Methodological Issues in Studying Intergenerational Influences on Consumer Behavior

George P. Moschis, Georgia State University
ABSTRACT - In studying intergenerational influences on consumer behavior the researcher is likely to be confronted with several methodological issues. Such issues may relate to (a) choosing the appropriate theory, (b) conceptualizing and measuring the relevant variables, (c) choosing the appropriate research design, and (d) interpreting the data. This paper highlights some issues related to these areas and whenever possible suggests ways of addressing them.
[ to cite ]:
George P. Moschis (1988) ,"Methodological Issues in Studying Intergenerational Influences on Consumer Behavior", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 569-573.

Advances in Consumer Research Volume 15, 1988      Pages 569-573

METHODOLOGICAL ISSUES IN STUDYING INTERGENERATIONAL INFLUENCES ON CONSUMER BEHAVIOR

George P. Moschis, Georgia State University

ABSTRACT -

In studying intergenerational influences on consumer behavior the researcher is likely to be confronted with several methodological issues. Such issues may relate to (a) choosing the appropriate theory, (b) conceptualizing and measuring the relevant variables, (c) choosing the appropriate research design, and (d) interpreting the data. This paper highlights some issues related to these areas and whenever possible suggests ways of addressing them.

INTRODUCTION

Researchers investigating intergenerational influences on consumer behavior are likely to be faced with a number of issues related to the methodology of studying such influences. Issues may range from those which relate to conceptual and theoretical underpinnings, to measurement and research designs. Although it would be rather difficult to address all the issues in this paper because of our limited knowledge of the area and space considerations, some of the most pressing areas deserve special attention.

The purpose of this paper is to present several key issues that researchers are likely to encounter, and to suggest ways of addressing them. Given that methodological issues in studying intergenerational influences can be present at various stages of the research process, those which are discussed in this paper are presented in the same order as the research process. Thus, issues related to appropriate theoretical and conceptual frameworks are discussed first, followed by issues related to the conceptualization and measurement of the influence processes, and lastly, the discussion focuses on issues related to the research design, data analysis and interpretation of the findings.

CONCEPTUAL FRAMEWORK

The theoretical and conceptual underpinnings of the approach can be found in theories of socialization. Specifically, the development of consumption-related orientations and the specific role of parents (agents of socialization) may be viewed in the context of a broader model of consumer socialization (Moschis 1987, Ward 1974). Previous reviews of the literature (e.g., Moschis 1987, Ward 1974) do not only suggest that the available data support the consumer socialization perspective as an approach to the study of intergenerational influences; they also suggest theoretical perspectives useful in understanding the processes of acquisition of consumer cognitions and behaviors.

It should be brought to the reader's attention, however, that the study of intergenerational influences is not always consistent with the socialization approach because of the assumptions made which may not always hold in the consumer socialization model. The study of intergenerational influences implicitly or explicitly makes a number of assumptions. First, the influence on consumer behavior "flows" from the older parent to the younger child. Second, the consumption-related orientations which are transmitted from one generation to the next are relatively stable and can be tracked; once they are transmitted they are held by both the influencer and the influencee. Third, intergenerational influences represent primarily the effects of external socia forces, parents in particular, on the person's consumer learning.

Research has shown that these assumptions may not always hold in the consumer socialization context. Specifically, there may be a reverse flow of influence from the child to the parent (the case of "reverse socialization"). Similarly, consumption patterns tend to change over a person's life cycle; and once they are acquired from parents they may undergo further formation and change. And, finally, the formation of select patterns of thoughts an actions may not necessarily be attributed to the person's social environment (parents), but rather to other factors. Thus, in conceptualizing intergenerational influences, the research must justify why intergenerational influences fit into a broader context of consumer socialization, and which theories and variables are related to this process.

Quite often the researcher's approach to the selection of variables is guided by strong suspicion and is based on his/her own speculations rather than theory and research in the area. The rationale for expecting a particular aspect of consumer behavior to be subject to intergenerational influences is often lacking.

Given that a number of approaches or theories can be used to study intergenerational influences, the researcher must decide on the appropriate framework. While elaboration on specific theoretical frameworks is beyond the scope of this research, the reader should be aware of the several theories which can be used to explain or study intergenerational influences (e.g., Moschis 1987), and how the specific theory may affect his or her approach. For example, interaction theories stress an active view of the individual in the socialization process. Researchers adopting this perspective would place greater emphasis on the learner s orientations toward the socialization agents, and how s/he interacts with them and is influenced by them. On the other hand, if one subscribes to connectionist theories, which view the socialization process as being externally controlled, the researcher's focus would be on the agent's behaviors and mechanisms used by the agent to influence the learner's behavior.

Before focusing on theoretical perspectives, the nature of the dependent variables that define the person's consumer role over his/her life cycle must be specified. The variables most relevant and desirable (from a societal perspective) must be identified and categorized to help us systematically conduct studies and accumulate research findings. A conceptual framework for classifying consumer behaviors is needed. While far from being adequate in terms of classifying a wide variety of consumption-related orientations, frameworks such as those presented by previous researchers (e.g., Moschis 1987) may well serve the purpose, particularly at the early stage of research. Researchers have a choice between conducting research in the area on an 7 hoc basis and using a more systematic approach to their investigation. Given that the field of consumer behavior is rich in dependent variables, researchers could arbitrarily select to study any number of variables. A more systematic approach to va able selection should include: (a) the definition and placement of all aspects of consumer behavior in a conceptual framework, and (b) the justification for studying select number of variables, where justification is primarily sought on specific theories.

The decision on the variables which define consumer behavior is not an easy one, given that we have limited understanding of the types of consumer behaviors that are likely to develop at various stages in the person's life. Perhaps one of the greatest difficulties in studying consumer socialization and consumer behavior over the life cycle is in understanding where certain changes in behavior as well as influences of socialization agents are likely to occur. Individuals are likely to acquire patterns of cognitions and behavior prior to reaching certain stages where theoretical explanations may account for differences in their thoughts and actions. Similarly, consumer socialization agents may have delaying effects.

INFLUENCE PROCESSES

Conceptual Issues

Researchers ideally should spell out the specific type(s) of influence in the transmission of specific skills, values, or behaviors from socialization agents (influencers) to influencee. A socialization agent (e.g., father, mother) may communicate certain information to the learner through various mechanisms. First, by performing certain acts an agent may consciously or unconsciously communicate certain norms and expectation. The communication in this case is likely to take place at a cognitive level, and influence and subsequent consumer learning is likely to be the result of observation or imitation of these behaviors. For example, a child may make a conscious effort to emulate the behavior of his father because the model's behavior is the most salient alternative open to him--i.e., he does the same thing his father does in an effort to be like him. Second, a socialization agent may influence the consumer behavior of others by using various reinforcement mechanisms, both positive and negative, in attempting to communicate to others certain desires. A parent, for instance, may reward certain behaviors which are consistent with such desires and punish others that are inconsistent. In an interpersonal setting, reinforcement may involve oven communication, such as praise for (positive) and complaining about (negative) using a product; it may also involve cognitive communication, where a person may dictate his/her desires to others by, for example, showing affection (positive reinforcement) and psychologically punishing them (negative reinforcement).

Finally, socialization agents may affect the consumer behavior of the learner through oven communication processes, often referred to as the "social interaction" mechanism. The social interaction mechanism is less specific, and it may involve a combination of modeling and reinforcement. During the course of the person's interaction with socialization agents, a person may acquire certain attitudes, values, and behaviors, which are often communicated explicitly.

Social interaction processes can have content and structure. Content includes expectations (norms) held by the communicator as to what the desirable or prescribed behaviors should be; or it can be information about consumption. The structure, on the other hand, has different connotations. When it refers to interpersonal processes it can be interpreted in the context of power and communication relations.

When researchers have to choose among the three socialization processes, they often can find little justification as to why certain processes should be included in, or excluded from the study. There is relatively little theory to suggest the skills, knowledge and behaviors acquired from specific agents, and much less to suggest the processes by which these are acquired. Furthermore, the decision on the various processes becomes even more difficult given that consumer-related orientations are likely to be transmitted via different processes in different socio-cultural settings (Moschis 1987).

Given our limited knowledge regarding the processes involved in intergenerational influences, it would be desirable to include all three processes wherever there is little theoretical or empirical justification to include specific ones, treating the type of influence operating as an empirical question. This process might help us accumulate some empirical evidence which might be useful for theory development.

Measurement Issues

Measures of intergenerational influence have been constructed from the point of view of the influencee, the agent, and the researcher. When such measures are approached from the influencee's standpoint, the focus is upon asking respondents to assess their response to the agent's behavior such as observations of parental behaviors and parental influence on their consumer decisions. Measures of influence can also be constructed from the influencer's perspective by asking individuals in the influencee's immediate environment to report such responses. For example, parents are asked to report on their influence on the child's behavior. While self-reported measures of influence may help researchers isolate the agent's effects from other third variables likely to produce similar responses, they rely on the respondent's ability to assess the extent of such influence.

Finally, researchers have used measures of agent's influence based upon the respondent's assessment of frequency and nature of interaction with the agent, inferring the agent's influence from statistically significant relationships between specific parent-child interactions and levels of consumer orientations. The aim is at falsifying the presence of influence rather than at confirming it. When a significant relationship is found the inference made is that variation in consumer orientation is due to the content of parent-child interaction. Lack of a significant relationship suggests that the agent may not influence the development of consumer orientations, or that the influence may not be linear.

There are several issues surrounding the measurement of frequency and nature of interaction (modeling, reinforcement, social interaction mechanisms). For example, while common operational definitions of modeling processes include source-learner similarity of behaviors and attitudes (Moschis 1987), most of the studies of consumer socialization have used measures of overt parent-child interaction about consumption. Considerably less attention has been devoted to the measure of the content of family communication, which is usually inferred from correlational data. Thus, for example, if one finds a relationship between parent-child interaction about consumption and the child's ability to price products accurately, the assumption is made that such communication focuses on prices. However, the child's ability to price products accurately may not necessarily develop as a result of family communications but may be due to a third variable (e.g., development of product desires), which can result both in greater awareness of product attributes and in child-initiated communication. The most perplexing measurement problem involves the need to establish functional patterns of communication and consumer behavior that cut across various discrete consumer orientations and socialization processes (McLeod and O'Keefe 1972). Measures focusing simply on time spent interacting with various socialization agents are not likely to produce meaningful results. Receiver-oriented categorizations of the agent's functions and ability to fulfill certain information needs combined with specific use patterns would seem to be the optimum strategy for measuring communication behavior (e.g., McLeod et al. 1982, McLeod and O'Keefe 1972).

Measures of the effects of socialization agents and antecedent variables, on the other hand, pose a rather different issue in that both relative and absolute measures can be used. Measures of relative effects of socialization agents can be assessed in terms of their intent, attitudes, and cognitions, by pitting the influencer's (agent's) orientations against the influencee's (learner's) orientations toward the object of communication. Thus, the emphasis shifts from the individual's orientation toward the object of communication (absolute measures) to co-orientation (relational measures), and from sampling individuals as units of analysis to sampling dyads, groups, or collectivities. Absolute measures, on the other hand, often consists of cognitive, affective, and behavioral outcomes measured in relation to some norm or expected direction. For example, learning about products (awareness) and developing skills at budgeting (behaviors) are used as unidirectional measures of effects of parent-child communications about consumption. When such measures are used, the assumption is often made that the agent's intention is to change the level of cognitions and behaviors to the desired direction.

Co-orientational assumptions must be checked before applying this measurement model. At times, intergenerational effects on consumer learning assessed using co-orientational variables are difficult to evaluate in terms of exactly the same elements of X (object of communication). Although it is very rare that both A and B would perceive exactly of same elements of X, some degree of similarity of their orientation toward the object of communication is necessary in order to apply the model sensibly at the co-orientational level. At times, the availability of symbols of X (often referred to as "codability") may hamper communication, as would "connotation," a term that describes variance between A and B in their respective usages of words to denote Xs (McLeod and Chaffee 1973).

Communication and influence effects on consumer learning may not only be in terms of "overlaps," in line with the co-orientational model, but also in terms of changes or developments of cognitive and behavioral patterns. For example, much of consumer socialization literature suggests that parent-child communications about consumption are related to the child's development of several cognitive and behavioral orientations about consumption (Moschis 1987). In addition, there may be other messages or cues that can be, often unintentionally, communicated from one person to the other during the course of interaction. Thus, it is often the effect rather than the perceived intent of the communication that should be measured. Because of the several ways family members can influence one another, several types of indices of communication effectiveness are often recommended in assessing interpersonal communication and influence. For purposes of measurement, it is necessary to assess a set of attributes of the role-incumbents that are particularly relevant to that role, and to obtain corresponding measures of the same attributes on the role-aspirant group. The particular set of attributes to be assessed is a function of the particular role under consideration. Such attributes include values, beliefs, attitudes and specific behavioral acts.

In order to assess the degree of intergenerational influence, the development of indices of communality between role aspirants and the role-incumbent groups is required. The degree of communality between the two groups can be indexed by a number of measures. For example, Tannenbaum and McLeod (1967) suggested several indices, including similarity of factor structure, relative factor salience, difference in concept judgments using the D statistic; differentiation between concepts, degree of role identification, homogeneity of judgments, and meaningfulness. In the area of consumer socialization, Rossiter and Robertson (1979) used confirmatory factor analysis to assess the similarity of factor structures between mothers and their children.

One of the difficulties encountered in trying to demonstrate the effectiveness of socialization agents in teaching behaviors is that learning involves at least two stages: acquisition and performance (e.g., Comstock et al. 1978). Failure to perform a behavior previously observed does not mean that acquisition has not taken place (Bandura 1973). Unfortunately, it is often impossible to determine whether a behavior has been acquired unless it is performed. However, a behavior previously acquired may not be performed immediately, but later if or when the appropriate eliciting conditions occur. Developmentalists have referred to this phenomenon as a "sleeper effect," while sociologists -- have referred to it as "anticipatory socialization."

Because consumer socialization involves the study of specific agents and their effects, a measurement-related issue involves the decision regarding the time lag between interaction with these agents and learning of consumer orientations. Comparison between short-term and long-term effects suggests possible differences in the effects of socialization agents as a function of time (Moschis 1987).

In sum, measurement issues must address the question what to measure (content or criterion behavior), how to measure and when to measure consumer socialization processes and effects.

Reliability and Validity

Reliability and validity are necessary conditions for sound consumer socialization research, if knowledge about the area is to be systematically accumulated over time. Issues of reliability and validity are magnified when the researcher works with groups of people at extreme stages in their life cycle -- i.e. children and elderly, or from different cultural settings. This is because people respond differently to questions or questionnaires. Thus, for example, an instrument could be found to be reliable and valid when adult subjects are used but it might be unreliable or invalid when applied to children and elderly.

Consumer socialization involves continuous adjustment to one's environment and learning new attitudes, values, and norms or modifying existing ones. Since not all orientations develop or change simultaneously, interrelationships among items expected to measure a particular construct are frequently low. Similarly, correlations of the same measure over time (test-retest reliability) are low as well because socialization involves the formation and change of orientations regarding a specific role. Thus, it is more likely to achieve higher reliability of constructs when working with subjects who are expected to have developed certain attitudes, values and behaviors than with subjects who are in the process of developing or modifying such orientations.

Another related problem that applies to both reliability and validity stems from the fact that responses to an instrument are likely to be significantly affected by the person's background characteristics. Factors such as race, social class and sex are likely to affect both -reliability and validity. For example, while Rossiter's (1977) objective test for measuring children's attitudes toward television commercials was found to be highly reliable, analysis of the instrument by income level suggested significant differences in responses (Bearden et al 1979). Similarly, Allen and Chaffee (1977) found that the two-dimensional model of communication structures fits the black subculture not as well as it does the white subculture from which it was originally derived.

Differences in responses to the instrument across social groupings could be due to a number of reasons. First, socialization may not occur with the same rate within groups of a social structure. Second, different norms, values, and beliefs may be emphasized in different sub-cultures. Third, sub-cultural variables may affect the respondent's communication skills and even the quality and meaningfulness of responses. Finally, third variables (e.g., I.Q.) may be related to both a specific sub-culture and item; making up the instrument (e.g., attitudes toward advertising).

The seriousness of several problems related to reliability and validity could be reduced if specific steps are taken. It appears that specification of the attitudes, values and behaviors that define the consumer role, or consumer behavior in general, has a lot to do with how persons from different backgrounds respond to measures of these variables. If the constructs and domains related to the focal variables are of nearly equal relevance to respondents, regardless of their background characteristics, reliability and validity should be improved. Second, data analysis by sub-culture might help us isolate differences due to cultural norms from differences due to measurement.

RESEARCH DESIGNS

The development of research designs appropriate to the study of intergenerational influences must deal with change over time and sequences of cause and effect (McLeod and O'Keefe 1972). Causal modeling using nonexperimental data gathered at a single point in time is one promising approach. When causal modeling is used, care must be taken in separating the effects of socialization processes whenever reciprocal causality exists. This can be accomplished in a number of ways including the use of instrumental variables and maximum likelihood estimates.

The ideal research design, however, is believed to be the panel design that has often been advocated but seldom used in consumer socialization studies. Using this design, data can be obtained from the person being socialized as well as from socialization agents such as parents. Over long periods of time the same respondents would again serve as data sources. A major advantage of this design is that it allows one to work strong tests of cause and effect relationships. In socialization studies, for example, parent-child correlations are often used as evidence of child's modeling of the parent's behavior, but may also be explained as reverse modeling -- i.e. the child's behavior influences the parent rather than vice versa (McLeod and O'Keefe 1972, Surlin et al. 1978). Cross-lagged correlation designs as well as causal modeling with and without lagged variables can be used with panel data.

Another issue deals with the appropriate sampling units. For example, one pressing need for research appears to be the need for studying the influence of family in the context of specific dyads, focusing upon specific pairs. Consumer socialization research (e.g., Moschis 1987) suggests that family influences have been examined only in the broad context of how parents affect the development of consumer behavior of their children. Ideally, research should address specific dyads and directions of influence such as mother-son, mother-daughter, father-son, and father-daughter. The study of family communications in the context of specific parent-child dyads may be useful in understanding the effects of specific interpersonal processes on the development of various types of consumer behaviors.

REQUIRED EVIDENCE FOR INFERRING CAUSALITY

Within the requirements for inferring causality known to researchers, one must show (a) correlation of the agent's with the influencee's consumer-related orientations, (b) time order of occurrence, and (c) elimination of other possible sources of influence. There are several issues surrounding the evidence of causality in the context of intergenerational influences.

With respect to agent-learner similarities of behaviors and attitudes which may be inferred from correlational analyses, the relationship does not necessarily imply causality. It is very likely that the reverse influence exists, or a simultaneous effect takes place. Furthermore, it is possible that the agent's influence may be in teaching norms and behavior which are not necessarily held by the agent.

Time order of occurrence is another condition which must be met. However, the issues here are: What is the necessary time lag before effects can occur? Does the acquisition/influence of certain consumer orientations require a greater time lag? Does influence vary over the life cycle and in different socio cultural settings? Unfortunately, there is little theory or research to help us answer these questions.

The last requirement, elimination of other possible influences, poses the greatest challenge to researchers. Many of the intergenerational effects may be direct, while other influences may be indirect via third variables, or they may be mediated in the socialization process.

One other issue that might take a considerable amount of research data to answer deals with how the person's acquisition of consumer behaviors relates to the development of other types of orientations, and how they relate to specific social structures and social systems. It is possible that consumer behavior is best understood in its relationships with other kinds of behavior as a dependent variable, and that consumer behavior may represent too narrow a scope for an adequate research program (McLeod and O'Keefe 1972). This would make both socialization processes and consumer behavior dependent upon the person's socialization into a more general role. For example, a fruitful area for future research might prove to be understanding the persons personality development throughout his life-span and how such development affects specific aspects of consumer behaviors. Similarly, changes in consumer orientations over the life cycle may reflect changes in the person's environment rather than effects of specific socialization processes and their antecedents. For example, one could check this assumption by developing global measure of the general adult public's behavior which could then be compared to those of the influencee's parents in producing certain effects on the learner's behavior.

SUMMARY

The study of intergenerational influences on consumer behavior is within the scope of the consumer socialization perspective. The consumer socialization approach not only presents a new way of studying consumer behavior, but it also reflects current thinking and trends in social sciences. However, given that the area is relatively new, there has been little data generated to make researchers aware of and address issues surrounding the process of studying consumer behavior using this particular perspective. It is hoped that this paper has raised the level of awareness of some key issues and has provided some direction for future research for achieving greater uniformity in methodologies used and improving the quality of research findings.

REFERENCES

Allen, Richard L. and Steven H. Chaffee (1977), "Racial Differences in Family Communication Patterns," Journalism Quarterly, Vol. 55 (Spring) pp. 8-13, 57.

Bandura, A. (1973), Aggression: A Social Learning Analysis, Englewood Cliffs, NJ: Prentice-Hall.

Bearden, William D., J.E. Teel, and R.R. Wright (1979), "Family Income Effects on Measurement of Children's Attitudes Toward Television Commercials," Journal of Consumer Research, Vol. 6 (December) pp. 308-311.

Comstock, G., S. Chaffee, N. Katzman, M. McCombs, and D. Roberts (1978), Television and Human Behavior, New York: Columbia University Press.

McLeod, Jack M., Carl R. Bybee and Jean A. Duvall (1982), "Evaluating Media Performance by Gratifications Sought and Received," Journalism Quarterly, Vol. 60(1), (Spring) pp. 3-12.

McLeod, Jack M. and Steven H. Chaffee (1973), 'Interpersonal Approaches to Communication Research," American Behavioral Scientist, 16 (April), pp. 469-99.

McLeod, Jack M. and Garrett O'Keefe, Jr. (1972), "The Socialization Perspective and Communication Behavior," in Current Perspectives in Mass Communication Research, G. Kline and P. Tichenor, eds., Beverly Hills: Sage Publications, pp. 121-168.

Moschis, George P. (1987), Consumer Socialization: A Life Cycle Perspective, Boston: Lexington Books.

Rossiter, John R. (1977), "Reliability of a Short Test Measuring Children's Attitudes Toward TV Commercials," Journal of Consumer Research, Vol. 13, (March), pp. 179-184.

Rossiter, John R. and Thomas S. Robertson (1979), "Children's Independence from Parental Mediation in Learning about OTC Drugs," 1979 Educators Conference Proceedings, Chicago: American Marketing Association pp. 653-652.

Surlin, Stuart H., Alan Wurtzel and Linda Whitener (1978), "Parental Control of Children's Television Viewing Behavior Support for the Reverse Modeling Principle," Paper presented to Mass Communication Division International Communication Association, Chicago, Illinois, April.

Tannenbaum, Percy H. and Jack M. McLeod (1967), "On the Measurement of Socialization," Public Opinion Quarterly, 21 (Spring), pp. 27-37.

Ward, Scott (1974), "Consumer Socialization," Journal of Consumer Research, 1 (September), pp. 1-16.

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