Predicting Home Energy Consumption: an Application of the Fishbein-Ajzen Model

Clive Seligman, University of Western Ontario
Don Hall, University of Guelph
Joan Finegan, University of Western Ontario
ABSTRACT - The present study employed the Fishbein-Ajzen model of attitudes z behavior to predict household energy consumption. Forty-eight couples answered a questionnaire about their attitudes toward energy consumption, and their intentions of consuming energy in the home during the next six weeks. Actual energy consumption was then computed from electrical meter readings. The results substantiate the model, and suggest that conservers & nonconservers do not share the same patterns of belief.
[ to cite ]:
Clive Seligman, Don Hall, and Joan Finegan (1983) ,"Predicting Home Energy Consumption: an Application of the Fishbein-Ajzen Model", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 647-651.

Advances in Consumer Research Volume 10, 1983      Pages 647-651


Clive Seligman, University of Western Ontario

Don Hall, University of Guelph

Joan Finegan, University of Western Ontario


The present study employed the Fishbein-Ajzen model of attitudes z behavior to predict household energy consumption. Forty-eight couples answered a questionnaire about their attitudes toward energy consumption, and their intentions of consuming energy in the home during the next six weeks. Actual energy consumption was then computed from electrical meter readings. The results substantiate the model, and suggest that conservers & nonconservers do not share the same patterns of belief.

The study of attitudes has been a central concern of psychology for many decades. Thus it is not surprising that psychologists interested in energy consumption behavior have also attended to this concept (see Olsen 1981 for a comprehensive review). The interest in attitudes derives largely from our assumption that attitudes influence behavior. Indeed, attitudes are typically defined as a predisposition to respond consistently (either favorably or unfavorably) toward the object of the attitude (Ajzen 1982). The link between attitudes and behavior has been studied extensively and remains a controversial topic (Zanna, Higgins, and Herman 1989). In fact, many studies in the psychological literature have suggested that behavior is not predicted very well from attitudes -(Wicker 1969). This negative approach, however, is not very helpful, because it ignores several, sophisticated discussions of the attitude-behavior relationship (Fishbein and Ajzen 1975; Kiesler, Collins, and Miller 1969). A fairer position would be that attitudes do nave utility for the prediction of behavior, as Ajzen (1982) has so persuasively argued.

Whereas mainstream social psychology seems to have been making substantial conceptual and methodological advances with regard to the attitude-behavior issue, research in the area of attitudes and energy behavior is more reminiscent of the attitude-behavior research of the 1950s and 1960s. For example, McDougall, Claxton, Ritchie, and Anderson (1931) in their review of consumer energy research conclude "...attitudinal variables have indicated little association with energy consumption" (p. 346). In addition to the reasons given by McDougall et al. (1981) themselves, there are several other good reasons why the past research has been unsuccessful in demonstrating strong relationships between attitudes and energy consumption. First, the term attitude has been used imprecisely. That is, studies purporting to measure attitudes have often been measuring other factors, e.g., knowledge of energy-related issues, energy awareness, beliefs, price consciousness, etc. Thus, attitudes have been implicated unfairly in some failures to correlate energy consumption with these other factors. Second, because houses and machines consume energy, it should be obvious that house factors, such as size, and climate factors, such as outside temperature, are the maj or determinants of home energy consumption. By comparison, the attitudes of the residents living in the houses seem inconsequential. But the real issue is: What are the effects of attitudes on the energy consumption or people living in similar housing. We know, for example, that the range of energy consumption of people living in identical housing can be as great as 2-1 in the winter and 3-1 in the summer (Socolow 1978). When house and climate conditions nave been held constant, substantial correlations have been round between energy consumption and some attitudes (Seligman, Kriss, Darley, Fazio, Becker, and Pryor 1979). Third, the failures to find high correlations between-energy consumption and attitudes have been almost entirely due to attempts to correlate global attitudes with specific behaviors. As Ajzen points out " attitudes toward an object ...are of little value if we are interested in predicting a particular action with respect to the object. To predict a single behavior we have to assess the person's attitude toward the behavior in question" (1982, p. 13). In other words, there is little reason to believe that we could predict a thermostat setting from one's attitude toward the energy crisis; instead we would have to assess one's attitude toward lowering the thermostat. For example, even though a person may be quite convinced that the energy crisis is real, s/he may not lower the thermostat because of the needs of a bed-ridden family member. Although this medical problem would not be reflected in their attitudes toward the energy crisis, it would likely emerge in their attitude toward adjusting the thermostat. Finally, the previous research has been almost exclusively atheoretical in terms of a theory of attitudes and its relationship to behavior.

The present study is an attempt to take into account the above concerns and to predict household energy consumption using a model of attitudes and behavior developed by Fishbein and Ajzen (1975). According to their model, behavior which is under volitional control is best predicted by an individual's intention to perform that behavior. The intention to perform a behavior, in turn,- can be predicted by two basic perceptions of the individual. The first is the individual's attitude toward performing the behavior, derived from his/her beliefs about the consequences of his/her performance of the behavior. The second is the individual's view of how those people who are important to him/her regard his/her performance of the behavior. Algebraically, the model may be expressed by:

B~BI   (1)

BI = (Aact) W1 + (SN) W2   (2)

where B = the behavior in question, BI = the behavioral intention, Aact = the attitude toward performing the action or behavior, SN is the subjective norm (the individual's perception that those who are important to him/her think he/she should or should not perform the behavior in question), and W1 and W2 are empirically determined regression weights reflecting the relative importance of Aact and SN in determining BI. The model further proposes that an individual's attitude toward a behavior may be determined by his/her beliefs about the consequences of his/her performance or the behavior (Bi) and his/her evaluations of these consequences (ei). This assumption may be algebraically expressed


where n = the number of salient beliefs held by the individual concerning the performance of the behavior. The subjective norms held by an individual concerning his/her performance of a certain behavior are determined by his/her normative beliefs, (i.e., his/her perceptions of what referents think he/she should do in the behavioral situation (NBj), and his/her motivation to comply with the advice of these referents (Mcj). This assumption may be algebraically expressed by:


where n = the number of salient referents with regard to the performance of the behavior. The model proposes that B can only be affected through BI which in turn must be influenced only through Aact and SN. Of course, this last proposition may be obstructed in cases where habit, exposure to new information, loss of volitional control, etc. enter into the situation (Fishbein and Jaccard 1973).

Perceptions such as those found in the Fishbein model have been useful in understanding environmental problem related behavior in several previous studies. Levenson (1974) found that beliefs about the consequences of pollution were positively associated with individual's membership in an anti-pollution group. Heberlein and Black (1976) found that the belief that the uses of led-free gasoline reduces pollution was associated with the actual use of lead-free gasoline. Hass, Bagley, and Rogers (1975) found that intentions to reduce energy consumption were positively associated with perceived noxiousness of an energy shortage. Attitudes concerning the use of nuclear energy were successfully predicted from beliefs concerning the consequences and the evaluation of those consequences (Otway and Fishbein 1976). While these studies employed portions of the Fishbein model in the relationships they examined, only Bowman and Fishbein (1978) used the entire model successfully to predict voting on a government act which would place increased safety restrictions on nuclear power stations.

In view of the support found for this model, it ,s expected that the following relations will hold for energy consumption behavior:


These relations will permit a comparison of the 'fine" components of the model (Bi s, ei s, NBj s. and Mcj s) between those who conserve home energy and those who do not. This will allow us to compare conservers v nonconservers regarding their beliefs about the consequences of energy consumption and the evaluation of these consequences. Most past research employing the Fishbein model has examined dichotic behavior (i.e., actions which can be performed in only one of two ways, such as voting "for" or "against", buying or not buying, etc.). The present study will attempt to extend the model to a continuous behavior (i.e., an action which can be performed to various degrees along a continuum).



Subjects were homeowners from "Glenwood Estates", a 161 unit, condominium complex in London, Ontario. The complex was arranged in separate rows, with each row consisting of six or seven attached units. Thus. some units would have neighbors on either side (i.e. interiors), whereas others would have neighbors on only one side (i.e., exteriors). All units were electrically heated and were similar in terms of number of rooms, and square feet of floor space.

The study was conducted during the 1¦79-80 Winter. Twelve male adults and 13 female adults, all t rom separate units, were contacted by telephone. In the course of the conversation, these pilot subjects responded to open-ended questions designed to elicit salient beliefs held about energy conservation in the home during the winter, and names of people and groups whose viewpoints on home energy conservation were the most salient. Respondents were asked to state the reasons they had for conserving (or not conserving), to list advantages and disadvantages for doing so, and to indicate referents who would approve or disapprove of this behavior or whose advice they would follow in this area.

A content analysis of these interviews revealed that subjects believed that home energy conservation would save money, help solve the energy crisis, be bothersome, make a house more (or less) comfortable, improve (or impair) their health and reduce personal energy waste. Among frequently mentioned referents were the government, media personnel, friends and neighbors, parents, spouses, utility and energy companies. These beliefs and referents were included in the final questionnaire.

Copies of this self-administered questionnaire were then distributed to 103 homes in the complex. The man and the woman of each household were instructed to complete the questionnaire individually without consultation with each other. Forty-eight couples correctly completed and returned the questionnaires. Subjects, on average, were in their early thirties (males M = 34.4, and females M = 31.0), had some post-secondary education, and had lived in their home a total of 2.8 years. The average family size was 3.3.


The first part of the questionnaire contained demographic items and assessed the conservation measures that the respondent had already taken to save energy (e.g., installing storm windows, turning off lights). Subjects were asked to rate on a 7-point scale whether they considered themselves to be "energy conservers" in the home, and to compare their household energy consumption to that of other households in Glenwood Estates.

Before any questions pertaining to the Fishbein-Ajzen model were asked, subjects were provided with a specific definition of "conserving energy in the home". That is, conserving energy was defined as "doing at least some at the following: turning off the lights not in use, lowering the thermostats in all rooms not being used, using electric appliances and hoe water as minimally as possible, and keeping the thermostats in rooms that are being used as low as you can tolerate".

Items measuring the theoretical constructs of the Fishbein-Ajzen model then followed. In accordance with Fishbein's and Ajzen's recommendations, these items specified explicitly the behavior (conservation), the target and context (in the home) and the time during which the behavior was to be performed (in six weeks). The scales that were used to assess BI, Aact, and SN, as well as examples of scales that tapped salient B s, e s, NB s, and Mc s, are shown below:

BI (Behavioral intention) I intend to conserve energy in my home over the next six weeks.

likely --:----:----:----:----:----:----:unlikely

Aact (attitude toward the act) Conserving energy in my home over the next six weeks would be:


harmful:----:----: -:--:--:--:beneficial

wise:----:----:----:----:----:----:-- foolish

SN (subjective norm) Most people who are important to me think I should conserve energy in my home over the next six weeks.

likely:----:----:--:----: -:----:----:unlikely

B(beliefs about the consequences of the act) Conserving energy in my home over the next six weeds would make me more comfortable in the house.

likely:----:--:----:--:--:--:---- :unlikely

e (evaluations of the consequences) Being comfortable in the house would be:


NB (normative beliefs) My friends and neighbors think I should conserve energy in my home over the next six weeks.

likely:----:----:----::--:----:--: unlikely

Mc (motivation to comply) With respect to conserving energy in my home over the next six weeks

I very much want to:---:--:----:----:I very much do not want to do want my friends and neighbors think I should do.

These items were listed in the following order: BI, Aact, SN, followed by a random presentation of corresponding pairs of B s and e s, and NB s and Mc s.

Behavior Measure

Total home energy consumption was defined as the amount of electricity used by a household over a specified period of time. The electrical meters (located on the outside of the condominium) were read twice; once at the beginning of the study, and again at the end of the six week period. Home energy consumption was measured by subtracting the initial reading from the final reading at six weeks.


Because of the difficulty in determining which member of a couple had more influence over household energy use (Seligman et al. 1979), the responses of husbands and wives were averaged to form the unit of analysis. This procedure did not seem unreasonable since husbands' scores correlated significantly with wives' scores on 20 out of the 28 items dealing with the model.

Analyses were conducted to see whether the location of the unit (i.e., interior vs. exterior) affected energy consumption. Subjects (n=14) who lived in exterior units did not use more electricity than subjects (n-34) who lived in interior units, F(1,47)=1.23. Furthermore, there were no differences between the two groups in the physical energy conservation measure taken, F(1,47)<1, the frequency with which lights were turned off, F(1,47)=2.13, and the amount of hot water used in daily activities, F(1,47)<1. Thus, the location of the house was not a factor in subsequent analyses.

Relationships Between Components of the Model

To test the assumption that BI is a function of Aact and SN, a standard multiple regression analysis was performed. As expected, Aact and SN accounted for a significant proportion of the variance in BI (R=.60, p .001). Yet of these two variables, only Aact contributed significantly to the prediction accuracy of the regression equation. In other words, a couples' attitude toward home energy conservation for the next six weeks predicted intentions to engage in this behavior, whereas their subjective norms did not.

The global measure of Aact was, as anticipated,correlated with SBi ei (r=.68, p<.001). Similarly, the global measure of SN correlated with SNB Mc (r= .5 1, p< .001).

The correlation between the total amount of energy used during the entire six week period and behavioral intention was marginally significant, r=.23, p<.06. Furthermore, when subjects' responses to the question, "do you consider yourself an energy conserver" were partialled out, this correlation was strengthened, r=34, p< .009.

Differences Between Conservers and Nonconservers

The degree to which subjects could be considered conservers was determined by summing their responses to three questionnaire items: do you intend to conserve energy in the-next six weeks, are you an energy conserver, and how does your energy consumption compare to that of other households.- On the basis of their total scores, subjects were divided into two groups; those above the median were defined as conservers and those below as nonconservers. Differences between these two groups in terms of their beliefs about the consequences of energy conservation (Bi), their evaluations of these consequences (ei), their normative beliefs (NB), and their motivation to comply with referents' wishes (Mc) were then tested with Hotelling's T2 statistic.

Conservers displayed a different pattern of beliefs compared to nonconservers, T2=18.47, F(6,41)=9.74, p<.05. As Table l(a) shows, conservers were more likely to believe that home energy conservation would increase personal comfort (p<.05), reduce personal waste (p<.01), and not be particularly bothersome (p<.05). In other words, conservers appeared to have a more positive attitude toward energy conservation than did nonconservers. The two groups shared similar opinions about the effects of energy conservation on health, and the likelihood that conservation would save money. Furthermore, each group believed that home energy conservation would help, although not a great deal, to solve the energy crisis.

Overall, conservers and nonconservers evaluated these beliefs similarly, T2=10.32, F(6,41)=1.53, n.s. Never the less, if one examines the evaluation of each belief separately, then differences between the two groups become apparent. Specifically, conservers rated both saving money and being comfortable more positively than nonconservers (p<.05 for each). As Table (1B) shows, the two groups did not differ in their evaluation of the other beliefs.

Hotelling's T2 statistic failed to uncover any over all differences between conservers and nonconservers in their normative beliefs, T2=7.49, F(6,41)=1.11, n.s., or in their motivations to comply with these beliefs, T2=3.88, F(6,41)<1, n.s. Nor were any differences found when the individual items were examined. These findings are consistent with earlier analyses which found that normative beliefs did not predict significantly any of the variance in BI.






The present results confirmed the proposed relationships between the components of the Fishbein-Ajzen model. Behavioral intentions were significantly predicted from a multiple regression of attitudes toward conservation and subjective norms, with attitudes toward conservation being the better predictor. The correlation between behavioral intention and energy consumption was marginally significant, but, admittedly, lower than one would have hoped for. This raises the question as to whether correlations in the .2 to .3 range are accurate representations of this relationship or whether there are additional concerns that must be taken into account to increase the relationship. The correlation between behavioral intention and energy consumption found in the present study is similar to the correlation found between attitudes and energy consumption by Becker, Seligman, Fazio, and Darley (1981). Both the present study and their study were conducted during the winter. The only studies to report high correlations between energy consumption and attitudes were conducted during the summer (Seligman, Kriss, Darley, Fazio, Becker, and Pryor 1979). One explanation for the summer/winter difference offered by Becker et al. (1981) has to do with the smaller range of energy consumption found in the winter (2-1) compared to the summer (3-1) in Northeast climates, All else being equal, truncated ranges weaken the correlation. Thus it is conceivable t had the correlation between energy consumption and attitudes or behavioral intentions-is underestimated by winter studies .

One might also consider whether energy consumption is the best measure to use to examine the attitude or intention -behavior relationship. That is, an individual's energy consumption is the result of many, different household behaviors, e.g., thermostat setting, installing insulation, using appliances, etc. Attitudes or intentions may be good predictors of these specific behaviors and still be relatively poor predictors of overall consumption. This is because there are other factors influencing the energy consumption of the house besides the behaviors of the individual whose attitudes and intentions we may have measured. For example, the energy consumption behavior of other people in the house (children, guests), unanticipated absences from the house or vice versa, and unusually cold weather which minimizes the differences between peoples' conservation responses all play a role.

The Fishbein-Ajzen model also allows us to contrast conservers and non-conservers regarding their beliefs about the consequences of conserving energy and their evaluations of these beliefs. Conservers viewed home energy conservation as more likely to increase comfort, reduce personal waste, and be less bothersome than did non-conservers. As well, conservers (vs. non-conservers) evaluated being comfortable and saving money more positively. The results concerning comfort are particularly interesting because they confirm the earlier findings of Becker et al. (1981) and Seligman et al. (1979) that attitudes toward personal comfort are an important consideration in home energy conservation. It should be pointed out that in the present study, beliefs and evaluations were elicited spontaneously from subjects, unlike the earlier research in which subjects responded to items generated by the experimenter.

One practical implication of these results is that attempts to reduce household energy consumption need not focus exclusively on peoples' attitude toward energy conservation. Rather, attention can also be directed at changing the relevant beliefs and evaluations that determine the global attitude toward conservation. Finally, it appears that the beliefs of other significant people have very little impact on the respondent's intention to conserve energy.


Ajzen, Icek (1982), "On behaving in accordance with one's attitudes", in Mark Zanna, E. Tory Higgins and C. Peter Herman Consistency in Social Behavior, The Ontario Symposium, Vol. 2, Hillsdale, New Jersey: Lawrence Erlbaum Associates, Publishers.

Becker, Lawrence J., Seligman, Clive, Fazio, Russell, H and Darley, John, M. (1981) "Relating Attitudes to Residential Energy Use," Environment and Behavior, 13, 590-609.

Bowman, Carole and Fishbein, Martin (1978) "Understanding Public Reaction to Energy Proposals. An Application of the Fishbein Model," Journal of Applied Social Psychology, 8, 319-340.

Fishbein, Martin and Ajzen, Icek (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison-Wesley.

Fishbein, Martin and Jaccard, James (1973), "Theoretical and Methodological Considerations in the Prediction of Family Planning Intentions and Behavior," Representative Research in Social Psychology, 4, 33-51.

Hass, J.W., Bagley, G. S. and Rogers, R. W. (1975), "Coping with the Energy Crisis: Effect of Fear Appeals upon Attitudes Toward Energy Consumption," Journal of Applied Psychology, 60, 754-756.

Heberlein, Thomas, A and Black, J. Stanley (1967), "Attitudinal Specificity and the Prediction of Behavior in a Field Setting," Journal of Personality and Social Psychology, 33, 474-479.

Kiesler, Charles A., Collins, Barry E. and Miller, Norman (1969), Attitude Change, New York: John Wiley & Sons, Inc.

McDougall, Gordon, H. G., Claxton, John D., Ritchie, J. R. Brent and Anderson, C. Dennis (1981), "Consumer Energy Research: A Review," Journal of Consumer Research, 8, 343-354.

Olsen, Marvin E (1981), "Consumers' Attitudes Toward Energy Conservation," Journal of Social Issues, 37, 108-131.

Otway, H. J. and Fishbein, Martin (1976), "the Determinants of Attitude Formation: An Application to Nuclear Power," Research-memorandum, International Institute for Applied Systems Analysis.

Seligman, Clive, Kriss, Mitchell, Darley, John, M., Fazio, Russell, H. Becker, Lawrence, J and Pryor, John (1979), "Predicting Summer Energy Consumption from Homeowners' Attitudes," Journal of APPlied Social Psychology, 9, 70-90.

Socolow, Robert H (1978), Saving Energy in the Home, Cambridge, Mass: Ballinger Publishing Company.

Wicker, Alan, W (1969), Attitudes versus Actions: The relationship of verbal and overt behavioral responses to attitude objects," Journal of Social Issues, 25, 61-78.

Zanna, Mark P., Higgins, E. Tory and Herman C. Peter (1982), Consistency in Social Behavior: The Ontario Symposium Vol. 2, Hillsdale, New Jersey: Lawrence Erlbaum Associates.