10
Alternative Hierarchies of the Attitude-Behavior Relationship: The Impact of Brand Commitment and Habit Sharon E. Beatty University of Alabama Lynn R. Kahle University of Oregon The theory of reasoned action and the low-involvement hierarchy model were tested using cross-lagged panel correlation. It was proposed that the low-involvement hierarchy model wouM-more accurately reflect the behavior of low-brand-committed individuals and that the theory of reasoned action wouM more accurately reflect the behavior of high-brand-committed individuals. Although the findings did not completely support these hypotheses, the general thrusts of the hypotheses appear to be supported Additionally, behavior was hypothesized and shown to influence habit for all individuals. INTRODUCTION One of the oldest and most important questions asked in social science and in consumer research is that of the relationship between attitudes and behaviors. This issue was addressed very early by LaPiere (1934). His research was interpreted as troubling because of the insufficient agreement between attitudes and actions, although this perspective has been questioned (Dillehay 1972). Through the years researchers have worked diligently to improve both the theory and measurement questions related to 9 1988,Academy of Marketing Science Journalof the Academyof MarketingScience Summer, 1988, Vol. 16, No. 2, 001-010 0092-0703 / 88 / 1602-0001 this issue, producing literally thousands of articles on the topic. The most widely examined paradigm on the attitude- behavior relationship is that of Fishbein's theory of reasoned action, wherein it is postulated that behavior derives primarily from intention, which derives primarily from a weighted combination of subjective norms and attitudes toward the act. The typical perspective of attitudes preceding behavior has been seriously challenged in some contexts, however (Bauer 1964; Bern 1972; Robertson 1971; Wright 1973). One perspective in consumer behavior (Krugman 1965) suggests that under conditions of low involvement or low concern, perceptual impact leads directly to behaviors, which subsequently influence attitudes. This perspective has led to a plethora of perspectives on the involvement construct and the postulation of a low-involvement hierarchy-of-effects model by some researchers (Calder 1979; Ray 1973; Robertson 1976). Another construct closely related to the attitude- behavior relationship which appears to have received little attention is habit. How is habit conceptually linked to attitudes and behaviors? It should be an important construct in better understanding the decision processes underlying changes in attitudes and behaviors for frequently purchased products. The purpose of this article is to investigate the relationship between attitudes and behaviors in the consumption of a frequently purchased product, specifically testing the applications of the theory of reasoned action and the low-involvement hierarchy model. These two attitude-behavior models will he assessed for two groups of individuals who exhibit either high- or low- brand commitment, using and illustrating cross-lagged panel correlation. In addition, the influence of habit in the attitude-behavior relationship will be assessed. JAMS 1 SUMMER, 1988

Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

Embed Size (px)

Citation preview

Page 1: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

Alternative Hierarchies of the Attitude-Behavior Relationship: The Impact of Brand Commitment and Habit

Sharon E. Beatty University of Alabama

Lynn R. Kahle University of Oregon

The theory of reasoned action and the low-involvement hierarchy model were tested using cross-lagged panel correlation. It was proposed that the low-involvement hierarchy model wouM-more accurately reflect the behavior of low-brand-committed individuals and that the theory of reasoned action wouM more accurately reflect the behavior o f high-brand-committed individuals. Although the findings did not completely support these hypotheses, the general thrusts of the hypotheses appear to be supported Additionally, behavior was hypothesized and shown to influence habit for all individuals.

INTRODUCTION

One of the oldest and most important questions asked in social science and in consumer research is that of the relationship between attitudes and behaviors. This issue was addressed very early by LaPiere (1934). His research was interpreted as troubling because of the insufficient agreement between attitudes and actions, although this perspective has been questioned (Dillehay 1972). Through the years researchers have worked diligently to improve both the theory and measurement questions related to

�9 1988, Academy of Marketing Science Journal of the Academy of Marketing Science Summer, 1988, Vol. 16, No. 2, 001-010 0092-0703 / 88 / 1602-0001

this issue, producing literally thousands of articles on the topic. The most widely examined paradigm on the attitude- behavior relationship is that of Fishbein's theory of reasoned action, wherein it is postulated that behavior derives primarily from intention, which derives primarily from a weighted combination of subjective norms and attitudes toward the act.

The typical perspective of attitudes preceding behavior has been seriously challenged in some contexts, however (Bauer 1964; Bern 1972; Robertson 1971; Wright 1973). One perspective in consumer behavior (Krugman 1965) suggests that under conditions of low involvement or low concern, perceptual impact leads directly to behaviors, which subsequently influence attitudes. This perspective has led to a plethora of perspectives on the involvement construct and the postulation of a low-involvement hierarchy-of-effects model by some researchers (Calder 1979; Ray 1973; Robertson 1976).

Another construct closely related to the attitude- behavior relationship which appears to have received little attention is habit. How is habit conceptually linked to attitudes and behaviors? It should be an important construct in better understanding the decision processes underlying changes in attitudes and behaviors for frequently purchased products.

The purpose of this article is to investigate the relationship between attitudes and behaviors in the consumption of a frequently purchased product, specifically testing the applications of the theory of reasoned action and the low-involvement hierarchy model. These two attitude-behavior models will he assessed for two groups of individuals who exhibit either high- or low- brand commitment, using and illustrating cross-lagged panel correlation. In addition, the influence of habit in the attitude-behavior relationship will be assessed.

JAMS 1 SUMMER, 1988

Page 2: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATYY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

THE THEORY OF REASONED ACTION

The theory of reasoned action has shown impressive heuristic utility in consumer behavior research, and researchers have learned much about attitudes and behavior from it. Much evidence supports the basic theory (Fishbein 1980), which asserts that the weighted combination of attitudes toward acts and subjective norms (which are like attitudes about others) lead to intentions, which in turn lead to behaviors. This conceptualization is illustrated in Figure la. But several additional considerations also appear noteworthy. For example, Mitchell and Olson (1981) have shown that attitude toward the advertisement can also influence intentions.

Because literally thousands of studies have dealt with the attitude-behavior relationship, a comprehensive review is beyond the scope of this paper. See Miniard (1981), Burnkrant and Page (1982) and Warshaw (1980) for more complete reviews on the adequacy of and the controversies surrounding Fishbein's model.

Habit

Arguing from social adaptation theory (Kahle 1984; Kahle, Kulka, and Klingel 1980), Wittenbraker, Gibbs, and Kahle (1983) and Kahle and Beatty (1987) found that habit is an important element in the attitude-behavior relationship, bringing into question the basic premise of the "reasoned action" aspect of Fishbein's theory. Triandis (1977) has also proposed that habit is an important element in attitude research, although he conceptualizes it differently than we do in the present study. A more standard definition of habit (cf. Mixon 1980) may be more useful in the present context. A habit is a well-learned schema with a behavioral component. Wittenbraker, Gibbs, and Kahle (1983, p. 410) describe the social adaptation theory rationale for conceptualizing habit:

This theory maintains that people develop subjectively adaptive schemata that guide behavior. As schemata are enacted more frequently in the context of stable perceptions

FIGURE la Conceptual Framework for the Theory of Reasoned Action

f I

Beliefs about consequences of

behavior X

I I

Normative Beliefs about behavior X

Attitude toward --4-

behavior X

Subjective norm § concerning

behavior X

4- Intention to perform behavior X

Behavior X

]lLfluence

Feedback

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

JAMS 2 SUMMER, 1988

Page 3: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

FIGURE lb Conceptual Framework for the Theory of Reasoned Action with the Addition of Habit

I I I I

Beliefs about consequences of

behavior X

Normative Beliefs about behavior X

!

I /

) Attitude toward behavior X

/ Subjective norm

r 8 behavior X

Intention to perform[___, Behavior X Habit X

Influence

Feedback

of subjective adaptivity, they begin to acquire characteristics of habits, such as consistency and resistance to change in the absence of external accommodative or internal organiza- tional forces.

Habit reflects a repetitively performed, stable behavior which is not actively deliberated upon at the time of the act. Therefore, for frequently purchased product categories, the inclusion of habit within the attitude- behavior relationship appears to be both theoretically and empirically important. Figure lb expands Fishbein's model by indicating that frequently repeated behaviors rasult in the formation of habit.

INVOLVEMENT AND COMMITMENT IN THE ATTITUDE-BEHAVIOR RELATIONSHIP

Involvement

The concept of involvement has received a good deal of attention in consumer research, where the emphasis has been on the nature of the construct's definition, dimensionality, and measurement (e.g., Bloch & Richins 1983; Houston and Rothschild 1977; Mitchell 1979; Smith and Beatty 1984; Traylor and Joseph 1984; Zaichkowsky 1985). Researchers have made various distinctions between types of involvement. For example, Beatty, Kahle, and Homer (forthcoming) make a distinction between ego involvement (importance of the product to the individual) and purchase involvement (concern for a particular

purchase). Houston and Rothschild (1977) made a similar distinction between enduring and situational involvement. The differences between these constructs are based on temporal perspectives, which are more relevant for major, discrete purchases than for frequently purchased products

where decisions are repeated over and over. Laurent and Kapferer (1985), providing an additional

perspective on involvement, established four antecedents or facets of involvement, including importance and risk consequences, risk probability, hedonic or pleasure value and the perceived sign value (identity of oneself with the product). These facets do not totally correspond with the enduring and situational involvement distinctions. They showed that these facets were differentially related to various behaviors. Of particular importance was their finding that all facets except hedonic value were significantly associated with extensive decision making.

The Low-Involvement Hierarchy Model

Fishbein's theory of reasoned action is somewhat similar to the hierarchy-of-effects model often used in consumer behavior and advertising to model the relationships between attitudinal elements and actions from an information processing perspective. This model came under attack by Krugman (1965) and others, who suggested that under conditions of low involvement. individuals may without "reasoned action" move directly from perceptual impact to behavioral enactment, and then to attitudinal processes.

JAMS 3 SUMMER, 1988

Page 4: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATIY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

Ray (1973) has provided one of the most complete explications of alternative hierarchy models. He suggested three hierarchies: (1) the standard learning hierarchy wherein a communication creates some cognitive change that gives rise to a change in affect and finally, a change in behavior (this model is typified by the Fishbein model), (2) a dissonance-attribution hierarchy in which behavior occurs first, followed by attitude change, then awareness and finally, comprehension, and (3) a low-involvement hierarchy, where awareness and minimal comprehension occur first, followed by behavior or trial and then attitude change.

Ray was able to provide a substantial degree of empirical confirmation with communication laboratory and field studies for both the learning hierarchy model and low involvement hierarchy model, but not for the dissonance model. These studies involved consumer reactions to repeated advertisements. Subjects were broken out by indirectly assessing involvement by observing (1) the number of times respondents indicated "don't know" when they evaluated a series of brands or (2) the number of points used on a brand rating scale. They found predicted differences based on these break-outs, although the connotative measures used were intention measures rather than measures of actual behavior.

As discussed by Robertson (1976), under high commitment the traditional learning hierarchy would be expected, whereas under low commitment the low- involvement hierarchy would be applicable. Robertson was defining commitment as the strength of the individual's belief system toward a product or brand. As discussed by Calder (1979), Krugman (1965), and Robertson (1976), the importance of the low involvement hierarchy model is the emphasis placed on behavior as the driving force behind attitudes. Calder believes that the key to low involvement lies in the feedback relationship between behaviors and attitudes. Thus, to understand this relationship it is necessary to study the impact of actual behavior rather than studying intention measures that serve as surrogates for behavioral measures.

Brand Commitment

A number of authors (Crosby and Taylor 1983; Lastovicka and Gardner 1977; Robertson 1976; Sherif, Sherif and Nebergall 1965; Traylor 1983) have linked the concept of commitment to involvement. Although some researchers have confused and overlapped the two constructs, we view them as two distinctive elements linked closely together. Brand commitment may be defined as an emotional or psychological attachment to a brand within a product class (Lastovicka and Gardner 1977) while involvement addresses a general level of interest or concern in an issue (i.e., with a product class or a purchase in that product class) without reference to a specific position or choice (i.e., brand).

Commitment appears to result from felt concern or ego involvement with the product or purchase decision (Beatty, Kahle and Homer forthcoming; Crosby and Taylor 1983; Zaichkowsky 1985; Zaltman and Wallendorf 1983). Crosby and Taylor (1983) suggest that a person first

becomes involved in an issue when important values or elements of one's ego are engaged. Later, these values or self images may be linked to a particular choice alternative (i.e., brand commitment). Beatty, Kahle, and Homer (forthcoming) established a strong linkage between ego involvement, purchase involvement, and brand commitment. Others also have established a definite link between involvement and commitment (cf. Selin 1986; Zaichkowsky 1985). 1

Brand commitment is conceptually similar to brand loyalty. A comprehensive review of brand loyalty has been provided by Jacoby and Chestnut (1978). These authors suggest that brand loyalty has been viewed and measured as a behavioral concept, an attitudinal concept, and simultaneously, a behavioral and attitudinal concept. They argue that composites combining both attitudes and behaviors more adequately capture the reality of the brand- loyalty construct. Jacoby and Chestnut (1978, p. 81) stress that "verbal reports of bias (i.e., statements of preference or intention to buy) are insufficient for defining brand loyalty." The emphasis in this paper is on attitudinal statements of commitment with no direct measures of behavior; thus the construct of brand commitment most adequately represents the construct of interest here.

Further, habit must be clearly distinguished from brand commitment. Jacoby and Kyner (1973) distinguish brand loyalty from repeat purchase behavior by stating that the former is a function of psychological (decision-making, evaluative) processes. Thus, this point stresses the conceptual thrust of this article - - high-brand commitment should be the result of reasoned action (i.e., behavior resulting from attitudes and intentions). On the other hand. habit, as defined here, reflects repeat purchase behavior of either a brand or a product class, which is not necessarily a direct function of psychological (decision-making, evaluative) processes.

HYPOTHESES

This study was designed to assess the applicability of the theory of reasoned action versus the low-involvement hierarchy model for individuals exhibiting either low- or high-brand commitment tendencies in their consumption of soft drinks. As indicated by Ray (1973), Calder (1979), Krugman (1965) and Robertson (1976), when individuals care little about their purchases, they will engage in minimal decision making, and attitudes will tend to flow from behavior. At high levels of ego or purchase involvement individuals will exhibit a reasoned action approach, that is, attitudes will lead to behavior.

As indicated in Laurent and Kapferer's (1985) study, and others (e.g., Houston and Rothschild 1977) greater involvement (based on a number of facets and incorporating both enduring and situational elements) leads to more extensive decision making (i.e., reasoned action is more tikely). 2 This perspective is also similar to Petty, Cacioppo and Schumann's (1983) findings that with higher involvement (defined as personal relevancy), individuals engaged in more extensive information processing, i.e., central versus peripheral route.

JAMS 4 SUMMER, 1988

Page 5: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

Extensive decision making (i.e., reasoned action) should result in higher brand commitment while the lack of extensive decision making (i.e., low involvement learning) should result in lower brand commitment (Jacoby and Kyner 1973). We are proposing that greater involvement (either or both ego and purchase) will lead to more extensive decision making which wilt lead to greater brand commitment.

The approach taken in this article involves treatment and study of the hierarchy models as decision models rather than communication models. Repetition and reactions to specific ads are not of interest here but instead interest is focused on differences found in the hierarchy for those individuals identified as high versus low committed. Brand commitment is the end result of a decision-making process. The goal is to examine what this process looks like. This goal is achieved by utilizing longitudinal survey research and cross-lagged panel correlation to assess the directions of influence.

The primary hypothesis is that the theory of reasoned action will best predict the high-committed individuals' attitude-behavior relationship while the low-involvement hierarchy model will best predict the low committed- individuals' attitude-behavior relationship.

Further, it is hypothesized that habit will be an important addition to both of these models because of the strongly repetitive nature of the product category studied. That is, when one engages in a frequently repeated behavior over time, the behavior becomes so natural and prefunctory that it becomes a habit, i.e., a behavior involving no predeliberation. This occurs regardless of whether the behavior is the result of a reasoned-action process or a low-involvement process. It is primarily dependent on the nature of the product category rather than on individual differences. Social adaptation theory was discussed earlier to provide a theoretical rationale, while a study on seat belt usage by Wittenbraker, Gibbs, and Kahle (1983) provides empirical support for the linkage.

Thus the following hypotheses are proposed: HI: For high-brand-committed individuals:

a. Attitude precedes Intention b. Subjective Norm precedes Intention c. Intention precedes Behavior

H2: For low-brand-committed individuals: a. Behavior precedes Attitude b. Hypotheses Hla, Hlb and Hlc should not

be true for these individuals. 3. For both groups, behavior which is frequently repeated

tends to lead to habit, i.e.: Behavior precedes Habit

METHOD

Subjects

The 201 participants in this study (approximately equal numbers of males and females) were student volunteers enrolled in various undergraduate business courses at three Western universities.

Topic

This study dealt with soft drink consumption. The primary target market for soft drinks is people under 35. This group accounts for 70% of all soft drink sales (English 1983). Thus, college sophomores and juniors are a particularly appropriate match for the topic area under study. Eighty-eight percent of the sample had consumed at least one soft drink in the previous week.

Questionnaire

The questionnaire, which respondents completed twice (four weeks apart), included multiple items designed to assess the Fishbein constructs, the related construct of habit, and the brand-commitment construct. The attitude (toward the acts) and subjective norm items were modeled after similar questions that Fishbein and Ajzen (1975) and Ajzen and Fishbein (1980, p. 256 ft.) have used. The brand commitment items evolved from a review of the literature on involvement, commitment, and brand loyalty. The three items represent a composite of attitudes and stated behaviors revolving around commitment and loyalty. The attitude, subjective norm, and brand commitment questions employed semantic differential-style 9-point scales with adjective pairings and response direction varied from item to item. After reversing negatively worded items, the unweighted sums of the items constituted the numeric definitions of the constructs identified. Habit and behavior consisted of frequency questions, while intention consisted of two 9-point scale items and a frequency question.3 Items from all three scales were summed to constitute the numeric definition of these constructs. Some researchers have measured these constructs differently, but this method does follow common practice (of. Bentler and Speckart 1979) and the suggestion of Ajzen and Fishbein (1980, p. 38): "We can simply ask the respondent to indicate how often he has performed the behavior in question." Many studies that have not used frequency questions have relied on single behavioral criteria rather than Fishbein and Ajzen's (1974) preferred multiple behavioral criteria. The relationships studied and ascertained deal with consumption of the product class, not with consumption of specific brands. Table 1 includes the items for each scale and indicates the item with the highest corrected item-total correlation.

Cross-Lagged Panel Correlation

Cross-Lagged Panel Correlation (CLPC) was introduced in modern times by Campbell and Stanley (1963) and by Pelz and Andrews (1964). It was in part inspired by Lazerfeld's 16-fold table technique. Much of the recent progress in CLPC's development can be attributed to David Kenny (1973; 1975; 1978; 1979; Kenny and Campbell 1984; Kenny and Harackiewicz 1979).

CLPC is a quasi-experimental design and a type of structural equation methodology useful for inferences about spuriousness. In the most basic ease in which CLPC could be used, two panel (longitudinal) variables are measured at two time points (el. Hilles and Kahle 1985).

JAMS 5 SUMMER, 1988

Page 6: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

TABLE 1 Scale Items for Composites

Construct Attitude (7 items)

Item I like to drink soft drinks. Disagree 1 ... 9 Agree" I enjoy having a soft drink with a meal. Often 1 ... 9 Never (reversed) Drinking a soft drink to keep me alert is Bad 1 ... 9 Good 1 like drinking a soft drink because it helps me to relax. Disagree 1 ... 9 Agree I like drinking a soft drink because it quenches my thirst. Agree 1 ... 9 Disagree (reversed) Drinking a soft drink is n o t health for me. Unlikely 1 ... 9 Likely (reversed) I enjoy drinking soft drinks because they taste good Agree 1 ... 9 Disagree (reversed)

Subjective Norm (3 items)

Most people who are important to me think that it is okay to drink soft drinks. Likely 1 ... 9 Unlikely (reversed) a My family thinks I should n o t drink soft drinks. Unlikely 1 ... 9 Likely (reversed) My friends think that I should drink soft drinks. Likely 1 ... Likely (reversed)

Intention (3 items)

I plan to drink at least one soft drink each day over the next week. Unlikely 1 ... 9 Likely" I intend to drink at least several soft drinks in the next week. Likely 1 ... 9 Unlikely (reversed) How many soft drinks do you intend to drink in the next week?

Behavior (4 items)

How many times in the past week have you consumed a soft drink? _ _ a On average, how many soft drinks did you drink a day in the past week with your friends?

How many soft drinks did you usually drink in the past week by yourself?.

Habit (4 items)

The above four behavior questions are asked again with addition of "out of force of habit" at the end, e.g.: How many times in the past week have you consumed a soft drink out of force of habit? a _ _

Brand Commitment

(3 items)

I consider myself to be highly loyal to one brand of soft drink. Disagree 1 ... 9 Agree a If my preferred brand or type of soft drink were not available at the store, it would make little difference to me if I had to choose another brand. Agree 1 ... 9 Disagree When another brand is on sale, I will generally purchase it rather than my usual brand. Likely 1 ... 9 Unlikely (reversed)

altem with highest corrected item total correlation

Suppose that variables A and B have been measured at Time 1 and Time 2. From these data six first-order correlations (see Figure 2) could be computed: two autocorrelat ions (rata2 and rb~b_,), tWO synchronous correlat ions (ralbl and razb2), and two cross-lagged correlations (ra~b2 and rbla:). In certain circumstances the cross-lagged correlations will equal each other if the relationship between them is spurious. Thus, inequality of the cross-lagged correlations in those circumstances implies nonspuriousness.

Over the years various authors have identified instances where inferential caution is necessary. Kenny (1975) has shown that stationarity and synchronicity are necessary assumptions for CLPC analyses. Stationarity means that the underlying structure of the variables is not shifting. Kenny has developed a Program, PANAL, to test for this change statistically (1978). This test was used to establish stationarity in the present study. Equality of pairs of synchronous correlations throughout a data set is consistent with stationarity. Synchronicity means that the variables that comprise the synchronous correlations occurred at the same time. In this study measuring

variables at the same time fostered synchronicity. Furthermore, after each set of questions subjects were asked if their views or responses had changed in the past week, and in all cases the p ropor t ion answering affirmatively was judged to be negligible. Thus, both synchronicity and stationarity are important precurgers of inferences about CLPC.

Testing for the significance of differences in CLPC research cannot be done with Fisher's z, as is usually the case with tests of differences between correlations, since the correlations are themselves correlated. Kenny (1975) recommends the Pearson-Filon test for differences between correlated correlations. This test, however, has very low power. It is preferable to set the alpha level at 0.1 when the sample size is smaller than 300. Thus, in the present study the Pearson-Filon test is used with an alpha level of 0.1.

CLPC shares a number of concerns with all types of research. Are the measures reliable, valid, and unitary? Are the measures tainted by social desirability or some other form of response bias? What is the appropriate time lag between measurements to optimize the effect of one

J A M S 6 SUMMER, 1988

Page 7: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

FIGURE 2 The Six Possible First-Order Correlations between Two Panel Variables, A and B

Time 1 Time 2

A rAIA2 A

r

ALTERNATIVE HIERARCHIES OE THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

J s,

!1 rB1B2 B

variable on the other? How should missing data be handled? On this issue Kenny (1975) recommends list- wise deletion in order to rule out sample differences as an explanation for observed differences in correlations. The data in Table 2 address several of these issues for the present study. We did use list-wise deletion here, which reduced our initial sample size from 201 to 187.

Rogosa (1980) and Cook and Campbell (1979) have noted that variables with larger autocorelations can in some cases spuriously appear to be effects in CLPC research. Obviously, it is often possible to circumvent or overcome this problem in a specific case (Kahle, Klingel, and Kulka 1981; Kenny and Campbell 1984). In the present study none of the reported differences are affected by the stability problem, for example. Kenny and Campbell (1984) have provided a current and balanced review on temporal data analysis, and they conclude that CLPC is still a very useful methodology.

Rogosa (1980) has identified instances in which CLPC leads to erroneous inferences while other methods do not. Kenny and Campbell (1984) have identified instances in which CLPC leads to more accurate inferences than other structural equation methods. CLPC is more appropriate for testing spuriousness and hypotheses; other structural equation methods are more appropriate for testing complex causal chains and complete theories. It is not an assumption of CLPC that all relevant variables have been conceptualized. Given the checkered history of research on the theory of reasoned action, the lack of requirement for this assumption is one advantage of CLPC in this research area.

RESULTS

Composites

Table 2 presents descriptive statistics for each composite, including Cronbach's coefficient alpha (a measure of internal consistency), the mean, the standard deviation, and the autocorrelations of the composites with themselves at two points in time (a measure of stability). The autocorrelations generally reflected stability with the exception of subjective norm and habit, whose autocorre- lations indicated some measurement problems. The coefficient alphas imply that the measures utilized are relatively consistent internally.

Cross-Lagged Results

In order to assess differences in the Fishbein model between low-brand-committed individuals and high- brand-committed individuals, two groups of respondents were formed by dichotomizing the distribution at the median on the brand commitment composite. To assess whether this split captured the hypothesized construct, the two groups were compared against another question which also asked for a commitment-type response: "If I have a choice I generally purchase one brand of soft drink." (Yes or No). The high-committed individuals were much more likely to agree with this statement than the low- committed individuals (82% vs. 37.5%; 2 = 41.10, p < .001), thus indicating a differentiation on brand commitment from another perspective.

JAMS 7 SUMMER, 1988

Page 8: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND COMMITMENT AND HABIT

TABLE 2 Descriptive Statistics for Composites

Composite Alpha Mean Time 1 2 1 2

Attitude .74 .79 36.08 36.39 Subjective Norm .59 .56 18.83 18.40 Intention .74 .75 13.89 14.62 Habit .81 .77 5.01 4.02 Behavior .81 .72 12.45 11.40 Brand Commitment .75 .75 11.90 10.99

1 6.41 4.84

10.68 11.6 15.57 7.01

SD 2 6.3

4.81 10.30 8.31

13.57 6.43

Auto Correlation

.56

.47

.75

.43

.62

.77

TABLE 3 Synchronous & Cross-Lagged Correlations

Grouping/Variables A B

Low n = 90 Commitment: Behavior �9 Habit .49 .57 High n = 97 Commitment: Subjective ~, Behavior .18 .28 Norm Intention ~ Behavior .54 .87 Intention P- Habit .34 .59 Behavior ~- Habit .88 .74

ap < .05 two-tailed test bp ~ . l two-tailed test

Synchronous Cross-Lagged Correlations Correlations

Time 1 Time 2 A1B2 rB1A2 Z

.45 .16 2.58 a

.29 .05 1.72 b

.78 .42 3.46 a

.52 .28 2.03 a

.50 .38 1.82 b

The synchronous and cross-lagged correlations, as well as the z values resulting from an application of the Pearson- Filon test to assess differences between the cross-lagged correlations, are presented in Table 3 for both the low. committed and high-committed groups. Only those relationships significant at the 0.1 level are reported here. Significant differences between cross-lagged correlations imply discrimination between composites and suggest directionality of influence.

For the high brand committed individuals, subjective norm and intention influenced behavior, while intention and behavior influenced habit. These relationships most closely resemble predictions from the theory of reasoned action; however, not all proposed linkages were verified. Attitude did not significantly enter into the model and subjective norm directly impacted upon behavior rather than intention. The important linkage between intention and behavior was verified. Thus, the empirical evidence supported Hlc but not H la or Hlb. For the low-brand- committed individuals the predicted relationship between behavior and attitude did not materialize; however, as predicted, none of the relationships hypothesized for high- brand-committed individuals were found either. Thus Hlb was supported but not H I a.

For both groups the behavior-habit linkage was clearly established, thus verifying the importance of habit within models examining frequently purchased product categories. This finding supports H3. Because this was the only significant relationship established for the low-committed individuals the indication is that these individuals are significantly different in their decision processing from the high-committed individuals and that behavior is the primary variable of interest rather than any of the attitudinal variables.

DISCUSSION AND CONCLUSION

Hypotheses I and 2 can not be fully accepted. However, the findings appear to support the general thrust of the hypotheses and clear differences between the two commitment groups were found. The lack of effect of attitude and subjective norm on intention for the high- committed individuals may be due to the repetitive nature of the behavior studied wherein more aspects of the decision process may be internal ized and thus nondecomposable.

Individuals exhibiting high commitment seem to be engaged in reasoned action, actions derived from intention

JAMS 8 SUMMER, 1988

Page 9: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTERNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND C O M M I T M E N T AND HABIT

and subjective norm. This finding tends to fit with advertisers' attempts at both creating positive-brand images and attitudes and creating a sense of drinking the "in" drink, (i.e, reference-group influence). This advertising would tend to facilitate the decision making of involved, committed individuals.

The failure to ascertain the behavior ~ attitude linkage suggests that attitudes for low-committed individuals simply have little effect; the emphasis is on behavior for these individuals. Thus, the decision process for each group seems quite different, while the behavior

habit relationship is similar between the two groups (i.e., hypothesis 3).

The primacy of behavior in the model for the low- committed individuals is consistent with the low- involvement hierarchy model; however, rather than behavior then influencing attitude, as predicted by the model, behavior directly influenced habit. This finding would suggest that habit's importance in low-involvement decision making has been underestimated. Further, habit appears particularly potent with soft drink consumption when it is noted that even high-brand-committed individuals tend to form habits based on both their intention and behavior. These findings suggest that the soft drink industry's concern with sampling and availability (English 1983) appears reasonable; regardless of the individual's commitment level, behavior influences the formation of habit, which is the ultimate element in both models and one important goal of advertisers as well.

The results for habit are particularly interesting in this study. The fact that habit did not function in the same manner as behavior implies that habit and behavior have discriminant validity with each other. Just as in an earlier study on seat belts (Wittenbraker, Gibbs, and Kahle 1983), habit replaced behavior as the final variable in the model. Habit is not a reasoned action. It may nevertheless derive from an action that at one point was reasoned (Day 1969). The reasoned action metaphor, presumed to operate completely in the Fishbein model, seems to be questionable for many behaviors and particularly, for individuals with little brand commitment. The range of "unreasonable" human behavior appears to be expanding continuously (Langer 1978; Nisbett and Ross 1980). A view of adaptation rather than reason as the central goal of human information processing increasingly seems plausible (Kahle 1984).

Although the theory of reasoned action did not fit well with the findings for either commitment group due to the prominence of habit and lack of importance of attitude, it did come closer to the findings for the high-committed group. The low-committed individuals appear to be using more of a low-involvement hierarchy model. Although the findings are not completely consistent with this model, low-brand-committed individuals appear to engage in little reasoned action in their consumption of soft drinks. The primacy of behavior for these individuals, as suggested by numerous researchers, was substantiated. The importance of influencing behavior directly for both groups of individuals appears reasonable given the findings from this research.

The implications of these f'mdings are that individuals' level of brand commitment can be utilized to differentiate the decision process by which individuals' attitudes and behaviors and related concepts interact. There were very clear differences noted between the two commitment segments. The nomological inclusion of brand commitment within the attitude-behavior domain appears to offer a potential avenue for further research. The results of this study must be cautiously interpreted, however. Our study included only one product category and was a survey rather than an experiment. Future studies should investigate other product or stimulus domains, using other metho- dologies, to ascertain the generalizability of results across broader domains.

FOOTNOTES

tAs pointed out by one reviewer, higher involvement may also lead to commitment to a high standard of quality or other such criteria rather than to commitment to a particular brand.

2There is no need to make a distinction between ego and purchase involvement when generalizing across situations for a frequently purchased product such as in this study due to the lack of any established temporal perspective.

3The frequency question involved asking individuals how many soft drinks they planned to drink in the coming week. Since the mean and standard deviation for this question were quite similar to the means and standard deviations for the 9-point scale items, there was no need to adjust the scales, and summation of the three scales was deemed appropriate.

REFERENCES

Ajzen, leek and Martin Fishbein. 1980. Understanding Attitudes and Predicting Social Behavior. Englewood Cliffs, N.J.: Prentice-Hall.

Bauer, Raymond A. 1964. "The Obstinate Audience." American Psychologist 19(May):319-28.

Beatty, Sharon E., Lynn R. Kahle and Pamela Homer (forthcoming). "The Involvement-Commitment Model: Theory and Implication." Journal of Business Research.

Bern, Daryl. 1972. "Self Perception Theories." In Advances in Experimental Social Psychology. Ed. Leonard Berkowitz. New York: Academic Press, 1-10.

Bentler, Peter M. and George Speckart. 1979. "Models of Attitude-Behavior Relations." Psychological Review 86 (September): 452-64.

Bloch, Peter H. and Marsha L. Richins. 1983. A Theoretical Model for the Study of Product Importance Perceptions." Journal of Marketing 47 (July): 69-81.

Burnkrant, Robert E. and Thomas J. Page, Jr. 1982. "An Examination of the Behavioral Convergent Diseriminant, and Predictive Validity of Fishbein's Intention Model." Journal of Marketing Research 19 (November): 550-561.

Calder, Bobby J. 1979. "When Attitudes Follow Behavior - - a Self- Perception/Dissonance Interpretation of Low Involvement. In Attitude Research Plays for High Stakes, Eds. John C. Maloney and Bernard Silverman. Chicago: American Marketing Association.

Campbell, Donald T. and Julian C. Stanley. 1963. Experimental and Quasi- Experimental Designs for Research. Chicago: Rand McNally.

Cook, Thomas D. and Donald T. Campbell. 1979. Quasi-Experimentation.. Design and Analysis Issues for Field Settings. Chicago: Rand McNally.

Crosby, Lawrence A. and James R. Taylor. 1983. "Psychological Commitment and its Effects on Post-Decision Evaluation and Preference Stability Among Voters." Journal of Consumer Research 9 (March): 413-31.

Day, George S. 1969. "A Two-Dimensional Concept of Brand Loyalty." Journal of Advertising Research 9 (September): 29-35.

Dillehay, Ronald C. 1972. "On the Irrelevance of the Classical Negative Evidence Concerning the Effect of Attitudes on Behaviors." American Psychologist 28 (October): 887-91.

English, Mary M. 1983. "Teens - - the Soft Drink Generation." Advertising Age M-33-M-34.

JAMS 9 SUMMER, 1988

Page 10: Alternative hierarchies of the attitude-behavior relationship: The impact of brand commitment and habit

ALTEILNATIVE HIERARCHIES OF THE ATTITUDE-BEHAVIOR RELATIONSHIP: BEATTY & KAHLE THE IMPACT OF BRAND C OM M IT M E NT AND HABIT

Fishbein, Martin. 1980. "A Theory of Reasoned Action: Some Applications and Implications." In Nebraska ,~vmposium on Motivation. Ed. Monte M. Page. Vol. 27. Lincoln: University of Nebraska Press.

_ _ and Icek Ajzen. 1974. "Attitudes Toward Objects as Predictors of Single and Multiple Behavioral Criteria." Psychological Review 81 (January): 59-74.

_ _ a n d _ .1975. Belief, Attitude, Intention, and Behavior: An Introduction to Theo(v and Research. Reading, Massachusetts: Addison-Wesley.

Hilles, W. Scott and Lynn R. Kahle. 1985. "Social Contract and Social Integration in Adolescent Development." Journal of Personality and Social Psychology 49 (October), 1114-21.

Houston, Michael J. and Michael L. Rothschild. 1977. "A Paradigm for Research on Consumer Involvement." Unpublished paper, Graduate School of Business, University of Wisconsin, Madison.

Jacoby, Jacob and Robert W. Chestnut. 1978. Brand Loyalty Measurement and Management. N.Y.: John Wiley & Sons.

Jacoby, Jacob and David B. Kyner. 1973. "Brand Loyalty vs. Repeat Purchasing Behavior." Journal of Marketing Research 10 (February): 1- 9.

Kahle, Lynn R. 1984. Attitudes and Social Adaptation: A Person-Situation Interaction Approach. Oxford: Pergamon.

and Sharon E. Beatty. 1987. "The Task Situation and Habit in the Attitude-Behavior Relationship: A Social Adaptation View." Journal of Social Behavior and Personality 2(2,part 1):219-232.

_ _ , David M. Klingel, and Richard A. Kulka. 1981. "A Longitudinal Study of Adolescents' Attitude-Behavior Consistency." Public Opinion Quarterly 45( Fall):402-14.

Richard A. Kulka and David M. Klingel. 1980. "Low Adolescent Self-Esteem Leads to Multiple Interpersonal Problems: A Test of Social Adaptation Theory." Journal of Personality and Social Psychology 39 (September): 496-502.

and D.B. Kyner. 1973. "Brand Loyalty versus Repeat Purchasing Behavior." Journal of Marketing Research 10(February): 1-9.

Kenny, David A. 1973. "Cross-Lagged and Synchronous Common Factors in Panel Data". In Structural Equation Models in the Social Sciences. New York: Seminar Press.

_ _ 1975. "Cross-Lagged Panel Correlation: A Test for Spuriousness." Psychological Bulletin 82 (November): 887-903.

1978. PANEL: A Computer Program for Panel Data Ana(vsis. Department of Psychology Research Report, University of Connecticut.

_ _ 1979. Correlation and Casuality. New York: Wiley. _ _ and Donald I. Campbell. 1984. "Methodological Issues in the

Analysis of Temporal Data." In Historical Social Psychology. Hillsdale, N.J.: Lawrence Erlbaum, 125-38.

_ _ and Judith M. Harackiewicz. 1979. "Cross-Lagged Panel Correlation: Practice and Promise." Journal of Applied Psychology 64:(August)372-79.

Krugman, Herbert E. 1965. "The Impact of Television Advertising: Learning Without Involvement." Public Opinion Ouarterly 29(Fa11):349-56.

Langer, Ellen J. 1978. "Rethinking the Role of Thought in Social Interaction." In New Directions in Attribution Research. Eds. John H. Harvey, William Ickes and Robert F. Kidd: Hillsdale, N.Y.

LaPiere, Richard T. 1934. "'Attitudes vs. Action." Social Forces 13(December):230-237.

Lastovicka, John L. and David M. Gardner. 1977. "Components of Involvement." In Attitude Research Plays for High Stakes. Eds. J.C. Maloney and B. Silverman. Chicago: American Marketing Association.

Laurent, Gilles and Jean Noel Kapferer. 1985. "Measuring Consumer Involvement Profiles," Journal of Marketing Research 22(February):41- 53.

Miniard, Paul W. 1981. "Examining the Diagnostic Utility of the Fishbein Behavioral Intentions Model." In Advances in Consumer Research. Ed. Kent B. Monroe. Ann Arbor: Association for Consumer Research, 42- 7.

Mitchell, Andrew A. 1979. "Involvement: A Potentially Important Mediator of Consumer Behavior." In Advances in Consumer Research. Ed. W.L. Wilkie Vol. 6. Ann Arbor: Association for Consumer Research.

Mitchell, Andrew A. and Jerry C. Olson. 1981. "Are Product Attribute Beliefs the Only Mediator of Advertising Effects on Brand Attitudes?" Journal of Marketing Research 18 (August): 318-22.

Mixon, D. 1980. "The Place of Habit in the Control of Action." Journal of the Theory of Social Behavior 10(October):169-86.

Nisbett, Richard E. and Lee Ross. 1980. Human Inference: Strategies and Shortcomings of Social Judgement. Englewood Cliffs, N.J.: Prentice Hall.

Pelz, Donald C. and Frank M. Andrews. 1964. "Detecting Causal Priorities in Panel Study Data." American Sociological Review 29(December):836- 48.

Petty, Richard E., John T. Cacioppo, and David Schumann. 1983. "Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement. ~ Journal of Consumer Research 10(September): I35-46.

Ray, Michael L. 1973. Marketing Communications and the Hierarchy-of- Effects. In New Models for Mass Communication Research. Ed. P. Clark. Beverly Hills, CA: Sage Publications, 147-76.

Robertson, Thomas S. 1971. Innovative Behavior and Communication. New York: Holt, Rinehart and Winston.

1976. "Low-Commitment Consumer Behavior." Journal of Advertising Research 16(April): 19-24.

Rogosa, D. t980. "A Critique of Cross-Lagged Correlation_" Psychological Bulletin 88(September):245-58.

Selin, Steven W. 1986. A Two-Stage Test of Selected Causal Antecedents of Recreation Program Loyalty. Unpublished dissertation. Eugene, Oregon: University of Oregon.

Sherif, Carolyn W., Musafer Sherif, and R.E. Nebergall. 1965. Attitude and Attitude Change: The Social Judgement Involvement Approach. New Haven: Yale University Press.

Smith, Scott M. and Sharon E. Beatty. 1984. "Development of a Generalized Involvement Scale. ~ In Scientific Method in Marketing: Philosophy, Sociology and History of Science Perspectives. Eds. Paul F. Anderson and Michael J. Ryan. Chicago: American Marketing Association.

Traylor, Mark B. 1983. "Ego Involvement and Brand Commitment: Not Necessarily the Same." Journal of Consumer Marketing t(2):75-9.

_ _ and WB. Joseph. 1984. "Measuring Involvement in Products." Psychology and Marketing 1:65.77.

Triandis, Harry L. 1977. Interpersonal Behavior. Monterey, Calif.: Brooks/ Cole.

Warshaw, Paul R. 1980. "A New Model for Predicting Behavioral Intentions: An Alternative to Fishbein." Journal of Marketing Research 17(May): 153- 72.

Wittenbraker, John, Brenda Lynn Gibbs, and Lynn R. Kahle. 1983. "Seat Belt Attitudes, Habits, and Behaviors: An Adaptive Amendment to the Fishbein Model." Journal of Applied Social Psychology 13(September- October):406-21.

Wright, Peter L. 1973. "Cognitive Processes Mediating Acceptance of Advertising." Journal of Marketing Research 10(February):53-62.

Zaichkowsky, Judith L. 1985. "Measuring the Involvement Construct in Marketing." Journal of Consumer Research 12(December):341-352.

Zaltman, Gerald and Melanie Wallendorf. 1983. Consumer Behavior Basic Findings and Management Implications. N.Y.: John Wiley & Sons.

ABOUT THE AUTHORS

SHARON E. BEATTY is assistant professor of marketing at the University of Alabama. Previously she was assistant professor of marketing at the University of Alabama and assistant professor of marketing at the University of Oregon, where she also received her Ph.D. She published articles in Journal of Marketing Research and Journal of Consumer Research among others and has presented a number of papers at national conferences. She was previously the director of marketing for a nationally franchised organization.

LYNN R. KAHLE earned his Ph.D. from the University of Nebraska-Lincoln. He currently is associate professor of marketing at the University of Oregon. His research has been published in the Journal of Consumer Research and the Journal of Personality and Social Psychology, among others. His most recent book is Attitudes and Social Adaptation.

JAMS 10 SUMMER, 1988