Two Competing Perspectives on Automatic Use: A Theoretical and Empirical Comparison
Sung S. Kim (
[email protected]) Assistant Professor, School of Business, University of Wisconsin-Madison 975 University Avenue, Madison, WI 53706, USA Phone 608-262-3481, Fax 608-263-3142
Naresh K. Malhotra (
[email protected]) Regents’ Professor, College of Management, Georgia Tech 800 West Peachtree Street NW, Atlanta, GA 30332, USA Phone 404-894-4358, Fax 404-894-6030
Sridhar Narasimhan (
[email protected]) Professor, College of Management, Georgia Tech 800 West Peachtree Street NW, Atlanta, GA 30332, USA Phone 404-894-4378, Fax 404-894-6030
1
Appendix A. Research Constructs and Measures Utilitarian Value Modifications of the scale were based on Mathwick et al. (2001). The items we used were: “All things considered, this website would provide very good value,” “Visiting this website would be worth my time and efforts,” and “It would be of value for me to visit this website.” We used seven-point scales anchored by “strongly disagree” (1) and “strongly agree” (7). Hedonic Value This construct was measured on seven-point scales anchored with “strongly disagree” (1) and “strongly agree” (7). Modifications to three items were based on Davis et al. (1992). The specific items used were: “Using this website is fun,” “Using this website is a joy to me,” “Using this website is enjoyable,” and “Using this website is very entertaining.” Social Value The three-item measure was modified from Perse (1990). The questions asked were: “Using this website makes people hold me in high regard,” “Using this website enhances the image which others would have of me,” and “Using this website helps me to show others who I am.” Each item was measured on a seven-point scale ranging from “strongly disagree” (1) to “strongly agree” (7). Usage Intention Usage intention was measured on seven-point scales. The items were developed based on Harrison et al. (1997). The questions asked were: “Would you intend to visit this website in the next month?” and “How certain are your plans to visit this website within the next month?” The first item was anchored with “unlikely” (1) and “likely” (7), whereas the second item was anchored with “uncertain” (1) and “certain” (7). Past Use Individuals’ utilization of an online information service in the previous period was measured by two selfreported items. One of the items was related to behavioral frequency: “On average, how frequently have you visited this website over the past month?” Six categories were given for this item (1 = less than once a month; 2 = once a month; 3 = a few times a month; 4 = a few times a week; 5 = about once a day; 6 = several times a day). The other item asked about the duration of IT use: “On average, how much time do you spend a week visiting this website over the past month?” This item was measured on seven categories (1 = less than 10 mins; 2 = 10 – less than 20 mins; 3 = 20 – less than 30 mins; 4 = 30 mins – less than 1 hr; 5 = 1 – less than 2 hrs; 6 = 2 – less than 3 hrs; 7 = 3 hrs or more). IT Use The scale of past use was used repeatedly to measure IT use in the next month. Because the items used were identical for past use and IT use, the measurement model was specified in a way that the measurement errors of the same indicators between the two factors were correlated. More specifically, the measurement error of the past frequency item was allowed to correlate to the measurement error of the current frequency item. Similarly, the measurement error of the past duration item was allowed to correlate to the measurement error of the current duration item. Covariates: 1. Age: (1 = younger than 18; 2 = 18-24; 3 = 25-34; 4 = 35-44; 5 = 45-54; 5 = 55-64; 6 = 65 years or older) 2. Gender: (1 = male; 2 = female) 2
3. Internet experience: (1 = less than 6 months; 2 = 6 to less than 12 months; 3 = 1 to 3 years; 4 = 4 to 6 years; 5 = 7 years or more) 4. Target system experience: (1 = less than 3 months; 2 = 3 to less than 6 months; 3 = 6 to less than 12 months; 4 = 1 to less than 2 years; 5 = 2 years or more)
3
Appendix B. Demographic Profiles of the Respondents Variables Age
Gender Internet experience
Target system experience
Sample size
Categories Younger than 18 18 - 24 years 25 - 34 years 35 - 44 years 45 - 54 years 55 - 64 years 65 years or older Male Female Less than 6 months 6 to less than 12 months 1 to 3 years 4 to 6 years 7 years or more Less than 3 months 3 – Less than 6 months 6 – Less than 12 months 1 – Less than 2 years 2 years or more
Sample A 0.9% 10.8% 20.8% 28.2% 24.6% 11.4% 3.3% 68.8% 31.2% 2.2% 6.0% 43.0% 38.4% 10.4% 10.3% 19.1% 32.2% 31.8% 6.6% 100.0% 886
Sample B 0.3% 5.1 13.9% 23.5% 33.8% 16.8% 6.6% 56.5% 43.5% 4.6% 12.4% 48.0% 27.2% 7.9% 23.1% 24.0% 21.5% 22.3% 9.1% 100.0% 1189
4
Appendix C. Past Use versus Habit/Automaticity One of the major assumptions in the proposed model (Figure 1) was that past use would be a good indicator of habit/automaticity. As a way to test this assumption, we examined a correspondence between past use and habit/automaticity in a separate study. In this separate study, we used the past-use scale described in Appendix A. Meanwhile, we developed a habit/automaticity scale. Initially, a pool of items was created that could reflect the extent to which the IT use in question is automatic. Special care was taken to make sure that the scale fit well with the descriptions for habit/automaticity in the literature, such as typical manners of acting in a stable context (Ouellette and Wood 1998), behaviors performed on a daily basis (Aarts et al. 1998), and actions requiring less conscious mental effort (Ronis et al. 1989). As a result, a pool of 12 items was created. Subsequently, two judges with expertise in scale development were asked to examine the potential items for their appropriateness. The following four items were selected as the most desirable measures: “To me, this website is a part of my daily life,” “I am dependent on this website in a variety of circumstances,” “I visit this website as a matter of routine,” and “To me, visiting this website is routine without a deliberate plan beforehand.” They were assessed by seven-point scales anchored with “strongly disagree” (1) and “strongly agree” (7). The target application chosen in this separate study was a personalized Web-based portal specifically developed for students at a large university in the Midwest. E-mail messages inviting recipients to fill out a Web-based survey were sent to 1,000 randomly chosen students. A total of 243 usable questionnaires were collected with 19 undeliverable messages, representing a response rate of 24.7%. Based on the collected data, we conducted confirmatory factor analysis via LISREL 8.7. We found that the model fit the data fairly well: χ2 (8) = 23.07, RMSEA = 0.088, CFI = 0.99, NNFI = 0.98, and SRMR = 0.016. Furthermore, the scales were shown to be reliable: The composite reliability and average variance extracted values for past use were 0.77 and 0.62, respectively, and the values for habit/automaticity were 0.96 and 0.86, respectively. However, the results indicated that the correlation between the two constructs was high (r = 0.91, p
