While questionnaire surveys allow researchers to collect exten-
sive information on various aspects of e-business, there are no strict rules on how they should be designed as well as what survey questions should be asked. As there are 34 questions in the survey seeking companies’ views on critical factors in making the decision to adopt e-business, the inclusion of all the questions’ feedback in regression analysis would render it ineffective due to the multicollinearity problem caused by the correlations between variables as well as the loss of the degrees of freedom. To overcome the above issues, the data analysis is carried out in two phases (Oliveira and Martins, 2010). In the first phase, factor analysis is conducted to identify the key factors representing the three groups of variables explaining the perceived benefits of e-business, obstacles to e-business adoption and business expectation. This helps to effectively extract useful information from the survey data and transform those data into a small number of variables/ components. Both the principal component method and maximum likelihood method using the varimax criterion are applied. The factor scores are estimated using the regression method. In the second phase, logistics regression is conducted using the factor scores obtained from the first phase as the inputs to further test the effect of different variables on the decision to adopt e-business. The logistics regression analysis in the second phase is much more manageable due to only a small number of variables involved2.
Note that the data for the variables specified by Eqs. (15) and (16) mainly concerning firms’ cost, revenue, expectations, etc. are not available from secondary sources. Therefore, the current study relies on an industry survey to collect data. There are many advantages of using the survey instrument. First, it allows for the collection of comprehensive information about the firms. Second, the survey questions can be designed to meet the study’s need for information, which is often not the case when using secondary data. Third, factors such as ‘business expectation’ and ‘policy’ suggested by the theoretical model cannot accurately be captured by a single numerical index or value. It would be better for the survey instrument to observe them from different aspects.
Despite its advantages, the use of survey for data collection is also subject to some issues. Especially, most companies would not be willing to provide confidential information about their business. Moreover, because survey participation is voluntary, participants are often not willing to spend considerable amount of time to answer questions seeking financial information such as sale revenue, cost and tax. To get around these issues, studies have to rely on indirect questions while conforming to the research ethics requirements. As a result, the variables identified by the theoretical model only serve as guidance for the design of the survey questions; the survey cannot ask for exactly what the theoretical model requires.
The questionnaire design was based on the literature as well as several discussions with industry experts and company managers during the Australian Logistics Council 2008 annual forum. There are five sections in the questionnaire asking questions on different aspects of companies and e-business adoption. The first section of the survey questionnaire seeks information on the firm’s profile including (i) firm size, (ii) year of establishment, (iii) existing technology, and (v) nature of business.