  • Intention-behaviour gap. Even strong intentions often go unrealised: people forget, change their minds, meet obstacles. The longer the time between measurement of intention and behaviour, the weaker the link.
  • Habits. For behaviours repeated often in a stable context (commuting by car, buying the same brand, eating fast food) past behaviour predicts future behaviour independently of intentions. Ouellette and Wood showed that in such settings habit guides behaviour automatically, while intentions matter more for new or infrequent behaviours[8].
  • Emotions and impulses. The model gives little room to affect, mood or impulsive choices, which matter in many consumer decisions.
  • Questionable multiplication. Products such as b × e depend on how the scales are scored (bipolar or unipolar). Changing the scoring can change correlations, so the multiplicative indices are statistically fragile.
  • Self-report. Intentions, beliefs and often the behaviour itself are measured by questionnaires, which brings social desirability bias and common-method variance.
  • Choice among alternatives. The basic model looks at one behaviour at a time; real consumers compare brands or options, a point raised by Sheppard and colleagues[4].
  • Separation of attitude and norm. In practice the two components are often correlated, and some argue that social expectations end up absorbed in attitudes.

References

Footnotes

  1. ↑ Fishbein M., Ajzen I., Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Addison-Wesley, 1975
  2. ↑ Ajzen I., Fishbein M., Attitude-behavior relations: A theoretical analysis and review of empirical research, Psychological Bulletin, 1977, pp. 888-918
  3. ↑ Ajzen I., Fishbein M., Understanding Attitudes and Predicting Social Behavior, Prentice-Hall, 1980
  4. ↑ Sheppard B.H., Hartwick J., Warshaw P.R., The theory of reasoned action: A meta-analysis of past research, Journal of Consumer Research, 1988
  5. ↑ Davis F.D., Bagozzi R.P., Warshaw P.R., User acceptance of computer technology: A comparison of two theoretical models, Management Science, 1989, pp. 982-1003
  6. ↑ Ajzen I., The theory of planned behavior, Organizational Behavior and Human Decision Processes, 1991, pp. 179-211
  7. ↑ Fishbein M., Ajzen I., Predicting and Changing Behavior: The Reasoned Action Approach, Psychology Press, 2010
  8. ↑ Ouellette J.A., Wood W., Habit and intention in everyday life, Psychological Bulletin, 1998, pp. 54-74
Theory of reasoned action — recommended articles
Attitudes and behaviour — Technology acceptance model — Cognitive component of attitude — Consumer buying behaviour — Expectancy theory — Social cognitive theory — Social marketing
Retrieved from CEOpedia