Technology acceptance: a meta-analysis of the TAM: Part 2 Shumaila Y. Yousafzai, Gordon R. Foxall and John G. Pallister Cardiff Business School, Cardiff University, Cardiff, UK Abstract Purpose – This paper is the second of two concerned with a meta-analysis of the technology acceptance model (TAM). This part aims to present a rigorous and quantitative meta-analytic review of 569 findings from 95 TAM studies as a basis for identifying gaps and providing guidelines for implementation management and conduct of future research. The paper also seeks to investigate the potential impact of methodological characteristics on the meta-analytic findings. Design/methodology/approach – The approach consists of meta-analysis following Hedges and Olkin’s procedures, moderator-analysis using homogeneity Q-values, analogue to ANOVA and weighted regression method. Findings – The dominant focus in empirical investigations of the TAM has been on modelling intention for its effect on self-reported usage behaviour, while the attitudinal construct has been neglected. This raises three questions: whether the exclusion of attitude from the TAM is beneficial for understanding of technology usage behaviour in mandatory settings; whether the revised TAM holds equally for mandatory and voluntary settings; and whether the emphasis on measuring intentions and self-report use rather than actual usage is warranted. An additional question answered in the meta-analysis is about the relative importance of PU and PEOU. Originality/value – The paper provides a rigorous meta-analysis to progress towards a unified view of the TAM. Keywords Technology-led strategy, Research, User studies, Modelling Paper type General review Meta-analysis is a technique that allows quantitative accumulation and analysis of descriptive statistics across studies without acquiring access to the original data. Meta-analysis has been criticized as “mixing oranges and apples” (Hunt, 1997, p. 61). Such a criticism assumes that meta-analysis aggregates findings of different phenomena. However, Cooper (1989) notes that a convergence or “triangulation” of findings from methodologically varying studies lends credence to the validity of an effect. When a relationship remains constant, although tested under a variety of circumstances, it is clearly robust. Coined and first applied by Glass (1976), meta-analysis received its impetus from Rosenthal (1984), Hunter et al. (1982) and Hedges and Olkin (1985). Although the techniques used may vary, all meta-analyses aim to derive a quantitative measure, the effect size (ES), of the relationship under study. Our meta-analyses were conducted using Hedges and Olkin’s (1985) procedures. The framework guiding our discussion and empirical investigation focuses on two types of empirically tested relationship. First are the relationships proposed in Davis et al.’s (1989) original model; second, the relationships that have been tested by later research. Subsequent discussion centres on examining the sign of the association that The current issue and full text archive of this journal is available at www.emeraldinsight.com/1746-5664.htm Technology acceptance 281 Received May 2007 Revised June 2007 Accepted July 2007 Journal of Modelling in Management Vol. 2 No. 3, 2007 pp. 281-304 q Emerald Group Publishing Limited 1746-5664 DOI 10.1108/17465660710834462