A Hierarchical Model of Source Effect in Retail Newspaper Advertising

James G. Barnes, Memorial University of Newfoundland, Canada
ABSTRACT - The relative source effects of retail store name, brand name and newspaper on consumer reaction to retail newspaper advertising are examined. Significant retail store and brand effects are found. Mean scores on three dependent variables reveal that retail advertisers can utilize different brand strategies in their newspaper advertising in order to obtain more positive consumer reaction.
[ to cite ]:
James G. Barnes (1978) ,"A Hierarchical Model of Source Effect in Retail Newspaper Advertising", in NA - Advances in Consumer Research Volume 05, eds. Kent Hunt, Ann Abor, MI : Association for Consumer Research, Pages: 235-242.

Advances in Consumer Research Volume 5, 1978      Pages 235-242

A HIERARCHICAL MODEL OF SOURCE EFFECT IN RETAIL NEWSPAPER ADVERTISING

James G. Barnes, Memorial University of Newfoundland, Canada

ABSTRACT -

The relative source effects of retail store name, brand name and newspaper on consumer reaction to retail newspaper advertising are examined. Significant retail store and brand effects are found. Mean scores on three dependent variables reveal that retail advertisers can utilize different brand strategies in their newspaper advertising in order to obtain more positive consumer reaction.

INTRODUCTION

It has been generally accepted that the effectiveness of an advertising message in accomplishing the objectives of the communicator will depend in part on the attitudes of the audience toward the originator of the message and toward various other sources involved in its transmission. The influence of the various sources in affecting consumers' reactions to advertising will depend on a number of attitudes towards these sources. These attitudes relate to feelings of affection, admiration, fear, power, and credibility. Hovland, Janis and Kelley (1953) concluded that such attitudes are the result of a learning process.

Research by marketing theoreticians into the concept of source effect has shown that the reactions of an audience to a communication are generally affected by the audience's perception of the communicator's expertise and trustworthiness in the communications context (Cox, 1963). Aaker and Myers (1975) have also observed that in choosing a particular spokesman to represent a brand, consideration should be given to the credibility (impartiality and expertness), attractiveness (prestige), and power of the source. They add that credibility and attractiveness are the two most appropriate dimensions of source for advertisers.

One failing, however, of much research which has been done in the advertising area on source effect is that studies have generally examined the effect of a single source. A number of studies have examined the effect on a consumer's reaction to a marketing communication of perception which a consumer has of the salesman delivering the message. Jacoby, et al. (1974) have examined the effect of a smiling source. McDougall and Fry (1975) investigated the effect of retail store image on consumer reaction to retail advertisements.

A number of authors have, however, commented that the concept of source effect should not necessarily be viewed in the context of the single source. Hovland, Janis and Kelley (1953) have stated that regardless of whether a source is perceived as the originator of the message, as the channel through which the message passes, or as an endorser who is featured in the message, the same basic principles apply in that the attitudes of members of the audience toward either of these sources may influence the manner in which the message is received and behavior which may result from receipt of that message. It should be clear that, in the advertising context, a number of sources are at work simultaneously in influencing the receipt of a marketing communications message. Fuchs (1964) examined the effects of two components (company and magazine) of what he considered to be a "total source" with several natural components. He went on to speculate that source is likely "n-dimensional" and that it is a differently-constructed total complex for each individual. This suggests that the way in which an individual perceives the several sources associated with a particular advertisement will depend on his or her past experience with the manufacturer of the product, with the retailer, and with the 'medium vehicle' used to transmit the message. since the experiences of individuals will vary, the total effect of the complex of sources will likewise vary across individuals.

Hansen (1972) also suggests that the effect of the various source components will be variable. He concludes that the individual's perception of the source will depend partly on the medium which carries the message and partly on the company which initiates it. The relative influence of Hansen's two source components will vary with the consumer's familiarity with each. He suggests that where a nationally-known manufacturer advertises in a large magazine, the source effects may be ascribed mainly to the company but where the advertiser is not well-known and the publication somewhat more dubious, the source effect may be based more on perception of the magazine.

This multi-leveled concept of source effect is largely substantiated in the research of Fuchs (1964). Using advertisements featuring unknown products, he found that subjects' attitudes towards these products were influenced by the company sponsoring the advertisements and by the magazines in which they were carried. He also found that the greatest attitude change towards the product was accomplished by the combination of high-prestige company and high-prestige magazine [Fuchs (1964) rated each company and magazine using five semantic differential-like scales: good-bad, reliable-unreliable, trustworthy-untrustworthy, prestigeful-not prestigeful, and reputable-disreputable. Companies and magazines which scored high on these scales were labeled 'high-prestige' and those which scored low were labeled 'low prestige'. Possibly more appropriate labels would have been 'positive-image' and 'negative-image'.], and the smallest attitude change toward the product was attributable to the combination of low-prestige company and low-prestige magazine. The attitude towards the product changed as a linear function of the prestige of the two source components.

These research results suggest that in retail advertising where the brand name of the advertised product is featured, consumer perceptions of the advertised product may be influenced by perceptions of the manufacturer. Where the consumer is not familiar with the brand name or where the brand name of the advertised product is not present, perception of the quality and value for the money involved in the purchase offer may be influenced by the consumer's attitude towards the retail store. It may be hypothesized further that only when the consumer is not familiar with either the brand name of the product or the retail store sponsoring the advertisements is the perception of the purchase offer likely to be influenced significantly by the medium vehicle which carries the advertisement.

THE HIERARCHICAL MODEL

It is useful to examine the several sources which may operate in an advertising context, and to distinguish sources of the product from sources of the message. In the case of a national advertiser whose advertising is carried in a popular consumer magazine, the company may be considered the primary source of both the product and the message. The magazine which carries the advertisement may be considered a secondary source of the message. If that advertisement had featured a well-known celebrity endorsing the product, that person could have been considered a tertiary source of the message. In the case of retail advertising where the brand name of the advertised product is supplied, we may regard the manufacturer of the product as the primary source of the product and the retailer as the secondary source of the product. Various middlemen may be involved in the actual physical movement of the product from manufacturer to retailer, but these are not relevant to the formation of consumer perceptions of the advertised product.

Also, in this branded retail advertisement, the retail store may be considered the primary source of the advertisement since it is seen by the consumer as the sponsor. The newspaper which carries the advertisement is a secondary source of the message and any persons featured in the advertisement as endorsers of the featured product should be considered tertiary sources of the message. Others, such as advertising agencies, may be involved in the preparation of the advertising message, but these are again not relevant to the formation of a consumer reaction to the advertised product.

In retail advertising where the brand name of the manufacturer of the product is not present, as is the case with much fashion advertising, the retail store is perceived as the primary source of both the product and the advertised message.

This study is designed to test for the presence of a hierarchical source effect in retail newspaper advertising. Because the use of an endorser in retail advertising is rare, the study concentrates on the relative impact of the manufacturer's brand name, the retail store's name, and the newspaper in influencing consumers' reactions to retail newspaper advertising. Previous research on the existence of a dual source effect (Fuchs, 1964) suggests that different source effects will operate in different contexts. For example, it is possible to test whether a brand name with a positive image is sufficient to offset a negative attitude towards a retail store in influencing a consumer's purchase decision. Similarly, it is possible to determine whether a retail store which was recently established or which enters a new market area can produce positive responses to its advertising by featuring brand name products which consumers hold in high regard. The different contexts in which a retail advertisement might appear are largely controllable by the retail advertiser. While the retailer must, for the most part, operate within the constraints of his existing store image, which may be positive, negative, or non-existent in the case of a new store, he does have some control over the brand names which he chooses to feature in his advertising, and over the newspapers in which the advertising appears. In making the selection of brand name products to appear in his advertising, the retailer may choose to feature brands which have a positive image, brands which are considered to have a negative image in the consumer's mind, or brands with which the consumer may be unfamiliar, or he may decide not to feature a brand name product in his advertising at all. Similarly, in many metropolitan markets, the retailer has a definite choice to make with respect to the newspapers in which his advertising is placed.

The implications for the retail advertiser should be obvious. Since the retailer is able to manipulate at least two of the source variables, the results of this study should permit questions such as the following to be answered:

1. Where a retailer has an established reputation and image in the trading area, is the newspaper selected likely to influence consumers' reactions to an advertisement?

2. Is it possible for an established retailer to detract from the effectiveness of his advertising by featuring in an advertisement a brand name which has a negative image?

3. In a situation where a retailer is moving into a new market and consequently has no established image, is it possible to improve reaction to newspaper advertising by featuring brand names which have a positive image and/or by advertising in a newspaper which has such an image?

4. Can a retailer whose store has an established negative image improve reaction to newspaper advertisements by featuring brand names which have a more positive image?

HYPOTHESES

In retail newspaper advertising, there are many different possible combinations of brand name, store name, and newspaper. It is, therefore, reasonable to hypothesize that different combinations of these variables will produce different consumer reactions. For example, a manufacturer's brand name may be present or absent; if present, the brand name may be known or unknown; and, if known, it may convey a positive or a negative brand image. Similarly, the name of the retail store appearing in the advertisement may be known or unknown; and, if known, the store image may be positive or negative. Finally, in cities where more than one daily newspaper operates, a selected newspaper may be held in high regard, or may have developed a negative image. A series of hypotheses have been developed based on the research reviewed above and on earlier study conducted by the author (Barnes, 1975).

The hypotheses to be tested in this research are as follows:

1. That the perceived source of the product will have a greater effect on the reaction than will the perceived source of the message;

2. That the brand name of the advertised product will have a significant effect on subjects' reactions;

3. That, in situations where no brand name is present or where the brand name in unknown, the retail store effect will predominate;

4. That the effect of the newspaper in which the advertisement appears will be significant only. in situations where the brand name is either absent or unknown and the retail store is unknown.

RESEARCH DESIGN

Independent Variables

The three main independent variables whose effect on consumer reactions to retail advertisements is examined in this study are the three elements of source effect discussed above. The three main treatments are, therefore, manufacturer's brand, retail store, and newspaper. In order to incorporate the various possible combinations of brand, store, and newspaper, the following levels of each of the three main treatment variables are incorporated into the study:

BRAND NAME:

1. positive brand image

2. negative brand image

3. unknown brand name

4. no brand name present

RETAIL STORE:

1. positive store image

2. negative store image

3. unknown store name

NEWSPAPER:

1. positive image as source of information

2. negative image as source of information

The research involves a 4 x 3 x 2 design which necessitated the preparation of twenty-four different versions of a print advertisement, each of which was exposed to subjects in one of the twenty-four cells of the design. In addition to testing the effects of these three main treatment variables, data were gathered on subjects' attitudes, interests and opinions concerning retail shopping and the mass media, attitudes towards specific stores, local newspapers and advertising in general, and certain demographic and socioeconomic information relating to the respondent and to her family.

Dependent Variables

In accordance with a suggestion made by Olson (1974), this study incorporates three dependent variables: perceived believability of the purchase offer; perceived value for the money; and motivation to act. The position of a respondent on each of these variables is considered to represent a component of her total reaction to the advertisement to which she is exposed. These three components of reactions have been included because this provides a more accurate indication of the subject's potential purchase behavior with respect to the advertised product. Insofar as the subject's reaction to the advertised purchase offer may be considered to represent her attitude toward the product in the context of the advertisement, the three-fold representation of that reaction may also be considered to correspond to Rokeach's conceptualization of attitude as "an organization of beliefs" (Rokeach, 1972). Rokeach conceived of each belief in an attitude organization as having three components: cognitive; affective; and behavioral. The dependent variables in this study may be interpreted as corresponding to these three components. The subject's believability of the advertisement may be considered to represent a person's knowledge about what is true or false, good or bad, desirable or undesirable. Perception of value for the money represents an affective dimension of the subject's reaction to the advertised product. The third dependent variable, motivation to act, clearly represents a behavioral dimension.

Questionnaire Design

The questionnaire employed in this study contained four major sections as follows:

1. Statements relating to subjects' activities, interests, and opinions concerning retail shopping and the mass media. Subjects indicated their level of agreement with each statement on a seven-point, Likert-type scale.

2. Questions relating to subjects' attitudes to- ward specific retail stores in the trading area, toward local newspapers, and toward advertising in general.

3. A retail newspaper advertisement to which subjects were asked to express their reactions in terms of the three dependent variables described above.

4. Demographic and socioeconomic questions relating to the respondent and to her family.

The three criterion variables (perceived believability, perceived value for the money, and motivation to act) were operationalized in the third part of the questionnaire. This part of the questionnaire contained twenty-one statements, each of which forms part of one of three scales, representing believability, value for the money and motivation to act. Subjects were asked to respond to each of these twenty-one statements on a seven-point, Likert-type scale ranging from definitely agree to definitely disagree.

The advertisement which appeared in the third part of the questionnaire featured sheets and pillow cases, as products with which female consumers would be familiar. The actual advertisement was designed by the advertising department of a local newspaper in such a way as not to reflect the advertising style of any particular retailer. The advertisement was prepared in twenty-four different versions, with each version featuring the name of a different retail store, a brand name for the advertised products, and the folio of a local newspaper.

The retail stores, brand names and newspapers featured in the advertisements were selected and labeled positive image' and 'negative image' based on research previously conducted in the area. Earlier store image studies were used to identify a local department store which engages in very little sale advertising and which has a very positive image in the market. This store was labeled 'positive image'. A discount store which employs a strong promotional advertising approach and which has a rather negative image was labeled 'negative image'. The assignment of these stores to their respective levels of the retail store variable was substantiated by the store image data gathered as part of this survey. A fictitious store name was used in one-third of the advertisements to represent the 'unknown' store level of the retail store variable.

A preliminary survey among female heads of households obtained data on the image of various brands of sheets and pillow cases. A brand which demonstrated high awareness and a high quality image was labeled 'positive image' and a second brand also demonstrating high awareness but a low quality image was labeled 'negative image'. A third fictitious brand name was used to represent the 'unknown brand' level of the brand name variable. One-quarter of the advertisements contained no brand name.

Earlier research studies on newspaper readership and image were used to assign the two local daily newspapers to the 'positive image' and 'negative image' levels of the newspaper variable. This assignment of the newspapers was also substantiated by data collected in this survey. Each of the twenty-four versions of the advertisement employed the same copy and prices.

Data Collection

The sample was drawn from the retail trading area of an Eastern Canadian city. This city was selected because it offers a broad range of large department and discount stores and because it has two daily newspapers. A total of 2,400 questionnaires was mailed to female heads of households selected randomly from recent voters' lists. Completed questionnaires numbered 429, of which 402 were useable. One hundred copies of each version of the advertisement were mailed and the 402 useable questionnaires returned resulted in cell sizes ranging from fifteen to twenty-five.

ANALYSIS

Subjects' responses to the twenty-one statements relating to the advertisement were analyzed using principal components analysis (Nie, 1975) and three factors resulted which were clearly interpretable as the three criterion variables discussed above. Sum scores on these factors constituted subjects' scores on the dependent variables. Results of this principal components analysis are presented in Table 1.

TABLE 1

FACTORS PRODUCED FROM PRINCIPAL COMPONENTS ANALYSIS OF 'ADVERTISEMENT' STATEMENTS (n=402)

The sixty "life style" statements in the first part of the questionnaire were also subjected to principal components analysis and eight factors were obtained with eigenvalues greater than unity. These eight factors are interpretable as dimensions of consumer behavior relating to retail shopping and the mass media and are important insofar as they may influence consumer reaction to advertising. The results of this principal components analysis are presented in Table 2. These eight "life style" factors and a number of demographic variables were included in the analysis as covariates since it is hypothesized that reaction to an advertisement is not a function only of the brand name, store name and newspaper associated with that advertisement, but also of certain demographic and life style characteristics of the respondent herself.

TABLE 2

FACTORS PRODUCED FROM PRINCIPAL COMPONENTS ANALYSIS OF 'LIFE STYLE' STATEMENTS (n=402)

The data were then submitted to the BMD-12V Multivariate Analysis of Variance and Covariance program (Dixon, 1973). Since the twenty-four versions of the advertisement were distributed randomly to subjects, there was no control over the size of cells into which the returned questionnaires would fall. In order to comply with the constraint of the BMD-12V program regarding equal cell sizes, it was necessary to select subjects at random to be eliminated from those cells containing more than fifteen subjects. In all, forty-two subjects were dropped, thereby creating a sample of 360 subjects, equally distributed over the twenty-four cells of the design.

It has been possible in this study, to control the exposure of subjects to the three main independent variables. However, it is also hypothesized that a subject's reaction to an advertisement is influenced by certain demographic and life style characteristics of the subject herself. While it is not possible to select and group together subjects with similar individual characteristics, it is possible, through use of analysis of covariance (ANCOV) to determine the effect on the dependent variables of subjects' positions on these individual characteristics.

The multivariate version (MANCOV) is a useful technique in situations where more than one dependent variable is involved and the researcher is interested in determining the effect of the various independent variables on the total set of dependent measures. In this study, while the effect of the independent variables on each of the dependent measures is of considerable interest, so too is the effect of these independent variables on the subject's overall reaction to a retail advertisement.

A series of analyses of covariance were performed containing a total of sixteen covariates: eight 'life style' variables and eight demographic variables.

Results

Table 3 contains the results of the analyses of covariance performed. In the first three columns of Table 3 are reported the results of the univariate analyses of covariance performed with each of the three elements of subject's reaction as the dependent variables. The fourth column contains the results of the multivariate analysis of covariance which indicates the effect of various independent variables on subjects' overall reaction to the advertisement. The univariate ANCOV, while producing indications of effect on subjects' reactions of the retail store, brand and newspaper variables, do not take into consideration the inter-correlations among the three criterion variables and do not consider whether there are significant differences between the levels of the independent variables in terms of overall reaction to the advertisements, that is, over the set of three criterion variables.

TABLE 3

SUMMARY OF UNIVARIATE AND MULTIVARIATE ANALYSES OF COVARIANCE (n=360)

The results of the four analyses indicate that no significant differences are evident between subjects exposed to the two levels of the newspaper variable. The newspaper in which the advertisements appeared had no significant effect on subjects' reaction to the advertisements. The other two main effects did, however, produce significant results.

Subjects' perceived believability scores were significantly affected by the retail store sponsoring the advertisement (p < 0.05) and by the brand name featured in the advertisement (p < 0.10). Two covariates were also significant, suggesting that the extent to which subjects believed the content of the advertisement was influenced by their general attitude toward advertising (p < 0.001) and by the extent to which they enjoy shopping (p < 0.10).

Perception of value for the money was significantly affected by only one main effect. The brand name appearing in the advertisement was significant at the 0.10 level. Only one covariate was significantly related to perception of value - subjects' orientation toward department stores (p < 0.10).

The extent to which subjects were motivated to act was also significantly affected by the brand name (p < 0.10). Three covariates were significant in this analysis. Motivation to act scores were significantly related to bargain orientation (p < 0.10), orientation toward department stores (p < 0.05), and enjoyment of shopping (p < 0.05). This analysis also produced a significant two-way interaction (store x brand) effect. Examination of the motivation to act mean scores reveals that the retail store with a positive image obtained its highest scores where it featured a positive-image brand name or where no brand name appeared. Both the negative-image retailer and the unknown retail store obtained high motivation to act scores from advertisements featuring positive-image brands. The ability of the store with a negative image to motivate subjects dropped off somewhat where a negative-image brand or an unknown brand was featured and declined dramatically where no brand name was present. This interaction effect is depicted in Figure 1.

FIGURE 1

TWO-WAY (S x B) INTERACTION MOTIVATION TO ACT: ANALYSIS OF COVARIANCE

The multivariate analysis (MANCOV) indicates that subjects' overall reaction to the advertisement was significantly influenced by the store sponsoring the advertisement (p < 0.05) and by the brand name featured in the advertisement (p < 0.10). All four of the co-variates which were significantly related to overall reaction were 'life style' variables: attitude toward advertising (p < 0.001); bargain orientation (p < 0.01); orientation toward department stores (p < 0.05); enjoyment of shopping (p < 0.10).

Although no significant newspaper effect was found in any of the analyses, mean scores on the three dependent variables by levels of the newspaper variable were examined primarily to explore the possibility that a newspaper effect may be operating where the brand and store effects are absent (that is, where the retail store is unknown and the brand is either unknown or absent). This comparison of mean scores for the two levels of the newspaper variable produced no significant t-statistics. Even where the retail store was unknown and the brand was unknown or absent there were no significant differences between the mean scores produced by the positive-image newspaper and those produced by the negative-image newspaper.

Much information can be obtained by examining mean scores on the dependent variable across combinations of store and brand. These results are presented in Table 4. In the case of the positive-image retail store, while t-tests failed to produce any significant differences among the mean scores of the four levels of the brand variable, there is a definite pattern in evidence. For example, the positive-image store consistently, over all three dependent variables, produced its best scores where its advertising featured either a brand name with a positive-image or no brand name at all. Clearly, in the latter case, the store is able to trade on its established positive store image. Reaction to its advertisements was less positive in those situations where it featured either a negative-image brand or a brand name with which respondents were unfamiliar.

The advertisements for the retail store with a negative image produced even more significant results. In general, this store's advertising was considered less believable than that of the positive-image store. Its most believable advertisements were those which featured the positive-image brand. In terms of indicating good value for the money, the negative-image retail store's advertising produced a significantly more positive reaction where it featured a brand name with an established positive image than where it featured no brand at all (p < 0.05). Similar results were found where motivation to act was the dependent variable. In this latter case, a significantly better reaction was obtained even where the negative-image brand or an unknown brand was featured as compared with no brand. It would appear that the negative image of the retail store is imputed to the advertised product in that situation where the product is unbranded.

In the case of a retail store concerning which the consumer has no established image, possibly a new store being established in a market area, it was found that its advertisements were perceived as significantly more believable when the positive-image brand was featured as compared with the situation where an unknown brand was present in the advertisement (p < 0.01). The same significant result holds true for the other two dependent variables. An advertisement for an unknown store was more likely to motivate a respondent to act even where it featured a brand with an established negative image (p < 0.10) or no brand at all (p < 0.05) as compared with advertisements which contained an unknown brand.

TABLE 4

MEAN SCORES ON THREE DEPENDENT VARIABLES BY LEVELS OF THE RETAIL STORE AND BRAND NAME VARIABLES (n=402)

CONCLUSIONS AND IMPLICATIONS

The brand name main effect was found to be significant at the 0.10 level in all four analyses of covariance. The retail store effect was significant at the 0.05 level in the perceived believability ANCOV and in the MANCOV. No significant newspaper effect was found. These results lend support to the hypothesis that in branded retail advertising, the consumer's reaction is most strongly influenced by those sources which are perceived to have responsibility for the product.

The fact that the newspaper effect was not significant in any case suggests that a consumer's reaction to a branded retail advertisement is not likely to be influenced by the newspaper in which that advertisement appears. A retailer would likely not detract from the attractiveness of his advertising by running advertisements in newspapers which have a somewhat negative image. That is not to say, of course, that all newspapers are equal. While the newspaper may not affect reaction to the advertisement, the retailer must still consider other aspects of the newspapers available to him including circulation, market coverage, cost per thousand, etc.

The most important implications of this study are for the retailer. A quality retail store with an established positive store image can detract from the success of its advertising by featuring negative-image brands or brands with which consumers are not familiar. Its best results will come from advertisements which feature brands which have a positive image or no brand at all. In this latter case, the store clearly benefits from its established quality reputation.

The opposite effect was found in the case of advertising for a lower quality store which is having image problems. This retailer would also be advised to feature established positive-image brand names in his advertising as these continually produced the most positive reaction. In fact, the negative-image retailer should use even negative-image brands or unknown brands rather than feature no brand at all. Where no brand is present, consumers apparently rely on the image of the store to evaluate the advertised product. Because of its negative image, a retail store which is held in low esteem by consumers should not feature unbranded merchandise in its advertising.

Finally, the retailer who is just establishing a store in a new market area, where he has no established store image, would also be advised to feature established positive-image brands in his advertising. Since the consumer has no image of the store, she apparently judges the advertised product on the basis of the brand name. In this study, advertisements for the unknown retail store produced a more positive reaction where no brand name was present than where either a negative-image brand or an unknown brand was present. Therefore, the retail store which is just establishing itself in a market should avoid the use of negative-image or unknown brands in its advertising.

REFERENCES

David A. Aaker and John G. Myers, Advertising Management (Englewood Cliffs: Prentice-Hall, Inc., 1975).

James G. Barnes, "Factors Influencing Consumer Reaction to Retail Newspaper "Sale" Advertising," 1975 Combined Proceedings (Chicago: American Marketing Association, 1975), 471-477.

Donald F. Cox, "The Measurement of Information Value: A Study in Consumer Decision Making," In W. S. Decker (Ed.). Emerging Concepts in Marketing (Chicago: American Marketing Association, 1963).

W. J. Dixon (Ed.), BMD: Biomedical Computer Programs (Berkeley: University of California Press, 1973).

Douglas A. Fuchs, "Two Source Effects in Magazine Advertising," Journal of Marketing Research, 1 (August, 1964), 59-62.

Flemming Hansen, Consumer Choice Behavior: A Cognitive Theory (New York: The Free Press, 1972).

C. Hovland, I. Janis and H. Kelley, Communications and Persuasion (New Haven: Yale University Press, 1953).

Jacob Jacoby, Edward W. Hart, Jr., George J. Szybillo, and Jacqueline Busato-Schach, "The Persuasive Impact of a Smiling Source," Working paper #138, Department of Consumer Psychology, Purdue University, 1974.

G. H. G. McDougall and J. N. Fry, "Source and Message Content Credibility in Retail Advertisements," Journal of the Academy of Marketing Science, 3(Winter, 1975), 60-68.

Norman H. Nie, et al., Statistical Package for the Social Sciences, 2nd ed. (New York: McGraw-Hill Book Company, 1975).

Jerry C. Olson, "Cue Properties of Price: Literature Review and Theoretical Considerations," Working paper, Pennsylvania State University, 1974.

Milton Rokeach, "The Nature of Attitudes," In Joel B. Cohen (Ed.). Behavioral Science Foundations of Consumer Behavior. (New York: Free Press, 1973).

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