HOT Topics! 
Publication of the Human Oriented Technology Lab
Carleton University

To Front Page : To HOT Lab

Interview with Dr Andrew Dillon on the Technology Acceptance Model and Usability

(c) 2005 by Maria Rasouli

Andrew Dillon is Dean and Professor of the School of Information at The University of Texas at Austin, where he also holds courtesy appointments in Psychology, and Management Information Systems. He received his MA in Applied Psychology from University College Cork, Ireland in 1987 and his PhD in Psychology from Loughborough University of Technology in the UK in 1991. He was a founding faculty member of the School of Informatics at Indiana University where he led the development of the Masters degree in Human-Computer Interaction before moving to Texas in 2002. He is an author of more than 75 papers and books, has served on editorial boards of several leading journals, including currently, Journal of the American Society for Information Science and Technology, Interacting with Computers, and the online Journal of Digital Information.

[I: ] Could you briefly describe the Technology Acceptance Model (TAM) and what its purpose is?

[A. Dillon: ] The Technology Acceptance Model was developed within the Management Information Systems community by Fred Davis and colleagues as a means of predicting adoption of IT. It is based around a simple set of variables that are viewed as determining a person's willingness to use a technology, which a researcher can establish by having people complete a short questionnaire. Multiple studies over the last 15 years or so have clearly supported the argument that one's intention to use is positively correlated with subsequent use, at least for most business situations.

[Editor: ] The TAM model is described pictorially and in more detail,  here.

[I: ] What empirical evidence is there to support the TAM model?

[A. Dillon: ] As I mentioned there have been multiple studies over the years and by now I would not be surprised if more than 100 tests of its utility have been published in literatures across the information field. I think HCI researchers were a little late in picking up on the model, largely because the focus on usability tended to blind people to the real question of actual use in discretionary environments.

[I: ] How do we measure "perceived ease of use" and "perceived usefulness" in the TAM? Are there alternative ways to measure these variables?

[A. Dillon: ] In TAM there is a set of specific items that ask users if they would find a tool under evaluation easy to use or useful for their work. The responses are recorded on a Likert scale. Obviously there are many other ways one could word items or collect such data. but the TAM approach seems to work.

[I: ] What are the strength and weaknesses of the TAM?

[A. Dillon: ] Well it is certainly cheap and cheerful, allowing a researcher the opportunity to collect reliable data on an important issue very quickly. I think its weaknesses lie not within it but in its application. While the behavioral intention to use certainly correlates with use in the many situations so far tested, some of us feel that this correlation is somewhat inflated by the lack of true discretion on technology use that one finds in the organizations or contexts in which the tool is used. After all, if I do not use a tool that is provided, I still have to get the work done, so unless I really have a choice over tools then I suspect my TAM scores for a provided tool are reflecting my acceptance of the inevitability of the tool being in my life. This is not true for all evaluations but certainly a significant proportion may not have adequately ruled this out. I have a lot of fun teaching people about TAM. The general response from students at first is that surely human action cannot be so easily reduced to a couple of key variables like this. When they see the data this either convinces them or drives some of them to dig deep into the issues in the hope of uncovering the limitations of this approach, which I see as a good way of driving research.

[I: ] The TAM has mostly been used to predict the use of new technology. Do you think the TAM could be used to predict consumers' acceptance of an ecommerce website?

[A. Dillon: ] Yes, there is no reason to limit the application of TAM; and data has already been presented confirming its value in e-commerce.

[I: ] In one of your papers you differentiated "perceived ease of use" in the TAM and "ease of use" as it is used in the usability literature. Could you elaborate on this?

[A. Dillon: ] This is crucial. TAM is perceptual --my ranking of how easy I consider a tool will be to use for my purpose. Usability evaluations capture this in part through satisfaction measures and we have data that reveals classic user satisfaction scores correlate significantly with TAM's perceived ease of use factor. The trouble is, perceptions of usability or ease of use do not correlate well with behavioral measures e.g., my performance scores or task completion scores such as errors made and time taken. In fact, people tend to be quite poor judges of their own competence with a tool,which is why the classic usability engineering approach emphasizes effectiveness, efficiency and satisfaction as three distinct measures to be taken when evaluating a product. The correlation between these three is very low. Just taking a perceived ease of use measure will not tell you if a person can actually use the tool well. To determine the latter, one has to observe use.

[I: ] Do you think the TAM could be used to guide usability practices in ecommerce? If the answer is yes, how could it be applied?

[A. Dillon: ] I think it can be used to guide design but only if one can make a reliable mapping between results and design decisions, which is beyond TAM's purpose and now requires expert interpretation by a user experience specialist. This is nothing new in HCI. It's just the same as saying that a great usability test does not only tell you what is wrong, it suggests useful ways of making things right! Sadly, we do not do enough in usability education and research to close this circle.

[I: ] The ISO definition of usability has three parts: effectiveness, efficiency and satisfaction. Do you think that it would be useful to expand the TAM to include some of these other factors? (e.g. perceived satisfaction or perceived efficiency).

[A. Dillon: ] One could do this but it would cease to be TAM. I think there is always room for refinement but I keep thinking the answer is beyond measures of user perception. You need to combine these with behavioral measures to get the full picture of what users are doing and and what users need.

[I: ] ISO doesn't include "usefulness" as part of the definition of usability. Should it be? Why?

[A. Dillon: ] Can you give me the space to answer this :) It is an interesting issue, and there are a couple of takes. One is that usefulness is implied in the definition and will be captured at least indirectly in the combined measures. I am not convinced of this. But the argument might be made that usability, as part of the product development lifecycle, presumes that the tasks being tested have been selected on utility or usefulness grounds, otherwise the test would be invalid. Of course this presumes a holistic approach to design which one can never guarantee.

[I: ] Thank you very much for your time.

[A. Dillon: ] You are very welcome.

To Front Page : To HOT Lab

Subscribe/Unsubscribe

Subscribe to HOT Topics and let us bring you interesting and  well-researched articles and interviews every month. Subscribe/Unsubscribe

To Front Page : To HOT Lab