Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO.

12, DECEMBER 2002 1135

Explaining Software Developer Acceptance of


Methodologies: A Comparison of Five
Theoretical Models
Cynthia K. Riemenschneider, Bill C. Hardgrave, Member, IEEE Computer Society, and Fred D. Davis

Abstract—Many organizations attempt to deploy methodologies intended to improve software development processes. However,
resistance by individual software developers against using such methodologies often obstructs their successful deployment. To better
explain why individual developers accept or resist methodologies, five theoretical models of individual intentions to accept information
technology tools were examined. In a field study of 128 developers in a large organization that implemented a methodology, each
model explained significant variance in developers’ intentions to use the methodology. Similar to findings from the tool adoption
context, we found that, if a methodology is not regarded as useful by developers, its prospects for successful deployment may be
severely undermined. In contrast to the typical pattern of findings in a tool context, however, we found that methodology adoption
intentions are driven by: 1) the presence of an organizational mandate to use the methodology, 2) the compatibility of the methodology
with how developers perform their work, and 3) the opinions of developers’ coworkers and supervisors toward using the methodology.
Collectively, these results provide surprising new insights into why software developers accept or resist methodologies and suggest
what software engineering managers might do to overcome developer resistance.

Index Terms—Software development, methodologies, intention models, technology acceptance model, theory of planned behavior,
diffusion of innovations.

1 INTRODUCTION

F UELED by the information technology (IT) revolution, the


Digital Economy has arrived in the United States—six
major economic studies have recently concluded that the
not used more? Research to date has yet to answer this
question [66]. Organizations that attempt to deploy meth-
odologies frequently encounter substantial resistance from
production and use of IT contributed half or more to the individual developers [39]. Therefore, the specific research
acceleration of U.S. productivity growth in the second half question driving the present research is: What are the
of the 1990s. Real business investment in software increased determinants of individual software developers’ intentions
from $82 billion in 1995 to $149 billion in 1999 [20]. Similar to use methodologies within organizations attempting to
trends are anticipated globally. However, some evidence deploy them? Previous research on this question is limited
suggests that software development is not improving as it and exploratory, and there is minimal established theory
should: Quality is increasing at a slower rate than expected, (e.g., [35], [38], [52], [53]).
productivity has actually declined 13 percent in recent According to Pfleeger [50], the field of software en-
years, and up to 75 percent of all development efforts are gineering needs to better understand the role of people in
considered failures [25], [65]. This “software crisis” [23] the adoption process, drawing upon social science models
threatens the continued expansion of the Digital Economy. as appropriate to further this understanding (p. 121):
Improving software development processes has been and
There is little reported in the software engineering literature
remains a top concern of IT management [11], [18], [28], [46]. about how each aspect relates to successful software
Methodologies are not a panacea for all software technology diffusion. Much of our understanding of these
development problems. Nevertheless, research indicates issues is anecdotal; we can draw stronger conclusions about
their use is generally related to increases in productivity successful technology transfer if our understanding is
and quality as well as reductions in time and effort [27], broadened using careful studies of representative situations.
[30]. Despite the existence of methodologies for several Accordingly, we examine five existing models of
decades, only about half of all organizations actually follow individual acceptance of information technology tools in
a methodology [22], [24], [53], [56]. Why are methodologies the domain of individual acceptance of methodologies. These
models include the Technology Acceptance Model (TAM),
TAM2, Perceived Characteristics of Innovating (PCI),
. The authors are with the Sam M. Walton College of Business Information
Systems Department, University of Arkansas, Business Building, Room Theory of Planned Behavior (TPB), and the Model of
204, University of Arkansas, Fayetteville, AR 72701-1201. Personal Computer Utilization (MPCU). Because each of the
E-mail: {criemen, whardgra, fdavis}@walton.uark.edu. five models of tool acceptance was originally derived from
Manuscript received 27 Apr. 2001; revised 22 Oct. 2001; accepted 22 Feb. more general theories of motivated human behavior, we
2002. expect them to generalize to some extent beyond the
Recommended for acceptance by Luqi.
For information on obtaining reprints of this article, please send e-mail to: domain of tool use intentions to encompass methodology
tse@computer.org, and reference IEEECS Log Number 114059. use intentions. There is not sufficient justification for
0098-5589/02/$17.00 ß 2002 IEEE
Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
1136 IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO. 12, DECEMBER 2002

assuming, without empirical validation, that the models public domain via books and Web sites, such as Structured
from the tool acceptance domain would be applicable to the Systems Analysis and Design Methodology (representing
methodology acceptance context. This is a key motivation the structured development paradigm) and the Rational
for why the present study was conducted. Unified Process (representing the object-oriented para-
Although usage of a technology tool is clearly different digm). Many companies create their own or extensively
from usage of a methodology, there are certain points of modify existing methodologies to better fit their organiza-
similarity, suggesting that some tool adoption determinants tion [19], [66]. The methodology used by the organization
may be applicable to methodology adoption decisions. For in this study was created by a consultant for its own
example, both tool and methodology use are workplace
internal use.
behaviors with potential consequences for job performance
The Software Engineering Institute’s Capability Maturity
and rewards, and both require effort and skill to learn and
Model (CMM) explains the importance of methodologies.
use. However, the five models examined may behave
The CMM measures the effectiveness of an organization’s
somewhat differently due to three typical differences
software development capability by identifying five levels
between the domain of individual IT tool acceptance, for
which these five models were developed, and the present of maturity that lay successive foundations for process
domain of methodology acceptance among software devel- improvement [49]. Level-1 indicates an immature software
opers. First, the degree to which adoption of the target organization in which development is inconsistent and
innovation has been mandated by senior management lacks common software development processes (i.e., meth-
tends to be greater for methodologies than for tools. Second, odologies are not used). At Level-2, an organization begins
the magnitude of behavioral change entailed by adopting a to demonstrate repeatability in their development process
methodology is greater than that of a tool. Third, the greater primarily through basic project management and measure-
emphasis on the use of project teams for organizing ments (i.e., the beginnings of a methodology). Level-3 is
software development work increases the relevance of characterized by increased efficiency, with fully documen-
coworker influences compared to use of individual IT ted software processes (i.e., a methodology is fully
productivity tools. As will be seen, the set of models deployed). At Level-4, software processes are quantitatively
examined here includes constructs reflecting the external understood and controlled. At Level-5, use of common
influences of organizational and group pressures on software processes has become routine and automatic.
individual intentions to use, as well as individual adopter As characterized by the successive maturity levels of the
concerns regarding the extensiveness of behavioral change CMM, standardizing development processes via methodol-
required. ogies is important for several reasons. Standardization of
In this study, a Fortune 1000 organization implemented a
the development process puts the burden on the process
custom methodology, which moved them from an environ-
rather than the developer. Thus, developers’ skills become
ment having no methodology at all to an environment
more interchangeable, which is critically important given
guided by the new methodology (the methodology is
described in more detail in the Appendix). In our analysis today’s shortage and high turnover of software developers.
of the five explanatory models, four determinants of Standardization can improve productivity [40] and, by
intentions to use the new methodology were found providing a common vocabulary for documentation, can
significant: usefulness, voluntariness, compatibility, and reduce maintenance effort, which consumes up to 80 percent
subjective norm. These findings indicate that a methodolo- of the IT budget [31]. In general, higher levels of maturity
gy’s usefulness and compatibility with existing work have been empirically linked to higher system quality,
processes heavily influence its acceptance. Also, contrary better match with functional requirements, and lower cycle
to common beliefs, acceptance of the methodology is not time, effort, and cost [27], [41], [49]. Unfortunately,
assured just because it is mandated by the organization, according to a recent CMM maturity profile, almost half
although developers do place value on the opinions of of all companies assessed are at Level-1 [56]. Because the
coworkers and supervisors in forming their use decision. CMM assessment process is generally voluntary and the
The remainder of the paper is organized as follows: participants are self-selected, it is projected that the true
Section 2 provides an overview of prior research in industry percentage of Level-1 organizations is closer to
methodologies. Section 3 discusses the models compared 75 percent [68]. Furthermore, more than 75 percent of
in this study. Section 4 describes the research methods used, organizations assessed at Level-1 never return to have a
and Section 5 provides the results. Finally, Section 6
reassessment, suggesting the initial attempt to adopt/
discusses the implications of the study and suggests future
develop a methodology may have been abandoned [1].
areas for research.
The adoption of a methodology represents a significant
change in work practices, and many developers are
2 BACKGROUND resistant to such change [50]. As a result, “The expenditure
2.1 Methodologies of time and effort in implementing a software development
We define a methodology as a comprehensive guide to methodology makes it one of the most serious areas of
developing a system. Examples of methodologies include concern in modern IS” [53, p. 647]. The difficulty organiza-
those that can be purchased, such as Ernst and Young’s tions face in successfully deploying methodologies is clearly
Navigator and Accenture’s METHOD/1, or acquired in the a problem warranting continued research attention.

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
RIEMENSCHNEIDER ET AL.: EXPLAINING SOFTWARE DEVELOPER ACCEPTANCE OF METHODOLOGIES: A COMPARISON OF FIVE... 1137

2.2 Prior Research on Software Developer variables of centralization, formalization, and management
Intentions to Use Methodologies support to be significant differentiators between individual
There has been little previous research specifically addres- adopters and nonadopters of object-oriented programming.
sing the determinants of individual software developers’ Agarwal and Prasad [3] also looked at programming tool
intentions to use or not use methodologies mandated by their acceptance, but focused on the intention to use a specific
organization. Pfleeger [50, p. 122] suggests that “researchers language, C, by a group of COBOL programmers. Using a
should look for examples of technology transfer and assess combination of the Theory of Planned Behavior, the
them to understand the key variables that make adoption fast Technology Acceptance Model, and several external factors
and effective.” An exploratory study by Roberts et al. [52], (such as training), Agarwal and Prasad were able to explain
[53] uncovered potential factors that might affect the approximately 76 percent of a COBOL programmer’s
implementation of methodologies. Using exploratory factor intention to use C (with a sample size of 50). Although
analysis on 88 survey items, the authors interpreted five these individual-level studies include constructs that
perceptual factors that may possibly be linked to methodol- appear among the five models of tool usage intention being
ogy acceptance among individual developers: organiza- adapted to the methodology usage context in the present
tional transition (preparedness and enforcement), study, they exhibit various limitations. Some previous
management support, process transition (understanding
studies did not explicitly model the determinants of
the new process), use of models by the process, and external
intentions and/or did not use established psychometrically
support. However, the Roberts et al. [52], [53] study was not
validated instruments. These previous studies focused on
theoretically grounded, employed psychometrically unvali-
the adoption of tools or techniques as opposed to the
dated measures, and did not assess the relationship between
the derived factors and individual developer usage inten- adoption of an entire methodology.
tions or behaviors [67]. In the context of object-oriented Overall, there is a dearth of research in the software
systems development, Johnson et al. [35] used Ajzen’s [7] engineering literature about the determinants of methodol-
belief elicitation pretesting procedure to identify specific ogy acceptance [50]. The only prior studies that directly
salient beliefs underlying the attitude, subjective norm, and pertain to individual-level adoption of methodologies are the
perceived behavioral control constructs of the Theory of preliminary exploratory studies by Roberts et al. [52], [53],
Planned Behavior. However, they did not empirically test Johnson et al. [35], and Khalifa and Verner [38]. Other related
the relationships between these elicited beliefs and the research addresses adoption of specific tools or techniques
TPB constructs, developer intentions, or usage of object- that may be used within a methodology, rather than adoption
oriented systems development. Khalifa and Verner [38] of a methodology per se. Other tangentially-related types of
developed a model of the determinants of software devel- studies, such as descriptive and contingency approaches,
opers’ usage of waterfall and prototyping methodologies have addressed methodology usage patterns and the
based partly on Triandis’ [60] model of human behavior. performance consequences of matching the methodology to
Although their model did not employ psychometrically the specific context, but have not studied the determinants of
validated scales of theoretical constructs, this is the most individual developer intentions to adopt a methodology
theoretically justified of the previously reported studies of (e.g., [4], [13], [21], [24], [26], [32], [34], [36], [45], [51]). The
individual developer adoption of methodologies. Khalifa present study goes beyond prior research by investigating an
and Verna [38] found that perceived process quality and individual’s behavioral intention to use methodologies mandated by
facilitating conditions significantly explained use of proto- their organization. Due to the absence of established theore-
typing (R2 ¼ :178) and waterfall (R2 ¼ :317) methodologies. tical models in the particular domain of individual-level
Some individual-level research has been done that did intentions to adopt methodologies, five established models
not directly address adoption of methodologies per se, but of individual-level intentions to adopt IT tools are adapted:
rather the adoption of specific tools (such as CASE) and TAM, TAM2, PCI, TPB, and MPCU. We will examine the
techniques (such as object-oriented programming) com- degree to which these tool adoption models generalize to the
monly used within a methodology [48], [64]. This is a key domain of methodology adoption.
distinction since adoption of methodologies represents a
much more radical change than adopting tools and
techniques [47], [53]. Further, tools and techniques can be, 3 EXPLANATORY MODELS
and often are, used outside the context of a methodology Several theoretical models have been introduced in the IS
[19], [63]. Chau [14] found ease of use, usefulness, and long- literature regarding the determinants of voluntary indivi-
term consequences to be significantly related to CASE tool dual intentions to adopt IT tools in the workplace. These
usage. Iivari [33] identified several variables significantly draw from various reference field paradigms, have many
linked to a broad-based measure of CASE usage, which similarities as well as differences, and have generally been
included effectiveness, potential, naturalness, encourage- validated in the context of individual-level, end-user
ment, and routinization. Kozar’s [39] qualitative interviews productivity tools, such as personal computers, word
suggested that compatibility and management support processing, spreadsheets, electronic mail, and graphics
were important in the decision to adopt a new technique software. Although it is not known how well any of them
for eliciting user requirements. Leonard-Barton [42] found will generalize to the domain of adopting a methodology,
that training, attitude toward innovations, and perceived our logic is that each model is an abstraction and
client preferences were statistically different between simplification of the true process by which developers form
adopters and nonadopters of a structured analysis techni- intentions to use methodologies and will inherently either
que. Sultan and Chan [57] found the organizational-level highlight or downplay particular mechanisms operating in

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
1138 IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO. 12, DECEMBER 2002

this domain. Thus, the models will provide alternative results of an innovation are observable by others. Moore
views into the specific phenomenon of methodology and Benbasat [44] also included a measure of voluntariness,
acceptance. In this section, each model is briefly summar- which is defined and measured the same by Venkatesh and
ized, focusing on the direct determinants of usage inten- Davis [62]. Although Moore and Benbasat [44] did not
tions (and not on antecedents of those direct determinants). report the relationship between the seven PCI constructs
Interested readers can find more detailed theoretical and usage or usage intentions, Agarwal and Prasad [2]
development and empirical validation of each model in reported empirical support for PCI as a model of intentions
the referenced articles. to use the World Wide Web.

3.1 Technology Acceptance Model (TAM) 3.4 Theory of Planned Behavior (TPB)
Davis et al. [17] and Davis [16] introduced the Technology The Theory of Planned Behavior, introduced by Ajzen and
Acceptance Model (TAM) to predict and explain knowledge his colleagues in social psychology [6], [8], [10], [55], posits
worker adoption of information technology applications in three primary determinants of intention: perceived beha-
the organizational workplace (e.g., electronic mail, word vioral control, subjective norm, and attitude. Perceived
processing, and graphics software). TAM posits that behavioral control refers to one’s perceptions of internal
intentions to use a target system are jointly determined by or external constraints on performing the behavior.
individuals’ perceived usefulness, which is the extent to Consistent with findings by Chau [14] and Taylor and
which the person thinks using the system will enhance his Todd [58], we represent distinct internal and external
or her job performance, and perceived ease of use, the subdimensions of perceived behavioral control as separate
extent to which the person perceives using the system will constructs. Subjective norm is defined and measured the
be free of effort. Across numerous studies, perceived same as in TAM2. Attitude refers to one’s degree of
usefulness typically has a strong influence on usage favorableness or unfavorableness towards performing a
intentions, and perceived ease of use often has a significant target behavior, and is conceptually equivalent to TAM’s
secondary effect (for recent review, see [61]). usefulness construct.

3.2 TAM2 3.5 Model of Personal Computer Utilization (MPCU)


TAM2 is an extension of TAM designed to encompass both Thompson et al. [59] introduced a model of personal
voluntary and mandatory usage situations by including two computer utilization (MPCU) based on Triandis’ [60] model
additional constructs. Subjective norm, defined as the of interpersonal behavior from social psychology. The
degree to which people think that others who are important model includes the social factors construct, which refers
to them think they should perform the behavior, was found to the individual’s internalization of the reference groups’
to be a significant determinant of intention for the use of subjective culture, and interpersonal agreements that the
tools in mandatory usage contexts [62]. Venkatesh and individual has made with others. MPCU’s affect construct
Davis [62] show that perceived voluntariness, defined as refers to the positive or negative emotional reactions an
the extent to which potential adopters perceive the adoption individual associates with a particular act. The Triandis [60]
decision to be nonmandatory, significantly moderates the model includes perceived consequences and Thompson et
direct effect of subjective norm on intention to use. al. [59] delineate three types: two near-term and one long-
term. Near-term consequences include complexity, defined
3.3 Perceived Characteristics of Innovating (PCI) and measured as essentially the reverse of perceived ease of
Moore and Benbasat [44], drawing from the extensive work use in TAM, and job fit, defined and measured equiva-
on diffusion of innovations by Rogers [54], published an lently to perceived usefulness in TAM. Career conse-
instrument to measure what they term “perceived char- quences, the dimension of long-term consequences of use,
acteristics of innovating” (PCI) regarding adoption of refers to outcomes that have a payoff in the future, such as
personal workstations. Seven constructs were represented increasing the flexibility to change jobs or increasing the
in the PCI instrument. These include relative advantage, opportunities for more meaningful work. Finally, Thomp-
which is the degree to which an innovation is perceived as son et al. [59] include facilitating conditions, which are
being better than its precursor, and complexity, which is the factors in the environment that make an act easy to do. This
degree to which an innovation is perceived as being construct parallels the external perceived behavioral control
difficult to use. Noting the conceptual equivalence, Moore
construct of TPB and we equate the two.
and Benbasat [44] used Davis’ [16] perceived usefulness
The five models have both common and unique
scale to measure relative advantage and ease of use scale to
constructs (Table 1). All models have a construct equivalent
measure complexity (in reversed direction). The PCI
to perceived usefulness (relative advantage in PCI, attitude
instrument also included measures of compatibility, result
demonstrability, image, and visibility. Compatibility refers in TPB, job fit in MPCU). TAM’s ease of use construct has a
to the degree to which an innovation is perceived as being reversed counterpart in complexity within both PCI and
consistent with the existing values, needs, and past MPCU. Subjective norm appears in TAM2 and TPB, and in
experiences of potential adopters; result demonstrability MPCU as social factors, although no such construct appears
refers to the degree to which an innovation is perceived to in TAM or PCI. Affect appears in MPCU, but not in the
be amenable to demonstration of tangible advantages. other models. The external subdimension of perceived
Image refers to the degree to which use of an innovation behavioral control appears in TPB, and in MPCU as
is perceived to enhance one’s image or status in one’s social facilitating conditions, whereas the internal perceived
system, and visibility refers to the degree to which the behavioral control subdimension is unique to TPB among

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
RIEMENSCHNEIDER ET AL.: EXPLAINING SOFTWARE DEVELOPER ACCEPTANCE OF METHODOLOGIES: A COMPARISON OF FIVE... 1139

TABLE 1
Construct Map

these models. PCI is the only model in this set that includes compatibility, result demonstrability, image, and visibility
compatibility, result demonstrability, image, and visibility, scales were adapted from Moore and Benbasat [44]. PBC-I,
although it omits subjective norm, internal and external PBC-E, and career consequences included items from Chau
perceived behavioral control, and career consequences. [14] and Thompson et al. [59]. Subjective norm items came
Voluntariness appears in TAM2 and PCI, but not in TAM, from Venkatesh and Davis [62] and Ajzen and Fishbein [9].
TPB, or MPCU. Finally, the career consequences construct Pilot testing of the instrument led to slight rewording of
appears in only MPCU among this set of models. Each some items and the elimination of the affect items, which
did not appear to be appropriate in the context of
model is tested as theorized by examining the hypothesized
methodologies (affect pertains to an emotional response).
direct determinants of behavioral intention.
The elimination of the affect construct is consistent with
prior studies of this type [58].
4 METHOD Measurement scales are provided in Table 2. Factor
4.1 Sample analyses confirmed the a priori factor structure for each
model.1 Cronbach alpha coefficients, as well as means,
A questionnaire was distributed to each of the 185
standard deviations, and intercorrelations of all constructs
application developers in a Fortune 1000 company
are provided in Table 3. Alpha reliabilities ranged from .77
approximately six weeks after the introduction of their
to .95, indicating acceptable internal consistency [12]. Not
new methodology (the methodology is described in the
surprisingly given their source, each individual construct
Appendix). One hundred twenty-eight developers com-
pleted the instrument for a response rate of 69 percent. The was significantly correlated with behavioral intention.
average age of the developers was 36 years with a reported
average of 10 years of development experience. The 5 RESULTS
average number of years the developers had been with
Each model (TAM, TAM2, PCI, TPB, MPCU) was tested
the company was five. Of those participating in the study,
individually using least-squares regression analysis. Re-
45 were female, 74 were male, and 9 did not report their
gender. The participants in the study were developers of gression diagnostics were checked for each model, which
all levels in the organization and they represented the led to the deletion of two data points whose residuals were
gamut of IT job functions in the organization. identified as outliers, violating the normality assumption
for residuals. Once these two points were deleted, no
4.2 Measures significant violations of regression assumptions were
The questionnaire was constructed using validated mea- found. Standardized variables were used as suggested by
surement scales from previous research, as discussed in Aiken and West [5]. Table 4 shows the results of each model
Section 3. The behavioral intention items were adapted test, delineating the varying names of the constructs as well
from standard scales such as those used by Ajzen and
Fishbein [9], Venkatesh and Davis [62], and Taylor and 1. Due to space limitations, factor analyses are not presented. All
constructs were validated in prior studies and in all cases in the present
Todd [58]. Usefulness and ease of use items were taken study, items loaded on the factors as expected with no cross loadings
from the original Davis [16] instrument. Voluntariness, exceeding .30.

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
1140 IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO. 12, DECEMBER 2002

TABLE 2
Measurement Scales

1. ADM is an acronym for Application Development Methodology, a pseudonym for the true methodology name, which is disguised in order to protect
the anonymity of the company where the data was collected.
2. The true company name was used on the actual instrument but is not used here to protect the anonymity of the company where the data was
collected.

as the beta coefficients and significance levels. Additionally, methodology adoption, the present empirical investigation
the R2 and adjusted R2 values are provided. Usefulness was needed. In order to offset the risk that there are
(p < :001) was the only significant variable in TAM, while potential determinants of methodology adoption that
usefulness (p < :001), subjective norm (p < :01), and volun- were not found significant for tool adoption, we included
tariness (p < :01) were significant in TAM2. In testing the all direct determinants of intention from each of the
PCI, we found relative advantage (p < :001), voluntariness specified models in order to more fully represent
(p < :001), and compatibility (p < :05) to be significant. In potentially relevant determinants.
TPB, attitude (p < :001) and subjective norm (p < :01) were Across the five models, four determinants of usage
significant. In MPCU, job fit (p < :001) and social factors intentions were found significant in at least one model:
(p < :01) were significant. usefulness (all five models), subjective norm (TAM2, TPB,
and MPCU), voluntariness (TAM2 and PCI), and compat-
ibility (PCI). None of the five original models contained all
6 DISCUSSION four significant intention determinants. In moving from the
To identify the significant determinants of individual domain of IT tools to methodologies, there was one main
software developers’ intentions to use a new methodology similarity in the pattern of significant intention determi-
introduced into their organization, we adapted five nants: Usefulness was a strong and highly significant
previously established theoretical models of individual- determinant of intentions to use the methodology in the
level user acceptance of IT tools: TAM, TAM2, PCI, TPB, present study, as typically found in the domain of tool
and MPCU. Despite some commonalities between these acceptance [2], [17], [43], [58], [59].
domains, there are also important differences. Compared There were several important differences in the pattern
to adopting an IT tool, adopting a methodology tends to of significant intention determinants in the present meth-
be more mandatory than voluntary, and more radical odology domain compared to previous tool adoption
than incremental. Because it was unclear a priori whether studies. Three constructs often found significant in previous
tool adoption models would be effective in explaining tool acceptance studies, ease of use (e.g., [17], [43], [59],

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
RIEMENSCHNEIDER ET AL.: EXPLAINING SOFTWARE DEVELOPER ACCEPTANCE OF METHODOLOGIES: A COMPARISON OF FIVE... 1141

TABLE 3
Intercorrelations, Means, Standard Deviations, and Cronbach Alphas for All Constructs

[62]), perceived behavioral control—internal [43], [58], and model determinants were significant. This shift in pattern
perceived behavioral control—external [43], [58], were of adoption determinants correspond to a large extent to
found nonsignificant in the present study. Compatibility the known differences between methodology and tool
has not typically been found significant in tool adoption usage contexts, namely, that methodology use is more
studies (e.g., [2]), but was found significant in the present radical and mandatory than tool use.
methodology domain. Subjective norm, which has been
6.1 Practitioner Insights
found significant in some tool intention studies (e.g., [58],
This research provides important new insights for mana-
[59], [62, Studies 3 and 4]) and nonsignificant in others (e.g.,
ging the effective introduction and deployment of a
[17], [43], [62, Studies 1 and 2]), was found significant in the
methodology within an organization. A common pitfall is
present research. Voluntariness and compatibility, both
for senior managers to assume that, once the organizational
significant determinants of methodology use intentions in
decision to adopt a methodology has been made, it is
the present study, were found nonsignificant as determi- sufficient to issue an official policy mandating use of the
nants of tool use intentions in some prior research [2] and new methodology in order to guarantee successful deploy-
significant in other previous studies (e.g., [15], [29], [33]). ment. Our findings disconfirm this assumption, demon-
Overall, these findings suggest that, as the behavioral strating instead that individual developer acceptance is far
domain changes from tool use to methodology use, there is from assured even in the presence of an organizational
a reduction in the relevance of how easy or hard the mandate. Other important drivers of methodology accep-
behavior is to perform and whether or not one possesses tance exist beyond the target user’s perception that the
adequate internal or external resources to perform it. There adoption decision is mandatory.
is an increase in the relevance of subjective normative
First and foremost, if a methodology is not regarded as
pressure to perform the behavior, the perception of a
useful by developers, its prospects for successful deploy-
formal mandate (the inverse of voluntariness), and the
ment may be severely undermined. Of all of the external
compatibility of the target behavior with individuals’
current ways of performing their work. This is consistent pressures impinging upon professional knowledge workers
with the observation that, compared to tool adoption, in an organizational setting, few are as influential as the
methodology adoption is a more fundamental and radical performance measurement and reward structure. There-
change to the behavioral processes in conducting one’s fore, to the extent a methodology is not useful, that is, it
work and is more apt to be a mandatory than voluntary does not enable developers to be more productive and
innovation. Thus, while the models as a group successfully achieve higher levels of performance in their job and,
generalized from the domain of tool adoption to metho- therefore, they are not likely to use it in a sustained manner,
dology adoption, there was a different pattern of which even if it is organizationally mandated. Demonstrating the

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
1142 IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO. 12, DECEMBER 2002

TABLE 4
Regression Analysis of Theoretical Models

individual productivity benefits (i.e., usefulness) of the could be altered to reward individual developers directly
methodology may help shape a developer’s intention to for organizational-level performance improvements.
adopt. However, in the words of Leonard-Barton [42, p. 7], The more compatible a mandated methodology is with
for many methodologies “the benefits from adoption are how developers perform their work, the more likely they are
long-term and they accrue to the organization as a whole; to form intentions to use it, especially if it is perceived to be
whereas, the costs are immediate and they are incurred by useful. Since a methodology is a comprehensive specifica-
individual analysts or programmers.” Such adverse effects tion of the steps and procedures to be used throughout all
on developer productivity, according to our model, would phases of the software development process [31], the more it
tend to strongly erode adoption intentions (i.e., the strong departs from the current work processes of any given
relationships between usefulness and intention as demon- developer, the less likely that developer will be to intend to
strated by the in each model suggest a reduction in adopt it. Professional knowledge workers such as software
usefulness will significantly reduce intention.). Positive developers usually evolve work practices that they find
consequences for the organization cannot offset negative effective and have come to rely on to produce effective work
consequences for the individual. To enhance the probability output. There is significant effort and uncertainty involved
of successful deployment, such mismatches between orga- in learning and routinizing new work patterns. Managers
nizational and individual productivity benefits must be can do several things to overcome this barrier to successful
eliminated, for example, by altering the nature of the deployment. Management would be wise to explicitly
methodology to provide greater individual performance demonstrate instances of the new methodology’s compat-
improvement. Or the performance measurement system ibility with existing work practices. If there are particular

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
RIEMENSCHNEIDER ET AL.: EXPLAINING SOFTWARE DEVELOPER ACCEPTANCE OF METHODOLOGIES: A COMPARISON OF FIVE... 1143

parts of a new methodology that especially depart from methodology acceptance in a variety of engineering
current practice, but which are not vital to the effectiveness domains beyond software engineering.2 This study sur-
of the methodology, these may be modified to be brought veyed developers shortly after the introduction of the
into greater alignment with current practice. Rather than methodology (i.e., postadoption). Are the same variables
attempting to convert from current practice to the new significant preadoption? Are the same variables significant
methodology in a single step (the “big bang” approach), it after a longer period of time? Overall, the present study lays
might be possible to design a migration path in which the foundation for at least three paths for future research:
developers adopt parts of the methodology incrementally, 1) improving the models (i.e., increasing explanatory
in each step experiencing improved productivity. Training power) in the existing domain of individual-level intentions
can be designed to undo existing work habits and routines to adopt methodologies, 2) extending the boundary condi-
while propelling developers along such a migration path. In tions of the models by adapting it to other domains (e.g.,
software development tools and techniques, and nonsoft-
any case, when selecting a methodology, managers should
ware development processes), and 3) investigating the time
take into consideration, all else being equal, how compatible
sensitivity of the models.
the methodology is to current work practices.
Our research shows that subjective norm exerts a distinct 6.3 Conclusion
influence on intentions to adopt a methodology above and Five established models of individual-level intentions for
beyond usefulness, voluntariness, and compatibility. This tool adoption (TAM, TAM2, PCI, TPB, MPCU) were used to
tells us that, even if developers regard a mandated provide insights into methodology adoption. This is the first
methodology as both useful and compatible with their study to empirically test these models in a methodology
current work practices, they may still avoid using it because context. Extending the boundaries of these models from the
coworkers or supervisors think they should not be using it. domain of tools to methodologies has demonstrated their
Developers may be motivated to behave in compliance with resilience in adapting to a new domain and the differences
such social influences if salient referents possess sufficient required by the new domain. This research also contributes
influence over them [62]. For example, developers who toward an integrative body of research needed in software
believe that a mandated methodology would be useful and
engineering by establishing the foundation for a new theory
compatible for them personally may work amongst peers
of methodology acceptance. By identifying the determi-
and supervisors who actively oppose its use. In order to
nants of acceptance, prescriptive models can be enhanced
sustain a balance of equitable social exchange, or to avoid
and contingency models of methodology use may be
damaging key work relationships, such developers may
extended from a focus on effectiveness of a methodology,
avoid using the methodology despite its merits. Manage-
if used, to a focus on whether or not a methodology will be
ment may be able to devise persuasive communication
used in the first place (e.g., [13], [21], [26], [32], [51]). From a
strategies designed to reduce the developers’ perceived
practical perspective, insights into acceptance provided by
strength of peers’ and coworkers’ counterimplementation
the study suggest better strategies for introducing meth-
desires, or to attenuate their motivation to comply with
odologies and avoiding situations where resources (money
such subjective norms.
and time) have been spent on methodologies that are not
6.2 Future Research Directions used [50]. By knowing the determinants of a developer’s
The present research shows that mandating methodology intention, management can take appropriate action to
use by software developers does not guarantee that the increase the adoption of a new methodology. Overall, from
methodology actually will be used. By demonstrating the both theoretical and practical perspectives, this study fills a
significant roles of usefulness, subjective norm, and significant gap in software engineering research.
compatibility above and beyond voluntariness, several
worthwhile avenues of future research are apparent. There APPENDIX
is a need for research to integrate the findings of the
various models presented here. Also, any additional DESCRIPTION OF THE METHODOLOGY
determinants of methodology acceptance should be identi- Description of methodology examined in this study. The
fied and incorporated into the model. The fact that data research site was a Fortune 1000 company with approxi-
was collected from a single organization may limit the mately 185 application developers. The methodology, based
generalizability of these findings, although the specific on the structured development paradigm, was custom-
methodology studied employs the traditional structured created for the company’s internal use under the leadership
development process and is therefore highly representative of an external consultant. In this case, the organization
of the domain of methodologies to which we would like to was moving from an environment of no prescribed
generalize [39]. Future research should attempt to confirm processes in place (i.e., Level-1 CMM; no methodology) to
these findings in other organizations. an environment guided by an organization-wide metho-
Future research needs to address whether the boundary dology (at least Level-2 CMM).
conditions of the models extend to the adoption of other The methodology is comprehensive, providing process
software development innovations such as CASE tools or directions for the complete life cycle. It includes specifica-
development techniques, and whether it will prove useful tions for the use of tools at various project stages by various
for other nonsoftware development process innovations, parties, including users. The methodology, while compre-
such as new organizational structures and engineering hensive, is adaptable to many different project types (e.g.,
processes. In fact, we were surprised at the lack of small, large; new development, maintenance). For example,
acceptance studies in the various engineering disciplines.
Perhaps this study can serve as a foundation for studies of 2. The authors wish to thank one of the reviewers for this suggestion.

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
1144 IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, VOL. 28, NO. 12, DECEMBER 2002

a director could use the methodology to launch a new [11] J. Brancheau, B. Janz, and J. Wetherbe, “Key Issues in Information
Systems Management: 1994-1995 SIM Delphi Results,” MIS
project and assign team members. The exact portions of the Quarterly, vol. 20, no. 2, pp. 225- 242, June 1996.
methodology, such as producing ERDs or prototypes to be [12] F. Brown, Principles of Educational and Psychological Testing. New
used by various types of developers, such as programmers York: Holt, Reinhart, and Whinston, 1976.
and database administrators, are determined by project [13] R. Burns and A. Dennis, “Selecting the Appropriate Application
characteristics (i.e., the methodology established a contin- Development Methodology,” DATA BASE, vol. 17, no. 1, pp. 19-
23, 1985.
gency model for methodology adaptation).
[14] P. Chau, “An Empirical Investigation on Factors Affecting the
All employees were provided a hardcopy of the Acceptance of CASE by Systems Developers,” Information &
methodology (approximately 150 pages) and access to an Management, vol. 30, pp. 269-280, 1996.
on-line version with embedded links to tools and templates. [15] W. Chin and A. Gopal, “Adoption Intention in GSS: Relative
When the new methodology was launched, all developers Importance of Beliefs,” Data Base Advances, vol. 26, pp. 42-64, 1995.
within the organization were instructed via a formal written [16] F. Davis, “Perceived Usefulness, Perceived Ease of Use, and User
Acceptance of Information Technology,” MIS Quarterly, vol. 13,
policy statement from the chief information officer to begin no. 3, pp. 318-339, Sept. 1989.
using the methodology for all projects. [17] F. Davis, R. Bagozzi, and P. Warshaw, “User Acceptance of
Epilogue: Effectiveness of the new methodology. For Computer Technology: A Comparison of Two Theoretical Mod-
this organization, measures of methodology effectiveness els,” Management Science, vol. 35, no. 8, pp. 982-1003, Aug. 1989.
six months after implementation revealed several major [18] G. Dickson, R. Leitheiser, J. Wetherbe, and M. Nechis, “Key
Information Systems Issues for the 1980’s,” MIS Quarterly, vol. 8,
improvements. Based on a comparison of 2,251 projects no. 3, pp. 135-159, Sept. 1984.
before the methodology with 280 projects after the [19] G. Dietrich, D. Walz, and J. Wynekoop, “The Failure of SDT
methodology, the company found: Diffusion: A Case for Mass Customization,” IEEE Trans. Eng.
Management, vol. 44, no. 4, pp. 390-398, Nov. 1997.
an increase in on-budget performance of 18 percent,
1. [20] Digital Economy 2000,U.S. Department of Commerce, Economics
a reduction in the number of late projects by 26 percent,
2. and Statistics Administration, http://www.ecommerce.gov, 2000.
[21] M. El Louadi, Y. Pollalis, and J. Teng, “Selecting a Systems
a reduction in average days late by 25 percent,
3. Development Methodology: A Contingency Framework,” Informa-
higher customer satisfaction, and
4. tion Resources Management J., vol. 4, no. 1, pp. 11-19, Winter 1991.
less training time required for new hires and for
5. [22] B. Fitzgerald, “Formalized Systems Development Methodologies:
transfers among teams. A Critical Perspective,” Information Systems J., vol. 6, pp. 3-23, 1996.
[23] W. Gibbs, “Software’s Chronic Crisis,” Scientific Am., pp. 86-95,
This comparison reveals a strong improvement in effec- Sept. 1994.
tiveness after the implementation of the methodology for [24] R. Glass, “A Snapshot of Systems Development Practice,” IEEE
many different projects. Software, vol. 16, no. 3, pp. 110-111, May/June 1999.
[25] R. Glass, “The Realities of Software Technology Payoffs,” Comm.
ACM, vol. 42, no. 2, pp. 74-79, Feb. 1999.
ACKNOWLEDGMENTS [26] B. Hardgrave, R. Wilson, and K. Eastman, “Toward a Contingency
Model for Selecting an Information System Prototyping Strategy,”
The authors are grateful to the Information Technology J. Management Information Systems, vol. 16, no. 2, pp. 113-136, 1999.
Research Center at the University of Arkansas for financial [27] D. Harter, M. Krishnan, and S. Slaughter, “Effects of Process
support of this research. Maturity on Quality, Cycle Time, and Effort in Software Product
Development,” Management Science, vol. 46, no. 4, pp. 451-466,
Apr. 2000.
[28] C. Hartog and M. Herbert, “1985 Opinion Survey of MIS
REFERENCES Managers: Key Issues,” MIS Quarterly, vol. 10, no. 4, pp. 351-
[1] I. Aaen and J. Damsgaard, “Software Process Improvement: What 361, Dec. 1986.
Management Tends to Forget,” Information Systems: Current Issues [29] M. Hebert and I. Benbasat, “Adopting Information Technology in
and Future Changes, Proc. IFIP WG8.2 and WG8.6 Joint Working Hospitals: The Relationships between Attitudes/Expectations and
Conf., pp. 89-110, Dec. 1998. Behavior,” Hospital and Health Services Administration, vol. 39, no. 3,
[2] R. Agarwal and J. Prasad, “The Role of Innovation Characteristics pp. 369-383, 1994.
and Perceived Voluntariness in the Acceptance of Information [30] J. Herbsleb, A. Carleton, J. Rozum, J. Siegel, and D. Zubrow,
Technologies,” Decision Sciences, vol. 28, no. 3, pp. 557-582, 1997. “Benefits of CMM-Based Software Process Improvement: Initial
[3] R. Agarwal and J. Prasad, “A Field Study of the Adoption of Results,” Technical Reports CMU/SEI-94-TR-013 and ESC-TR-94-
Software Process Innovations by Information Systems Profes- 013, Software Eng. Inst., Carnegie Mellon Univ., Aug. 1994.
sionals,” IEEE Trans. Eng. Management, vol. 47, no. 3, pp. 295-308, [31] J. Hoffer, J. George, and J. Valacich, Modern Systems Analysis and
Aug. 2000. Design, second ed. Reading, Mass.: Addison-Wesley, 1999.
[4] R. Agarwal, A. Sinha, and M. Tanniru, “Cognitive Fit in
[32] G. Howard, T. Bodnovich, T. Janicki, J. Liegle, S. Klein, P. Albert,
Requirements Modeling: A Study of Object and Process Meth-
and D. Cannon, “The Efficacy of Matching Information Systems
odologies,” J. Management Information Systems, vol. 13, no. 2,
Development Methodologies with Application Characteristics
pp. 137-162, Fall 1996.
—An Empirical Study,” The J. Systems and Software, vol. 45,
[5] L. Aiken and S. West, Multiple Regression: Testing and Interpreting
pp. 177-195, 1999.
Interactions. Newbury Park, Calif.: Sage, 1991.
[6] I. Ajzen, “From Intentions to Action: A Theory of Planned [33] J. Iivari, “Why are CASE Tools not Used?” Comm. ACM, vol. 39,
Behavior,” Action Control: From Cognition to Behavior, J. Kuhl and no. 10, pp. 94-103, Oct. 1996.
J. Beckmann eds., New York: Springer Verlag, pp. 11-39, 1985. [34] R. Johnson and B. Hardgrave, “Object-Oriented Systems Devel-
[7] I. Ajzen, Attitudes, Personality, and Behavior. Chicago: Dorsey Press, opment: Current Practices and Attitudes in Industry,” J. Systems &
1988. Software, vol. 48, no. 1, pp. 5-12, 1999.
[8] I. Ajzen, “The Theory of Planned Behavior,” Organizational [35] R. Johnson, B. Hardgrave, and E. Doke, “An Industry Analysis of
Behavior and Human Decision Processes, vol. 50, pp. 179-211, 1991. Developer Beliefs About Object-Oriented Systems Development,”
[9] I. Ajzen and M. Fishbein, Understanding Attitudes and Predicting The DATA BASE for Advances in Information Systems, vol. 30, no. 1,
Social Behavior. Englewood Cliffs, NJ: Prentice-Hall, 1980. pp. 47-64, 1999.
[10] I. Ajzen and T. Madden, “Prediction of Goal-Directed Behavior: [36] M. Jones and K. Arnett, “Current Practices in Management
Attitudes, Intentions, and Perceived Behavioral Control,” Information Systems,” Information & Management, vol. 24, no. 2,
J. Experimental Social Psychology, vol. 22, pp. 453-474, 1986. pp. 61-69, 1993.

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.
RIEMENSCHNEIDER ET AL.: EXPLAINING SOFTWARE DEVELOPER ACCEPTANCE OF METHODOLOGIES: A COMPARISON OF FIVE... 1145

[37] M. Keil, “Pulling the Plug: Software Project Management and the [61] V. Venkatesh, “Creation of Favorable User Perceptions: Exploring
Problem of Project Escalation,” MIS Quarterly, vol. 19, no. 4, the Role of Intrinsic Motivation,” MIS Quarterly, vol. 23, no. 2,
pp. 421-443, Dec. 1995. pp. 239-260, June 1999.
[38] M. Khalifa and J. Verner, “Drivers for Software Development [62] V. Venkatesh and F. Davis, “A Theoretical Extension of the
Method Usage,” IEEE Trans. Eng. Management, vol. 47, no. 3, Technology Acceptance Model: Four Longitudinal Field Studies,”
pp. 360-369, Aug. 2000. Management Science, vol. 46, no. 2, pp. 186-204, Feb. 2000.
[39] K. Kozar, “Adopting Systems Development Methods: An Ex- [63] I. Vessey and R. Glass, “Strong vs. Weak Approaches to Systems
ploratory Study,” J. Management Information Systems, vol. 5, no. 4, Development,” Comm. ACM, vol. 41, no. 4, pp. 99-102, Apr. 1998.
pp. 73-86, 1989. [64] I. Vessey, S. Jarvenpaa, and N. Tractinsky, “Evaluation of Vendor
[40] M. Krishnan, T. Mukhopadhyay, and D. Zubrow, “Software Products: CASE Tools as Methodology Companions,” Comm.
Process Models and Project Performance,” Information Systems ACM, vol. 35, no. 4, pp. 90-105, Apr. 1992.
Frontiers, vol. 1, no. 3, pp. 267-277, 1999. [65] R. Whiting, “Development in Disarray,” Software Magazine, vol. 18,
no. 12, p. 20, Sept. 1998.
[41] M.S. Krishnan, C.H. Kriebel, S. Kekre, and T. Mukhopadhyay,
[66] J. Wynekoop and N. Russo, “Systems Development Methodolo-
“An Empirical Analysis of Productivity and Quality in Software
gies: Unanswered Questions,” J. Information Technology, vol. 10,
Products,” Management Science, vol. 46, no. 6, pp. 745-759, 2000.
pp. 65-73, 1995.
[42] D. Leonard-Barton, “Implementing Structured Software Meth- [67] S. Yadav, N. Shaw, L. Webb, and C. Sutcu, “Comments on ‘Factors
odologies: A Case of Innovation in Process Technology,” Interfaces, that Impact Implementing a System Development Methodology,’”
vol. 17, no. 3, pp. 6-17, May/June 1987. IEEE Trans. Software Eng., vol. 27, no. 3, pp. 279-281, Mar. 2001.
[43] K. Mathieson, “Predicting User Intentions: Comparing the Tech- [68] E. Yourdon, “Where’s the Basis for Year 2000 Optimism?”
nology Acceptance Model with the Theory of Planned Behavior,” Computerworld, vol. 32, no. 7, p. 68, 1998.
Information Systems Research, vol. 2, no. 3, pp. 173-191, 1991. [69] R. Zmud, “An Examination of ‘Push-Pull’ Theory Applied to
[44] G. Moore and I. Benbasat, “Development of an Instrument to Process Innovation in Knowledge Work,” Management Science,
Measure the Perceptions of Adopting an Information Technology vol. 30, no. 6, pp. 727-738, June 1984.
Innovation,” Information Systems Research, vol. 2, no. 3, pp. 192-222, [70] R. Zmud, “Diffusion of Modern Software Practices: Influence of
1991. Centralization and Formalization,” Management Science, vol. 28,
[45] C. Necco, C. Gordon, and N. Tsai, “Systems Analysis and Design: no. 12, pp. 1421-1431, Dec. 1982.
Current Practices,” MIS Quarterly, vol. 11, no. 4, pp. 461-475, Dec.
1987. Cynthia K. Riemenschneider is an assistant
[46] F. Niederman, J. Brancheau, and J. Wetherbe, “Information professor of information systems in the Walton
System Management Issues for the 1990s,” MIS Quarterly, vol. 15, College of Business at the University of Arkan-
no. 4, pp. 475-500, Dec. 1991. sas. Her research has appeared in Information
[47] W. Orlikowski, “CASE Tools as Organizational Change: Investi- Systems Research, Information and Manage-
gating Incremental and Radical Changes in Systems Develop- ment, The Journal of Computer Information
ment,” MIS Quarterly, vol. 17, no. 3, pp. 309-340, Sept. 1993. Systems, and others. She currently conducts
[48] P. Palvia and J. Nosek, “An Empirical Evaluation of System research on IT and small businesses, IT adop-
Development Methodologies,” Information Resources Management tion, and people in IT.
J., vol. 3, no. 3, pp. 23-32, 1990.
[49] M.C. Paulk, C.V. Weber, S.M. Garcia, M.B. Chrissis, and M. Bush,
“Key Practices of the Capability Maturity Model,” Technical
Bill C. Hardgrave holds the Edwin and Karlee
Report CMU/SEI-93-TR-25, Version 1.1, Software Eng. Inst.,
Bradberry Chair in Information Systems and is
Carnegie Mellon Univ., Feb. 1993.
the executive director of the Information Tech-
[50] S.L. Pfleeger, “Understanding and Improving Technology Trans- nology Research Center in the Walton College
fer in Software Engineering,” J. Systems and Software, vol. 47, nos. 2- of Business at the University of Arkansas. His
3, pp. 111-124, July 1999. research interests include software develop-
[51] D. Ratbe, W. King, and Y. Kim, “The Fit Between Project ment, software process improvement, object-
Characteristics and Application Development Methodologies: A oriented development, and adoption of software
Contingency Approach,” J. Computer Information Systems, vol. 40, development innovations. He has published in
no. 2, pp. 26-33, 1999-2000. such journals as The Journal of Management
[52] T. Roberts, M. Gibson, and K. Fields, “System Development Information Systems, Communications of the ACM, IEEE Software, The
Methodology Implementation: Perceived Aspects of Importance,” DATA BASE for Advances in Information Systems, Information &
Information Resources Management J., vol. 12, no. 3, pp. 27-38, July- Management, Journal of Systems and Software, and Educational &
Sept. 1999. Psychological Measurement, among others. He is a member of the IEEE
[53] T. Roberts, M. Gibson, K. Fields, and R. Rainer, “Factors that Computer Society.
Impact Implementing a System Development Methodology,”
IEEE Trans. Software Eng., vol. 24, no. 8, pp. 640-649, Aug. 1998. Fred D. Davis is a professor, David D. Glass
[54] E. Rogers, Diffusion of Innovations, third ed. New York: Free Press, Chair, and the department chair of Information
1983. Systems at the Walton College of Business,
[55] D. Schifter and I. Ajzen, “Intention, Perceived Control, and Weight University of Arkansas. His research on user
Loss: An Application of the Theory of Planned Behavior,” acceptance of information technology and com-
J. Personality and Social Psychology, vol. 49, pp. 843-851, 1985. puter-assisted decision making has appeared in
[56] Software Engineering Institute, Process Maturity Profile of the MIS Quarterly, Management Science, Decision
Software Community 1999 Year End Update, Mar. 2000, http:// Sciences, and Organizational Behavior and
www.sei.cmu.edu/sema/pdf/2000mar.pdf. Human Decision Processes.
[57] F. Sultan and L. Chan, “The Adoption of New Technology: The
Case of Object-Oriented Computing in Software Companies,”
IEEE Trans. Eng. Management, vol. 47, no. 1, pp. 106-126, Feb. 2000.
[58] S. Taylor and P. Todd, “Understanding Information Technology . For more information on this or any computing topic, please visit
Usage: A Test of Competing Models,” Information Systems our Digital Library at http://computer.org/publications/dlib.
Research, vol. 6, no. 2, pp. 144-176, 1995.
[59] R. Thompson, C. Higgins, and J. Howell, “Personal Computing:
Toward a Conceptual Model of Utilization,” MIS Quarterly, vol. 15,
no. 1, pp. 125-143, Mar. 1991.
[60] H. Triandis, “Values, Attitudes, and Interpersonal Behavior,”
Nebraska Symp. Motivation, 1979, H. Howe and M. Page eds.,
pp. 195-259, 1979.

Authorized licensed use limited to: ALTRAN. Downloaded on March 10,2021 at 14:21:06 UTC from IEEE Xplore. Restrictions apply.

You might also like