The effects of sponsor relevance on consumer reactions to internet sponsorships.

Sponsorship has been acknowledged as one of the newest and most popular forms of Internet advertising. Internet sponsorships generated $1.87 billion in 2001, accounting for one-third of all Internet advertising revenue for that year (Internet Advertising Bureau, 2002). Despite the growing interest

in sponsorships, past research has primarily focused on event sponsorships, and corporate sponsorships to some degree. This focus may be due in part to a lack of understanding about the nature of Internet sponsorships. Like event sponsorships, Internet sponsorships are typically brief text ads (e.g., "Today's sponsor is") that identify the sponsor's brand and product (e.g., "Samsonite luggage") (Rodgers and Thorson 2001). These similarities suggest that extending sponsorship research to nontraditional contexts such as the Internet is a logical and appropriate line of inquiry.

Although sponsorship is viewed as a unique form of advertising (Gardner and Shuman 1987), there has been no general clarification about how to make sponsorship accountable--either on-or off-line. Scholars and practitioners assume that sponsorships have advantages over other forms of promotion (Meenaghan 1991) because they are small, flexible, and can improve the image of companies that use them (Hastings 1984). One question that remains unanswered is exactly how to structure sponsorships for maximum persuasive impact.

The idea of sponsorship linking has been suggested as a way of structuring sponsorships, but only recently has the concept of linking been empirically examined. Sponsorship linking broadly refers to the nature of the link between a company and its brand (i.e., the sponsor) and the thing being sponsored (i.e., the sponsee) (Otker and Hayes 1987). Much of the empirical research on sponsorship assumes that congruent sponsorship links are more effective than incongruent sponsorship links, but sponsorship research that has tested the congruity effect has yielded inconsistent findings. The "newness" of the Internet as a potential outlet for sponsorships further complicates this issue. What is the nature of an Internet sponsorship link and to what extent does this link influence consumer reactions to this new form of online advertising?

The primary goal of this research is to examine the effects of sponsorship linkage in a new context--an Internet Web site--and the role of sponsor relevance in these effects. The moderating effects of two variables, site credibility and intent to return to the site, are examined in relation to sponsor relevance. This was accomplished by conducting two experiments using relevant and irrelevant Internet sponsorship links embedded in a newspaper Web site. Experiment 1 utilized a group of undergraduate students and Experiment 2 replicated the first experiment using a sample of nonstudent adults.

INTERNET SPONSORSHIPS AND ASSOCIATIVE LINKS

Associative links provide the theoretical framework for understanding Internet sponsorship effects. An associative link refers to the perceived strength of the sponsor-sponsee matchup. Research in this area assumes that strong associative links are more persuasive than weak associative links. Theories on associative learning suggest that stronger sponsorship alignments help to establish explanatory links that aid consumers in determining the sponsor-sponsee connection. This process is presumed to occur through spreading activation, one of the dominant models of semantic processing in cognitive psychology. Spreading activation is believed to work within a semantic memory network that consists of a set of interconnected nodes, with each node representing a concept (Anderson 1983). In the context of sponsorship, both the sponsor and sponsee represent individual nodes (or concepts) within memory. When a strong sponsor-sponsee link is presented to the memory system, the individual nodes become interconnected and activation spreads throughout the network. This spread of activation serves to make the related areas of the memory network available for further cognitive processing of the sponsorship.

Although strong associative links are thought to ignite the spread of activation, sponsorship studies testing this effect have yielded mixed findings. One example is a survey by D'Astous and Bitz (1995). They hypothesized that strong company-event associations would be rated more positively than weak company-event associations. This hypothesis was confirmed, but only for commercial (and not philanthropic) sponsors. McDaniel (1999) has also explored an aspect of sponsorship linkage by matching event and brand in terms of involvement. He experimentally paired three sponsors (Taco Bell, Marlboro, and Toyota) with three sports events (PBA Bowling, NHL Hockey, and the Summer Olympics). Contrary to McDaniel's predictions, the results showed that only one of the congruent-involving sponsors outperformed the incongruent-involving sponsors in terms of attitude toward the ad. In fact, attitude toward the ad did not change at all and became more negative for two of the congruent sponsors.

These findings suggest that consumers do attend to the sponsorship link, although the exact nature of the link remains unclear. What, for example, is the link between Toyota and the Olympics or Marlboro and PBA Bowling--two sponsor match-ups examined in the study just cited? Research on associative links suggests that the nature of the sponsor-sponsee link must be made clear so that its meaning becomes associated in consumers' minds. According to associative theories, spreading activation can be strengthened when the sponsor and sponsee share semantic features (Sperber and Wilson 1995).

Drawing on this notion, the current research defined sponsorship linkage as the perceived relevance between the sponsor-sponsee link in terms of two related semantic features--the product's descriptive element and its alliance. A health company sponsoring health content in a health section of a Web site is one example of this sort of sponsorship link. Johar and Pham (1999) have shown that consumers use the relatedness of a sponsor and sponsee in sponsor identification. Likewise, a recent study on sponsorships of the Indianapolis 500 revealed that sponsors with logical links to the automotive industry had greater economic gains in their share prices than sponsors with illogical links (Cornwell, Pruitt, and Van Ness 2001). Both of these studies demonstrate the critical nature of the sponsorship link. Thus, defining sponsorship links in terms of relevant semantic features should help to crystallize the connection between the sponsor and sponsee by making the link more apparent to potential consumers. Relevant sponsorship links should increase the likelihood that the link will influence memory, attitudes, and behavioral intentions--the three dependent variables examined here.

INTERNET SPONSORSHIPS AND LINKAGE EFFECTS

Sponsor Relevance and Memory Effects

Companies often use sponsorship as a marketing tool to enhance brand awareness (Rajaretnam 1994). Theories on associative learning suggest that the perceived relevance of the sponsor and sponsee will influence memory. For example, the congruity principle assumes that congruent information is remembered and preferred over incongruent information (Osgood and Tannenbaum 1955). This assumption is supported by associative links, which suggest that spreading activation within memory will be greatest when sponsors and sponsees connect on a relevant criterion. Research on event sponsorships has revealed a positive relationship between sponsor recall and the perceived relevance of the sponsorship link. Stipp and Schiavone (1996) and Stipp (1998) conducted telephone interviews with 479 individuals who had watched the 1992 Summer Olympics and found that consumers who could recall Olympic sponsors also perceived a strong relationship between the sponsoring brand and the Olympics. A similar relationship has been found by other studies on event sponsorships. For example, Gwinner and Eaton (1999) found that sponsor recall was heightened when participants perceived a relationship between the image of the event and sponsored brand. Thus, by matching the sponsor and sponsee on a relevant dimension, it is expected that consumers will have greater recall for sponsorships with relevant linkages. This leads to the following prediction:

   H1: Sponsor recall will be higher for a relevant than an
   irrelevant sponsorship linkage.

Sponsor Relevance and Attitudinal Effects

The primary goal of sponsorship is to generate positive attitudes toward companies that sponsor (Cornwell and Maignan 1998). An attitude is an individual's overall evaluation of an object, issue, or person. Attitudes generally refer to how favorably/unfavorably or how positively/negatively an object is viewed. Attitudes have served as a dependent variable in studies that have examined the effects of event sponsorships (e.g., Lee, Sandler, and Shani 1997). The findings of two separate studies revealed that the mere presence of a sponsorship influenced individuals' perceptions of the sponsoring company (Dean 2002) and its perceived citizenship (Dean 1999). Although much of this research assumes that evaluative processing of sponsorships is natural and automatic to some degree, this process is likely much more complex. The limited content of sponsorships implies that some effort in processing will be necessary for evaluation to take place. In these circumstances, situational factors such as the perceived relationship between the sponsor and sponsee can influence the evaluation.

Research on associative links suggests that highly accessible attitudes are more likely to influence brand evaluations than less accessible attitudes. Accessibility in this context refers to the likelihood that an individual will use a specific piece of information, in this case the sponsorship link, in forming an evaluation. In relation to sponsor relevance, an attitude will be more accessible to the extent that the association strength between the sponsor and sponsee is relevant. It is expected that consumer responses to relevant sponsors will be more accessible and, hence, more positive than responses to irrelevant sponsors. An experiment by Becker-Olsen (1998) supports this point. She manipulated prosocial news clippings as having either logical or illogical sponsorship links. Her findings revealed that attitudes were more positive toward all but one of the logical sponsors, presumably due to the perceived relevance of the sponsor-sponsee link. Thus, the following hypothesis is posited:

   H2: Attitude toward the sponsor will be higher for a relevant
   than an irrelevant sponsorship linkage.

Sponsor Relevance and Intended Behavioral Effects

The concepts of attitude and behavior are inextricably linked to persuasion. Attitudes are evaluative tendencies regarding some feature of the environment, and behaviors are the observable actions relating to those evaluations. Studies in advertising generally assume that more favorable brand evaluations can result in increased purchase intentions for the brand. A number of studies have demonstrated the ability of sponsorships to influence purchase intentions of female consumers (McDaniel 1999), Olympic viewers (Crimmins and Horn 1996), and sports fans (Madrigal 2000). However, not all authors agree that sponsorships can influence purchase behaviors (Hansen and Scotwin 1995), presumably due to their limited size and limited ability to persuade (McDonald 1991).

Research on associative links suggests that behavioral intentions for sponsorships can be enhanced with a relevant sponsor-sponsee link. It was argued earlier that relevant linkages help the spread of activation in memory by making sponsorship attitudes more accessible. Once the attitude is accessed, the attitude-to-behavior process is more likely to result. A relevant sponsorship linkage is postulated to act as a critical determinant of the attitude-to-behavior transfer. Cornwell, Pruitt, and Van Ness (2001) examined the value of logical sponsorship links for Indianapolis sponsors on product sales and stock prices. They found that Indianapolis 500 winners sponsored by companies with a direct link to the automotive industry yielded significantly greater sales and stock price increases than did sponsors with an indirect link. This discussion leads to the following hypothesis:

   H3: Purchase intent for the sponsored product will be higher
   for a relevant than an irrelevant sponsorship linkage.

Sponsor Relevance and Context Effects

Much research in advertising has been devoted to understanding context effects on audience processing of advertisements appearing in those contexts. Briefly, this research has established that the context in which ads appear is at least as important as the ads themselves. Because the context constitutes the emotional environment within which advertisements are evaluated and processed, the context is critical to determining how ads will he processed by the audience (Schumann and Thorson 1990).

A Web site serves as the context in the present study. Although consumers have intuitions about relevance that can help them distinguish relevant from irrelevant information, the context of a message can make these intuitions difficult to elicit (Sperber and Wilson 1995). In the context of sponsorships, this process is further complicated by the fact that so little information is conveyed about the meaning of the sponsorship. Selecting a context that is relevant to the sponsor's product or brand can enhance this interpretative process. Hirschman and Thompson (1997) have suggested that consumers' relationships to nonadvertising forms of mass media are essential to the manner in which meaning is derived from ads. The authors assert that the meaning of an advertisement depends on the combined interpretation of two sources: an advertising source and a nonadvertising source. Applied to sponsorships, consumers who think poorly of a Web site or believe its content lacks credibility will use these evaluations when forming attitudes toward companies that sponsor the Web site. In effect, the evaluations formed about the Web site will "rub off" on evaluations made about the sponsor. Known as image transfer, this effect has been demonstrated in situations where an event-brand sponsorship link was formed between a sporting event's image and the sponsoring brand. Gwinner and Eaton (1999) conducted an experiment and found that the image transfer process was enhanced when the event and brand were matched on either an image (the event and brand share a similar image) or functional (the product is used in the sponsored event) basis.

These findings suggest that companies that sponsor Web sites that consumers find credible will be able to capitalize on the sponsorship linkage better than companies that do not. That is, sites with high perceived credibility will likely transfer positive evaluations from the site to the sponsor in cases where the sponsorship linkage is relevant. Sites with low perceived credibility likely will not do so. Likewise, individuals who are devoted to a Web site and intend to return to it will likely hold positive attitudes toward the site as well. This evaluation should then transfer from the site to the sponsor in the manner just described. Although not a direct test of image transfer, Madrigal (2000) found through survey analysis that fans that were devoted to and identified with a particular sports team were also more likely to purchase products from team sponsors. The findings suggest that one's devotion to the context in which sponsors appear will influence evaluations of sponsors appearing in those contexts. To test for this context effect, the following hypothesis was proposed:

   H4: Web site credibility and intent to return to the site will
   moderate sponsorship linkage to influence attitude toward the
   sponsor, such that higher credibility and intent to return to the
   site will translate to more positive evaluations for relevant
   sponsors.

EXPERIMENT 1

Method

Experimental Design

The hypotheses were tested using a 2 (sponsorship linkage) X 2 (section) within-subjects factorial experiment. Sponsorship linkage has two levels: relevant and irrelevant. While there are many ways to examine sponsorship linkage, this study used the linkage between the sponsor's product and a specific section in the Web site. A relevant sponsorship link implies a close and natural connection between the sponsor's product and the Web site's section, such that the link is appropriate or fitting. An irrelevant sponsorship link implies the opposite. The second variable, section, is a replication variable with two levels, health and travel, which were selected because of their popularity among Internet sponsors (Yung 1998). A replication variable was selected to increase generalizability beyond a single section. Both sponsorship linkage and section were within-subjects factors, which means that each participant was exposed to two sponsors: a relevant sponsor and an irrelevant sponsor in the health and travel sections.

Pretest

The purpose of the pretest was to determine which products and sponsors to use in the research. This was accomplished by compiling a list of unfamiliar words to be used as fictitious sponsor names and then pairing these names with a list of health and travel products. Unknown brand names were used to minimize prior attitudes and associations that familiar brand names might evoke. Forty individuals who were recruited through a mall intercept method evaluated the sponsor/product pairs in terms of liking, familiarity, and health or travel orientation. Each participant received five dollars. The relevant sponsor/product pairs selected for the health section were Jasil herbal teas and Birele vitamins; the irrelevant pairs were Zenan cameras and Bisk travel maps. The relevant sponsor/product pairs for the travel section were Situs luggage and Dree travel service; the irrelevant pairs were Jolo weight loss service and Fanon stationary bikes. Using a variety of low-and high-ticket products and services should enhance the ability to generalize across product/service type.

Stimulus Materials

One Web site was designed and four versions of the Web site were created to control for ordering effects. Each Web site contained four pages: one for the study's directions, one for the health section, one for the travel section, and a final screen instructing participants to fill out the paper-and-pencil survey. To make the Web site look as authentic as possible, the site's logo (i.e., The Daily Times) appeared at the top of both screens in each section, followed by the section heading and logo. The sponsorship ("Today's sponsor is: brand/product") appeared just below the section heading, followed by the reporter's byline and news story. Each page contained a "click here" button at the bottom of the screen to advance to the next page. A professional graphic designer was paid to design the Web site and logos. The news stories were pretested using a group of 15 advertising students to control for tone, length, involvement, credibility, and health/travel-relatedness. The final Web site was pretested using the same advertising students to ensure that all links worked properly. Each student was paid $20 for his or her time.

Sample

A total of 110 undergraduate students (20% male, 80% female) from a large Midwestern university participated in the study. Each participant received $10. Using a relatively homogenous group of students as the initial sample should reduce the effects of factors other than the manipulated independent variable. A student sample was also deemed appropriate given recent attempts to attract a younger audience to newspaper Web sites (Albers 2002).

Procedure

The experiment was conducted in a computer laboratory in several sessions. All computer settings, including the screen's resolution, color palettes, font size, and refresh rates, were standardized for control purposes. Participants were randomly assigned to one of four versions of the Web site to control for selection bias. Participants were instructed to read the news stories carefully, but at their usual pace. To increase the likelihood that the stories would be read, participants were told that they would be asked questions about the stories. Participants were given nine minutes to read the stories, which was determined by averaging the reading times of a slow and fast reader of the four stories prior to the experiment.

Independent Variable

The independent variable, sponsorship linkage, was manipulated at two levels: relevant and irrelevant. Two sponsors were selected for each level so that effects could be attributed to relevance level rather than idiosyncrasies of an individual sponsor. As noted above, pretest results determined the sponsors and products to be used (see above). Participants were exposed to a relevant and irrelevant sponsor of different brands in both the travel and health sections.

Dependent and Moderating Variables

There were three dependent variables and two moderating variables. The dependent variables were sponsor recall, attitude toward the sponsor ([A.sub.sp]), and purchase intent (PI). The two moderating variables were Web site credibility and intent to return to the Web site. The dependent and moderating variables were measured in the following order.

Recall. Four memory measures--two unaided and two aided--were used to assess sponsor recall. Participants were first asked to list all brands and/or products or services that they could recall (unaided). The two aided (i.e., closed-ended) items asked participants to select the sponsor/product they remembered seeing, followed by a choice of four sponsor's names, one correct and three decoys. This process was repeated for the relevant and irrelevant sponsors. All correct answers received + 1 for a total recall score that ranged from 0 to 4.

Attitude Toward the Sponsor ([A.sub.sp]). Participants were then asked to rate their attitude toward the sponsors ([A.sub.sp]) on three, five-point semantic differential scales: bad/good, disliked/liked, unfavorable/favorable ([alpha] = .92).

Purchase Intent (PI). Next, purchase intent (PI) was measured using three, five-point semantic differential scales, which included the sponsor's name and product followed by the endpoints: I'm likely to make a purchase/I'm unlikely to make a purchase, I would like to have more information/I would not like to have more information, and I'm interested in it [the brand/product]/I'm not interested in it [the brand/product] ([alpha] = .73).

Web Site Credibility. Web site credibility was measured next with eleven, seven-point semantic differential scales: not qualified/qualified, not believable/believable, not experienced/experienced, not knowledgeable/knowledgeable, untrustworthy/ trustworthy, unbiased/biased, uncompromising/compromising, nor reputable/reputable, unethical/ethical, not objective/ objective, and not credible/credible (Haley 1996) ([alpha] = .89).

Intent to Return to the Site. Last, intent to return to the Web site was assessed using three Likert scales, which ranged from (1) unlikely to (5) very likely. The statements included the following: What is the likelihood that you will return to this newspaper sometime in the near future? What is the likelihood that you will subscribe to this newspaper? What is the likelihood that you will return to the health/travel section of this newspaper? Cronbach's [alpha] for the three items was .84.

Manipulation Check

Although the stimulus materials were pretested to ensure a strong sponsorship linkage manipulation, participants in the study also evaluated sponsorship linkage on a five-point scale by responding to the statement, "How well did this brand's product or service fit with the section it sponsored?" The scale ranged from 1 (not very well at all) to 5 (very well). To assess sponsorship linkage, a repeated measures analysis of variance (ANOVA) was conducted. Results revealed a significantly closer link for the relevant (M = 3.91, SD = 1.23) than the irrelevant (M = 2.77, SD = 1.07) sponsors, thus indicating a successful manipulation of sponsorship linkage, F(1, 206) = 45.90, p < .001.

Results

Statistical Procedures and Subsidiary Analyses

Participants responded to each level of sponsorship linkage (relevant/irrelevant) on multiple dependent measures. Therefore, a multivariate analysis of variance (MANOVA) is the logical statistical procedure to use provided the following assumptions are met. First, the assumption of sphericity (i.e., homogeneity of variance) must be met. Second, two or more of the dependent variables must correlate (Lomax 2001). To determine whether the two provisions were met, two subsidiary analyses were undertaken. To test for sphericity, a one-way MANOVA was conducted using sponsorship linkage as the within-subjects factor and the three dependent measures and two covariates as the criterion variables. The findings confirmed that sphericity was met for the student sample (Box's M = 20.51, F = 1.33, p = .17). A correlation analysis of the three dependent and two moderating variables revealed a significant correlation among four of the variables for the student sample (see Table 1), thus confirming the second set of provisions. Confirmation of both subsidiary analyses indicates that a one-way MANOVA is appropriate here.

A third subsidiary analysis was performed to determine whether to include the replication variable (section) in the statistical analyses that follow. If no significant differences occur on the dependent measures, then it is appropriate to drop the replication variable and, in effect, "collapse" the data across the two levels of sponsorship linkage. A one-way MANOVA using section as the within-groups factor and the three dependent measures as the criterion variables yielded no significant results, and no interaction effect was found when entering sponsorship linkage as a second within-subjects factor in the analysis. In short, the results confirm that neither the health section nor the travel section influenced consumers' reactions to the sponsors or Web site, so the replication variable was dropped from additional analyses. The remaining findings were reported across the two levels of sponsorship linkage.

Hypothesis Tests

Main Effects. Hypothesis 1 predicted that relevant sponsors would be remembered better than would irrelevant sponsors. This hypothesis was tested by conducting a one-way MANOVA using sponsorship linkage as the within-groups factor, the three dependent variables as the criterion variables, and the two moderating variables as covariates. The overall MANOVA was significant, Wilks's [lambda] = .70, F(3, 211) = 30.75, p < .001. Relevant sponsors were remembered better than were irrelevant sponsors, F(1, 213) = 38.74, p < .001, thus supporting H1. Hypothesis 2 predicted that relevant sponsors would also be liked better than would irrelevant sponsors. This hypothesis was also confirmed, F(1, 213) = 16.59, p < .001 (see Table 2). The third hypothesis, which predicted that relevant sponsors would yield higher purchase intentions than would irrelevant sponsors, was also supported, F(1, 213) = 45.92, p < .001. All means and standard deviations can be found in Table 2.

Moderating Effects. Hypothesis 4 predicted that Web site credibility and intent to return to the Web site would moderate any main effects resulting from sponsorship linkage. Using the same MANOVA procedure described above, the results showed that Web site credibility moderated sponsorship linkage, Wilks's [lambda] = .96, F(3, 211) = 2.99, p < .05, on [A.sub.sp,], F(1, 213) = 7.02, p < .01. Intent to return to the Web site did not moderate sponsorship linkage (p = .27), but approached near significance for [A.sub.sp] (p = .08) (see Table 2). Thus, only partial support was found for H4.

Discussion

The results of Experiment 1 suggest that sponsors that closely align with a relevant section of a newspaper Web site will be more persuasive than irrelevant sponsors. Recall was higher for relevant than irrelevant sponsors, as was sponsor liking and purchase intentions. In addition, Web site credibility (but not intent to return to the site) had a moderating effect on sponsor evaluations, where higher credibility scores translated to more positive evaluations for the relevant sponsors. These results suggest that companies would do best to select Web sites that have relevance to their product. The results also demonstrate the importance of selecting credible Web sites that have a loyal following of users. When consumers express positive attitudes toward a Web site, those attitudes can apparently transfer to the companies that sponsor relevant content.

Inasmuch as these findings support the predicted hypotheses, the findings might be attributed to the fact that the sample was comprised of a group of college students whose attitudes about e-newspapers and Internet sponsorships may differ from an older, more demographically diverse sample of nonstudent adults. Younger consumers are more avid readers of e-newspapers (Albers 2002), and may therefore be more prone to seeing sponsorships in e-newspapers. If this is true, then we would expect that students may be more accepting and perhaps less discriminating of seeing alternative forms of advertising in e-newspapers than, say, a heterogeneous group of nonstudents. Given this possibility, it was decided to replicate the first experiment using a more demographically diverse group of nonstudent adults.

EXPERIMENT 2

Method

Sample

Experiment 2 used the exact method employed in Experiment 1. The independent and dependent variables were the same, as were the stimulus materials and experimental procedures. The only difference between the two experiments was the sample. Experiment 1 used a group of students, whereas Experiment 2 used a group of nonstudent adults. A total of 86 adults were recruited from professional clubs and organizations (e.g., Audubon Society, Garden Club, African American Institute) of a small Midwestern city. Five surveys were dropped because of missing data. This left a total of 81 adults (79% female, 21% male). Ages ranged from 20 to 83, with an average age of 54 (SD = 15.93). Household income ranged from $25,000 or less to $70,000 or more, with an average household income of $50,000, thus confirming that samples 1 and 2 were demographically different. The host club or organization was paid $20 for each member that participated in the experiment.

Manipulation Check

The same statistical procedures used in Experiment 1 were also used in Experiment 2. Subsidiary analyses were run, as described above. The results confirmed that a one-way MANOVA is the appropriate statistic to be used (see Table 3). The replication variable was examined with the same procedure described above. As in Experiment 1, the results were not statistically significant and the replication variable was subsequently dropped from further analysis. Last, the same statistical procedure was run to determine whether the manipulation of sponsorship linkage was successful. Findings revealed that participants perceived a significantly stronger link for relevant (M = 3.44, SD = 1.32) than irrelevant (M = 2.84, SD = 1.00) sponsors, F(1, 151) = 11.80, p < .001.

Results

As in Experiment 1, all hypotheses in Experiment 2 were analyzed using a one-way MANOVA within-groups design. Similar to Experiment 1, the findings revealed a significant main effect for sponsorship linkage on all three dependent variables, thereby supporting the first three hypotheses, Wilks's [lambda] = .87, F(3, 147) = 7.14, p < .001. For example, the results showed that the relevant sponsors were more effective than the irrelevant sponsors in terms of sponsor recall, F(1, 149) = 14.14, p < .001; [A.sub.sp,] F(1, 149) = 3.82, p < .05; and PI, F(1, 149) = 5.65, p < .05. The means and standard deviations are displayed in Table 4. Consistent with H4, Web site credibility moderated sponsorship linkage, Wilks's [lambda] = .92, F(3, 147) = 4.39, p < .01, on [A.sub.sp,] F(1, 149) = 11.83, p < .001. Intent to return to the site also moderated sponsorship linkage, Wilks's [lambda] = .90, F(3, 147) = 5.26, p < .01, on [A.sub.sp,] F(1, 149) = 14.87, p < .001 (see Table 4).

GENERAL DISCUSSION

This research examined the effects of sponsor relevance on consumers' reactions to Internet sponsorships. Sponsorship linkage was presented as a means of structuring Internet sponsorships. Contrary to past research on event sponsorship, the findings demonstrated a consistent pattern of findings. Relevant sponsors were more likely than irrelevant sponsors to elicit stronger recall, brand evaluations, and purchase intentions. It is important to note that these findings were replicated across four product types and two Web site sections. The findings were also replicated for both student and nonstudent adults.

A theoretical implication of the findings is the notion that associative links are critical to the success of Internet sponsorships. Relevant sponsorship links were apparently more successful than irrelevant sponsorship links at creating an associative link. Although not specifically tested, the findings suggest that the nature of the associative link may be at least as important as the link itself. In the present study, the sponsor-sponsee link was defined in terms of the perceived relevance between the product's descriptive element (travel service) and its alliance (travel section in a Web site). By matching sponsors and sponsees in terms of shared semantic features, participants were presumably able to connect related concepts on the basis of their relevance. Exactly how this process occurs is not entirely clear from the current study. Theories on associative learning suggest that relevant sponsors may ignite the spread of activation to a greater extent than irrelevant sponsors do. Future studies will obviously need to examine this process in greater detail. Nevertheless, the current research illustrates the importance of matching a sponsor and sponsee in terms of features that consumers can readily identify in an effort to help them make sense of the sponsorship.

Practically speaking, these findings suggest that companies should select Web sites (and content within Web sites) that are directly related to their products or services. In the present study, there were no obvious benefits for brands that sponsored irrelevant Internet content. This finding seems to challenge past sponsorship research that asserts that there are benefits to simply having a sponsorship presence at an event. Indeed, the results of the current research suggest that selecting an appropriate sponsee may be more complex than previously thought. In addition, while the concept of sponsorship linkage has been applied primarily to event sponsorships, the findings suggest that Internet sponsorships can also benefit from being structured in terms of relevance. Despite their limitations in terms of content, relevantly paired sponsors and sponsees can prompt greater persuasion than irrelevant sponsors and sponsees.

This research also examined the moderating effects of two Web site variables on sponsor relevance. The findings revealed that the perceived credibility of the site, and in some cases, an intent to return to the site, moderated sponsor relevance. How consumers perceived the contexts within which Internet sponsorships appeared clearly mattered. Presumably, contexts that lack credibility will hurt a sponsor's chance of success even if the sponsorship linkage is relevant. The results found here suggest that sponsorship links interact with other factors that can potentially influence the evaluation of Internet sponsorships. These factors will need to be articulated and examined in future studies. Nevertheless, this study helps to establish the importance of the context (i.e., sponsee) in which an Internet sponsorship appears. When consumers felt positively about the Web site in which the sponsorship was placed, those positive feelings transferred to more positive evaluations of the sponsoring brand. This image transfer effect has been demonstrated in other sponsorship studies, but not in the context of a Web site.

On a practical level, these findings demonstrate the importance of sponsoring content that consumers like and perceive as credible. Although this may seem like an obvious or logical thing to do, Internet sponsors do not currently appear to take context into consideration, as indicated by countless irrelevant Internet sponsorships. As Web pages become more cramped by Internet ads, it may become increasingly difficult for on-line advertisers to get users' attention. Sponsoring valued content may be a viable option in overcoming these practical difficulties.

LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH

Inasmuch as these findings demonstrate the role of relevance in sponsorship effects, a number of limitations should be noted. First, the hypothetical associations used in the experiments were designed to represent an extreme (although not unrealistic) manipulation of sponsorship linkage. For that reason, the sponsorship linkages provided participants with only one kind of sponsor/product information. Additional studies are needed to test the effects of other types of sponsor/product information. In addition, only one type of brand association (i.e., descriptive) was tested here. Future studies will need to test alternative brand associations (e.g., functional, symbolic, image) to determine their effects on sponsorship linkages.

In addition, sponsorship linkage was conceptualized in terms of an alliance between a sponsor's product and a section within a Web site. Alternative conceptualizations should be considered in follow-up studies. What, for example, is the effect of aligning a brand with a relevant/irrelevant chat room, virtual scoreboard, or book review? It will also be important to use actual brands in follow-up studies. The use of hypothetical brands may have provided knowledge structures that were less stable, and therefore more susceptible to change, than those of actual brands.

This research has also asserted that Internet sponsorships are a natural extension of traditional sponsorships since these two forms of promotion look alike. Whether this is actually the case is an issue that will need to be examined and tested in follow-up studies. In addition, although Internet and traditional sponsorships may look the same, it is unclear whether they share similar functions. Operationally speaking, Internet sponsorships differ from traditional sponsorships in terms of what is being sponsored. With traditional sponsorships, companies primarily sponsor events. Companies that sponsor Web sites typically sponsor content. The sheer volume of content found on-line implies that relevant sponsorship linkages may be easier to achieve in an Internet context. A logical next step is to test the effects of Internet and traditional sponsorships to determine which are better for what purposes.

Last, this research tested the effects of sponsorship linkages on memory, attitudes, and purchase intentions. Not all authors agree, however, that these are appropriate measures of sponsorship effects (see McDonald 1991). Perhaps this is due in part to the fact that past research on event sponsorships has yielded inconsistent findings when it comes to awareness and preference measures. The current research demonstrates that these measures can provide meaningful evaluations of how well an Internet sponsorship will perform. Whether this is due to the nature of sponsorship linkages or some inherent characteristic of Internet sponsorships is an issue that will need further development.

CONCLUSION

In short, this study represents the first attempt to examine the role of relevance in consumer reactions to Internet sponsorships. Relevant sponsorship linkages clearly benefited the brands that used them. The context in which Internet sponsorships appeared also proved to be an important factor to consider when selecting appropriate outlets to sponsor. Although more research is needed to better understand the effects of Internet sponsorships, the concepts presented here may be useful for scholars and practitioners interested in better understanding the boundaries of this unique on-line advertising strategy.

TABLE 1
Correlations Among the Three Dependent and Two Moderating Variables
(Student Sample)

                                         Correlations

                            1       2        3        4        5

Sponsor recall (1)         1.00    .15 *   .13      .03      .06
Attitude toward the               1.00     .32 **   .19 **   .14 *
  sponsor (2)
Purchase intent (3)        1.00   -.02     .05
Web site credibility (4)          1.00     .09
Intent to return to the           1.00
  Web site (5)

Note: n = 220 except in cases where data were missing.

* p < .05

** p < .01.

TABLE 2
Means and Standard Deviations for the Dependent Variables
and Covariates (Student Sample)

                                Attitude
                  Sponsor       toward        Purchase
                  recall        sponsor        intent

                Mean    SD    Mean     SD    Mean    SD

Irrelevant      1.00   1.20   11.89   2.77   6.70   2.66
  sponsor (a)
Relevant        1.99   1.13   13.45   2.74   9.28   2.93
  sponsor (b)

                                Intent to
                  Web site      return to
                 credibility     Web site

                Mean     SD     Mean    SD

Irrelevant      55.56    9.83   7.07   2.81
  sponsor (a)
Relevant        56.50   10.30   7.25   3.35
  sponsor (b)

Note: The sample size may be less than 110 for each
level of sponsorship linkage in cases where data were missing.

(a) n = 109.

(b) n = 108.

TABLE 3
Correlations Among the Three Dependent and Two Moderating Variables
(Adult Sample)

                                       Correlations

                            1      2       3         4         5

Sponsor recall (1)         1.00    .03    .20 *   -.07       .08
Attitude toward the               1.00    .04      .34 **    .38 **
  sponsor (2)
Purchase intent (3)                      1.00      .04       .12
Web site credibility (4)   1.00                   1.00       .27 *
Intent to return to the                                     1.00
  Web site (5)

Note: n = 162 except in cases where data were missing.

* p < .05

** p < .01.

TABLE 4
Means and Standard Deviations for the Dependent Variables
and Covariates (Adult Sample)

                                Attitude
                  Sponsor       toward        Purchase
                  recall        sponsor        intent

                Mean    SD    Mean     SD    Mean    SD

Irrelevant       .94   1.01   12.21   2.49   6.76   2.64
  sponsor (a)
Relevant        1.67   1.35   13.17   2.83   7.93   3.15
  sponsor (b)

                                Intent to
                  Web site      return to
                 credibility    Web site

                Mean     SD     Mean    SD

Irrelevant      53.79   11.21   6.64   3.41
  sponsor (a)
Relevant        54.37   12.41   7.42   3.49
  sponsor (b)

Note: The sample size may be less than 81 for each level
of sponsorship linkage in cases where data were missing.

(a) n = 77.

(b) n = 76.

The author thanks Pulitzer Technologies for generously funding this research.

REFERENCES

Albers, Rebecca Ross (2002), "Attracting Younger Readers," Presstime (May), available at www.naa.org/Presstime/.

Anderson, John R. (1983), "A Spreading Activation Theory of Memory," Journal of Verbal Learning and Verbal Behavior, 22 (3), 261-295.

Becker-Olsen, Karen L. (1998), "Corporate Sponsorship: A Look at the Moderating Effects of Fit and Reach on Image and Purchase Intentions," Ph.D. dissertation, Lehigh University.

Cornwell, T. Bettina, and Isabelle Maignan (1998), "An International Review of Sponsorship Research," Journal of Advertising, 27 (1), 1-21.

--, Stephen W. Pruitt, and Robert Van Ness (2001), "The Value of Winning in Motorsports: Sponsorship-Linked Marketing," Journal of Advertising Research, 41 (1), 17-31.

Crimmins, James, and Martin Horn (1996), "Sponsorship: From Management Ego Trip to Marketing Success," Journal of Advertising Research, 36 (July/August), 11-21.

D'Astous, Alain, and Pierre Bitz (1995), "Consumer Evaluations of Sponsorship Programmes," European Journal of Marketing, 29 (12), 6-22.

Dean, Dwane Hal (1999), "Brand Endorsement, Popularity, and Event Sponsorship as Advertising Cues Affecting Consumer Pre-Purchase Attitudes," Journal of Advertising, 28 (3), 1-12.

--(2002), "Associating the Corporation with a Charitable Event Through Sponsorship: Measuring the Effects on Corporate Community Relations," Journal of Advertising, 31 (4), 77-87.

Gardner, Meryl P., and Philip Shuman (1987), "Sponsorship: An Important Component of the Promotion Mix," Journal of Advertising, 16 (1), 11-17.

Gwinner, Kevin P., and John Eaton (1999), "Building Brand Image Through Event Sponsorship: The Role of Image Transfer," Journal of Advertising, 28 (4), 47-57.

Haley, Eric (1996), "Exploring the Construct of Organization as Source: Consumer's Understandings of Organizational Sponsorship of Advocacy Advertising," Journal of Advertising, 25 (2), 19-35.

Hansen, Flemming, and Lene Scotwin (1995), "An Experimental Enquiry into Sponsoring: What Effects Can Be Measured?" Marketing and Research Today, 23 (3), 173-181.

Hastings, Gerard B. (1984), "Sponsorship Works Differently from Advertising," International Journal of Advertising, 3, 171-176.

Hirschman, Elizabeth C., and Craig J. Thompson (1997), "Why Media Matter: Toward a Richer Understanding of Consumers' Relationships with Advertising and Mass Media," Journal of Advertising, 26 (1), 43-60.

Internet Advertising Bureau (2002), "IAB Internet Advertising Revenue Report," available at www.iab.net/adrevenue/ IAB_PWC_2001Q4.pdf.

Johar, Gita Venkataramani, and Michael Tuan Pham (1999), "Relatedness, Prominence, and Constructive Sponsor Identification," Journal of Marketing Research, 36 (3), 299-312.

Lee, Myung-Soo, Dennis M. Sandler, and David Shani (1997), "Attitudinal Constructs Toward Sponsorship: Scale Development Using Three Global Sporting Events," International Marketing Review, 14 (3), 159-169.

Lomax, Richard G. (2001), An Introduction to Statistical Concepts for Educational and Behavioral Sciences, Mahwah, NJ: Lawrence Erlbaum.

McDaniel, Stephen R. (1999), "An Investigation of Match-Up Effects in Sport Sponsorship Advertising: The Implications of Consumer Advertising Schemas," Psychology and Marketing, 16 (2), 163-184.

McDonald, Colin (1991), "Sponsorship and the Image of the Sponsor," European Journal of Marketing, 25 (11), 31-38.

Madrigal, Robert (2000), "The Influence of Social Alliances with Sports Teams on Intentions to Purchase Corporate Sponsors' Products," Journal of Advertising, 29 (4), 13-24.

Meenaghan, Tony (1991), "The Role of Sponsorship in the Marketing Communication Mix," International Journal of Advertising, 10 (1), 35-47.

Osgood, Charles K., and Percy H. Tannenbaum (1955), "The Principle of Congruity in the Prediction of Attitude Change," Psychological Review, 62, 42-55.

Otker, Ton, and Peter Hayes (1987), "Judging the Efficiency of Sponsorship," European Research, 15 (4), 53-58.

Rajaretnam, Javalgi (1994), "The Long-Term Effects of Sponsorship on Corporate and Product Image: Findings of a Unique Experiment," Marketing and Research Today, 22 (1), 62-74.

Rodgers, Shelly, and Esther Thorson (2000), "The Interactive Advertising Model: How Users Perceive and Process Online Ads," Journal of Interactive Advertising, 1 (1), available at jiad.org/v011/n01/rodgers/(accessed June 2003).

Schumann, David W., and Esther Thorson (1990), "The Influence of Viewing Context on Commercial Effectiveness: A Selection-Processing Model," Current Issues and Research in Advertising, 12 (1), 1-24.

Sperber, Dan, and Deirdre Wilson (1995), Relevance Communication and Cognition, 2nd ed., Cambridge, MA: Blackwell.

Stipp, Horst (1998), "The Impact of Olympic Sponsorship on Corporate Image," International Journal of Advertising, 17 (1), 75-87.

--, and Nicholas P. Schiavone (1996), "Modeling the Impact of Olympic Sponsorship on Corporate Image," Journal of Advertising Research, 36 (4), 22-28.

Yung, Katherine (1998), "Sponsoring the News Online," Dallas Morning News (December 28), 1D, 3D.

Shelly Rodgers (Ph.D., University of Missouri-Columbia) is an assistant professor in the Department of Advertising at the Missouri School of Journalism.