An Examination of the Psychometric Properties of a Conservation-Oriented Consumption Scale

Katryna Malafarina, University of Minnesota
Jeff Jass, University of Minnesota
ABSTRACT - Due to increased academic focus on measuring attitudes and behaviors concerning energy conservation matters, the psychometric properties of a "Conservation-Oriented Consumption" scale (Allen, Calantone and Schewe, 1982) are assessed. A review of the development of the current measure reveals the basis for a four factor conceptualization ("general feelings of effectiveness", "perceived change in consumption", "perceptions of others" and "locus of blame") of conservation-oriented consumption proposed by Allen et al. The present study supports a three factor solution and demonstrates the reliability, discriminant and construct validity of the scale.
[ to cite ]:
Katryna Malafarina and Jeff Jass (1994) ,"An Examination of the Psychometric Properties of a Conservation-Oriented Consumption Scale", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 536-542.

Advances in Consumer Research Volume 21, 1994      Pages 536-542

AN EXAMINATION OF THE PSYCHOMETRIC PROPERTIES OF A CONSERVATION-ORIENTED CONSUMPTION SCALE

Katryna Malafarina, University of Minnesota

Jeff Jass, University of Minnesota

ABSTRACT -

Due to increased academic focus on measuring attitudes and behaviors concerning energy conservation matters, the psychometric properties of a "Conservation-Oriented Consumption" scale (Allen, Calantone and Schewe, 1982) are assessed. A review of the development of the current measure reveals the basis for a four factor conceptualization ("general feelings of effectiveness", "perceived change in consumption", "perceptions of others" and "locus of blame") of conservation-oriented consumption proposed by Allen et al. The present study supports a three factor solution and demonstrates the reliability, discriminant and construct validity of the scale.

In his State of the Union Address on February 15, 1993, President Clinton gave Americans a prelude to his proposed economic plan in which the public discovered that they may be subjected to a broad-based energy conservation tax which would impact all forms of energy, including petroleum, natural gas, coal, hydroelectricity and nuclear power. One implication of this proposed tax is the need for Americans to conserve energy in many areas of life; from home heating to gas consumption. This issue, once a hot topic for researchers in marketing in the late 1970's and early 1980's, is once again attracting interest within the field (Malafarina and Loken 1993).

With these factors in mind, it appears to be an appropriate time to revitalize the impact that research in marketing exercises in this area. Several methodological alternatives, including experimentation and survey research, are available in accomplishing this objective. When the latter route is chosen, it is critical for the administered tests to exhibit acceptable levels of reliability and validity.

This study recognizes the importance of solid measurement practices if this area of marketing is to be revitalized successfully. Malafarina and Loken (1993) conducted a content analysis of social marketing studies (half of which related to environmental and energy conservation issues) and determined that the area is underdeveloped in terms of methodology. This study evaluates the reliability, discriminant validity and convergent validity of a scale created by Allen, Calantone and Schewe (1982) which assesses consumers' attitudes toward energy conservation through the construct "conservation oriented consumption". This scale was chosen for two reasons. First, academic literature dealing with measurement of energy conservation topics is quite limited. The "conservation oriented consumption scale" was one of the few pieces which reported its actual scale and fully described the conceptualization behind the scale. Second, the scale is an extension of an empirically tested scale created by Allen (1982). The second scale (Allen, Calantone and Schewe 1982) replicated the findings from the first scale (Allen 1982) and furthermore created new items which tapped an additional construct.

This paper will briefly review the literature addressing energy issues. Next, the "conservation oriented consumption" scale will be described and the theory behind it will be explored. Furthermore, the psychometric properties of the scale will be assessed. Finally, the relevance of the scale in relation to academic pursuits will be discussed.

LITERATURE REVIEW

Many studies were conducted a decade ago assessing varying aspects of consumers' attitudes and interests toward energy conservation. The December, 1981 issue of the Journal of Consumer Research was devoted entirely to studies addressing energy conservation and the environment. Other studies in marketing literature have included analyzing energy consumption patterns by the stage of the family life cycle (Fritzsche 1981), using the Fishbein-Ajzen model to predict home energy conservation (Seligman, Hall and Finegan 1982) and assessing public policies on energy conservation (Rudelius, Weijo and Dodge 1984).

Two interesting results can be explicated from these studies. First, the extent to which an individual accepts personal responsibility for conserving energy is an important factor influencing attitudes and behavior (Allen, Callentone and Schewe 1982; Allen 1982; Awad et al 1982 and Ellen, Wiener and Cobb-Walgren 1991). Second, conservers and nonconservers do not share the same attitudes and beliefs (Seligman, Hall and Finegan 1982; Allen 1982; Allen, Callentone and Schewe 1982).

Development of Conservation-Oriented Consumption Scale

The scale utilized by Allen, Calantone and Schewe (1982) in their study measuring international consumers' Conservation-Oriented Consumption is composed of four unique dimensions: "Perceived Change in Consumption" (PCC), "General Feelings of Effectiveness" (GFE), "Locus of Blame" (BLAME) and "Perceptions of Others" (OTHERS). Previous research has identified these factors as strong indicators of Socially Conscious Consumption (SCC) (Kinnear et.al 1974; Webster 1975; Allen and Dillon 1979; Allen 1982; Millstein 1977; Yankelovich 1974). An examination of these earlier studies provides insight into the focus of the current measure of interest.

Perceived Consumer Effectiveness

Berkowitz and Lutterman (1968) define a Socially Responsible Person as one who participates in community activities, strives to meet her obligations, possesses the ideals of the American core culture and feels unalienated from society. In their study the authors attempt to create an understanding of the construct "Social Responsibility" by reporting a number of behavioral and attitudinal correlates to their "Social Responsibility Scale" (SRS). Their results demonstrate the measure's ability to identify subjects possessing traditional, conservative American ideals. Of particular interest is the socially responsible citizen's belief that his individual actions influence events in the community as a whole. Persons possessing social responsibility hold fast to the belief that their own actions influence their surrounding community. This concept is reflected in later work as Perceived Consumer Effectiveness.

Attempting to improve upon this research, Kinnear, Taylor and Ahmed (1974) implement their "Index of Ecological Concern" to explore the characteristics of consumers that contribute to socially conscious purchasing patterns. Ecological concern is operationally defined as consisting of two dimensions: Buyers' attitudes must express concern for the ecology while purchase behavior must be consistent with maintenance of the environment (Kinnear and Taylor 1973). Perceived Consumer Effectiveness, defined as the extent to which the respondent believes that an individual consumer can be effective in pollution abatement, is found to be a powerful indicator of ecological concern. "Those who felt strongly that consumers could be useful in pollution abatement demonstrated higher than average concern...as consumers perceive that individuals can be increasingly effective in pollution abatement, they will show more concern for the ecology" (Kinnear et. al. 1974, p. 22). Consumers who felt that individual action was influential in the broader arena exhibited greater concern for the issue at hand.

In his conceptualization of the Socially Conscious Consumer, Webster (1975) departs from the "traditionally" responsible citizen depicted in the Social Responsibility Scale (Berkowitz and Lutterman 1968). According to Webster, a Socially Conscious Consumer is one who "takes into account the public consequences of his or her private consumption or who attempts to use his or her purchasing power to bring about social change" (p.188). Perceived Consumer Effectiveness is envisioned in this study as an attitudinal variable. High levels of PCE reflect an attitude held by individuals who view the world as an interconnected system where the actions of one person have an influence on the entire system. Perceived Consumer Effectiveness demonstrates a significant influence on the behavioral construct SCC: Individuals scoring high on Perceived Consumer Effectiveness reported participating in more socially conscious behaviors than subjects exhibiting low PCE. In addition, Webster (1975) offers a full description of the Socially Conscious Consumer, profiling the group's characteristics beyond PCE. These consumers are accepting of the views of others, are engaged in behavior that is somewhat counter to social norms, are somewhat insensitive to social pressures and are willing to engage in purchase behavior that may not be popularly accepted but is consistent with their own ideas of responsibility. It is argued that while the social responsibility scale identifies traditional social values, the socially conscious consumption measure taps "modern" consumer values. It is interesting to note that "perceived consumer effectiveness" is the only researched construct to demonstrate a significant relationship with both of these measures. Because of its focus on actual behavior, Webster (1975) recommends that subsequent research concentrate on examining the dimensions of socially conscious consumption.

In accord with the recommendation of Webster (1975), Scott (1977) continues the stream of research focusing on socially conscious consumption. She does, however, introduce an additional dimension into the conceptualization of the construct. Arguing that traditional persuasive messages may produce only marginal changes in perceived consumer effectiveness, Scott (1977) introduces self-perception based strategies into the SCC arena. Specifically, Scott demonstrates the effectiveness of the foot-in-the-door technique in influencing both verbal and overt socially conscious behavior. Subjects who agreed to display a "Conserve Resources-Recycle" sign in their window were more likely to perform a subsequent envelope-addressing task. The importance of this study lies in the reconceptualization of PCE as a self-perception variable.

Self Perception and Conservation Issues

Self-Perception theory is an attributional approach to understanding attitudes. Bem (1972) originally discussed the theory, stating that to the extent that external causes for an act are absent, the individual engaged in the act will infer his attitude toward the topic on the basis of his behavior.

Belk, Painter and Semenik (1981) also demonstrate that if consumers display an internal locus of control (a belief in personal responsibility for their lives) rather than an external locus of control (a belief that chance or others control their lives) they generally display more ecological activism. The overall perspective from these two studies is that if a consumer believes individuals have the ability to solve energy problems, then they are more likely to report positive attitudes towards energy conservation.

Allen (1982) suggests that energy related issues may represent a situation where consumers are highly receptive to self-perception based labeling approaches. In his study those exposed to an attribution message perceived they had made more changes in their consumption to help with the energy problem than those in the persuasive condition. These results support a viewpoint in which PCE incorporates a self-perception dimension.

Allen and Dillon (1979) incorporate both a self-perception view and the more traditional conceptualization of Perceived Consumer Effectiveness in their scale measuring socially conscious consumption. Rather than suggesting that the reconceptualized interpretation of PCE may in fact be an entirely unique construct, Allen and Dillon propose a dual-dimensional model of Perceived Consumer Effectiveness. One dimension, "perceived change in consumption" (PCC), is envisioned as the self-perception component of PCE as it measures an "individual's self-perceptions that he/she has made in consuming and using products as a response to a specific social problem" (Allen and Dillon 1979, p. 551). It is argued that socially conscious consumers are more likely to perceive that they have changed their consumption habits in order to help alleviate the social problem. The second PCE dimension, here labeled "concern and effectiveness", measures an individual's general attitude of concern about a given social issue, as well as the person's own feelings of effectiveness in responding to the problem. (Allen 1982 subsequently labels this dimension "general feelings of effectiveness".) This dimension corresponds to the conceptual definitions applied to the entire construct, perceived consumer effectiveness, as envisioned by Kinnear et. al (1974) and Webster (1975).

Allen and Dillon (1979) constructed a measure of PCE from a pool of thirty-six items believed to tap the various dimensions of socially conscious consumption. Using factor analysis with Varimax rotation the item pool was reduced to a manageable fifteen item scale. Four factors were generated from the analysis: OTHERS (25%), BLAME (16.2%), PCC (12.6%) and Concern and Effort (8.4%) (numbers in parentheses reflect percentage of common variances).The "perceptions of other" (OTHERS) construct is based on an argument presented by Yankelovich (1974). In his study Yankelovich posits a model that outlines five stages of concern individuals experience when considering social issues that have cost consequences. A person at the fourth stage is concerned about the relevant issue yet will fail to take action if it is perceived that others are not acting in a responsible way. Similarly, Belk, Painter and Semenik (1981) report that individuals are more likely to conserve energy if they believe that other individuals are also conserving energy. The "perception of others" construct, therefore, identifies subjects' outlook concerning how others consume in socially conscious manners.

The "locus of Blame" (BLAME) construct is adopted from a study completed by Milstein (1977). Reporting results from a number of focus groups, Milstein concludes that consumers do not make increased efforts at energy conservation due to skepticism and cynicism. He reports that approximately 67% of subjects surveyed blame either big business, Politicians or the government for any energy problems. Allen and Dillon (1979) and Allen et. al (1982) reason that to the extent that individuals blame these parties for creating the energy problem, they are not likely to consume in an energy conscious way.

In a later study, Allen (1982) reports results from a factor analysis with Varimax rotation for a similar twelve item scale measuring socially conscious consumption. In this paper, Allen adheres to his conceptualization of perceived consumer effectiveness by including both PCC and GFE in his scale. He also replaces BLAME with an additional "Others" construct. The factors, labels, eigenvalues and percent of variance explained are reported as follows (Factor, label, Eigenvalue, % of Variance): 1, GFE, 2.364, 19.7%; 2, OTHERS, 1.934, 16.1%; 3, Others must act, 1.19, 9.9%; 4, PCC, 1.02, 8.5%.

This latter study more convincingly supports the proposition suggested by past research that perceived consumer effectiveness is strongly tied to SCC (28% of variance explained by PCE in 1982 study versus 12.5% in 1979 study).

Projective Techniques

Because energy consciousness is deemed as "politically correct", many individuals will respond to measures with a positive bias which may not reflect their true attitudes. Therefore, scales measuring energy conservation attitudes should try to incorporate items which may indirectly assess consumers' attitudes so that they will respond with their true attitudes and not those deemed as socially desirable. These types of items can be obtained through projective techniques and are included on this scale.

Although Allen, Calantone and Schewe (1982) do not mention projective techniques specifically, we believe they do play a role in two of the subscales: Locus of Blame and Perception of Others. Projective techniques are often used in measurements designed to assess responses which may have social implications (Aaker and Day 1986). If individuals believe that a response may reflect upon their self image in some manner or that giving a certain response may be deemed socially appropriate, they may not give accurate answers. Third person techniques are particularly useful for revealing true attitudes (as compared to socially desirable ones). "Third person techniques usually ask what friends, neighbors or the average person would think about a particular situation. The researcher can then observe, to some extent, the respondent's projection of their own attitudes onto the third person, thus revealing more of their "true feelings" (Aaker and Day 1986). The Perception of Others subscale is particularly subject to this type of technique.

The Conservation Oriented Consumption Scale

The preceding logic provides the basis for the conceptualization of the measure examined in this study (Allen et. al. 1982). The four identical factors utilized by Allen and Dillon (1979) (GFE, PCC, OTHERS, and BLAME) are presented as indicators of conservation-oriented consumption (a new label for socially conscious consumption). In addition, the dual-dimensional structure of perceived consumer effectiveness is adhered to. Seventeen items are included in the measure with six items measuring GFE (G1 - G6), two measuring PCC (P1 and P2), five tapping BLAME (B1 - B5) and four measuring OTHERS (O1 - O4). Alpha coefficients are reported for the four unique samples drawn for the study as an indication of reliability. Alpha's ranged from .65 to .69 for GFE, .42 to .48 for BLAME, and .43 to .60 for OTHERS. These coefficient values are offered as indicating an acceptable level of reliability for the scales.

EMPIRICAL MEASUREMENT ANALYSIS

Sample

The Conservation Oriented Consumption scale (Allen et. al. 1982) as well as the convergent and discriminant scales (Awad et. al. 1982) were administered to 192 undergraduate and graduate business students: 143 undergraduates and 49 graduates. A one-way ANOVA with unequal sample sizes was conducted to ensure that the respondents were similar in their responses. An overall F-test revealed that the null hypothesis was non-significant, indicating that the sample means were equal and that subjects' responses could be viewed as equivalent. Because the p-value approached significance (p=.051), orthogonal contrasts were run on the classes to determine if there were any differences between graduate and undergraduate students or between undergraduate classes.

No significant differences were found between the four undergraduate classes, but a slight significant difference was found between the graduate and undergraduate classes. Specifically, the graduate classes had a slightly higher mean then the undergraduate classes on the overall scale. It was hypothesized that the differences in mean score were due to the age difference in the students. The older (graduate) students are likely to have more experience with energy consumption behaviors and more awareness of energy conservation issues. To examine these possibilities, we conducted a discriminant analysis on the following: awareness of a potential gas tax in the state, awareness of a possible energy conservation tax, purchase experience with compact refrigerators, purchase experience with microwave ovens and purchase experience with new or used cars. The authors anticipated that there might be some potential differences between graduate and undergraduate students and therefore included questions assessing these five items at the end of the questionnaire. The two analyses pertaining to potential taxes were conducted solely to determine whether or not there were any significant differences between students in their knowledge of related energy issues. The three questions relating to purchase behaviors were analyzed solely to determine whether any significant differences existed in experience with purchasing products that are affected by consciousness of their impact on energy conservation (ie-larger cars consume more gas than do smaller cars and different appliances vary in the level of electricity they consume). Two significant differences were obtained; graduate students indicated a higher level of awareness than did undergraduates and also had greater purchase experience with microwave ovens. Additionally, a one-way ANOVA was conducted on both the discriminant and convergent scales (to be discussed later) between all the classes and no significant differences were obtained. Because the sample was relatively equivalent, the sample was randomly split to conduct an exploratory and confirmatory factor analysis. Although there is a slight difference between undergraduates and graduates, this issue is not a problem since both groups are still represented in each sample.

Reliability

Several methods exist for estimating reliability including, test-retest, equivalent forms and split-halves. We chose to examine reliability through split halves, using Cronbach's Alpha as a measure of reliability. Coefficient Alpha is "the expected correlation of one test with another test of the same length when the two tests purport to measure the same thing" (Nunnally1978, p.215). From table 2, alpha's appear to be reasonable (Nunnally 1978) and are equal to, and often greater than, the coefficients reported by Allen. This provides substantial evidence for the reliability of the scale. (Note the items included in particular runs, however, and how alpha values changed).

Item-to-total correlations and inter-item correlations

Item-to-total correlations were examined to determine which items could potentially be dropped from the scale. On an initial analysis, it was apparent that items B1 and B3 (the first and third statements on the BLAME scale) should be dropped from the scale because their Item-to-total correlations were -.3199 and -.1859, respectively. Additionally, the "Alpha if Item-deleted" column on the analysis indicated that alpha would be increased if these items were dropped. Furthermore, an analysis of the correlation matrix revealed consistently low or negative correlations with these items. A second reliability analysis was then run, eliminating items B1 and B3, which increased alpha to .7849. An analysis of the output indicated that B2 (second statement on the BLAME scale) could also be dropped due to low Item-to-total correlations and an indication that alpha would be increased if the item were deleted. This item was dropped and another reliability run was conducted which further increased alpha to .7931. In addition to these empirical indicators, we referred back to the theory establishing the BLAME construct for further direction. From examination of the theory , the items, and the responses, it was concluded that subjects may not have comprehended the wording of these items. The reliability analysis, therefore, identified items B1, B2 and B3 as possible items to be excluded. Items G2 (second item on the GFE scale) and item O1 (first item on the OTHERS scale) could have potentially been deleted due to low Item-to-Total correlations and low inter-item correlations, but were retained due to their consistency with the theoretical basis (Nunnally, 1978).

TABLE 1

CONSERVATION ORIENTED CONSUMPTION (ALLEN, CALATONE, AND SCHEWE, 1982)

TABLE 2

Structure of Scale-Exploratory Analysis

To conduct our exploratory stage of factor analysis, we used two extraction methods; Alpha and Principal Axis Factoring (PAF). These two methods were chosen from the available methods because both are good methods for assessing the underlying dimensions of a scale. Varimax rotation was used for both factoring methods (Kim 1975).

Six factors were identified from the initial factor run that exhibited eigenvalues greater than one. Because six factors were not consistent with the theory and because an analysis of the factor loadings revealed no consistent pattern of interpretation, we forced an additional run to four factors. These runs revealed that the factor loadings could still not be interpreted in accordance with theory. In fact, no meaningful interpretation was possible with the loadings. B1, B2 and B3 in particular demonstrated no underlying commonality amongst themselves or with the other BLAME items. For instance, in a four factor PAF solution B1 loaded on factor four, B2 loaded on factor two, and B3 loaded on factor one. Consistent with our reliability analysis, factor analysis recommends that these items be eliminated from the scale. Subsequent analyses were run without these items.

TABLE 3

At this point the authors examined the theory underlying the scale and concluded that the problem with the factor analysis may have been due to the two PCE subscales. As previously discussed, Allen and Dillon (1979) were the first researchers to propose this two-dimensional view of perceived consumer effectiveness. Additionally, only work completed by Allen and his colleagues (Allen et. al, 1982; Allen, 1982) supports this conceptualization. From our analysis, we decided to run another factor analysis with three factors. When we ran this factor analysis, our loadings produced meaningful interpretations: All of the items loaded clearly onto theoretically expected factors. From this analysis we concluded that only three factors were meaningful and that the PCE scale could not be delineated into two subscales. The three factor solution is further explained when one considers the two item construction of the PCC subscale. Additional indicators are recommended for this construct. The PCE items loaded on the first factor, the BLAME items loaded on the third factor and the OTHERS items loaded on the second factor. The following results from a factor analysis restricted to three factors and utilizing PAF extraction and Varimax rotation exemplify our findings. Results are reported as follows (Factor number, label, Eigenvalue, % of Variance explained): 1, PCE, 3.51, 25.1%; 2, OTHERS, .9, 6.4%; 3, BLAME, .77, 5.5%.

As our results indicate, the PCE factor exhibits the largest Eigenvalue. Consistent with previous research, PCE is determined to be a powerful indicator of socially conscious consumption.

Validity of the Scale

We specifically addressed construct validity, discriminant validity and convergent validity. Construct validity concerns a hypothesized relationship between a measure of a construct and a particular observable variable (Nunnally 1978). Peter (1981) further states that "if a construct were hypothesized to have three dimensions, a factor analysis of a purported measure of the construct which produces three meaningful factors could be interpreted as supportive evidence of construct validity." Our factor results support the conceptualization of PCE as a single dimensional construct. While Allen et. al purport that PCE contains two dimensions, our results suggest that PCC and GFE are too closely related to be considered separate entities. Construct validity is established for this conceptualization (Webster, 1975; Kinnear et. al, 1974) of the latent variable.

Discriminant validity reflects the extent to which the measure is unique and not a reflection of other variables and is determined by low correlations between the measure of interest and other measures that are not measuring the same variable or concept (Campbell and Fiske 1959). Discriminant validity was assessed through correlations with a Cynicism and a Home Economics scale created by Awad et al. (1982). The Awad et al scale was used for validation purposes for two reasons. First, as mentioned previously, there is a paucity of reported scales dealing with energy conservation attitudes. Although some scales have been used to measure energy consumption patterns by stage of family life cycle (Fritzche 1981), scales such as this were not appropriate for validation purposes because they did not specifically deal with attitude measurement. The Awad et al scale was intentionally designed to measure attitudes. Second, on a conceptual basis, the Awad et al. scale could be divided into subscales to use for both discriminant and convergent validity. In Awad et al.'s study, cynicism and home economics were shown to be negatively related to energy conservation. Table 3 displays the correlations between the portions of the Allen et. al scale and the discriminant and convergent scales. As is evidenced, support for discriminant validity is established.

Convergent validity measures the degree to which attempt to measure the same concept using two or more different measures yield the same results and is determined by high correlations between the measures (Campbell and Fiske 1959). Convergent validity was assessed through correlations with two scales created by Awad et al (1982): Concern for Supply and Avowal of Social Norms. Both of these scales were shown to be positively related to energy conservation.

As is made apparent by the above low correlations between scales, convergent validity was not established. Although convergent validity was not established, an analysis of the "Concern for Supply" scale and the "Avowal of Social Norms" scale (Awad et. al. 1982), provides a few explanations for the current results. First, Awad et al. initially administered the scale to individuals in association with five specific behaviors: use of air conditioning, use of washer/dryer, use of dishwasher, use of heating and use of lights. For example, the statement "I am always careful about my use of electricity" was rated five times, once for each type of behavior. Subjects in this sample were only given the statement to rate once, and the subject was left to interpret the meaning of the behavior. Therefore, the subject may have viewed this statement as reflecting a "general" behavior or may have viewed the statement in relation to only one specific behavior such as using the dishwasher. This type of interpretation could have occurred for any of the three "Avowal of Social Norms" questions. Since the scale was initially meant to assess specific behaviors, any misinterpretation of the question could lead to lower convergent validity.

TABLE 4

Second, the items in the "Concern for Supply" scale use terminology which may confuse the subjects. For example, subjects could respond to this statement, "The most important reason for conserving electricity is to be sure there will be an adequate supply for our children" in several ways, depending upon their interpretation of the salient aspects of the statement. If a subject believes that ensuring an "adequate supply for our children" is an important, but not the most important reason for conserving energy, the subject would tend to disagree with this statement while in fact he/she may still believe that we need to maintain the supply for the future. Additionally, simply replacing "for our children" with "for ourselves in the future" could have an impact on how people respond to this question. The other two statements have similar problems in that while a subject may have a concern for supply, the subject's focus may be on concern for supply on a personal basis and not as a concern for future generations.

These three items seem to intertwine the aspects of concern for supply and future generations and ignore aspects related to concern on a more individual basis. Although the correlations from this study are positive, which is consistent with the positive relation Awad et al. found, they are not sufficiently large enough to justify convergent validity. The previous analysis of the individual items, however, demonstrates how wording in the Awad et. al scale may have produced our low correlations.

Confirmatory Factor Analysis

The second set of data were used to confirm the results of the exploratory stage. Comparable results were achieved using this split sample technique. The following output summaries detail the properties of the refined conservation-oriented consumption scale (without items B1, B2 and B3). Scale reliability is reported as follows (Scale items, coefficient alpha): Overall scale, .8069; PCE, .8167; BLAME, .6945; OTHERS, .6974.

It is apparent from the results that coefficient alpha for the overall and subscales increased over the exploratory analysis results, providing further evidence for the reliability of the scale.

Maximum Likelihood is the most common confirmatory factor analysis approach (Stewart 1981) and was therefore used for confirmatory purposes. The three factor model was supported, with every item loading on the same factor as in the exploratory analysis (PCE-factor 1; OTHERS-factor 2; BLAME-factor 3). One difference was noted, however. In this confirmatory run OTHERS was identified as the factor explaining the greatest percentage of variance. Results are reported as follows (Factor number, label, eigenvalue, % of variance explained): 1, PCE, 1.415, 10.1%; 2, OTHERS, 3.5, 25%; 3, BLAME, 1.31, 9.4%. Noticeably stronger Eigenvalues are produced in this run for OTHERS and BLAME than in the exploratory stage.

The convergent and discriminant validity tests support the results from the exploratory analysis. Discriminant validity is established while no support is forwarded for convergent validity (Table 4).

Conclusion

As conservation issues receive greater attention by the government, the media and academia, these topics will increasingly influence the products consumers purchase and the services they use. Measures such as Allen, Callantone and Schewe's (1982) scale measuring conservation-oriented consumption will prove useful in monitoring these shifting attitudes. We believe the scale to be relevant due to increased interest in the area on the part of marketing researchers and because of the salience of energy issues in the minds of American citizens produced, in part, by the recent passage of Clinton's budget plan which includes a 4.3 cent gas tax.

Anticipating the potential resistance to the new tax, Clinton has emphasized that Americans should be willing to accept personal environmental responsibility for their individual actions. However, Ellen, Wiener and Cobb-Walgren (1991) report that consumers high in PCE do not favor government intervention. This would seem to pose a conflict for Clinton, which could explain why Clinton has emphasized that Americans should take personal responsibility for actions. By focusing on the voluntary aspect of energy conservation in a positive manner, resistance to intervention may be dispelled (Ellen, Wiener, and Cobb-Walgren, 1991). Allen, Calantone and Schewe (1982) have identified a dimension, "perceived consumer effectiveness" (PCE) as an indicator related to personal responsibility. Whereas personal responsibility merely reflects an individual's willingness to accept accountability, PCE reflects a more active construct in which an individual not only takes responsibility, but also acts on it (Ellen, Wiener and Cobb-Walgren 1991). This construct (PCE) should help researchers and practitioners in determining which consumers will believe that their actions can make a difference and who should then behave proactively on their energy conservation concerns.

The present study contributes to this stream of research on conservation oriented consumption by providing an empirical test of several psychometric properties of Allen, Calantone and Schewe's (1982) measure of PCE and presenting evidence for reliability, construct validity and discriminant validity. While we feel that the study does not provide overwhelming evidence for these properties, it does support the use of the scale in the following areas: (1) it provides further evidence for the construct PCE, (2) it enhances the reliability of the scale from the findings of Allen et. al (1982), and (3) it provides evidence for the discriminant capabilities of the scale.

This study additionally supports the use of projective techniques in scales. Although Allen, Calantone and Schewe did not initially emphasize this aspect of their scale, it can be seen that the inclusion of these items accounted for a significant proportion of the variance.

This study does, however, recognize the need for additional empirical investigations concerning this measure. Future research may assume four distinct challenges. First, because the current results differ in regards to the dimensionality of PCE, future research may wish to further investigate the construct in an effort to establish its dimensionality (uni-dimensional versus two-dimensional). Secondly, an attempt to assess the nomological validity of conservation-oriented consumption should be examined in future work. Additionally, it is recommended that this scale be used to assess consumers' attitudes concerning potential energy conservation programs. Through these efforts stronger evidence for construct validity may be established for this measure. Finally, it is hoped that researchers examining other socially related issues will contribute to this area of scholarship in marketing by developing reliable and valid measures.

REFERENCES

Aaker, David A and George J. Day, (1986), Marketing Research,John Wiley and Sons, Inc: New York, 126-130.

Allen, Chris T. (March 1982). "Self-perception based strategies for stimulating energy conservation". Journal of Consumer Research. 8: 381-390. 5

Allen, Chris T., Roger J. Calantone and Charles D. Schewe (1982). "Consumers' attitudes about energy conservation in Sweden, Canada, and the United States, with Implications for policymakers," Journal of Marketing and Public Policy. 1: 57-67.

Allen, Chris T., William Dillon (1979). "On Receptivity to Information Furnished by the Public Policymaker: The Case of Energy", in 1979 Educator's Proceedings, Chicago, IL: American Marketing Association. pp. 550-556.

Anderson, Thomas W. and William Cunningham (1972). "The Socially Conscious Consumer," Journal of Marketing. 36: 23-31.

Awad, Ziyad A., Roger H. Johnston, Jr., Shel Feldman, and Michael V. Williams. (1982) "Customer attitude and intentions to conserve electricity,"Advances in Consumer Research. 10: 652-654.

Bem, Daryl J, (1972), "Self Perception Theory," in Advances in Experimental Psychology, ed. Leaonard Berkowitz, New York, N.Y.: Academic Press. pp 1-62.

Berkowitz, Leonard and Kenneth Lutterman (1968), "The Traditional Socially Responsible Personality," Public Opinion Quarterly, 32: 169-85.

Campbell, Donald T. and Donald W. Fiske, (March 1959) "Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix," Psychological Bulletin,56: 100-122.

Ellen, Pam S., Joshua L. Wiener, and Cathy Cobb-Walgren, (1991) "The Role of Perceived Consumer Effectiveness in Motivating Environmentally Conscious Behaviors," Journal of Public Policy and Marketing, 10(2): 102-117.

Fritzsche, David J. (1981) "An analysis of energy consumption patterns by stage of family life cycle," Journal of Marketing Research. 18: 227-232.

Kim, Jae-On, "Factor Analysis," SPSS Statistical Package for the Social Sciences, Second Edition, (1975) New York: McGraw Hill, Chapter 24.

Kinnear, Thomas, James R. Taylor and Sadrudin A. Ahmed (1974), "Ecologically Concerned Consumers: Who are They?," Journal of Marketing, 38: 20-24.

Malafarina, Katryna and Barbara Loken, (1993), "Progress and Limitations of Social Marketing: A Review of Empirical Literature on the Consumption of Social Ideas ", Advances in Consumer Research, forthcoming.

Milstein, Jeffrey S. (1977) "Attitudes, Knowledge, and Behavior of American Consumers Regarding Energy Conservation with some Implications for Governmental Action", Advances in Consumer Research Vol. 4, pp. 315-321.

Nunnally, Jum (1978), Psychometric Theory, New York: McGraw Hill.

Peter, Paul J. (1981), "Construct Validity: A Review of Basic Issues and Marketing Practices," Journal of Marketing Research, 18, 133-145.

Rudelius, William, Richard Weijo and Gary Dodge (1984), "Marketing energy conservation to homeowners: an action program from public policy research," Journal of Public Policy and Marketing. 3: 149-166.

Rummel, R.J. (1970), Applied Factor Analysis, Evanston: Northwestern University Press.

Scott, Carol, A. (1977), "Modifying Socially-Conscious Behavior: The Foot-in-the-Door Technique", Journal of Consumer Research. 4: 156-64.

Seligman, Clive, Don Hall, and Joan Finegan (1982), "Predicting home energy consumption: an application of the Fishbein-Ajzen model," Advances in Consumer Research. 10: 645-651.

Stewart, David W. (1981), "The Application and Misapplication of Factor Analysis in Marketing Research", Journal of Marketing Research. 18: 51-62.

Webster, Frederick E. (1975), "Determining the Characteristics of the Socially Conscious Consumer", Journal of Consumer Research. 2: 188-196.

Yankelovich, Daniel (1974), "How Opinion Polls Differ from Social Indicators," in Social Indicators and Marketing, Chicago: American Marketing Association, 54-66.

----------------------------------------