Printer Friendly

The effects of consumer education on consumer search.

The Effects of Consumer Education on Consumer Search

Consumer information and education programs have been suggested as public policy alternatives to regulation for improving consumer and marketplace efficiency. However, acquisition and processing of information are complex and time-consuming tasks for most consumers, further confounded by the growing number of complicated consumer goods and innovations in the marketplace. Technological innovations in production and communication have made a wide variety of new goods available, have led to the creation of many new models or styles of familiar goods, and have increased options with respect to how and where goods can be purchased, all of which increase the complexity of consumers' choices.

Some evidence that use of information improves both consumer and marketplace efficiency exists (Sproles, Geistfeld, and Badenhop 1978; Marvel 1976; Devine and Hawkins 1972; Nelson 1970), and it has been suggested that consumer education can, in turn, improve consumers' information acquisition and use skills. Many consumer education courses emphasize the benefits to be derived from engaging in prepurchase information search and attempt to help consumers become more effective in their search activities and in evaluating the relative merits of various information sources. However, few attempts have been made to assess the overall impact of consumer education programs on subsequent consumer behavior such as information search.

This paper presents an empirical investigation of relationships between selected consumer and marketplace characteristics and consumers' allocation of search time among four specific information sources (product test reports; dealer sales representatives; advertisements; friends, relatives, and acquaintances) with respect to their most recent major household appliance purchases. Discussion centers chiefly on relationships between respondents' participation in various types of consumer education activities and the extent to which they use specific information sources.

LITERATURE REVIEW

Existing research findings indicate that a positive relationship exists between the overall amount of information search undertaken and certain aspects of consumer and marketplace efficiency. Researchers have demonstrated, for instance, that information improves consumers' abilities to evaluate product quality (Sproles, Geistfeld, and Badenhop 1979) and is associated with lower monopoly power (Nelson 1970) and less price dispersion (Ratchford 1980; Morris and Bronson 1970; Stigler 1961). However, prior work relating to information acquisition deals mainly with factors affecting the overall amount of search consumers undertake. There has been little examination of factors that might affect the allocation of overall search effort among information sources. Yet information sources are known to vary in content and in the extent to which and the manner in which they are used (Langrehr 1979). Consumer and marketplace efficiency may be affected by the choice of information source. A better understanding of the factors that affect search time allocation decisions might, in turn, aid in understanding the contribution of such factors to consumer efficiency in the marketplace and to effective policy decisions.

Previous research with respect to consumer education suggests that a positive relationship exists between participation in consumer education programs and knowledge levels (Garman 1979; Moschis and Churchill 1977; Kelso 1975; Stanley 1975). However, the few attempts that have been made to investigate the relationship between consumer education and actual consumer behavior have met with little success (Bowers 1979; Hawkins 1977). Notable exceptions include a pilot study by Staelin (1978) and an experiment by Olson, Bisogini, and Thonney (1982) in which positive relationships were found between consumer education and product safety behavior and food choices, respectively.

SPECIFICATION OF VARIABLES, HYPOTHESES, AND ESTIMATING EQUATION

The theoretical model that serves as the framework for this research is based on Becker's household production model and is summarized in the Appendix. The following time allocation functions were estimated:

t sub li = f(CE, w, V, CC, PE, PR, U), where t sub li is time allocated to production of commodity 1 (information) by searching information source i, CE is a vector of consumer education activities, w represents the wage rate, V represents unearned income, CC represents a vector of consumer characteristics expected to be related to search effort, PE is a vector of variables representing prior experience in the marketplace and with purchasing household appliances, PR represents perceived risk, and U represents the consumer's perceived urgency of the purchase situation. The theoretical and operational definitions of variables entered in the model, hypothesized relationships between dependent and independent variables, and the empirical specification of the estimating equation follow.

Dependent Variables: Time Allocated to Information Source Use

Information search consists of the acquisition of information about a set of alternatives in a given choice situation. It may be deliberate acquisition for a specific purchase or unintentional exposure to messages in the environment. It may also be internal (memory scan) or external (purposeful acquisition of information from the environment). Only deliberate, external search was considered in this study.

The vector of dependent t sub 1 variables represents time allocated to prepurchase acquisition of consumer product information from various sources. Principal information sources identified by Maynes (1976) include advertising, sellers, friends, acquaintances, and independent product test reports (such as those found in Canadian Consumer, Consumer Reports, Consumers' Research, and Protect Yourself magazines). Stigler (1961) contends that search cost is proportional to the number of searches undertaken because the chief cost of search is time. Thus, if the value of a consumer's time1 remains constant across activities, it can be argued that search time is likely to be proportional to the number of searches conducted within a source category (for example, advertisements or product test reports). In addition, our respondents were asked to recall information search activities relating to a purchase that may have been made as long as 14 months prior to data collection (see the first paragraph of the Data, Results, and Discussion section). Evidence presented by Bishop, Jenrenaude, and Lawson (1975) and Fraisse (1968) suggests that recall of time spent on activities is less reliable than recall of frequency of occurance of an activity. Therefore, the number of searches of a particular source conducted is considered to be a reasonable proxy, on average, of search time allocated within an information source category.2 Specifically, the t sub 1 vector is represented by the following:

t sub li = (t sub 11, t sub 12, t sub 13, t sub 14), where t sub 11 is time allocated to search of noncommercial product test reports and is measured by the number of reports referred to prior to purchase, t sub 12 is time allocated to search involving appliance dealers and is measured by the number of dealers canvassed in person or by telephone, t sub 13 is time spent referring to advertisements about the product category and is measured by a respondent-reported estimate of the number of advertisements referred to prior to purchase, and t sub 14 is time allocated to consultation with friends, relatives, and acquaintances and is measured by the number of these individuals consulted. 1 In this study, predicted wage rates are used as a measure of the opportunity cost (value) of time. The technique used to predict this opportunity cost is described in the section on the wage rate. 2 Clearly major differences in time spent per search across information source categories may exist. That is, it is likely that a search involving a visit to a dealer will take considerably more time than a search by referring to an advertisement. It is, therefore, inappropriate to make comparisons, for instance of parameter estimates, across equations.

Independent Variables

The set of independent variables representing consumer and marketplace characteristics was chosen largely from factors identified in previous studies as potential contributors to search time allocation decisions. However, because the main purpose of this paper is to examine the relationships between participation in consumer education activities and the use of various information sources, detailed discussion is reserved for those variables relating to consumer education. Other variables are discussed briefly so results pertaining to consumer education variables may be examined in relation to the entire model.

Consumer education

The difference between the terms information and education is not distinct. However, Thorelli (1978) suggested that generic material, such as that relating to general buying strategies (for example, "how to" test drive an automobile), is correctly termed education. Material relating to specific products or services without generalizing among purchase alternatives was regarded by Thorelli as information.

Six types of consumer education are included in this research. The first is formal, institutional consumer education courses such as those offered in some high schools (denoted as CE1 in the empirical model). These courses tend to be general in nature, focusing on consumer skills relating to resource management. The second (CE2) includes consumer education workshops, seminars, or short courses relating to general buying strategies, such as those sponsored by educational institutions (excluding those specified above), social service agencies, consumer organizations, and adult education programs. The third (CE3) includes programs intended to disseminate printed consumer education materials relating to general buying strategies (such as pamphlets dealing with comparison-shopping skills). The fourth (CE4) involves those printed consumer education materials that deal with buying strategies specific to the product category. CE1, CE2, CE3, and CE4 are dichotomous variables equal to one if the respondent had participated in the consumer education activity and equal to zero otherwise.

The fifth type of consumer education (CE5) is the frequency of readership of consumer information and/or action columns in the popular press that may have served to increase consumers' awareness of certain issues and/or purchasing strategies, thereby contributing to their knowledge bases. The last type of consumer education (CE6) is the frequency of readership of consumer periodicals such as Canadian Consumer or Consumer Reports, which present educational "how to" material in addition to technical, product-specific testing information. CE5 and Ce6 are entered as categorical variables where: "Read Consumer Magazines Some" and "Read Consumer Columns Some" are equal to one if the respondent reads periodicals or columns sometimes and are equal to zero otherwise, and "Read Consumer Magazines Often" and "Read Consumer Columns Often" are equal to one if the respondent reads magazines or columns frequently and are equal to zero otherwise. Infrequent readership is the null condition in both cases.

Consumer education, particularly generalized forms such as courses and workshops, often stresses the importance of effective prepurchase search and would, therefore, be expected to increase search efforts of participants. Some forms of consumer education attempt not only to acquaint consumers with availability of lesser-known forms of information, but also to help consumers evaluate the relative quality of various information sources in the hope of limiting their reliance on biased and/or subjective sources and instead encouraging use of more objective and trustworthy sources.

If one accepts the normative evaluations of information source quality of Maynes (1976) and Cox (1967), friends and advertising may be considered biased and/or subjective sources, whereas product test reports are likely to be both more objective and trustworthy with respect to the data they provide. Therefore, one might expect to notice differences with respect to the effect of consumer education on the use of different information sources. It should be noted, though, that different information sources provide different types of information and may be a good source of one type but a poor source of another. For instance, it may be true that advertising is biased with respect to quality-related information but may be the most accurate and up-to-date source of price and location information. Similarly, product test reports are likely to be highly accurate and trustworthy with respect to quality evaluations but lack adequate pricing information.

Wage rate

The respondent's wage rate is used as a measure of the opportunity cost of time. Respondents employed outside the home (apporximately 46 percent of the sample) were asked to provide either their hourly wage rates or their annual, monthly, or weekly salaries and an estimate of the number of hours worked per week from which an hourly wage rate is computed. For respondents not employed outside the home, no market wage rate was observed, so one had to be estimated. However, wage rates imputed on the basis of those observed for employed respondents would suffer from selectivity bias as observed wage rates are conditional on the individual's participation in the labor force. To correct for this bias, an instrument was constructed from determinants of the respondent's employment status, a procedure developed by Heckman (1974). The probability that a respondent is employed is estimated as a function of factors associated with her home and market productivity. The inverse of the Mills ratio generated from this estimation is the instrument used to correct for selectivity bias. A second equation is then estimated for respondents who reported wage rates in which observed wages are a function of factors affecting the respondent's market wage and the inverse of the Mills ratio. In the event of missing values for any of the variables in this regression, the variable mean for employed respondents is substituted. Parameter estimates from this estimation are then used to predict wage rates for all respondents. This predicted wage is then used as the opportunity cost of respondents' time in estimating the time allocation functions.

According to the behavioral implications pertaining to the relationship between time and market goods that flow from the theoretical model, a change in the relative cost of an input will lead to reallocation of household resources. For example, an increase in the wage rate, ceteris paribus, is expected to lead to a substitution of market-purchased information or less time-intensive forms of information search for time-intensive types of search. One would expect, therefore, to see a positive relationship between wage rate and the number of product test reports consulted, the number of advertisements referred to, and the number of friends consulted, but a negative relationship between wage rate and the number of dealer visits.3 3 Theory also leads us to expect that an increase in the value of time will result in a decrease in the total amount of time spent searching. However, total search is not estimated in this study and because the units in which search is measured for different information sources are not comparable, the effect of a change in the opportunity cost of time on the total number of searches undertaken is not clear.

Unearned income

A respondent-reported estimate of the amount of income, in dollars, received by the household from sources other than the respondent's salary or wages is used to examine income effects. Previous studies, which have chiefly tested the effect of total income on extent of search undertaken, show mixed results. Some find a positive relationship between income and the extent of search (Carter, Andrews, and Hanna 1984; Claxton, Fry, and Portis 1974; Katona and Mueller 1985; Newman and Staelin 1971), which invites the conclusion that information is a normal good. Others find a negative relationship, often leading the researchers to conclude that income represents the opportunity cost of time (Ferber 1955; Mincer 1963; Punj and Staelin 1983). In this study, the wage rate is used as the best estimate of the opportunity cost of time while unearned income is used to determine income effects.

Consumer characteristics

Age and education have been found in previous studies to be related to the extent of prepurchase search activity, but the relationships have been inconsistent (Carter, Andrews, and Hanna 1984; Katona and Mueller 1985; Westbrook and Fornell 1979). It can be argued, though, that age and education can affect consumers' information processing skills and/or their tastes for information. Family size and the population density of the area in which respondents live are factors that might be expected to have an impact on consumers' abilities to use certain information sources by affecting their mobility and/or access to the information sources.

Age is entered as the age of the respondent in years, and education is the number of years of education experienced by the respondent. 9The latter variable is converted from the categorical variable reported in the original data set to a continuous variable by using the median of each category as the number of years of education received by the individual.

Family size is the number of individuals of all ages living in the household. Population density is the population of the area in which the respondent lives. This variable is also converted from a categorical to a continuous variable in the same manner as education.

Prior experience

Marketplace experience is represented by the respondent's tenure in the local marketplace (that is, the lenght of time in years she has lived in her current community). Product experience is represented by the number of appliances of a particular type the respondent had owned prior to the one purchased during the 14-month period preceding data collection. If, for example, a respondent had owned five television sets prior to purchasing her most recent one, prior experience would be entered in the data set as "5." Experience is expected to contribute to the consumer's knowledge of the marketplace, of the product, and of decision-making skills. Previous research indicates that prior purchasing experience is negatively associated with the overall amount of search undertaken (Punj and Staelin 1983). Because information sources are known to vary in content and consumer use (Maynes 1976), relationships between prior experience and use of different information sources may vary.

Perceived risk t Cox (1976) proposed that "the amount and nature of risk will define consumer information needs and consumers will seek out sources, types and amounts of information that seem most likely to satisfy their particular information needs" (p. 604). Risk is operationalized in this study, following the example of Cunningham (1967), as the product of respondents' degree of certainty that an event would happen and their evaluations of the amount of "danger" involved if the event did happen, each measured on a five-point Likert-type scale. Four types of risk--economic, performance, physical, and social--are identified, each of which has a total potential risk score of 25. The final total risk index consists of the sum of these four risk measures and so has a total potential score of 100.4 4 Cox (1967) argued that the type of risk (performance, economic, physical, or social) associated with a purchase situation will affect the sources of information that consumers use. An attempt was made to estimate a model in which each type of risk was entered separately but the function failed to converge. Therefore, it was necessary to use a single composite measure to overall risk.

Urgency of purchase

The urgency of the purchase sutuation as determined by (1) the respondent's perception of the urgency of her need for the appliance, (2) the situation that led to the purchase, and (3) the time available to shop for the item is expected to affect the overall amount of search undertaken and, perhaps, the source of information used. Each of tg these three aspects of urgency is measured on a four point scale, and the three scores are summed to arrive at a total urgency index. A categorical variable is created from the index where total urgency scores of 9 to 12 were entered as "high urgency," and scores of 3 to 4 represent "low urgency," which is the null condition.

Censoring bias correction factor

An instrumental variable, lambda, is also entered into the model to correct for sample censoring bias. Only respondents who both answered the questionnaire and had purchased one of the specified appliances during the 14 months prior to the survey date could be included in the analysis. If respondents who had purchased during the specified time are systematically different from nonrespondents and/or nonpurchasers,the model will suffer from sample censoring bias, and parameter estimates will be biased. Therefore, an instrumental variable is constructed that, when included in the final estimation, corrects the sample censoring problem. Specifically, the probability that a member of the original sample both responded and purchased was estimated as a function of family income size, hours employed outside the home, age, total family income, age of youngest child, education, and province of residence using a probit procedure. The inverse of the Mills ratio (lambda) generated from the probit procedure is included among the explanatory variables in the final time allocation estimations.

Estimating Equation

The empirical model for estimating information source use is: t sub li = a sub 0 + a sub 1 (wage rate) + a sub 2 (unearned income) + a sub3 (age) + a sub 4 (education) + a sub 5 (family size) + a sub 6 (population density) + a sub 7 (tenure) + a sub 8 (previous ownership) + a sub 9 (CE1) + a sub 10 (CE2) + a sub 11 (CE3) + a sub 12 (CE4) + a sub 13 (CE5-some) + a sub 14 (CE5-often) + a sub 15 (CE6-some) + a sub 16 (CE6-often) + a sub 17 (perceived risk) + a sub 18 (moderate urgency) + a sub 19 (high urgency) + a sub 20 (lambda), where the t sub li's are the dependent variables (time spent searching information category i) and consist of a set of ordered, discrete categories representing number of searches conducted of information source i as described in Table 1, and the a sub j are regression coefficients.

DATA, RESULTS, AND DISCUSSION

Data

The data used in estimation of the time allocation functions were collected by administering a self-completed questionnaire to a randomly drawn sample of 1,000 English-speaking members of an all-female national consumer panel operated by a major consumer products marketing firm in Canada. Six hundred and twenty-three questionnaires were returned, and three hundred and forty-five of these qualified for inclusion in the study with respect to purchase of one of a specified set of household appliances (including washing machines, clothes dryers, refrigerators, stoves, dishwashers, freezers, microwave ovens, food processors, air conditioners, televisions, stereo equipment, and video cassette recorders) during the 14-month period immediately prior to receipt of the questionnaire. Respondents included in the final sample are generally older, better educated, less transient, and had higher incomes than the general female English-speaking Canadian population. Summary statistics describing the sample can be found in Table 1 through Table 3.

Table : Dependent Variables

Table : Continuous Independent Variables

Table : Categorical Independent Variables

Analysis and Results

Analysis consists of separate estimation of the time allocation functions for each of the four information sources. This requires that one of two assumptions be made. The first possibility is that the allocation of time to one information source is independent of time allocations to other information sources. If, in fact, time allocation to one source is dependent upon time allocations to the remaining sources, then the set of explanatory variables for the t(li) variable being estimated should include the t(li) variables for the remaining sources.

One implication of this dependency, if it exists, is that each of the four estimating equations suffers from specification error due to omission of relevant explanatory variables, and reported coefficients would be biased. Further, inclusion of the other t(li) variables among the set of independent variables in each of the estimations implies that a simultaneous system exists that would be properly estimated using a regression technique such as indirect least squares or two-stage least squares. The effects of this bias are unclear. That is, there is no clear a priori evidence as to whether the coefficients would be overestimates or underestimates of the true parameters. Further, it should be noted that respecification as a simultaneous system would lead to an underidentification problem.

Alternatively, because the set of explanatory variables for each time allocation function is identical, it can be argued that Zellner's Seemingly Unrelated Regression applies. Judge et al. (1985) show that when the set of explanatory variables in a set of simultaneous equations is identical, the estimates obtained from a two-stage least squares estimation are identical to those obtained from OLS. However, it is not known if this property holds for maximum likelihood estimation.

The limited, discrete, and ordered nature of the dependent variables requires use of an estimation technique that generates efficient estimates under these conditions. The method employed in this study is ordered probit, a maximum likelihood technique for estimating regression models in which the dependent variable consists of a set of ordered, discrete categories.

Parameter estimates are reported in Table 4. They can be interpreted as follows: "a one unit change in the independent variable produces a change of B(i) standard deviation units in the probability of observing a particular value for the dependent variable" where B(i) is the parameter estimate from the probit estimation corresponding to the independent variable, X(i) (Zick 1980, p. 132). Consequently, ordered probit coefficients for the focus (consumer education) variables and for other variables for which parameter estimates are statistically significant at the .10 level or better are transformed into marginal effects for easier interpretation.5 These marginal effects are reported in Table 5.

Use of product-specific consumer education publications, moderate and frequent readership of consumer periodicals, and moderate readership of consumer information and action columns are all positively related to search of test reports at statistically significant 5The formula used to calculate marginal effects is basically the same as that derived by Zick (1980, p. 132) as follows: partial derivative of P with respect to x sub 1 = [1/2 pi e -LFP superscript 2/2] B sub i where: P = the probability of conducting a certain number of searches of a given information source, X(i) = the independent variable for which the marginal effect is being calculated, B(i) = the parameter estimate from the probit estimation corresponding to the independent variable, X(i), LFP = B(0) + B(1) (wage) + B(2) (unearned income) + B(3) (age) + B(4) (education) + B(5) (family size) + B(6) (population density) + B(7) (tenure) + B(8) (CE1) + B(9) (CE2) + B(10) (CE3) + B(11) (CE4) + B(12) (CE5) + B(13) (CE6) + B(14) (perceived risk) + B(15) (urgency) + B(16) (lambda). A marginal effect was calculated for each respondent and the sample means are reported in Table 5. levels. None of the consumer education variables displays a statistically significant relationship with dealer search. Positive and statistically significant relationships exist between use of advertising as a source of information and use of product-specific consumer education publications and both moderate and frequent readership of consumer columns. Use of product-specific consumer education publications and frequent readership of consumer columns are positively and significantly related to search involving discussions with friends, relatives, and acquaintances. It is interesting to note, however, that in no case is the relationship between participation in either high school- or community-based consumer education courses or workshops and search effort statistically different from zero.

In order to determine whether the inclusion of variables accounting for participation in consumer education contributes to the explanatory power of the model, a reduced model is estimated from which all consumer education variables (CE1 to CE6) are omitted. An F-test is then conducted to test the null hypothesis that none of the coefficients with respect to consumer education variables is significently different from zero. In only one case, that of the model for dealer search, is the null hypothesis accepted. For the other three models, it is concluded that inclusion of the education variables do improve the explanatory power of the model.

Other results of interest are those pertaining to experience and urgency of purchase variables. Tenure (marketplace experience) is negatively associated with search involving product test reports. Age and previous ownership (product experience) are negatively related to the number of friends and relatives consulted. These results suggest that, at least in some circumstances, experience reduces the propensity for consumers to engage in deliberate, external search. For instance, it appears that one's own experience may be substituted for the experience of friends, relatives, and acquaintances in the marketplace and/or in purchasing appliances in general. The results also suggest that marketplace and purchasing experience may make consumers more efficient at conducting search.

Contrary to expectations, results for the opportunity cost (wage rate) variable are not statistically significant nor are they always of the expected sign.

Unearned income is statistically significant only in the dealer search model. Recall that, because the value of time is accounted for in the wage rate variable, the parameter estimates on the unearned income variables represent the pure income effects. These results suggest, then, that dealer search is a normal good but that search involving the remaining information sources is unresponsive to changes in income.

The relationships between population density and search involving advertisements and search involving friends, relatives, and acquaintances are positive and statistically significant, while no relationship statistically different from zero exists between population density and search of product test reports or search of dealers. These results suggest that larger population centers offer more opportunities for consumers to consult with friends, relatives, and acquaintances and to increase exposure to advertising but that there is no increase in accessibility of dealers or product test reports as population density increases.

High and moderate levels of urgency surrounding the purchase situation are positively associated with consultations with friends and relatives. This suggests that where the need for an appliance is urgent and/or time for search is short, friends and relatives who offer sufficiently easy access to information are the preferred source. On the other hand, a statistically significant negative relationship exists between high levels of urgency and use of advertising as a source of information that, at first blush, seems counterintuitive. A priori expectations are that urgent need for the appliance and shortage of search time would create a need for just the kind of information best provided by advertising (price and availability) and so lead to greater use of advertising as an information source. However, the information provided by advertising may not be so much deliberately sought as acquired through unintentional exposure during the prepurchase period while viewing television, listening to radio, or reading newspapers and magazines. If this is the case, it is logical that, when confronted with an urgent purchase, consumers acquire less information relevant to the purchase from advertising than when the purchase is not urgent. Alternatively, the information available from advertising is not of the type that is needed and/or desired when the purchase situation is urgent.

Increased search is believed to contribute to marketplace efficiency because it is associated with less price dispersion (Kelso 1975; Devine and Hawkins 1972; Stigler 1961) and lower monopoly power (Nelson 1970). Increased search has also been shown to be associated with better decision making (Ratchford 1980; Sproles, Geistfeld, and Badenhop 1978, 1979), a key element of consumer efficiency. Because one of the objectives of consumer education efforts is to encourage more active prepurchase information search (British Columbia Ministry of Education 1983), the results of this study suggest that policy-makers who have promoted consumer education as a means of improving consumer and marketplace efficiency with minimal direct intervention in the marketplace have reason to be encouraged. However, it seems that education delivered via printed consumer education materials, such as pamphlets, newspaper columns, and consumer periodical articles relating to general and product-specific buying strategies, is preferred to that delivered in the classroom.

Consumers who voluntarily participate in consumer education activities might also be predisposed to conducting extensive information search and may, in fact, have engaged in consumer education as part of their search process. No attempt was made in this study to determine whether or not consumer education participation had been mandatory for any of the respondents and, if so, whether or not any difference in search behavior existed between those respondents who had participated in consumer education voluntarily and those whose attendance was mandatory. Langrehr (1979) raises the question as to whether differences might exist in consumer competency levels of students who are required to take a consumer education course and those who do not take the course. His study, using a small sample, concludes that a required consumer education course (as opposed to an elective course on economic principles) has a positive effect. Differences may also exist between those who are required to take the course and those who do so voluntarily. These differences in consumer education program characteristics merit further attention.

Consumer education, like general education, is likely to contribute to consumers' decision-making abilities in two ways. It may create awareness of and preference for information that would be expected to increase the amount of time consumers spend seeking information. It may also improve the efficiency with which consumers handle information that would likely decrease the time consumers must devote to gathering a given amount of information. These two counteractive effects are not examined separately in this study. However, ignoring the possibility that searchers choose to participate in consumer education, a net positive relationship between consumer education and the amount of information search conducted within a source category might be interpreted as indicating that the taste and preference for information effect is dominant. A net negative relationship, on the other hand, may suggest that the efficiency effect is dominant. Further study of the separate effects of education on tastes and preferences for information and efficiency of search also merits attention in future research.

The desired end result of public policy initiatives is (or should be) maximization of public welfare--that is, efficient resource management by consumers and producers alike. In this work, prepurchase information search is assumed to have a positive effect on consumer and marketplace efficiency. Better evidence of the link between search and decision-making efficiency would make studies such as this one more meaningful. Further effort should be directed toward development of reliable and useful measures of time and search efficiency or productivity. Sproles, Geistfeld, and Badenhop (1978, 1979) and Sproles (1983) have investigated the area of consumer efficiency and have made some recommendations that warrant reinforcement. These include (1) further development of criteria for determining choice efficiency for objective evaluation of product quality, (2) assessment of choice efficiency for decisions relating to products of different involvement levels, and (3) investigation of the efficiency. The need for resolution of the methodological problems associated with measurement of search behavior itself could be added to these recommendations.

An integral part of evaluating search efficiency is assessing objectively, and preferably quantitatively, the relative quality of information sources. It would be most useful to have more than the intuitive, qualitative evaluations of information source quality that were available for this research.

CONCLUSIONS

This research provides tentative but encouraging results with respect to evaluating the impact of policy initiatives on consumer behavior. The usefulness of the model is demonstrated. The potential applications of the model for evaluating the impact of other policy initiatives deserve attention.

A number of positive relationships among participation in consumer education involving written educational materials and the extent of search involving product test reports, advertisements, and friends are reported. To the extent that increased search has been shown to improve consumer decision-making and marketplace efficiency, consumer education may be said to be of benefit to consumers.

For three of the four information sources considered, consumer education received from informal educational materials performed better, statistically, than consumer education received in a classroom setting. These findings carry interesting implications with respect to the relative efficacy of the two general types of consumer education efforts and, consequently, for decisions about program funding and program planning and implementation. However, due to the failure to distinguish between the effects consumer education has on preferences for spending time conducting search and the efficiency with which consumers conduct search, this conclusion should not be overemphasized without more detailed study of this issue.

It is further concluded that a need for development of effective measures of choice efficiency, of search behavior, and of information source quality exists. This would make the results of investigations, such as the one reported here, much more meaningful.

The results of this study relate to search behavior associated with major household appliances. How these results would be affected if the study were extended to include relatively smaller purchases is unclear, a priori. The household appliances included in this study exhibit at least some of the characteristics of goods that Engel and Blackwell (1982) refer to as "high involvement." These characteristics include high purchase cost, high ongoing operation and maintenance costs, technological complexity and, therefore, uncertainty and risk. Engel and Blackwell suggest that high involvement purchase decisions are likely to involve more extensive external information search than low involvement purchase decisions. They further suggest that consumer education is likely to modify the criteria by which purchase alternatives are evaluated for high involvement purchases but not for low involvement purchases. This suggests that consumers are likely to engage in less search overall and that consumer education is likely to have less influence on search behavior for smaller purchases than for those examined in this study.

According to economic information theory, as laid out by Stigler (1961), the optimal amount of search for consumers to undertake is that for which the marginal benefits of search are equal to the marginal costs. Benefits depend, in part, upon the cost of the good to be purchased: the higher the cost of the good, the greater the potential savings from search. Costs depend, in part, upon the efficiency with which consumers search: the more efficient they are, the less it costs them to obtain a given amount of information (or the more information they obtain for a given cost). This suggests two possible outcomes. The optimal amount of search for small, less costly items is less than for large, more costly ones because the marginal benefit of search is likely to be lower for the less costly items. In addition, if consumer education causes consumers to become more efficient at search, and, hence, lowers their marginal costs of search, the optimal amounts of search would be expected to be greater for all items, including smaller, lower-cost goods. The evidence available from this study is insufficient to predict which of the possible outcomes would dominate, although it seems likely that purchase decisions involving smaller, less costly items than those examined in this study would involve less extensive search, overall. No conclusions are possible with respect to the effect that consumer education might have on search of individual information sources when small, less costly goods are involved. This is another area of interest for further study.

The characteristics of consumer services are considerably different from those of goods, as are the contributions different information sources make to the decision process for services. This study offers no evidence that explains search behavior with respect to services. This, too, is a fertile area for future research.

It is suggested that time allocated to searching one information source may not be independent of time allocated to searching other sources. However, the implications of any interrelationship among time allocation decisions, if or where they exist, are not clear. One might imagine, for instance, that if the consumer were to refer to product test reports first, the information gained there might narrow his or her search to a few top-rated brands and hence reduce subsequent search involving one or more of the remaining information sources. If, however, the consumer first asked friends' advice and then obtained information from a product test report that conflicted with that advice, subsequent search might be increased in order to resolve the disparity. That is, the effect of one search on subsequent searches of the same or alternate sources might well depend on the order in which searches are conducted, the degree of congruity of information gained from alternate searches, and the degree of confidence that consumers attach to a particular source. These questions are clearly beyond the scope of the present study but present intriguing and important areas for future research.

APPENDIX

Theoretical Model

The theoretical basis of this research is Becker's (1966) household production model. Becker assumed that households derive satisfaction from the consumption of commodities that are produced by combining goods and services purchased in the marketplace, householders' time, and the services of household capital. Purchases of goods and services are constrained by the household's income, and time allocation decisions are constrained by the time available (i.e., 24 hours per day). Mathematically, the model can be expressed as follows: (1) U = u(I, Z; TP) subject to (2) I = F sub 1 (X sub (li), t sub (li); HK), (3) Z = F sub 2 (X sub 2 t sub 2) (4) summation of I from 1 to N P sub 1 X sub 1 + P sub 2 X sub 2 = Y = T sub W W + V, and (5) summation of I from 1 to N + sub 1, + T sub 2T + sub W = T, where: I = a stock of consumer-held information, service flows from which contribute to consumer satisfaction,6 Z = all other commodities in the household's commodity set, TP = a set of fixed tastes and preferences represented by socioeconomic characteristics, X sub (li) = a vector of market-purchased inputs to the production of information, t sub (li) = a vector of time inputs to information production, HK = a stock of fixed human capital inputs, X sub (2) = market-purchased inputs to production of all other commodities, t sub (2) = time inputs to production of other commodities, P sub (li) = the price of market-purchased inputs to information production, P sub (2) = the price of market-purchased inputs to production of other commodities,

6A stock is a quantity of a commodity existing at a moment in time that provides flows of services. Consumers derive satisfaction from consumption not of the stocks of inputs per se, but rather from the services flowing from the stocks. The flow of services is assumed to be proportional to the stocks. Y = total household income that, Becker argues, is comprised of income earned through employment in the workforce (t sub (w) W where T sub (w) represents the time spent in the workforce and W represents the wage rate) and income derived from sources other than employment (V), and T = the total time available.

Following standard procedures for constrained maximization of the utility function (equation (1)) subject to equations (2), (3), (4), and (5), the following time allocation functions are derived: (6) t sub li = f sub 3 (P sub 1, P sub 2, W, V; HK, TP) for i = 1 to n.

In the context of the empirical model specified, several substitutions are made. Taste and preference variables include perceived risk (PR) and elements of the Consumer Characteristics (CC) vector such as family size. The stock of human capital is proxied by prior experience (PE), the elements of the Consumer Education (CE) vector, and some elements of the Consumer Characteristics (CC) vector such as age and education. Urgency (U) and population density, and element of the Consumer Characteristics (CC) vector, are considered to be indicators of the price of market-purchased inputs to the production of information, since the cost of conducting search is related to the size of the area in which one lives and to the urgency of the purchase situation. It is often assumed, in studies that use cross-sectional data as this one, that the price of all other goods is fixed at a point in time and, therefore, is not included among the explanatory variables.
COPYRIGHT 1989 American Council on Consumer Interests
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1989 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Fast, Janet; Vosburgh, Richard E.; Frisbee, William R.
Publication:Journal of Consumer Affairs
Date:Jun 22, 1989
Words:7168
Previous Article:Individual consumption within the household: a study of expenditures on clothing.
Next Article:An indirect test of children's influence on efficiencies in parental consumer behavior.
Topics:


Related Articles
Consumer response to telecommunications deregulation: the equal access decision.
Time as a direct source of utility: the case of price information search for groceries.
Consumer information and education effects on knowledge and choice of fire resistant upholstery.
Shop around when borrowing.
Sources of information, consumer attitudes on nutrition, and consumption of dairy products.
Consumer vulnerability to fraud: influencing factors.
Consumers' use of nutritional labels while food shopping and at home.
Relationships among Information Search Activities When Shopping for a Credit Card.
Attribute framing in former price comparison advertisements (1).
Financially distressed consumers' information search for retirement plans.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters