December 14, 2002
1
A Unified Model of
IT Use Choices: Contributions from TAM, TTF, and CSE
Diane M. Strong*
Worcester Polytechnic
Institute
Invited Presentation
First Annual
Workshop on HCI Research in MIS
Barcelona, Spain
2002
*This is joint work with Mark
T. Dishaw, University of Wisconsin Oshkosh
December 14, 2002
2
General Research Objective
- Understand the software
utilization choices of end users, by using and extending existing models
- Task-technology
Fit (TTF) models
- Technology Acceptance
Model (TAM)
- Individual Abilities
Constructs, e.g., Experience, Computer Self-efficacy
- Conduct a series
of studies testing the models and combinations of them
December 14, 2002
3
Task-Technology Fit
Models
December 14, 2002
4
1. TTF Model Study
Operationalize the TTF model
in the software maintenance context
- Task Model
- Vessey's debugging model (planning, knowledge building, diagnosis,
modification activities) plus coordination
- Technology Model
- Henderson & Cooprider Functional Case Technology Model (Production
and Coordination functionality)
December 14, 2002
5
Dimensions of Fit
- Fit along two dimensions
- Production Fit:
how well the tool��s production functions support software maintenance
activities
- Coordination Fit:
how well the tool��s coordination functions support maintenance coordination
activities
- Compute Fit using
an interaction approach (Venkatramen,
1989)
(Dishaw & Strong, 1998)
December 14, 2002
6
2. Add Experience to
TTF
Operationalize Individual Abilities
as:
- experience with
the task
- experience with
the technology
Tool experience and its interaction
with tool characteristics is significant
Task experience not significant
Adjusted R2
of 0.63
(Dishaw & Strong, Forthcoming)
December 14, 2002
7
3. Combined TAM and
TTF
- TAM: beliefs about
the technology, i.e., perceived usefulness and perceived ease of use
- TTF: matching of
the technology to the needs of the task to deliver benefits
- TAM + TTF: addresses
both technology beliefs and rationally computed fit to task
- Tool experience
as an individual ability
- Path model, rather
than regression
- Fit as latent variable,
rather than computed as interaction
December 14, 2002
8
TTF-TAM Combined Model
December 14, 2002
9
Combined TAM / TTF
Results
Better results than either
TAM or TTF alone
Utilization variance explained:
- 36% with TAM
- 41% with TTF
- 51% with TAM/TTF
(Dishaw and Strong, 1999)
December 14, 2002
10
4. Add Computer Self-efficacy
(Work-in-progress)
- CSE may be a better
predictor of individual ability for new tools than is tool experience
- Generalize TTF assessment
beyond software maintenance tasks and tools
- Develop an instrument
for assessing problem-solving tasks, and the support of such tasks with
software
- Test previous TTF
and TAM/TTF models with a new dataset
December 14, 2002
11
Computer Self-Efficacy
- Derived from the
Social Cognition literature, and is based on Bandura��s work on self-efficacy
- A specialized definition
of Self-efficacy, i.e., a person��s belief in their ability to accomplish
a specific task
- A judgment of one��s
ability to use a computer
December 14, 2002
12
Adding CSE to TTF/TAM
December 14, 2002
13
Model Operationalization
- Software maintenance
TTF is generalized by changing the questionnaire items since
- Task model is well
grounded in the problem solving and cognitive science literature
- Technology model
is grounded in the literature on information technology support functionality
- Add Compeau &
Higgins (1995) 10-item, single factor measure of CSE
December 14, 2002
14
Item and Scale Testing
- Item Testing using
a panel of faculty, advanced students, and professionals
- Pilot Study using
a small number students and professionals in the university
December 14, 2002
15
Data Collection
- Use revised instrument
- Subjects are students
in several classes after the completion of an ordinary assignment
- Currently, have
136 data points from:
- Operations Management
simulation class doing modeling
- Programming class
doing 3 GL program maintenance
- Programming class
doing OO program maintenance
- Business analysis
class doing statistical modeling
December 14, 2002
16
Data Analysis
Using Amos 4.0, test the
models
- TTF
- TTF plus CSE
- Combined TAM/TTF
- Combined TAM/TTF
plus CSE
December 14, 2002
17
General TTF Model
- Chi Sq. 26.77,
d.f. 17, p=0.061
- AGFI = 0.89, GFI
= 0.95
December 14, 2002
18
General TTF Model with
CSE
- Chi Sq. 27.24,
d.f. 22, p=0.202
- AGFI = 0.91, GFI
= 0.96
December 14, 2002
19
Lessons for a Unified
Model:
Importance of Task
- Traditional HCI
focuses on Usability, with little or no Task emphasis
- TAM adds Usefulness,
which implicitly includes Task
- TTF has explicit
Task focus, which adds to the explanatory power
December 14, 2002
20
Lessons for a Unified
Model:
The Fit Construct
- Beyond production
and coordination Fit to additional dimensions of Fit
- Beyond a point
estimate of Fit to a process of Fitting over time (as in implementation)
- Beyond individual
level models (TTF, TAM) to organizational level models, e.g., for Enterprise
systems
December 14, 2002
21
Lessons for a Unified
Model:
Experience and CSE
- Measure Experience
and Self-efficacy for both Task and Technology
- Self-efficacy theory:
As Experience increases, Experience dominates abilities as measured
by Self-efficacy
- Need to better understand
relationship between Experience and Self-efficacy
December 14, 2002
22
References to the Studies
Study 1: Dishaw, M.
T. and D. M. Strong, "Supporting Software Maintenance with Software
Engineering Tools: A Computed Task-Technology Fit Analysis",
Journal of Systems and Software, Vol. 44, No. 2, December 1998,
pp. 107-120.
Study 2: Dishaw, M.
T. and D. M. Strong, "The Effect of Task and Tool Experience on
Maintenance CASE Tool Usage", Information Resources Management
Journal, Forthcoming.
Study 3: Dishaw,
M. T. and D. M. Strong, "Extending the Technology Acceptance Model
with Task-Technology Fit Constructs", Information & Management,
Vol. 36, No. 1, July 1999, pp. 9-21.
Study 4 (in-progress): Dishaw,
M. T., D. M. Strong, and D. B. Bandy, ��Extending the Task-Technology
Fit Model with Self-Efficacy Constructs��, Proceedings of the Americas
Conference on Information Systems, August 9-11, 2002, Dallas, TX,
pp. 1021-1027.