Theoretical background and model development
The overarching theoretical framework for this study will be the Unified Theory of Acceptance and Use of Technology (UTAUT, Venkatesh et al. 2003). The UTAUT was designed to explain the Behavioral Intentions to use and the Use Behavior of information systems. The theory holds that four key constructs are direct determinants of usage intention and behaviour:
- Performance Expectancy: The degree to which an individual believes that using the system will help him or her to attain gains in job performance.
- Effort Expectancy: The degree of effort an individual associates with the use of the system.
- Social Influence: The degree to which an individual perceives that important others believe he or she should use the information system.
- Facilitating Conditions: The degree to which an individual believes that an organizational and technical infrastructure exists to support use of the system.
The effect of these four key constructs on usage intention and behaviour is moderated by Gender, Age, Experience (with the system), and Voluntariness of Use, i.e. these moderating variables specify when the effects of the key constructs on the dependent variables will be weaker or stronger (Baron and Kenny 1984).
The UTAUT was developed through a review and empirical consolidation of eight models which had previously been employed to explain information systems usage. It is most closely related to the Theory of Reasoned Action (Fishbein and Ajzen 1975) and its extension, the Theory of Planned Behaviour (Ajzen 1991). These are two of the leading theories of action in the social sciences, which form the theoretical basis of over 800 studies published in the PsycINFO and Medline databases (Francis et al. 2004).
In a longitudinal study which applied the UTAUT to research software adoption in the financial services industry and customer service management software adoption in the retail electronics industry, the R² of Behavioral Intentions and Use Behaviour reached .77 and .53, respectively (Venkatesh et al. 2003). That is, the model was able to explain 77% and 55% of the statistical variance of these dependent variables, which indicates a very good model fit and external (predictive) validity of the theory.
By itself, the UTAUT will form the basis for understanding the reasons, facilitating conditions and inhibitors of adoption of Web 2.0 tools for researchers. However, in order to also understand the implications for scholarly communication, the model needs to be extended. Thus, the Use Behaviour of Web 2.0 tools for researchers is hypothesized to influence Sharing and Re-Use Behaviour of scientific data, Discovery Techniques of scientific data and literature, Publication Behaviour, and Research Findings Communication Behaviour.
Model operationalization
In order to operationalize the constructs and adapt the theory’s measurement model to the context of the present study, we will perform triangulation, i.e. a qualitative enquiry which incorporates three different viewpoints and methods.
First, we will draw on the combined expertise of the research team. The team members, who are active researchers in the natural, social and library sciences, as well as practitioners and promoters of Open Science, digital curators and architects of Web 2.0 tools for researchers, will have in-depth exploratory discussions on the meaning of each of the constructs in the model to achieve a common understanding of them. Second, we will review existing literature, blog posts, essays and online discussion on Web 2.0 tools for researchers, with specific attention paid to reasons for and inhibitors to adoption, and implications for scholarly communication behaviour. Third, we will conduct a limited number of semi-structured, qualitative interviews with members of our target group. Interviewees will be selected to represent a heterogeneous spread across academic disciplines, academic vs. corporate researchers, and other demographic criteria.
The insights generated by this triangulation will be used as a pre-test of the internal validity of our theoretical model, to find ways to quantify the Scholarly Communication measures, and finally to formulate the wording of the scale items of the measurement model. For example, as a result of this triangulation, Performance Expectancy could be a multi-item construct measured as a perceived gain (or decrease) in scholarly merit, and as a perceived improvement in access to high-quality information; Facilitating Conditions could be a multi-item construct measured as the perceived support and training by the researcher’s library, and the financial cost of use of these tools, and the IT budget available to the researcher.
Empirical study and quantitative analysis
Following the model operationalization, we will conduct a large-scale empirical study. The rationale for doing this is that it will enable us to perform statistical analysis of the collected data. Using a structural equation modeling approach (described further below) will allow us to go beyond a qualitative, anecdotal, or phenomenal understanding. It will enable us to quantify the strength of the effect of each promoter, inhibitor and facilitator of the adoption of Web 2.0 tools, as well as the relative importance of the factors vis-à-vis each other. This can lead to prescriptions as to which inhibitors to tackle first, and which promoters and facilitators to focus on. It will also provide empirical evidence as to which degree the use of Web 2.0 tools influences scholarly communication.
The study design will be online-survey based and cross-sectional. In general, a longitudinal study design would be preferable to a cross-sectional one for studies like these, as it would measure the intra-individual changes in the independent variables and their effects on the dependent variables over time, instead of differences between respondents and the effects of these differences. A longitudinal design would thus be better at ruling out the effects of “hidden moderator variables” that could account for the differences between respondents in a cross-sectional design. However, based on our experience, we feel that the time frame for the empirical study of (at most) six months is realistically not long enough to capture not only significant changes in the respondents’ attitude towards and use of Web 2.0 tools, but also the effects of these changes on long-term variables like publication and communication behavior. Thus, the aforementioned qualitative triangulation will be even more important to detect potential hidden variables before conducting the empirical survey.
The sampling technique will be a stratified random sample, with participants randomly filtered to achieve representative quotas for different academic disciplines, age groups, gender, and academic vs. corporate researchers. Using appropriate incentives will be instrumental in securing a sufficient response rate. Moreover, the wording for the survey invitations will have to be general enough (and not refer specifically to “Web 2.0 tools for researchers”) so as not to create a self-selection bias, in which researchers who are a priori more interested in these tools than the average researcher are strongly overrepresented in the sample. To control for such a self-selection bias, we will perform a second “adoption sampling” study: We will select a smaller, random representative sample of researchers who have not participated in the survey, and search for these people on the Web 2.0 resources included in the study to obtain a real-world measure of adoption. Due to people not using their real name on these services, or electing to keep their profile hidden, this “adoption sampling” approach will systematically underestimate the “true” adoption, whereas the “self-reported usage” will systematically overestimate the “true” adoption. Together, these two data points will help to establish a minimum vs. maximum extent of use. They will also provide a rich data base for descriptive statistics on the demographics, disciplines, and subject areas of the researchers who use these Web 2.0 tools.
The relationship between the constructs in our extended UTAUT model will then be analysed using structural equation modeling techniques. Specifically, we will be using Partial Least Squares (PLS) analysis (Fornell and Cha 1994; Chin 1998). PLS allows a simultaneous testing of hypotheses, taking indirect and moderating model effects into account. When compared to covariance-based structural equation modeling approaches such as LISREL and AMOS, PLS enables single-item measurement as well as multi-item measurement and the modeling of constructs as either reflective or formative. As a distribution-free method, PLS has fewer constraints and statistical specifications than covariance-based techniques.
The model will then also be calculated for sub-samples, divided by different classes of Web 2.0 tools. This means that we can pinpoint the promoters, inhibitors, and facilitating conditions for specific Web 2.0 tools (e.g. only social networks, or only folksonomy/tagging/literature management tools), and the effect of only these tools on scholarly communication behaviour vis-à-vis other tools.
References
Ajzen, Icek (1991), "The Theory of Planned Behavior," Organizational Behavior and Human Decision Processes, 50, 179-211.
Baron, Reuben M. and David A. Kenny (1986), “The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations,” Journal of Personality and Social Psychology, 51 (December), 1173-1182.
Chin, Wynne W. (1998), "The Partial Least Squares Approach for Structural Equation Modeling," in Modern Methods for Business Research, George A. Marcoulides ed. New York: Lawrence Erlbaum Associates, 295-336.
Fishbein, Martin and Icek Ajzen (1975), Belief, Attitude, Intention and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley.
Fornell, Claes and Jaesung Cha (1994), “Partial Least Squares,” in Advanced Methods of Marketing Research, Richard P. Bagozzi, ed. Cambridge: Blackwell, 52–78.
Francis, Jillian J. Martin P. Eccles, Marie Johnston, Anne Walker, Jeremy Grimshaw, Robbie Foy, Eileef F.S. Kaner, Liz Smith, and Debbie Bonetti (2004), Constructing Questionnaires Based on the Theory of Planned Behaviour: A Manual for Health Services Researchers. University of Newcastle.
Venkatesh, Visvanath, Michael G. Morris, Gordon B. Davis and Fred D. Davis (2003), “User Acceptance of Information Technology: Toward a Unified View”, Management Information Systems Quarterly, 27 (September), 425-478.
