Research Findings

To empirically prove this model, we surveyed 253 Internet users. The characteristics of the respondents are outlined in Table 1. For interested readers, the research methodology used in this study is detailed in Appendix A, "How Was This Study Conducted," and Appendix B displays the measurement used. The results largely supported our propositions. A schematic representation of the Model of Consumer Acceptance of Virtual Stores that includes the standardized path coefficients is displayed in Figure 2.

Table 1: Characteristics of Respondents.

	Frequency	Percent (%)	Cumulative (%)
Gender
Male	121	47.8	47.8
Female	132	52.2	100
Age
Less than 20	32	12.6	12.6
20 – 29	126	49.8	62.5
30 – 39	75	29.6	92.1
40 – 49	13	5.1	97.2
More than 50	7	2.8	100
Individual Annual Income
Less than $15,000	66	26.1	26.1
$15,000 to 29,999	51	20.2	46.2
$30,000 to 44,999	76	30.0	76.3
$45,000 to 59,999	35	13.8	90.1
More than $60,000	25	9.9	100
Shopping Online
Never shopped	52	20.6	20.6
Have shopped	201	79.4	100
Computer Skills
Poor	18	7.1	7.1
Fair	61	24.1	31.2
Good	99	39.1	70.4
Very good	75	29.6	100

Figure 2: The Model of Consumer Acceptance of Virtual Stores. (Coefficients associated with structural paths represent standardized estimates; parenthesized values represent standard errors.)

The results suggested that one of the proposed CSFs, information richness, does not positively affect PU. The development of this hypothesis and associated scale was based on IRT, which is a predictive theory to understand users' communication media choices (Daft & Lengel, 1986). However, the criteria for rich communication proposed by IRT were questioned by a number of empirical studies (e.g., Lee, 1994; Markus, 1994). Computer-mediated communications, such as electronic mail, prove to be popular despite the fact that, according to IRT, they provide lean communication instead of rich communication (Lee, 1994). One of the weaknesses of IRT is that it measures information richness solely by the properties of media and ignores the interactions between people and the social context in which the communication occurs. Therefore, the authors suspect that the insignificant linkage between IR and PU is due to the measurement weakness.