Xinshu Zhao. The Persuasion Handbook: Developments in Theory and Practice. Editor: James Price Dillard & Michael Pfau. Sage Publications, 2002.
Following an effort to summarize research on humor in advertising, Weinberger and Gulas (1992) concluded that “the broad question of humor’s effectiveness is unanswerable” (p. 35) due to too many contingencies in terms of Persuasion goals, message factors, the audience, the product advertised, and equivocal findings of the studies. The even broader question of what we know about the Persuasive effects of advertising generally may be equally unanswerable, at least within the confines of a single chapter. This chapter, therefore, does not purport to summarize everything we know about the Persuasive effects of advertising. Its more modest objective is to present a framework for the organization of advertising-related Persuasion research. To illustrate the utility of the framework, selected research studies are reviewed.
Theories versus Variables
Many advertising researchers aim to build theories that predict and explain human behavior in a wide variety of Persuasive situations. Those researchers often define areas of research in terms of the theories they test: spiral of silence, agenda setting, third-Person effect, and framing. From this Perspective, theory itself is the objective.
To theoretically oriented researchers, the history of advertising research is a history of one dominant theory following another (Alwitt & Mitchell, 1983; Maloney, 1994). An admittedly oversimplified history of the field would be something like this: Hovland, Janis, and Kelley’s (1953) source credibility theory during the 1950s, McGuire (1968) and others’ learning theory during the 1960s, Krugman’s (1965, 1966-1967) involvement theory during the 1970s, and Petty and Cacioppo’s (1981) Elaboration Likelihood Model during the 1980s.
The regular emergence and replacement of one dominant theory with another seems to have stopped during the 1990s. During that decade, for the first time in nearly half a century, there was no single dominant theory in the field of advertising. Even before the midpoint of the decade, Maloney (1994) had anticipated that “advertising researchers in this century’s closing decade may seem confused and floundering” (p. 48). Now, after the end of the decade, things do not seem any brighter. There is still no dominant theory or theories. The field seems no nearer to an answer for “how advertising works” than it was 10, 20, or 30 years ago.
This gloomy view ignores the other tradition of the field. Unlike many other areas of Persuasion research, advertising has a parallel applied focus that is manifested in the existence of the advertising industry. From the beginning, the industry had a need for credible knowledge to guide practitioners’ daily decision making. Pragmatically oriented advertising researchers were initially inspired to address questions such as “Should we run a 30-second or 60-second ad?” and “Should we use humor in this ad?” Such researchers often define their sub-areas of interests in terms of the variables they study including ad length research, celebrity effect research, and clutter research. One particular type of industry research, copy testing, attempts to predict advertising effectiveness in terms of a variety of variables, such as recall and ad liking, before the ads are released. While those re searchers often evoke theories to predict or interpret results, the utility of theory lies in its implications for message design rather than for explanation.
A quick glance at issues of the Journal of Advertising Research shows that there have always been a substantial number of studies in this tradition. This literature has been largely ignored by previous academic syntheses of the field, as all of those syntheses were organized according to theories and psychological/behavioral processes (e.g., Alwitt & Mitchell, 1983; Cafferata & Tybout, 1989; Clark, Brock, & Stewart, 1994). Implicitly, variables-oriented researchers rejected the assumption that answers to grand questions were possible. Instead, they attempted to make advertising questions more tractable by making them more narrow.
In that sense, the lack of a dominant theory is not an indication of floundering and confusion but instead a sign of progress. While fewer studies seek to produce the next dominant theory, new theories continue to be proposed, although they are often variable specific. The meaning transfer theory for celebrity effect (McCracken, 1989), the four-color theory for humor effect (Spotts, Weinberger, & Parsons, 1997), and the preceding-succeeding theory for clutter and serial effect (Zhao, 1997) are just a few examples. Many researchers have come to realize that theories, although useful, apply appropriately only within certain boundaries. Within such boundaries, published studies continue to use the theories, including source credibility theory, the Elaboration Likelihood Model, and every other theory in between. From a practical point of view, this signals the beginning of real successes for the theories; they are beginning to have an impact in the real world.
The focus on variables, especially independent variables, is an important distinction that separates much advertising research from other Persuasion research. Furthermore, even those advertising studies aimed at testing theories inevitably have variables. So a typology of advertising studies based on variables would not exclude those theory-oriented studies, while an organizing scheme based on theories would exclude many of the variable-oriented studies. Accordingly, this chapter proposes a typology of advertising effect studies according to the types of variables they investigate.
| Table 25.1 A Typology of Advertising Studies | |||||
| Independent Variables | Media | ||||
| Outdoor | Radio | Television | Computer | ||
| Directly manipulable Internal characteristics | |||||
| Directly manipulable External relations | |||||
| Indirectly manipulable | |||||
| Measured | |||||
A Typology of Advertising Variables
From an advertiser’s point of view, two factors may serve as starting points: the independent variable and the type of media in which the independent variables exert their effects. Although the type of media is itself a type of independent variable, it is granted special status here for reasons discussed in what follows.
Classifying the Independent Variables. Table 25.1 displays a four-category typology of independent variables based on their manipulability (in the leftmost column). From the Perspective of a potential Persuader, a most salient characteristic of an independent variable is the degree to which it can, in principle, be controlled. Previous writers have cast this distinction in terms of active versus attribute variables (Kerlinger, 1986) and, equivalently, manipulated versus measured variables. Manipulable variables include all those features of the advertising process that can be shaped by one or more message sources (e.g., ad placement, type of appeal), whereas measured variables are features of the product or audience that cannot be altered (e.g., age of the audience members).
As Table 25.1 shows, we propose some additional distinctions within these two broad categories. Manipulable variables are of two types. The phrase “manipulable internal characteristics” refers to features of the message or its implementation. Loosely speaking, these are variables most closely associated with the creative side of advertising. “Manipulable external relations” are relational, and they have two aspects. The first is the advertisement’s relationship with other messages (e.g., news, entertainment, other ads) that surround the ad. Traditionally known as the media strategy variables, they are usually not directly controllable by an advertiser or advertising agency alone. The second aspect of external characteristics of advertising is its comparative relationship with other nonadvertising strategies. Advertisers have the choice of using advertising versus not using advertising at all or of using advertising versus using other strategies such as public relations (PR) news releases and coupon promotions. Advertisers may also choose the level of advertising budgets. Traditionally known as ad-related marketing variables, they are often decided by marketers/producers in consultation with their advisers and business partners such as advertising agencies, PR agencies, retailers, and marketing research firms.
Some psychological and behavioral variables, such as attitude toward an advertised product and involvement in the product, are often treated as dependent variables in advertising research. However, message senders may hope to influence those variables in order to produce downstream changes in purchase intention or behavior. And these variables are only indirectly manipulable through more controllable variables such as creative and media strategies. In Table 25.1, they are labeled indirectly manipulable variables.
The final row of Table 25.1 pays homage to the fact that some aspects of the advertising process are functionally uncontrollable. These variables are characteristics of the products themselves or of the members of the target audience. Age, gender, and educational level are beyond the control of advertisers. Thus, this row represents measured variables.
Types of Media. Today, there are several different types of media from which advertisers can choose. One might argue that media type is another directly manipulable independent variable. That is true, but media are more than that. Different media evoke different sensory channels, evoke different levels of interactivity, and require different ways of managing fundamental parameters such as time and space. Furthermore, particular independent variables are differentially relevant to different media. Image and color are important for print, television, and computer-related media but have no meaning for radio. Interactivity is fundamental for computer-related media but becomes a near constant for some other media. The clutter is a spatial concept for print media but a sequential concept for broadcast media. The juxtaposition of these five media (see Table 25.1) with the four types of independent variables yields a 4 × 5 matrix that is used to organize the remainder of this chapter.
Effects of Directly Manipulable Internal Variables
Celebrity Endorsement in All Media. The analysis of 110 announcements of contracts of celebrity endorsements by various companies indicate that, on average, the announcements have led to higher stock prices for the companies. This suggests that the investors generally view the contracts as worthy investment in advertising (Agrawal & Kamakura, 1995).
Negative Strategy in Political Ads for Various Media. Based on a meta-analysis of 52 studies published between 1984 and 1998, Lau, Sigelman, Heldman, and Babbitt (1999) concluded that, contrary to the conventional wisdom, there is little evidence that negative ads produce more positive attitudes toward or win more votes for the sponsoring candidates relative to positive ads. There is also little indication that negative ads produce more negative attitudes toward or take more votes from the opposing candidates relative to positive ads. And people do not necessarily dislike the negative ads. The study also uncovered little support for the popular belief that negative ads discourage voters from participating in the political processes.
The effects of attacking strategy on voters’ candidate preference depend on political involvement and attention to political news, according to Faber, Tims, and Schmitt (1993), whose study was not included in Lau et al.’s (1999) meta-analysis. Faber et al. (1993) found that, as a result of being exposed to negative ads, those who are highly involved are more likely to change their candidate preference (in either direction) than are those who are less involved. Nevertheless, after being exposed to negative ads, those who read more newspaper news are less likely to change their candidate preference than are less frequent readers.
Humor in Print and Broadcast Ads. Summarizing 20 years of research on the effects of humor, Weinberger and Gulas (1992) concluded that humor attracts attention but does not harm comprehension despite the belief that humor confuses the audience. Humor enhances ad liking, while it has little effect on source credibility, brand attitude, purchase intention, or purchase behavior. Research conducted subsequent to the Weinberger and Gulas review has suggested that humor does have a positive impact on ad attitude, brand attitude, and purchase intention (Zhang, 1996) and that people exposed to humorous television ads are more likely to consider the surrounding programs as entertaining (Perry, Jenzowsky, Hester, King, & Yi, 1997).
The effects of humor are moderated by many factors. Relevant humor is more effective than nonrelevant humor (Weinberger & Gulas, 1992). Humor is more effective when existing, feeling-oriented, low-involvement, and/or low-risk products are advertised (Weinberger & Gulas, 1992; Weinberger, Spotts, Campbell, & Parsons, 1995). The kind of audience also has an impact on humor’s effectiveness (Weinberger & Gulas, 1992). In one study, participants with a positive prior attitude toward a brand were positively affected by humor in terms of ad attitude, brand attitude, purchase intention, and purchase behavior, while participants with a negative prior attitude were negatively affected (Chattopadhyay & Basu, 1990). In an experiment using print ads, participants lower in need for cognition were more likely to be affected by humor in terms of developing a more positive ad attitude and brand attitude as well as higher purchase intention (Zhang, 1996). Humor in event promotions increased attendance to social events but helped little in attracting attendees to business events (Scott, Klein, & Bryant, 1990). Furthermore, it has been found that the effect of humor on brand attitude is mediated by ad attitude (Zhang, 1996) and by cognitive responses (Chattopadhyay & Basu, 1990).
Copy Vividness of Print Ads. More vivid copies tend to generate a more favorable attitude toward an advertised brand and make the audience more likely to choose the brand when the ads are shown together with other ads for competing brands and when the audience is engaged in issue-relevant thinking. The advantages of the vivid copies tend to disappear when, with other conditions unchanged, the ads are not shown together with other ads for competing brands (Heath, McCarthy, & Mothersbaugh, 1994).
Spokesperson in Print Ads. A company may choose a corporate official, such as the chief executive officer, or an outside noncommercial authority as the spokesperson in its ad. Straughan, Bleske, and Zhao’s (1996) experiment shows that the audience is more likely to be interested in a message from a chief executive officer than in the identical message from an outside authority. The higher interest leads to more of the desired attitude change, which in turn leads to more of the desired behavioral change.
Attribute Versus Relation in Print Ads. An ad may focus on the attributes of a product or on the product’s relation to people and its use. Malaviya, Kisielius, and Sternthal (1996) found that, when a commercial focused on attribute was placed together with ads for competing brands, the commercial tended to generate more favorable attitudes than did a commercial focused on relation. When the commercial was placed together with ads of noncompeting brands, however, the relation strategy produced more favorable attitudes than the attribute strategy.
Message Appeal or Argument Strength in Print Ads. Stronger argument has a positive effect on ad attitude, brand attitude, and purchase intention. But the individuals in more need for cognition are more likely affected by argument strength in terms of developing a more positive ad attitude and brand attitude as well as higher purchase intention (Zhang, 1996).
Celebrity Endorsement in Print Ads. While Petty, Cacioppo, and Schumann (1983) showed that issue-relevant thinking can eliminate the positive effects of spokespersons’ fame, competitive setting may also matter. In Heath et al.’s (1994) experiment, spokespersons’ fame led to a more favorable attitude toward an advertised brand and made the audience more likely to choose the brand when the ads were shown together with other ads for competing brands, even when the audience was engaged in issue-relevant thinking. This positive celebrity effect disappeared when ads were not shown together with other ads for competing brands.
The audience’s Perception and attitude toward a celebrity and its attitude toward the ad become less and less favorable as the number of products endorsed by the celebrity increase (Tripp, Jensen, & Carlson, 1994). Furthermore, according to Till and Shimp (1998), the negative information about a celebrity resulted in a decline in attitude toward an endorsed brand, but only when the celebrity endorser was a fictitious figure. That general relationship was further moderated in varying degrees by three other variables: association set size, timing of the negative information, and strength of the link between brand and celebrity.
Association Strategy in Print Ads. When an ad associates the advertised brand with another object, the result could be assimilation or contrast (Meyers-Levy & Sternthal, 1993). Assimilation refers to the phenomenon whereby consumers evaluate the advertised brand more like the associated object than they would if there had been no association between the brand and the object. An advertiser may take advantage of this phenomenon by associating its brand with an already positively evaluated object. Contrast refers to the opposite phenomenon whereby consumers evaluate the advertised brand more unlike the associated object than they would if there had been no association between the brand and the object. An advertiser may take advantage of this phenomenon by associating its brand with an already negatively evaluated object. Meyers-Levy and Sternthal’s experiment shows that contrast occurs when (a) the consumers devote a substantial amount of cognitive resources to the processing of the ads and (b) there is little overlap between the advertised brand and the associated object. The absence of either of these two factors leads to assimilation.
Youth Appeal in Cigarette Print Ads. The youth appeal in Joe Camel ads appeared to encourage adolescents to pay more attention to the Camel ads than to other cigarette ads (Fox, Krugman, Fletcher, & Fischer, 1998). duration (length) on recall. Furthermore, because few interaction (moderating) effects with other independent variables were found, the researchers argued that Perhaps the main effect of length should be the focus of the attention after all.
Emotional Appeals in Television Ads. In Hitchon and Thorson’s (1995) experiment, higher emotional appeals produced more positive ad attitude than did lower emotional appeals. The effect of emotional appeals on brand recall depended on the audience’s product involvement; the emotional appeals produced higher recall when the involvement was high, but emotion had little effect on recall when the involvement was low.
Music and Sound in Television Ads. Olsen (1995) compared three strategies: background music throughout an ad, background silence throughout the ad, and background silence only when key information is presented. The results show that the audience’s recall of the key information was the highest when silence appeared only with key information. The effect was even stronger, however, when the key information was the last item in a series.
The Size and Type of Web Banner Ads. We b users clicked an animated banner ad more quickly and were more likely to recall the ad than to recall a nonanimated ad (Li & Bukovac, 1999). Web users were more likely to click a larger banner ad and clicked it more quickly.
Effects of Directly Manipulable External Variables
To Ad or Not to Ad (various media). Public service announcement (PSA) campaigns have Between-Information Interval Within Radio Ads. An ad often contains multiple pieces of information. Olsen (1997) investigated the effect of the between-information time interval together with two other factors: (a) whether the listeners’ attention is focused on the ad or diverted and (b) whether there is background music or background silence. The results showed that as the between-information interval increased from 0 second to 1, 2, and 3 seconds, listeners’ recall of the information also increased linearly in all but one situation. The only exception occurred when the listeners’ attention was diverted and there was background silence and when the interval increased from 2 seconds to 3 seconds. In that situation, the recall decreased, and did so significantly, by about 50%.
Length of Television Commercials. A number of studies (Fabian, 1986; Mord & Gilson, 1985; Patzer, 1991) have suggested that 15-second ads are 50% to 90% as effective in creating learning and attitudinal change as are 30-second ads. Singh and Cole (1993) argued that the effect size of ad length depends on repetition and message content (i.e., emotional vs. informational). Indeed, their experiment showed that 30-second ads generated better brand memory and attitude than did 15-second ads when the appeal was emotional; when the appeal was informational, 15-second ads were generally as effective as 30-second ads. It was also found that the length effect on brand recall diminished as repetition increased. Pieters and Bijmolt (1997) analyzed data based on 2,677 television commercials of naturally varying lengths aired between 1975 and 1992 in The Netherlands, and they reported positive effects of traditionally relied on donated rather than paid media. In part due to the limitations that often come with the donations, the recent trend has been to switch to the paid schedule at the risk of losing future donations from the media. A three-market field experiment (Murry, Stam, & Lastovicka, 1996) found that the donated schedule was as effective and cost-efficient as the paid schedule in Persuading the youths to avoid drinking and driving, thereby reducing the counts of incapacitating and fatal highway accidents. Given this knowledge, the researchers suggested that the public service campaign managers not abandon the donated media for the paid media.
Despite this recommendation, in 2000, the U.S. decennial population census used paid advertising for the first time in history in hopes of increasing participation. A study conducted by two researchers from the U.S. Bureau of the Census and U.S. Treasury Department (Bates & Buckley, 2000) found that exposure was significantly related to being knowledgeable about the census but was not significantly and directly related to the likelihood of completing the census forms. Furthermore, this lack of relationship was across the board with regard to factors such as race/ethnicity.
When a company wants to spread a message, it may choose to buy advertising space or to work with media reporters to generate news coverage. The message in an ad, however, is less likely to be believed by the audience than the identical message appearing in a news story, according to Straughan et al.’s (1996) experiment. This lower credibility leads to less of the desired attitude change, which in turn leads to less of the desired behavioral change.
Ad Budget for All Media. Based on a meta-analysis of 389 real-world split cable-television experiments, Lodish et al. (1995) reported that increasing the advertising budget, in relation to those of competitors, do not increase sales in general. Increasing the advertising budget, however, did seem to help increase the sales of the products/brands newly introduced to a market but not the sales of the established products/brands. Furthermore, prime-time gross rating points, which are nearly entirely dependent on budgets, have a highly significant and positive relation with sales (Lodish et al., 1995). Cobb-Walgren, Ruble, and Donthu (1995) reported that brands with higher advertising budgets yielded substantially higher levels of brand equity, which in turn led to significantly greater brand preferences and purchase intentions.
Clutter of Print Ads. Kent and Allen (1994) focused on a particular type of clutter (i.e., the other ads for directly competing brands). The study found that competitive clutter leads to less claim recall, but only when the advertised brands are unfamiliar.
Ha (1996) saw three dimensions in the concept of clutter in magazines—quantity, competitiveness, and intrusiveness—and manipulated each of them in her experiment. She also measured the participants’ Perceptions of the three dimensions as three mediating variables. In addition, she measured six dependent variables: the reading of, memory of, involvement in, and attitude toward the focal ad under test; the resistance toward competitive ads; and the attitude toward the focal brand. Of the several dozens of independent-dependent, independent-mediating, and mediating-dependent relations, half a dozen turned out to be statistically significant, among which a couple were surprises in terms of the direction of the effects. A higher quantity of ads led participants to Perceive a higher quantity, a higher Perceived quantity or a higher Perceived competitiveness led to more negative attitudes toward the magazine that carried the ads, and a higher Perceived intrusiveness was associated with less memory of the ads. Nevertheless, a higher Perceived competitiveness was associated with more readers of the focal ad and higher involvement in the ad message.
Frequency (repetition) for Television Ads. Although the main effect of repetition was not Singh and Cole’s (1993) central concern, their experimental results showed a general increase in brand recall and claim recall, as the frequency increased from 1 to 4 and then to 8. Singh, Mishra, Bendapudi, and Linville’s (1994) experiment found that airing the same ad twice generated more recall than airing it once. In Hitchon and Thorson’s (1995) experiment, the participants reported 17%, 21%, 53%, and 81% brand-name recall when commercial repetitions were 2, 4, 6, and 12, respectively. However, they also found that the attitudes toward the ad became more negative as the repetition increased.
An analysis of field data from three Super Bowl broadcasts found that every additional ad for a brand added 3.5 Percentage points to brand recall, 6.5 Percentage points to brand recognition, and one third of a Percentage point to ad liking on a scale from 0 to 100 (Zhao, 1997).
Spacing of Television Ads. When multiple television ads are used for the same brand, how far apart should the ads be placed? The answer depends on whether the advertiser’s objective requires immediate or delayed reaction from the audience. According to Singh et al.’s (1994) experiment, long lag (four other ads between the two focal ads) produced more day-after recall, while short lag (one ad in between) produced more short-delay recall.
Clutter of Television Ads. While an earlier experiment (Webb & Ray, 1979) reported a decreased brand recall as a result of the increased clutter of television ads, a later experiment (Brown & Rothschild, 1993) found no such effect on brand recall or recognition. Brown and Rothschild (1993) speculated that the negative clutter effect might have diminished since the Webb and Ray (1979) study, which seemed to be consistent with Johnson and Cobb-Walgren’s (1994) experiment that also failed to find a statistically significant clutter effect on recall or recognition.
Sizable negative effects of clutter on brand recall and recognition, however, were found in two quasi-experiments published later. One (Pieters & Bijmolt, 1997) was based on 39,000 consumers’ reactions to 2,677 television commercials aired between 1975 and 1992 in The Netherlands, and the other (Zhao, 1997) was based on postgame interviews with more than 1,000 viewers of three Super Bowl broadcasts between 1992 and 1994. Extending clutter effect to attitudes, the ads in more crowded “pods” (commerical breaks) were less likely to be liked (Zhao, 1997).
Johnson and Cobb-Walgren (1994) found that the viewers with slower cognitive processing speed were more negatively affected by clutter in terms of recognition and recall of the brand name and ad message. It was also argued that the concept of television clutter should be divided into two parts: the ads preceding an advertisement and the ads succeeding the advertisement (Zhao, 1997). Indeed, while both have negative effects, the effects of preceding ads were found to be much larger.
Serial Position (presentation order) of Television Ads. Thorson, Reeves, Schleuder, Lang, and Rothschild (1985) and Thorson and Reeves (1986) reported that the commercials’ presentation order had more significant effects than did content variables such as arguments, executional style, brand, and product. Later studies continued to confirm the primacy effect, showing that the first positions in a pod generate better memories (Burke & Srull, 1988; Cameron, Schleuder, & Thorson, 1991; Pieters & Bijmolt, 1997; Stewart, Pechmann, Ratneswar, Stroud, & Bryant, 1985).
The situation with regard to the recency effect is more complicated. Controlled experiments continued to show the recency effect— that the last few positions generate better memories than do the middle positions—thereby further confirming the famous U-shaped curve that psychologists had demonstrated decades ago. Some researchers, however, argued that the recency effect might be a phenomenon limited to laboratory settings, where memory was tested immediately after the presentations of the ads, a setting that is quite different from most of the real-world advertising situations (Zhao, Shen, & Blake, 1995). They cited previous nonadvertising experiments (Craik, 1970; Glanzer & Cunitz, 1966) reporting that, when memory tests were delayed and distracting tasks were inserted in between, the advantage of the last positions disappeared and the U-shaped curve was replaced by a monotonically declining line. Indeed, in two advertising studies (Pieters & Bijmolt, 1997; Zhao, 1997) based on large-scale field data, it was the monotonically declining line— not the recency effect or U-shaped curve— that showed up for serial effects on memory.
Furthermore, serial position was redefined as a one-for-one exchange between the stronger negative effects of preceding ads and the weaker negative effects of succeeding ads. And it was argued that the monotonically declining curve and the lack of recency effects should be expected in field studies. A linear negative effect of serial position on ad liking was also found (Zhao, 1997).
Context of Television Ads. Context can be defined in many different ways. Cameron et al.’s (1991) experiment looked at the context in terms of the presence or absence of news teasers. The participants paid more attention to commercials following news teasers. But the news teasers did not produce a detectable effect on visual or verbal recognition of the commercial contents. Perry, Jenzowsky, King, et al. (1997) looked at the context in terms of program content and found that humor in television programs had a negative effect on the recall of the advertised products.
Context of Web Ads. Stevenson, Bruner, and Kumar (2000) looked at Web page complexity in terms of number of items, colors, and animation, and they compared ads in a complex Web page to ads in a simpler page. The complexity had a negative impact on users’ attitude toward the site, the ads in the site, and the advertised brand. It also reduced the users’ intention to purchase the advertised brand. The complexity did not, however, seem to significantly reduce users’ attention to the ad.
Effects of Indirectly Manipulable Variables
Media Use Behavior (all media). People who watch many television channels or listen to many radio stations are most likely to avoid commercials (Speck & Elliott, 1997). African Americans who watch more television tend to have a more positive attitude toward advertising than do those who watch less television. The same relation between television watching and attitude toward advertising does not exist among Caucasians, according to Bush, Smith, and Martin (1999).
Attitude Toward Advertising in General (print and broadcast media). Consumers’ positive attitude toward advertising in general is positively associated with involvement with specific advertisements (James & Kover, 1992). Those people who think advertising in general as interesting, useful, or believable are less likely to avoid ads, while those who view advertising as excessive, annoying, or a waste of time are more likely to avoid ads (Speck & Elliott, 1997).
Ad Attitude for Print and Broadcast Ads. Based on interviews of nearly 15,000 consumers after five matched pairs of commercials were aired, the Advertising Research Foundation’s Copy Research Validity Project (Haley & Baldinger, 1991) found that liking of ads predicts product sales far better than do any of the other indicators often measured in copy testing. Biel and Bridgwater’s (1990) smaller scale study showed a similarly stronger effect of ad liking than of any other independent variables.
While it had been known that more positive ad attitude leads to more positive brand attitude, Brown and Stayman (1992) showed that this effect is both direct and indirect. Ad attitude exerts this effect both directly and indirectly through the mediating variable of brand cognition.
Brand Equity for Various Media. Brand equity is a complex concept with multiple dimensions (Aaker, 1991; Aaker & Biel, 1993). Cobb-Walgren et al.’s (1995) definition of brand equity focuses on the consumer Perception dimension, making it very close to brand attitude, hence something indirectly manipulable by advertisers. Their study found that higher brand equities generated significantly greater brand preferences and purchase intention.
Mood While Processing Television Advertising Messages. A positive mood assists an individual in remembering brand names better than does a neutral mood (Lee & Sternthal, 1999).
Within-Brand Versus Between-Brand Processing of Television Advertising Messages. A television ad viewer may focus on the characteristics of an advertised brand, a psychological phenomenon called within-brand processing. Or the viewer may be engaged in comparing the advertised brand to other competing brands, a process called between-brand processing. Kent and Machleit (1990) reported that between-brand processing leads to increased brand recall, while within-brand processing leads to better brand recognition.
Television Program-Related Psychological and Behavioral Variables. Individuals who pay more attention to television programs are more likely to watch television commercials (Krugman, Cameron, & White, 1995).
Murry, Lastovicka, and Singh (1992) reported that viewers’ liking of programs positively influenced ad attitude and brand attitude, while program-induced affect (feelings or mood) had no effect on these same attitudes. Coulter (1998) also found a positive effect of program liking on ad attitude, but program liking was seen as a mediating variable that was directly affected by program cognition and program-induced affect, and program-induced affect was found to be directly affected by program cognition.
Furthermore, when the commercial is in the first position in a pod, higher liking of the program produced more positive ad attitude (Coulter, 1998) and more positive brand attitudes (Murry et al., 1992) as compared to commercials in other pod positions. Also, according to Coulter (1998), the effect of program liking on ad attitude is stronger when the ad and the program have congruent emotional contents.
Attitude Toward Web Sites. Chen and Wells (1999) first proposed this concept. In Stevenson et al.’s (2000) experiment, users with more positive attitudes toward a Web site were more likely to have a positive attitude toward an ad in the site, a positive attitude toward the brand advertised, and a stronger purchase intention regarding the brand.
Effects of Measured Variables
Race (all media). African Americans watched more television and had more positive attitudes toward advertising than did their Caucasian counterparts (Bush et al., 1999).
Age (print and broadcast). As people get older, their cognitive processing speed declines, and consequently their brand and ad message memory (recall and recognition) decline significantly (Johnson & Cobb-Walgren, 1994). Older people are also more likely to avoid newspaper ads but less likely to avoid radio ads than are younger people (Speck & Elliott, 1997).
Gender (various media). Women have a more positive attitude toward advertising in general than do men (Bush et al., 1999). Men are more apt to change channels during television commercial breaks than are women (Krugman et al., 1995).
Male homosexuals (gays) tend to have more negative attitudes toward advertising than do female homosexuals (lesbians), while lesbians are less concerned with appropriate homosexual portrayals in advertising than are gays (Burnett, 2000).
Sexual Orientation (all media). Compared to other consumers, homosexual consumers read different newspapers and magazines, watch different television shows, listen to different radio programs, and are more likely to use catalogs and online resources (Burnett, 2000). They also tend to have a more negative attitude toward advertising.
Income (print and broadcast media). According to Speck and Elliott’s (1997) survey, people with higher income are more likely to avoid ads in magazines, in newspapers, and on television.
Product Involvement (television ads). While advertisers may indirectly manipulate ad involvement, situation involvement, or even brand involvement, they may have little room to manipulate product involvement. Hitchon and Thorson (1995) reported that higher involvement with a product produced a more positive attitude toward an advertised brand, while product involvement produced no discernible effect on brand recall.
Family Communication Pattern (television ads). Researchers have identified two types of communication between parents and their children (McLeod & Chaffee, 1972). In concept-oriented communication, parents foster open two-sided communication, encouraging their children to discuss their ideas and develop an independent Perspective. In socio-oriented communication, parents are concerned with maintaining control over their children’s thoughts, behavior, and exposure to outside influences. A four-category typology was developed based on those two dimensions (Moschis, 1987): Laissez-faire parents are neither socio- nor concept oriented; protective parents have a high level of socio-oriented communication but a low level of concept-oriented communication; pluralistic parents are the opposite, having a high level of concept-oriented communication but a low level of socio-oriented communication; and consensual parents have high levels of communication of both types.
Rose, Bush, and Kahle’s (1998) survey of mothers in the United States and Japan found that laissez-faire mothers had the most positive attitudes toward and the lowest mediation of their children’s exposure to television advertising. Pluralistic and consensual mothers had the highest mediation of and most negative attitudes toward advertising. The responses of protective mothers were between those extremes.
Web User Mode. Web surfers are no more likely to click banner ads or remember the ads than are information seekers, according to Li and Bukovac’s (1999) experiment.
More Relevant Research Questions
Most of the research questions in advertising research started with whether—whether negative strategy wins votes for the sponsoring candidate, whether a celebrity endorser helps to Persuade consumers to buy the advertised brand, and so on. The expected answers are either yes or no. While those studies are all quantitative in terms of methodology, the questions are by nature qualitative or “dichotomous.” More recent advertising studies were more likely to ask contingency questions involving moderating variables—whether the negative strategy is stronger when the voters are more involved, whether the celebrity effect is weaker when the product is more involving, and so on. Those questions probe in more details and require comparisons of magnitudes of the effects.
The next stage of development might be to ask more “quantitative” questions starting with how much—how much celebrity endorsers affect memory, attitude, or behavior; how much stronger or weaker the celebrity effects are than the effects of ad length, and so on. These would be entirely new types of questions that require entirely new types of answers, which may lead to entirely new types of theories. For practitioners and policy makers, “precision” answers to such “precision” questions should be more useful in the cost-benefit calculation of the daily decision making.
Two other recommendations might seem obvious due to the changes in the economic environment in general and in the advertising media in particular, but they merit mention nonetheless. First, there should be, and most likely will be, more studies on computer-related advertising. The trend is already quite clear in the most recent conferences and journals in the field. Second, there should be, but might not be, more studies on advertising effects in countries other than the United States, especially in the emerging economies. The financial cost and the cultural and language barriers make it difficult for U.S.- or Western Europe-based researchers to conduct research in those developing countries. The differences in research traditions and orientation make it difficult for the researchers based in those countries to produce the large amount of empirical research needed to participate in dialogue with the Western research community.
The Crises of Relevance
The preceding review and comments are based on the assumption that the prevailing methodologies can validly inform the goals of the advertising community. Unfortunately, the apparent mismatch between the goals and the methods of the advertising research troubled some observers during the 1990s, as it did during previous decades. As discussed previously, the research community has dual goals. One goal is applied—to create immediately useful knowledge that may help industry practitioners, public policy makers, and others make better decisions in their professions.
The second goal is theoretical—to advance the basic understanding of advertising processes, especially advertising’s Persuasive processes and effects, even if the knowledge might not have an immediate, clear, or specific application. For this second goal, it is also implied that ultimately it is the real-world processes and effects that should concern us, not some peculiar phenomena that occur only in laboratory settings with nonrepresentative ads on nonrepresentative subjects.
Accordingly, Sheth (1972) argued for “testing of theories in naturalistic and realistic settings” (pp. 565-566) as the appropriate methodology for the field. The advertising studies during the next few years failed to meet Sheth’s standard. Jacoby (1976) complained that much of the consumer research, of which the advertising research is considered a sub-field, “is not worth the paper it is printed on or the time it takes to read it” (p. 2). Preston (1985) reported a growing “detachment” of advertising research from the actual “marketing and advertising context” (p. 3) such that the selection of the major concepts and the manipulation of the major variables under study “digressed from actual advertising practice” (p. 5). Eight years after Preston’s observations, Wells (1993, p. 489) saw the same thing, finding that the prevailing methodology in the field was moving “away from the real world,” using students to represent consumers and laboratory to represent the environment (pp. 491-492). Resolution of this crisis of relevance requires that advertising researchers forsake traditional methodologies (Wells, 1993).
McQuarrie (1998) conducted a meta-analysis of more than 400 experimental studies of advertising published between 1965 and 1997. He measured those studies’ methodological detachment from advertising goals in six respects: (a) forcing exposure rather than arranging for nonfocal attention to embedded advertisements, (b) failing to measure choices, (c) not incorporating competitive advertisements into the design to allow for interference, (d) taking immediate measurements instead of allowing for decay, (e) not arranging for repeated exposures, and (f) exclusively using fictitious or unfamiliar brands. McQuarrie’s findings should be alarming. Advertising studies were relatively detached throughout the Period between 1972 and 1998, and the detachment had grown even higher during the more recent years.
Most of the advertising researchers, however, did not appear to be alarmed by this finding or by the critiques offered by Jacoby (1976), Preston (1985), and Wells (1993). There has been no elaborated or systematic defense on behalf of the prevailing practice. The researchers did appear to be aware of the criticism, as some succinct statements scattered or buried in some articles’ methodology or conclusion section might suggest. Those statements are usually along the lines that, because this is a theory testing or basic research, external validity is not that important. Here is an example from a highly respected researcher (Kamins, 1990):
Although one could criticize the current research on the grounds of limited external validity (i.e., use of laboratory experiment and a student sample), the focus of the research was directed toward theory testing, therefore placing the value of internal validity as more important than external validity.
As discussed earlier, the advertising-related Persuasion research is the child of a marriage between the advertising industry and experimental psychology, a marriage symbolized in the 1920 hiring of John B. Watson, then a leading experimental psychologist, by J. Walter Thompson, then America’s largest advertising agency. From the very beginning, it appeared to be a marriage of convenience rather than of love. Watson ended up producing no research in advertising (Maloney, 1994). For a long time, consumer psychologists insisted on the traditions from experimental psychology in terms of both topic and methodology. In topic selection, they insisted on searching for universal theories. In methodology, they paid little attention to external validity, namely generalizability. Both mismatch the basic needs of the advertising industry (and public policy makers as another important patron), leading to a troubled marriage and an unhappy child.
As is the case with any troubled marriage, at least one party has to compromise or the marriage will not last. A divorce in this case would be easy. All it would take is for advertising practitioners and public policy makers to stop paying attention to advertising research. Many of them never paid attention to it anyway. Over the years, more and more advertising researchers are selecting their topics based on independent variables, a sign of compromise in one of the two major points in dispute. With regard to methodology, however, many of the researchers continue to ignore the criticism while doing what they always have done. Until more advertising researchers place more value on the generalizability and ultimate utility of the knowledge they create, the mismatches between the goals and the methods of the advertising research are likely to continue. So too are the crises of relevance and the troubles in the marriage.