The problem of complexity has received relatively little attention until recently, but the studies that did address this issue (e.g., Casali & Gaylin, 1988; Schutz, 1961; Spence & Lewandowsky, 1991) show that complexity affects performance on various tasks. The following research questions are proposed based on the literature review detailed in the last section, addressing the impact of complexity on online communication effectiveness:
To propose a reliable and affordable instrument for measuring objective complexity of Web sites.
To propose a reliable and direct instrument for measuring perceived complexity of Web sites.
To propose the relationship between objective complexity and perceived complexity of Web sites.
To propose the impacts of objective complexity and perceived complexity on Web sites’ communication effectiveness.
To dynamically chart the optimal complexity based on objective and perceived complexity.
Figure 7-1 tentatively illustrates the proposed framework of this research. The objective complexity measure is proposed based on Berlyne’s (1971, 1974) stimulus complexity theory and Wood’s (1986) task complexity theory. Objective measurement alone, however, is usually handicapped by inaccuracy. In the situation where “Everybody wants to be able to measure (Web site) complexity, but nobody quite knows how it should be done” (Maddox, 1990, p. 705), what is lacking, however, is a simple, generalizable measure of perceived complexity that can be used in assessing the complexity of a Web site from the consumers’ (as oppose to designers’) perspective. Introducing measure of the Web surfers’ own perceived complexity, independent of the objective components actually imbedded, has been suggested as a way of keeping the intended framework from being tautological (Abelson & Levi, 1985).
Figure 7-1: Proposed framework
Operationalization of carefully built-up constructs is an important part of research in consumer behavior. It has also traditionally been one of the more intractable problem faced by researchers. This step is particularly important to the future exploration of the issue of complexity. In the following sections, feasible measures for capturing the concept of objective and perceived Web site complexity are proposed based on the structure of the framework proposed in the earlier section. In so doing, this study represents a small step forward in this direction by conceptualizing feasible scales to measure objective and perceived Web site complexity in marketing situations.
In this framework, objective complexity is defined as a Web site with multiple dimensions (Boulding, 1970) in the form of component, coordinative, and dynamic complexity (Wood, 1986). Component complexity refers to the number of distinct information cues, i.e., Web design elements, that must be processed in the performance of a task, i.e., Web surfing, while coordinative complexity describes the form, strength, and interdependencies of the relationships between the information cues. Dynamic complexity arises from changes in the relationships between information cues over time, particularly during task performance. Because objective complexity obscures the perception and understanding of information cues, it is believed to significantly degrade task performance (Banker et al., 1998). This psychological view, I believe, is particularly appropriate for studying objective level of complexity of Web sites. A Web site, as an entity composed of lots of design elements, has multiple dimensions parallel to the component, coordinative and dynamic complexity dimensions proposed by Wood (1986). Figure 7-2 outlines the sample measures of component complexity of Web sites.
Figure 7-2: Proposed objective complexity measure
These measures adopted the method used in Berlyne’s previous complexity research— objective complexity measure is proposed based on the amount of material measure in Berlyne’s stimulus complexity (1971; 1974) and on several other marketing-related complexity studies, such as advertising (e.g., Anderson & Jolson, 1980; Morrison & Dainoff, 1972) study. Specifically, the design elements such as the number of frames, the number of colors, the number of banner ads, the number of animated ads, the number of screenfuls, etc., are used to proximate the amount of material measure. Based on Wood’s (1986) complexity dimension, the scales, therefore, are given to count the amount of the occurrence of the elements (component complexity), the relationship between these occurrences (coordinative complexity) and impact of the navigation states on the relationships between these occurrences (dynamic complexity).
Component complexity of a Web site is a direct function of the number of distinct implementations that are executed in the Web site and the number of distinct information cues that must be processed in these design implementations. A formula which captures the component complexity of Web site could be proposed as:
where n = number of distinct implementations in each design element j, Wij = number of information cues to be processed in the performance of the ith implementation of the jth design element, p = number of design elements in the Web page, and OBC1 = component complexity.
While component complexity captured the objective occurrence of the designed elements, it was limited by the assumption that the design elements are all independent. The basic rationale of component complexity is that the larger amount of the occurrences (smaller amount of the blank) will introduce more component complexity. While this assumption serves the purpose of objectivity, it loses sight of the possible interaction between those elements. Hence, another dimension of objective complexity, coordinate complexity, should also be considered. Coordinative complexity refers to the nature of the relationship between design elements, which proximates the irregularity of arrangement, heterogeneity of elements, incongruity measures proposed in Berlyne’s (1971, 1974) stimulus complexity theory. A formula that captures the coordinative complexity of Web site is:
where n = number of distinct implementations in each design element, ri = number of precedence relations between the ith implementation and all other implementations in the Web site, and OBC2 = coordinative complexity.
In addition to the static complexity of the implementations of design elements and information cues needed to perform a Web surfing task, the navigation function of the Web sites requires individuals frequently to adapt to the changes as they are surfing through layers of Web sites. This third dimension of Web site objective complexity, which I call dynamic complexity adapted from Wood (1986), is due to changes in the states of the navigation, which has an effect on the relationships between design elements. This definition is consistent with the concept of incongruous juxtaposition in Berlyne’s (1971; 1974) stimulus complexity theory. A simple index of dynamic complexity would, therefore, be the sum across specific surfing frames for any or all of the indices for the two dimensions of (static) complexity. The usage of navigation aid such as share boarders, site map, smart agent assistance, is viewed as a way to decrease the cognitive efforts required from the users. Therefore, the employment of navigation aid serves to decrease the complexity of the Web site. In formulating the dynamic complexity, I subtract the navigation aid from the total dynamic complexity score (a pretest may be conducted to ask users to weigh the navigation aid based on how much easier each aid makes their surfing task). A formula for calculating these differences from the OBC1 and OBC2 indexes is given below:
where OBC1 = component complexity measured in standardized units, OBC2 = coordinative complexity measured in standardized units, f = the number of layers over which the Web site is measured, NAV = navigation aid, d = number of the navigation aids employed in the site, and OBC3 = dynamic complexity.
Hurt and Hibbard (1989) use the self-report measurement approach to measure the perceived characteristics of microcomputer innovations. The self-reporting procedure is easy and inexpensive to administer and normally has high reliability (Hurt, Joseph, & Cook, 1977). In this study, Rao’s (1985) measurement of the perceived complexity of the task are modified and adopted to measure the perceived complexity (Figure 7-3). Stress was laid on the fact that the response needed is the respondents’ own perception of the complexity of the Web site they visited.
Figure 7-3: Proposed perceived complexity measure
One way to capture the optimal complexity of the Web site is to dynamically chart the optimal level based on objective and perceived complexity. In addition, I propose the use of a perceived optimal level of complexity measure to gauge this construct as a validation. Following items are adopted and modified from Franz and Robey’s (1986) study (Figure 7-4).
Figure 7-4: Proposed optimal complexity measure
As alluded to in previous sections, the reason that I introduced measure of the Web surfers’ own perceived complexity, independent of the objective components, is in response to the need for using a combination of both the system-centered approach and the user-centered approach. As the perceived complexity measures consumers’ subjective evaluation of the Web site, the relationship between objective and subjective measures of complexity is presumed to be strongly, however, not perfectly correlated while both objective complexity and subjective complexity will have an invert-U relationship with optimal complexity, i.e., moderate level of perceived complexity depicts maximum optimal complexity and that moderate level of objective complexity depicts maximum optimal complexity. Hence, I propose the following propositions.
P1: Perceived complexity does not equal, but highly correlates to objective complexity.
P2: An invert-U relationship exists between objective complexity and optimal complexity.
P3: An invert-U relationship exists between perceived complexity and optimal complexity.
These proposed relationships have been charted in Figure 7-5.
Figure 7-5: Proposed relationships between objective, perceived, and optimal complexity
Further, communication effectiveness is conceptualized as consumers’ information search and subsequent liking of their Web surfing experiences. This identification is deemed appropriate as search has been identified in previous studies as consequences of consumers’ online loyalty (Srinivasan, Anderson, & Ponnavolu, 2002; Urbany, Kalapurakal, & Dickson, 1996). Several studies have shown that complexity of task environment positively impacts on the communication effectiveness. Rayne (1976) conducted two experiments with tasks of different levels of complexity and used two processes tracing techniques, explicit information search and verbal protocols. Staelin and Payne (1976) reported on three marketing studies and investigated the relationships between different information-seeking patterns and task characteristics. Rubin (1977) also investigated information seeking in different contexts and found greater information search in a complex context. Schwartz (1977) found that task environment complexity as well as cognitive complexity were positively correlated with the amount of information acquired. Radzicki (1975) found the same relation between amounts of information acquired and perceived risk, as well as task complexity.
Some researchers report that, with increases in task complexity or difficulty, information search increases up to a maximum level and then decreases as the consumer proves unable to handle the information load (Sieber & Lanzetta, 1964; Streufert, Suedfield, & Driver, 1965). Berlyne (1971, 1974, 1975) also discovers that there is an inverted-U relationship between ratings of liking and complexity. Although previous research points to the inverted-U relationship between complexity and communication effectiveness, none of the investigation addresses the different types of complexity and their relationship with communication effectiveness. In this study, based on the previous proposition, both objective complexity and perceived complexity are expected to have inverted-U relationships with communication effectiveness, while optimal complexity is expected to have an almost linear relationship with the communication effectiveness. Hence:
P4: An inverted-U relationship exists between objective complexity and communication effectiveness.
P5: An inverted-U relationship exists between perceived complexity and communication effectiveness.
P6: A linear relationship exists between optimal complexity and communication effectiveness.
These proposed relationships have been charted in Figure 7-6.
Figure 7-6: Proposed relationships between objective, perceived, optimal complexity and communication effectiveness
Finally, the greater the cognitive complexity of an individual, the more abstract the conceptual schemes used on processing information, and the greater the amount of information search exhibited (Rao, 1985). Fertig (1969) and Miller (1969) investigated the relationship between cognitive complexity and amount of information acquired and found a strong positive correlation. So did Larreche (1975) in an applied marketing situation, Karlins and Lamm (1967) in a social work scenario, and Schneider and Giambra (1971) in two laboratory experiments. Schroder, Driver and Streufert (1967) investigated the relationship in a large number of experimental settings and found that, in every case, cognitively complex individuals were significantly more active in searching for information than less complex ones. Hence:
P7: Web users’ cognitive complexity moderates the relationship between perceived complexity/objective complexity and optimal complexity; it also moderates the relationship between perceived complexity/objective complexity and communication effectiveness.
For example, consumers with higher cognitive complexity is expected to demand a higher objective and perceived complexity levels to reach the maximum optimal complexity level in comparison to consumers with lower cognitive complexity. Figure 7-7 depicts the proposed moderating effects of cognitive complexity on the relationship between objective complexity and optimal complexity.
Figure 7-7: Proposed moderating effects of cognitive complexity on the relationship between objective complexity and optimal complexity