SOCIAL DIMENSIONS OF INTERACTIVE ADVERTISING

Introduction

The Internet is the consummate modern symbol of communication-driven societal transformation in the world today. Along with the rapid increase of the Internet population, the Internet advertising market has grown exponentially. For example, total online advertising expenditure in the U.S. reached $8.2 billion in the year 2000 (Internet Advertising Bureau, 2001).

The uniqueness underlying the rapid proliferation of the Internet seems to emanate from the fact that it has a substantive impact on both interpersonal and mass-mediated dimensions of communication. On the net, the traditional boundaries between interpersonal and mass-mediated communication are blurred (Morris and Ogan, 1996). Since the beginning of the Web, many scholars and practitioners have highlighted the integrative nature of the Internet, and the concept of “interactivity” has centered on discussions regarding the functions and applications of the medium in advertising and communication fields (i.e. Cho 1999; Rafaeli and Sudweeks 1997; Morris and Ogan 1996; Pavlik 1996; Rafaeli 1988). Particularly in advertising studies, scholars have focused on interactivity as a measure of advertising effectiveness on the Internet (Macias, 2000; Wu, 2000; Cho, 1999). For example, Cho and Leckenby (1997) found that interactivity plays a crucial role as a criterion for advertising effectiveness on the Internet. They empirically tested and verified that “a higher degree of interactivity yields better advertising effects (i.e., favorable attitude toward the target ad, favorable attitude toward the brand, and high purchase intention)”. In addition, Wu (2000) also found that a higher degree of perceived interactivity leads to more favorable attitude toward the website.

Although the interactive characteristics of the Internet have received considerable attention from a variety of fields, scholars are still in the dark about the nature of the interactive communication process on the Internet. Despite the uniqueness of the Internet as an electronically networked environment, most Internet advertising studies dealing with interactivity still tend to consider Internet users “a mass of disconnected individuals hooked up to the media but not to each other” (Katz, 1957). Research perspectives of the existing studies have been confined to interaction between medium/message and an individual user, while the social interconnections between Internet users have received minimal attention. This neglect stems partly from the fact that the term “interactivity” has been defined narrowly, which in turn has prevented a consideration of interaction as a part of social processes.

The objective of this study is twofold. The first objective is to clarify the conceptual problems resulting from the narrow definition of interactivity, and to conceptualize interactivity as a comprehensive one mirroring the social dimension of interaction. The second objective is to show empirically how user-medium interactivity is embedded in the social processes of communication on the Internet. Employing social network analysis, this study tries to demonstrate how people’s perceived interactivity to the Internet is influenced by the personal social networks that they sustain and manage over the Internet, and how personal factors, such as opinion leadership and the need for cognition, are related to various characteristics of personal social networks. An examination of the relationships between the social interconnections of individuals and relevant personal factors leads to a more elaborate understanding of the dynamic processes embedded in building “interactivity” with potential consumers – a process that is crucial for interactive advertising.

Interactivity Definitions Revisited

According to Cho (1999), “interactivity” has been defined in three dominant ways: 1) human - human interaction 2) human - machine interaction and 3) human - message interaction. Human-human interaction refers to social interaction entailing face-to-face/mediated communication between sender and receiver, while human-machine/human-message interaction is the behavioral/psychological interaction type occurring between human and non-human entity. The third type of interactivity, human-medium/message interaction, has been of particular importance in Internet advertising research literature.

The research trend focusing on the human-medium/message interaction can roughly be classified into two different perspectives. The first one is the approach viewing interactivity as a given technological characteristic of the medium. For example, Steuer (1992) classified a variety of media based on three interactivity levels (high/medium/low). Based on three criteria (speed, range, mapping), he viewed interactivity as a consequence depending on technological factors, though he did analyze the roles of participants in determining interactivity level. He defined interactivity as “the extent to which users can participate in modifying the form and content of a mediated environment in real time” (p. 84).

The second approach is to regard interactivity as being dependent on the psychological state of an individual user (i.e. Wu, 1999; Newhagen, Cordes, and Levy, 1995). This approach regards both subjective and system efficacy as crucial factors for interactivity, but focuses more on the perceived aspects of interactivity by an individual user. Wu (2000) defined perceived interactivity as “the extent to which a person perceives he or she controls over the interaction process, his or her communicative counterpart (a person, a mass-mediated environment, or a computer-mediated environment) personalizes and responds to his or her communicative behavior” (41).

Although these approaches shed light on some important technological/user-centric aspects that underlie the interactivity concept, they rest on attribute-based perspectives that consider interactivity a static attribute, not a dynamic social process. Rafaeli (1988) has criticized the attribute-based treatments of the interactivity concept: “This technical tit-for-tat reciprocity…does not have an obvious reflection on the social relations involved. Even taken together, the technological improvements should not be mistaken as providing or even regulating interactivity” (p 116). These approaches tend to reduce interactivity either to a given technological characteristic or a psychological state. By doing so, they miss the dynamic inter-subjective/interpersonal relations surrounding individuals.

From “interACTIVE” to “INTERactive”: Interactivity as a Social Process

By definition, human interaction is social and always entails “reciprocally sustained involvement” (Goffman, 1957). Even dyadic interaction, the simplest unit of interaction, cannot be fully explained by the participants’ characteristics because relational properties exist, which affect and constrain individual actors’ behaviors.

Without a consideration of its social aspects, interaction can seem to be simple aggregates of fragmented mechanical reactions. Many researchers tend to regard the following factors as sufficient criteria for determining interactivity: immediacy of response, diversity of user options, feedback, bi-directionality, etc. Although these criteria illuminate some important aspects of interactivity, they are not enough to differentiate socially meaningful interaction from fragmented reactions. Rafaeli(1988) writes:

Consider the exchange between a person and a vending machine: (1) Sign on machine catches the person’s attention. (2) Person inserts coin in machine. (3) Machine dispenses a candy bar. Is this an interactive situation? Are vending machines interactive media? The vending machine encounter, while clearly bi-directional, and perhaps even reactive to a very tasty degree, lacks qualifications for interactivity (p. 121).

To distinguish interaction from individual reaction, one must contextualize interactivity within social processes, which are composed of the interrelations between various psychological and technological factors. In other words, interactivity is not a static attribute of either an individual or a given medium, but a consequence generated in the course of interactive communication processes, which are affected by personal, social, and technological factors.

Understanding the Relational Perspective: Social Network Analysis

From the sociological perspective, “Behavior is interpreted in terms of structural constraints on activity, rather than in terms of inner forces within units that impel behavior in a voluntaristic, sometimes teleological, push toward a desired goal” (Wellman, 1988, p. 20). This sociological tradition can be traced back to French sociologist, Emile Durkheim’s (1897) important argument: “Sociological method as we practice it rests wholly on the basic principle that social facts must be studied as things, that is, as realities external to the individual.”

Social influence may be either 1) a social psychological one, such as internalized norms/values, or 2) a structural one that is a sociological property external to individuals. Even the number of people in a group may be an important sociological constraint that influences the actors’ behaviors in the interaction situation. Simmel (1950) writes, “It will immediately be conceded on the basis of everyday experiences, that a group upon reaching a certain size must develop forms and organs which serve its maintenance and promotion…” (87).

Understanding the sociological perspective of structural influences that are external to individuals is especially crucial for Internet usage studies. Compared with other traditional media, the Internet is unique in that it provides a new communication environment that merges the roles of the information-intermediary and the medium of interpersonal/social communication. Individual Internet usage is embedded into the electronically networked social environment (Wellman, Salaff, Dimitrova, Garton, Gulia, and Haythornthwaite, 1996).

An analysis of the socially embedded nature of the Internet usage requires taking a relational perspective different than traditional attribute-based research methods. Rogers and Bhowmik(1970) define the relational perspective as “… a research approach in which the unit of analysis is a relationship between two or more individuals” (p. 524). In this approach, relations between individual actors are not treated as given assumptions, but as variables. Social network analysis is the most prominent and advanced quantitative research methods based on the relational perspective. Scott (2000) writes, “Social network analysis emerged as a set of methods for the analysis of social structures, methods that allow an investigation of the relational aspects of these structures” (p. 38). Social network analysts try “to describe social structure in terms of networks and to interpret the behavior of actors in light of their varying positions within social structure” (Marsden, 1990, p. 436). Wellman (1988) notes that the primary focus of this approach is “how the patterned relationships among multiple alters jointly affect network members’ behavior. Hence it is not assumed that network members engage only in multiple duets with separate alters” (20). In this distinct approach, various relational properties between individuals, groups/organizations, or nation states are operationalized as quantitative variables, which can be measured and analyzed mathematically as well as statistically.

Social network analysis consists of two different approaches: the ego-centric approach and the whole network approach. The ego-centric approach “considers the relations reported by a focal individual. These ego-centered networks provide an Ptolemic views of their networks from the perspective of the persons at the centers of their network” (Garton, Haythornthwaite, and Wellman, 1997). On the other hand, the whole network approach focuses on a network of a certain population, which is based on specific boundary specifications. Examples include clubs and formal organizations (Scott, 2000). This study employs the ego-centric (or personal) approach because the network variables measured from the ego-centric approach can be used with other traditional attribute-based variables. The virtue of this approach is that it enables researchers to investigate various individual attributes in the context of ongoing social relationships.

Hypotheses: Perceived Interactivity, Personal Networks, and Personal Factors

This study examines the social process linking various personal and social factors with the users’ psychological perceptions of the Internet as a distinct medium. It measures perceived interactivity, a concept proposed by Wu (2000) that reflects the users’ psychological perceptions of the Internet. However, in this study, the concept is applied to the Internet in general. Perceived interactivity reflects the users’ general expectancy of interactivity toward the Internet, which may provide long-term contexts for various short-term studies of perceived interactivity toward specific websites.

The networked nature of the Internet suggests that the behavioral/psychological aspects of individual users may be conditioned, situated, and influenced by their electronically mediated social surroundings since their potential opportunities for communication and social support are constrained by the properties of his/her social network. “Despite the limited social presence of CMC [computer-mediated communication], people find social support, companionship, and a sense of belonging through the normal course of CSSNs [computer-supported social network] of work and community, even when they are composed of persons they hardly know” (Wellman, Salaff, Dimitrova, Garton, Gulia, and Haythornthwaite, 1996, 220). Hence, an Internet user who frequently interacts with others may be exposed to more resources and social support that are available through the Internet. With the accumulation of these online experiences and his or her history of interactions offline, he or she may become more communication-oriented, and develop a higher expectancy level of interactivity to the medium.

In order to measure the personal network properties, this study first employs a concept/variable called “social network density,” which mirrors the quantitative aspects of individual users’ personal social networks. Density refers to “the general level of linkage among the points in a graph” (Scott, 2000, 69). In the social network analysis, point refers to individuals or individual entities, and graph to graphical representation of the linkages among points. Marsden (1990) defines the operational definition of network density as “the mean strength of connections among units in a network” (453). Network density reflects the proportion/strength of connections between network members. Second, the study measures the frequency of contacts over the Internet (e.g. email, instant messenger) between network members. Frequency of contacts between network members indicates the degree of online communication activities with which the network members are involved. In addition to overall frequency of contacts between network members, this study measures contact frequency between a focal individual and other network members in two ways: 1) frequency of sending message 2) frequency of receiving message. Distinguishing the two types of frequency enables researchers to identify the directions of communication activities with which each focal individual is involved.

Given the discussion, the following hypotheses are proposed:

H1: The density of online social network to which an Internet user belongs significantly affects his/her level of perceived interactivity toward the Internet.

H2: The frequency of contacts over the Internet between network members significantly affects the level of perceived interactivity toward the Internet.

H2a: The Frequency of sending message is positively related to perceived interactivity.

H2b: The Frequency of receiving message is negatively related to perceived interactivity.

Relationships between personal psychological/behavioral factors and the properties of personal social network are also examined in relation to the social communication processes. What kinds of personal factors are significantly related to the properties of personal networks on the Internet? What kind of people tend to have relatively dense personal networks and to contact others frequently on the Internet? In this study, two variables reflecting personal characteristics are examined in relation to social contexts: opinion leadership and the need for cognition. Katz and Lazarsfeld (1955) originally found that opinion leaders are more active communicator than non-leaders, in that they tend to be more exposed to mass media. Their findings suggest that opinion leaders who engage with Internet-related information/issues may participate more actively in online communication activities than others.

Need for cognition is a construct popularly employed in many consumer behavior studies. Need for cognition reflects the tendency to “engage in and enjoy thinking” (Cacioppo and Petty, 1982, p. 116). This cognitive characteristic may be related to the properties of personal network, in that communication activities require considerable cognitive and emotional efforts, and a cognitively active person may be a more active communicator than a cognitively less active person. Given the discussion, the following hypotheses are suggested:

H3a: Opinion leaders are likely to have denser personal networks than non-leaders.

H3b: Opinion leaders are likely to have higher frequency of contacts than non-leaders.

H4a: Need for cognition is positively related to the density of personal networks.

H4b: Need for cognition is positively related to the frequency of contacts between network members.

The following hypothetical diagram visually illustrates these proposed hypotheses:

Methods

Sampling Procedure

To test the hypotheses shown above, an online survey using the Cold Fusion database technology was conducted. Samples were drawn using the cluster sampling method. First, 8 states out of the total 52 states in the U.S. were randomly selected. Thirty cities from each state were also randomly picked. The e-mail addresses of prospective participants living in the 240 cities selected then were found using the e-mail search engine at Yahoo.com and Lycos.com. A recruiting electronic message for this online survey was sent to the prospective participants directing interested people to the URL of the online questionnaire (http://www.ciadvertising.org/student_account/fall_00/adv391k/junon93/database/expectancy.html). The total number of e-mails sent is 2320, and the response rate was 4.4%. The number of total respondents for this study was 102.

Measurements

Respondents were asked a series of questions about general Internet usage, perceived interactivity, personal networks of individuals, opinion leadership, need for cognition and demographic information. Perceived interactivity was measured by 10 Likert items with a 5-point scale (1 = Strongly Disagree, 5 = Strongly Agree). The alpha reliability on this scale was .89. Regarding personal social network, a variety of items were measured. First, each respondent was asked to report various information about five people with whom they typically interact over the Internet. Information included whether the respondent was a focal person of his/her personal network, the person’s first names/initials, and his or her demographic information such as age and gender. Second, closeness and frequency of contacts between respondent and network members as well as between network members were measured using five point scale items (e.g. 1= Very Rarely, 5= Very Often; 1= Not Close, 5= Very Close). Network density variable was obtained by dividing the sum of the items measuring closeness of relationships. Frequency of contact variable was calculated by summing items indicating contact frequency between a focal individual and other network members, as well as between other network members. Measures for frequency of sending and receiving messages were calculated by summing five items indicating either sending or receiving frequency between a focal individual and other network members. Opinion leadership was measured employing the 7 items with a 5 point scales developed by Childers (1986). Cronbach alpha of this measure is .92. Need for cognition was measured based on 10 items with 5 point scales developed by Cacioppo and Petty (1982). Cronbach alpha of this measure is .92.

Results

Hypothesis 1 states that a focal individual’s perceived interactivity is influenced by the density level of the personal network in which network members operate. Multiple regression analysis results in Table 1 show that network density is a significant predictor for respondents’ perceived interactivity level at .05 level, controlling for respondent’s age and the amount of time spent in Web usage, which may exert influence on perceived interactivity. Based on the results, it is possible to infer that people who belong to denser personal social network on the Internet are likely to have a higher level of perceived interactivity toward the Internet in general. Hypothesis 1 is supported.

Table 1. Multiple Regression: Predicting Perceived Interactivity

	Perceived Interactivity
Predictor Variables	B	Beta	t
Network Density	12.74	.43	2.41*
Age	-1.05	-.08	.82
Time Spent in Web Usage	2.26	.17	1.28
Overall Frequency of Contacts in Network	-.16	-.53	1.83
Frequency of Sending Messege	.75	.81	2.81**
Frequency of Receiving Message	-.53	-.56	1.95*
Constant	35.84

R = .42, R-squared = 18%, F = 3.17**, * p £ .05, **p £ .01

Hypothesis 2 states that the perceived interactivity level of a focal individual is significantly affected by the frequency of contacts between network members including the focal individual (ego). Table 1 shows that the overall frequency of contacts between network members is not a significant predictor for a focal individual’s perceived interactivity level. Although the overall frequency of contacts is not a significant predictor, the separate measures indicating either frequency of sending or receiving message are statistically significant (frequency of sending message at .01 level, frequency of receiving message at .05 level). Regression coefficients of both frequency measures show interesting results in that the directions of both measures are different from each other. In other words, the frequency of sending message is positively related to the level of perceived interactivity (.75), while the frequency of receiving message is negatively related to the level of perceived interactivity (-.53). Based on the results, it is possible to infer that people sending messages more frequently than receiving are likely to have a higher level of perceived interactivity toward the Internet than those who receive messages more frequently than they send messages. Hypotheses 2a and 2b are supported by the results, while Hypothesis 2 is not supported. With this multiple regression equation, 18% of variance is explained.

Table 2. Multiple Regression: Predicting Network Density

	Network Density
Predictor Variables	B	Beta	t
Opinion Leadership	.01	.34	2.81**
Need for Cognition	.00	.06	.53
Time spent in Web usage	.00	.01	.05
Age	-.05	-.10	1.04
Constant	.00

R = .39, R-squared = 15%, F = 4.16**, * p £ .05, **p £ .01

Hypothesis 3a states that opinion leadership as a personal factor of a focal individual significantly affects the network properties, such as density. Table 2 illustrates the multiple regression results showing the relationships between personal factors and social network density, controlling for respondents’ age and the amount of time spent in Web usage. As shown in Table 2, opinion leadership is the only predictor that is statistically significant at .01 level. Although opinion leadership has a statistically significant relationship with network density, the effect of opinion leadership on network density is too small to have any practical meaning. Hypothesis 4a also states that need for cognition as a personal factor affects network density. Based on the results shown in Table 2, Hypothesis 3a and 4a are not supported.

Table 3. Multiple Regression: Predicting Contact Frequency

	Contact Frequency
	B	Beta	t
Opinion Leadership	.42	.12	.93
Need for Cognition	.53	.19	1.54
Time spent in Web usage	7.00	.15	1.54
Age	-.69	-.02	.16
Constant	7.28

R = .35, R-squared = 12%, F =3.12**, * p

SOCIAL DIMENSIONS OF INTERACTIVE ADVERTISING