liven...

Av 5'.

n

£53 ..

v a
.30..

.4 4 1.: Andrlﬂrrnw )1.”

.V. .frn‘ilh If...

. «nut.

: »
31352:”

A:

v§ .13. 5:
4%..) at.

r

 

. m if
my... ,

. .
in?! .Vﬁig‘mh
Ct} t; \ . . | ‘

.15! 5! I

.3“.

2.1.73.1“:

. ‘E I;
xvii.
i‘hﬂv‘

3.; Bl".
..1)\II.K.‘!|

b
in in 215,-. 3

iii»; 5....» . {7.5.2.1. .:

.v {inwdndwuiiixltitaiit

it; .

lint...

) r5341... 9!.

>1)... viii“.

9..

0:51.: u”! 5.3!... I’ldm...
‘98! .1... 5:35};

.8. t-.. $15.93 $1.573." In

3...: 3:93.)...» ).vs

. {7‘}! a}

\I‘

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

DYNAMIC USER EXPERIENCE OF INFORMATION
TECHNOLOGY INNOVATIONS: A SELF- REGULATORY
PERSPECTIVE

presented by

Chi Cong Dang

has been accepted towards fulfillment
of the requirements for the

doctoral degree in psychology

 

 

WWZM

/7Major Professor’ ﬁgnature

1 1/18/2009

 

Date

MSU is an Affirmative Action/Equal Opportunity Employer

 

OI-I-u-u-n-----u-o—--o-u----a-I-—o--o--—o-n-.-.-.—--.—-.-o-I-o-o-.-I-ICO-I-c-u-o-u-I-o-n-c-u—o---—~—-—

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Aoc&Pres/CIRCIDatoDue.indd

 

DYNAMIC USER EXPERIENCE OF INFORMATION TECHNOLOGY
INNOVATIONS: A SELF- REGULATORY PERSPECTIVE

By

Chi Cong Dang

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Psychology

2009

ABSTRACT
DYNAMIC USER EXPERIENCE OF INFORMATION TECHNOLOGY
INNOVATIONS: A SELF- REGULATORY PERSPECTIVE
By
Chi Cong Dang

In this dissertation, a model of within-person dynamic experience of information
technology (IT) innovations that was built on social cognitive theory was presented. Data
from an eight-week study that sampled 58 users’ daily experience of Palm Pilot software
as an IT innovation were analyzed by HLM and regression techniques. Results showed
that innovation self-efﬁcacy and outcome expectation affected and were affected by IT
usage dynamically within persons. The effect of innovation self-efﬁcacy and outcome
expectation on usage operated through behavioral intention and implementation
intention, whereas the inﬂuence of usage frequency on innovation self-efﬁcacy and
outcome expectation were partially mediated through efficacy disconﬁrmation and
outcome disconﬁrmation, respectively. Signiﬁcant individual differences were observed

in within-person processes. Research and practical implications were discussed.

ACKNOWLEDGEMENTS

This dissertation was completed with the caring support and assistance of many
individuals. I am very thankful for their contributions and encouragement over the years.
My deep gratitude goes to Dr. Daniel Ilgen, my main advisor on this dissertation and
throughout graduate school years, for his guidance, assistance, and encouragement. I
greatly appreciate his caring advice, hard work, and ﬁnancial support, without which this
research would not have been possible. Dan has been a great model on both professional
and personal levels. His teaching and caring will be with me for many years to come,
and I. sincerely hope to use many learned lessons for greater purposes.

I would like to thank Drs. Richard DeShon, Kevin Ford, Remus Ilies, and Linda
Jackson for their valuable comments. Their feedback has been very valuable and helpful
in producing this work on ﬁrmer theoretical and analytical grounds. Many thanks go to
Jay Park, Gordon Schmidt, and numerous undergraduate students who have assisted in
the data collection and coding process as well as Marcy Schafer and Mark Yampanis,
who helped formating and editing this dissertation

I would like to express my gratitude to Michigan Statement University for its
ﬁnancial support and the wonderful Main Library, where I have spent countless hours
and considered a second home. I also want to thank all the faculty in the Organizational
Psychology Program at Michigan State for their support, lessons, and gentleness that they
have shared during my graduate school.

On a more personal level, my heart-felt gratitude goes to my parents and siblings
for their encouragement and steadfast support. They have given me roots and encouraged

me to fly far and high with the knowledge that they are always there in time of needs. I

iii

am especially grateful to my wife, Maria Daane, for her unwavering support and
unconditional love. She encouraged me to follow my calling and embarked on this
journey with me at a great personal sacriﬁce. Her love and support have been strong and
steadfast through all the ups and downs, as we went through our graduate school together,
got married, and had our two young children, Julia and Jonathan. In this long, exciting
journey, I ﬁnd it very comforting and reassuring to have such a wonderful companion.
Now as this joumey is drawing to a close, I sincerely hope that it will open up other

journeys that we both will enjoy together with our two children.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................. vi
LIST OF FIGURES ........................................................................... vii
INTRODUCTION .............................................................................. 1
Review of Extant IT Innovation Adoption Literature ................................. 4
Theory of Reasoned Action and Theory of Planned Behavior ....... 4
Technology Adoption Model ............................................. 11
Social Cognitive Theory .................................................... 19
Strengths of Extant Research ...................................................... 23
Limitations of Extant Research .......................................... 24
Self-Regulated Use ofIT Innovations: A Dynamic Perspective. . . . ........29
Behavioral Intention ........................................................ 33
Implementation Intention .................................................. 34
Self-Efﬁcacy and Eﬂicacy Disconﬁrmation ............................. 38
Outcome Expectation and Outcome Disconﬁrmation ............... 47
METHODS .................................................................................... 53
Overview of the research .................................................................. 53
Sample ........................................................................................ 53
IT Innovation ................................................................................. 54
Training ...................................................................................... 55
Procedure ..................................................................................... 58
Measures ..................................................................................... 60
Time .......................................................................... 60
Outcome Expectation ...................................................... 60
Innovation Self-eﬁ‘icacy .................................................. 60
Behavioral Intention ..................................................... 61
Implementation Intention ............................................... 61
Outcome Disconﬁrmation ................................................. 61
Efﬁcacy Disconﬁrmation ................................................ . 61
IT Usage ...................................................................... 62
DATA ANALYSIS .......................................................................... 63
RESULTS ...................................................................................... 65
DISCUSSION .................................................................................. 90
Summary of Findings ....................................................................... 90
Research Contributions and Implications ................................................ 93
Practical Implications ....................................................................... 98
Boundary Conditions and Future Research .............................................. 101

Conclusions ................................................................................... 103

APPENDICES ................................................................................. 105

BIBLIOGRAPHY .............................................................................. 155

vi

Table 1.

Table 2.

Table 3.

Table 4.

Table 5.

Table 6.

Table 7.

Table 8.

Table 9.

Table 10.

Table l 1.

LIST OF TABLES

Descriptive Statistics of Studied Variables ................................. 66

HLM Analyses of Effects of Behavioral Intention and Implementation
Intention on Usage Frequency ................................................................. 70
HLM Analyses of Effects of Outcome Expectation and Innovation Self-
Efﬁcacy on Behavioral Intention .............................................. 73
HLM Analyses of Effects of Outcome Expectation and Innovation
Self-Efﬁcacy on Implementation Intention .................................. 75
HLM Analyses of Effects of Outcome Expectation and Innovation Self-
Eﬂicacy on Usage Frequency .................................................. 77
HLM Analyses of Effects of Outcome Expectation and Innovation Self-
Efficacy on Usage Frequency over Time ..................................... 79
HLM Analyses of Innovation Self-Efﬁcacy Growth over Time ........... 81
HLM Analyses of Effects of Usage Frequency and Efficacy
Disconﬁrmation on Innovation Self-Eﬂicacy ............................... 83
HLM Analyses of Effects of Usage Frequency on
Efﬁcacy Disconﬁrmation ....................................................... 85
HLM Analyses of Effects of Usage Frequency and Outcome
Disconfirmation on Outcome expectation ...................................... 87
HLM Analyses of Effects of Usage Frequency on Outcome

Disconﬁrmation .................................................................. .89

vii

LIST OF FIGURES

Figure 1. Theory of Planned Behavior ...................................................... 7
Figure 2. Technology Adoption Model .................................................... 12
Figure 3: Traditional Model of IT Innovation Adoption Research ...................... 25

Figure 4: Testable Self-Regulatory Model of Dynamic User Experience of
IT Innovations ........................................................................................... 31

Figure 5. Hypothesized Self-Regulatory Model of Dynamic User Experience of
IT Innovations ..................................................................... 67

Figure 6. Revised Self-Regulatory Model of Dynamic User Experience of IT
Innovauons ........................................................................ 91

viii

INTRODUCTION

Organizations have increasingly invested in information technology (IT) in the
past decade. American businesses are estimated to spend over 50% of their capital budget
on IT investments (Rockart, Earl, & Ross, 1996). Organizations may invest in IT to
improve workers’ productivity (Dewett & Jones, 2001), enhance the quality of work life
(Curley & Pyburn, 1982)), improve decision-making (Dewett & Jones), or deliver
superior products and customer services (Ives & Learmonth, 1984).

Increasing investments in and implementation of IT innovations, however, often
do not result in the degree of beneﬁt for which they are intended (Stratopoulos &
Dehning, 2000). The failure has been known as the “productivity paradox” (Landauer,
1995; Sichel, I997; Stratopoulos & Dehning, 2000), a phenomenon in which increased
implementation of IT innovations in the workplace is not associated with increased
organizational performance. For example, it is estimated that $300 billion were invested
in implementing enterprise resource planning (ERP) worldwide in the 19905 (James &
Wood, 2000). However, only about 50% of ERP implementations met organizations’
expectations (Adam & O’Doherty, 2003).

A number of scholars (e.g., de Vries, Midden, & Bouwhuis, 2003; Muir, 1994;
Will, 199!) suggested that the disappointing performance is not due to mechanical
malfunctions or poor design. Often, it is users’ underutilization of innovations that are
responsible for the lack of improvement (Johansen & Swigart, 1996; Moore, 1991;
Norman, 1993; Weiner, 1993). Most users employ only a small number of features out of
many available in IT innovations and use them infrequently (J asperson, Carter, & Zmud,

2005; Ross & Weill, 2002). As the beneﬁts of IT innovations are contingent on the

frequency and duration of their usage, it is important to understand what influences the
frequency and duration with which individuals use IT innovations.

Given the importance of IT in organizations, a signiﬁcant amount of research
effort has been devoted to understanding the psychological processes that explain IT
innovation usage (Legris, Ingham, & Collerette, 2003). Most of the existing studies have
relied on the Theory of Planned Behavior (TPB; Ajzen, 1991; Ajzen & F ishbein, 1977),
the Technology Acceptance Model (TAM; Davis, 1989), and the Social Cognitive
Theory (SCT; Bandura, 1997), with their belief and expectation constructs to understand
users’ behaviors. Existing research has made signiﬁcant contributions in explaining
intentions to use IT innovations. For example, Venkatesh, Morris, Davis, and Davis,
(2003) reported the TAM-based model accounted for 70% of variance in usage intention.
Yet researchers have predicted usage behavior less successfully, explaining around 30%
of self-reported usage.

Despite its contributions, existing IT adoption research has a number of important
limitations. First, most studies have used weak criteria, examining general behavioral
intention to use IT innovations and/or self-reported usage frequency and duration. In a
review, Legris et a1. (2003) reported that only one out of 22 studies measured usage
frequency objectively. The second limitation is that most studies hold the view that
behavioral intention is the only proximal determinant of IT usage. This assumption is
incompatible with our current knowledge. Although behavioral intention is an important
precursor to behaviors, it is often not automatically translated into actions. The path from
intention to behavior is often long. People need implementation plans and cognitive

resources to enact their behavioral intention (Gollwitzer, 1999). Additionally, they often

face many distractions and competing thoughts. They need to shield themselves from
distracting alternatives and focus attention on intended activities. It is thus not surprising
that behavioral intention predicts self-reported IT usage only modestly.

The third limitation, perhaps most importantly, is that past research has largely
i gnorcd the temporal dimension of IT usage. Beliefs and expectations are not ﬁxed; they
change over time. Beliefs and expectations regarding IT innovation usage are maybe
even more malleable (MeLone, 1990), as they are attached to IT innovations, with which
users are unfamiliar and may not have prior expectations. Additionally, usage behaviors
change over time (Rogers, 1995). They wax and wane. Yet, existing research has not
addressed the dynamic change in IT expectations and usage despite their realities in user
expenence.

In the present study, I drew on self-regulation theory from the social cognitive
perspective (Bandura, 1997; Zimmerman, 2000) to build a model to explain the dynamic
usage of IT innovations and evaluate its validity. The model examined how beliefs
influence and are influenced by usage behavior iteratively over time. Additionally, the
model incorporates implementation intention as a second proximal determinant of usage
behaviors, beside behavioral intention. The present work focused on predicting actual
usage behaviors by using objective measures of IT usage frequency and duration.

To examine the dynamic nature of IT usage, this research employed a longitudinal
design in which data were collected at multiple points in time. Palm Pilots were used as
an IT innovation as well as a measuring device through which longitudinal research data
were collected. Palm Pilots were chosen for a number of reasons. First, because Palm

Pilots are relatively small and increasingly becoming a more integrated device that can

perform many functions (email, phone, voice-commanded computing, etc. . .), it is a
computing device that will. be used more often in the future. Second, Palm Pilots have a
log application that is able to record usage behavior objectively over time. Finally, Palm
Pilots can administer and collect process measures such as intentions and beliefs on a
daily basis.

In the remainder of the presentation, IT innovation adoption literature is reviewed.
A model of dynamic user experience of IT innovation based on the self-regulation
literature (Bandura 1997; Zimmerman, 2000) is then presented. The methods for
conducting an empirical test of the model are discussed next. Finally, the research results
are presented and discussed.
Review of Extant I T Innovation Adoption Literature

Research has largely relied on the Theory of Reasoned Actions (TRA; Ajzen &
Fishbein, 1977), the Theory of Planned Behavior (TPB; Ajzen, 1991), the Technology
Adoption Model (TAM; Davis, 1989), and the Social Cognitive Theory (SCT; Bandura,
1997) to investigate the psychological process that explains IT innovation adoption. The
following sections reviewed this body of research. I brieﬂy describe each theory and
highlighted some key studies from each perspective. I then evaluate the strengths and
limitations of the existing research.

T heory of Reasoned Action and Theory of Planned Behavior

Theory of Reasoned Action (Ajzen, 1991; Ajzen & Fishbein, 1977) is a well-
validated theory of human behavior in social psychology. Its central tenet is that
behaviors are governed by the actors’ conscious intentions and beliefs about behaviors’

consequences. Two sets of beliefs—attitude toward and subjective norm regarding a

behavior—are assumed to be key forces that drive the behavior in question. Attitude
toward a behavior is deﬁned as a favorable or unfavorable evaluation of the behavior,
reflecting the actor’s subjective perception of consequences of the behavior. The theory
assumes that humans are motivated to seek pleasure and avoid pain. It, thus, predicts that
positive attitude toward a behavior leads to a higher likelihood of exhibiting that
behavior, whereas negative attitude is associated with a lower likelihood. In turn, attitude
toward a behavior is determined by the actor’s salient beliefs about the probability that
the behavior is associated with speciﬁc consequences and the subjective evaluations of
those consequences.

Subjective norm regarding a behavior reﬂects the social nature of beliefs. It refers
to perceived social pressure to perform the behavior and thus reﬂect the actor’s subjective
beliefs of behavioral consequences imposed by the social environment. Approving norm
is predicted to lead to the intentions to perform the behavior, whereas disapproving one
tends to inhibit it. The theory also assumes that subjective norm is inﬂuenced by the
actor’s salient normative beliefs and their motivation to comply with those beliefs.

Of all key expectations, the TRA assumes that the actor’s intention toward a
behavior, behavioral intention, is the most proximal determinant of the behavior of
interest. Behavioral intention is viewed as the actor’s motivation to exhibit the behavior.
It inﬂuences the behavioral choice, the level of effort, and the persistence that an actor
expresses. Strong intention toward displaying a behavior is predicted to lead to a higher
likelihood of engaging in the behavior. Behavioral intention is assumed to be jointly
determined by attitude toward and subjective norm regarding the behavior. It is also

expected to fully mediate the effects of these two variables on actual behaviors.

The TRA works well to explain behaviors when the behaviors in question are
under the actor’s volitional control, the condition in which the actor is reasonably able to
execute his or her intentions (Ajzen, 1991). However, when behaviors are beyond the
actor’s volitional control, the relationships between behavioral intention and behaviors
are weakened because the actor’s ability to execute the intended behaviors is constrained.

To extend the TRA to explain behaviors in situations in which volitional control
is limited, Ajzen (1991) added a new construct, called perceived behavioral control, to
the model and changed the name of the theory to the Theory of Planned Behavior (See
Figure l ). Perceived behavioral control (PBC) closely resembles Bandura’s (1986) self—
efﬁcacv construct (Ajzen, 1991). In fact, Ajzen suggested that much of what we know
about perceived behavioral control is from Bandura’s research on self—eﬁ’icacy.
According to the TPB, perceived behavioral control (deﬁned as perception of ease or
difﬁculty of performing a given behavior) inﬂuences the behavior directly and through
behavioral intention. Ajzen offered two rationales for the direct effect of perceived
behavioral control on actual behavior. First, perceived behavioral control often reﬂects
actual behavioral control and thus will explain actual behavior over and above behavioral
intention. Second, if intention or motivation to perform a behavior is held constant,
greater conﬁdence in one’s ability or perception of a high level of behavioral control is
more likely to bring the effort of performing the behavior to successful ends. The theory
further assumes that perceived behavioral control is determined by more salient control
beliefs and perceived probabilities that those beliefs control the behavior.

A number of studies have used the TRA and the TPB framework to study IT

innovation adoption. Davis, Bagozzi, and Warshaw (1989) were some of the ﬁrst

STE—Em .5555 he E85. 3 2:3,.—

 

 

 

SSS—om

 

 

 

 

83:85
ESSEom

 

 

35:00
_Eo_>mnom
838qu

 

 

 

 

 

 

A

 

 

 

Eo Z
958 Sam

whom—on 35:8
mo mcoumﬁgm
ﬂozon SEED

 

 

 

 

 

 

 

Sauce.

 

bmﬁoo
8 839502
£23m “Sanchez

 

 

 

 

 

e28 assess
mo mcocmﬁgm
e23 353:2

 

 

 

 

 

researchers to apply the TRA in IT innovation adoption research. In the study, the authors
examined usage of a word processor among 107 MBA students. At the beginning of a
semester, students were given one-hour of training on how to use the word processor.
They were then given a questionnaire measuring the TRA variables. The same
questionnaire with the addition of a two-item self-reported measure of the frequency of
their usage during the semester was administered at the end of the semester (14 weeks
later). Behavioral intention, attitude, and subjective norm regarding the use of the word
processor were operationalized in accordance to Ajzen and Fishbein’s (1977)
recommendations. This entailed probing users of the word processor to provide salient
beliefs regarding the beneﬁts and norms associated with use of the word processor. Scale
items were then framed with reference to a speciﬁc target (the word processor), action
(usage), and context (the MBA program). Noticeably, subjective norm was measured
with only one item. Results showed that attitude toward using the word processor was
independently related to behavioral intention at both times ([3 = .55 at Time I and .48 at
Time 2), but subjective norm was not (B = .07 at Time 1 and .10 at Time 2). Behavioral
intention at Time 1 was moderately related to self-reported usage frequency at Time 2 (r
= .35), and behavioral intention measured at Time 2 was strongly related to usage
frequency at Time 2 (r = .63).

In another study that relied on the TRA/TPB framework, Taylor and Todd (1995)
assessed computer usage in a computer resource center at a university among 786
potential student users. They followed Ajzen and Fishbein’s (1977) guidelines in
designing scales and measured attitude, subjective norm, perceived behavioral control,

and behavioral intention regarding the use of the center a month after the semester

started. The frequency and duration of usage by student users, who also reported the
number of projects they worked on in every visit to the computer resource center, were
tracked and recorded by the center attendants. The authors used structural equation
modeling to analyze the data. Frequency, duration, and number of worked projects were
used to estimate usage as a latent variable, although no rationale was offered to explain
why the number of worked projects was viewed as an indicator of computer usage (My
guess is that they needed a third indicator in order to estimate a latent factor.) The rest of
the constructs were modeled as manifest variables (no reason was offered to explain why
they were not modeled as latent variables as well.) Results showed that attitude,
subjective norm, and perceived behavioral control had independent positive effects on
behavioral intention, explaining 57% of its variance. Behavioral intention and perceived
behavioral control independently affected usage behavior, accounting for 36% of usage
variance.

In another study, Harrison (1997) studied how well the TPB predicted behavioral
intention to adopt new IT innovations among 97 managers who played substantial roles
in making adoption decisions for their own ﬁrms. IT innovations in this study were
considered as new IT systems that their ﬁrms would like to adopt to help them compete
in the market. Following recommendations by Ajzen and Fishbein (1977), the authors
developed scale items and measured attitude, subjective norm, and perceived behavioral
control in an elaborate process. Like other studies, results showed that attitude, subjective
norm, and perceived behavioral control had independent effects on behavioral intention
to adopt the IT innovation. Attitude accounted for an increment of 27% of variance;

subjective norm, 9%; and perceived behavioral control, 4%.

In a ﬁeld study, Chau and Hu (2001) examined the effects ofthe TPB constructs
on behavioral intention to use telemedicine technology in a sample of 408 physicians in
Hong Kong. Telemedicine was deﬁned as the use ofinformation and telecommunication
technology to deliver healthcare service of needed expertise and information among
geographically disperse populations of patients and physicians. The researchers used the
measures developed by Taylor and Todd (1995) to assess the TPB constructs. They found
that attitude (B = .63) and perceived behavioral control (13 = .22) had independent
positive relationships with behavioral intention, whereas subjective norm did not.
Together, the TPB constructs accounted for 32% of variance in behavioral intention.

Taken together, evidence from previously discussed studies suggests that
constructs in the TRA and the TPB explain IT adoption and self-reported time usage
reasonably well. The independent effects of attitude and perceived behavioral control are
consistent across studies. The independent effect of subjective norm, however, was less
consistent. Perhaps, the nature of social settings may explain the presence or absence of
the independent, direct effect of subjective norm. In some technology adoption
situations, workers are required to adopt new innovations. They may use innovations in
order to comply with mandates from their supervisors, rather than because of their own
feelings and beliefs. In those settings, subjective norm is likely to have a direct effect on
innovation usage, independent of the effect of attitude. In other situations, users adopted
IT innovations on a voluntary basis, as they could choose whether or not to adopt the
innovations (Davis et al., 1989). In those settings, the effect of subjective norm is more
likely to be fully channeled through attitude as attitude represents a set of beliefs and

feelings users have about innovations.

10

T echnologv Adoption Model

Widely used in technology adoption research, the TAM (Davis, 1989) is derived
from the TRA and is considered as an application of the TRA to explain adoption
behavior (Figure 2). The TRA explains behaviors reasonably well, but it requires
researchers to construct customized scales to measure salient beliefs that are speciﬁc to
the IT innovations in question. To get around this problem, Davis developed the TAM
with the goal of providing a small set of internal beliefs, attitude, and intentions and a set
of common generic measures that can be used across technologies and settings to explain
adoption behaviors. Currently, the TAM is the dominant and most popular theory in IT
innovation research, as the majority of existing studies have relied on the TAM
framework (Legris et al., 2003; Venkatesh, 2000; Venkatesh, Davis, & Morris, 2007).

According to the TAM, perceived usefulness and ease of use are two key beliefs
that inﬂuence adoption behaviors. Perceived usefulness refers to the degree to which
using an IT innovation would enhance users’ outcomes (Davis, 1989; Davis et al., 1989).
It is predicted to inﬂuence adoption behavior directly and through attitude toward
adoption behavior. According to Davis et a1.(1989), perceived usefulness has a positive
inﬂuence on attitude toward adoption behaviors because valued outcomes (perceived
usefulness), through associative Ieaming and cognitive-affective consistency
mechanisms, become associated with positive affect toward the means that users utilize
to obtain those outcomes (attitude toward adoption behaviors). In addition to its effect on

attitude, perceived usefulness also affects adoption behaviors directly. Davis et a1. (1989)

11

.232 gamete. hue—258a. "N «.53...—

 

 

 

xocozcoum
omamD

 

 

 

:oEBE
_So_>mnom

 

33 mo ommm

 

 

 

 

 

bungee.

 

 

 

 

 

 

a $0533:
woﬁooaom

 

 

 

12

suggests that the direct effect happens because the cognitive decision to enhance
performance and outcomes is the key force that drives user’s intentions to engage in
certain behaviors. This inﬂuencing process sometimes occurs without fully activating the
affect associated with the behavior-reward association (Davis et al., 1989). Speciﬁcally,
in the context of technology adoption, people make decisions to carry out usage
behaviors by recognizing the association between behaviors and outcomes. They do not
always evaluate how these outcomes meet their hierarchy of goals and needs and thus do
not fully activate the subjective affect associated with such outcomes every time. In other
words, people form intentions toward adoption behaviors they believe will increase their
perfomiance, and the effect of such belief is over-and-above users’ feelings or attitudes
that are associated with the behaviors (Davis et al., 1989).

Ease of use is deﬁned as the degree to which users perceive a particular IT
innovation to be free of effort (Davis et al., 1989). By deﬁnition, it is similar to perceived
behavioral control in the TPB. Yet, at the time of its conception, Davis et a1. were not
aware of the relationship, because F ishbein and Ajzen had not introduced the concept of
perceived behavioral control. Davis et al. assumed that ease of use inﬂuences intention
and behavior through two mechanisms. First, ease of use has an instrumental effect.
When a product is easy to use, user’s effort saved due to increased ease of use may be
redeployed, and it enables a person to accomplish more work. Ease of use thus
contributes directly to perceived usefulness (Davis et al., 1989). Second, ease of use
inﬂuences the sense of self-efﬁcacy regarding a user’s ability to carry out behaviors
needed to operate the innovation in question (Davis et al., 1989). High levels of ease of

use tends to lead to a heightened sense of self-efﬁcacy. Because self-efficacy is thought to

13

operate autonomously from instrumental determinants of behaviors due to inborn drives
for competence and self-determination (Bandura, 1986), ease of use is thus posited to
inﬂuence behavioral intention directly, over and above perceived usefulness.

Differing from the TRA, subjective norm is not included in the TAM. Subjective
norm is thought to inﬂuence behaviors directly through compliance processes or
indirectly via attitudes through the internalization and identiﬁcation process. In the
context of IT adoption, decisions to adopt and use IT innovations by managers and
professionals is voluntary in many cases. In those cases, subjective norm is thought to
inﬂuence adoption behaviors through the process of internalization and identiﬁcation.
Only in a small number of cases, workers may use IT innovations in order to comply with
mandates from their supervisors; and the compliance process is thought to be the main
mechanism to drive social norm’s effect. Yet, standard measures of subjective norm do
not differentiate compliance from internalization and identiﬁcation (Davis et al., 1989). It
is, thus. difﬁcult to disentangle the direct effects of subjective norm on behavioral
intention from indirect effects via attitude. Because of subjective norm’s uncertain
theoretical and psychometric status, it is excluded from the TAM (Davis et al., 1989).

In the initial formulation of the TAM, attitude was thought to mediate the effect
of perceived usefulness partially and ease of use fully. However, it was dropped from the
model later for a number of reasons (Davis et al., 1989; Venkatesh & Davis, 2000). First,
excluding attitude improved the parsimony of the model. Second, attitude only partially
mediated of the effect of perceived usefulness on behavioral intention. Third, the
independent relationship between attitude and behavioral intention was not consistent

when perceived usefulness was included in the model (Davis et al., 1989; Taylor & Todd,

14

1995; Venkatesh & Davis, 2000). Finally, there was a strong link between perceived
usefulness and behavioral intention (Davis, 1989; Davis et al., 1989; Venkatesh, 2000;
Venkatesh & Davis, 2000).

To help evaluate the TAM, Davis (1989) developed generic measures of ease of
use and perceived usefulness that can be used across a variety of IT. The measures
contained four items each and are found to be psychometrically sound. Factor analysis of
scale items invariably yielded two factor solutions (e. g., Venkatesh, 2000; Venkatesh &
Davis, 2000). Yet these two factors were not independent. In a series of studies, Davis
(1989) reported their correlation ranged from .10 to .60. The scales of ease of use and
perceived useﬁilness were internally reliable (Chin & Gopal, 1995; Davis; Davis et al.,
1989; Mathieson, 1991). Legris et al. (2003) reported that coefﬁcient alphas for perceived
useﬂilness exceeded .83 in 22 studies they reviewed, whereas coefﬁcient alphas for ease
of use were greater than 0.79. The availability of valid generic measures is an important
contribution of the TAM because it allows researchers to compare results across studies
and supports cumulative theoretical development. In addition, the availability of the
scales may also explain why the TAM has become increasingly popular among
researchers.

The majority of existing studies have used the TAM to examine IT innovation
usage and adoption. Most evaluated the effects of ease of use and usefulness on
behavioral intention. For example, Venkatesh (2000) examined the validity of the TAM
in four samples of 246 workers from three different organizations in which new IT
innovations were recently introduced (58 sales clerks using online help desk software in a

retail electronic store, 41 workers and 104 workers of two different branches in the same

15

real estate company using new property management software, 43 employees switching
to the Windows Operating System in a ﬁnancial service ﬁrm). All research participants
were trained to use the software after the innovations were adopted by the organizations.
Perceived usefulness, ease of use, and behavioral intention regarding the use of the new
software were assessed shortly after the training (time 1), one month (time 2), and three
months after the training (time 3). In his analysis, Venkatesh pooled data across samples
and found that perceived usefulness (13 = .49, .52, .54, at time 1, 2, and 3, respectively)
and ease ofuse (B = .29, .20, .17, respectively) independently predicted behavioral
intention at each of the three measurements points. Together, ease of use and perceived
usefulness accounted for about 35% of variance in behavioral intention.

The relationships of ease of use and perceived usefulness with behavioral
intention were also found to be consistent across many other studies. In a review, Legris
et al. ( 2003) examined 22 studies that yielded 28 possible relations between each of the
TAM constructs and behavioral intention. Out of 19 reported relations, perceived
usefulness had a positive relation to behavioral intention 16 times and a non-signiﬁcant
relation three times. Ease of use was not related to behavioral intention 3 times and was
signiﬁcantly related 10 times out of the 13 reported times. It should be noted these
relations refer to partial correlations that studies reported in the results of their multiple
regression and structural equation modeling analyses. Meta-analysis for estimating
overall effect sizes of the relations between the TAM constructs and behavioral intention
could not be conducted because most studies in IT adoption research did not provide the
zero-order correlation matrix. Out of 22 studies reviewed, correlation matrix were

provided in only three studies (Legris et al., 2003).

16

In studies where IT usage was investigated (e. g., Taylor & Todd, 1995; Venkatesh
& Davis, 2000), the vast majority of the studies assessed usage through self-report
measures with one to three items asking users for the frequency and/or duration of their
IT usage. Generally, the effect of perceived usefulness on self-reported usage was found
to be more consistent, whereas the inﬂuence of ease of use was not. For example, in
Davis et al. (1989)’s study discussed earlier, where MBA students’ use of word
processing was studied, results showed that self-reported usage at the end of the semester
was related to perceived usefulness (r = .65) and ease of use (r = .27) measured at the
beginning of the semester. Perceived usefulness and ease of use also independently
predicted self-reported usage frequency and together accounted for 45% of its variance.
However, when measured together to predict self-reported usage at the end of the
semester, perceived usefulness was correlated and independently accounted for variance
in self-reported usage frequency, whereas ease of use was neither related nor accounted
for unique variance.

In another study, Davis (1989) surveyed 112 Canadian IBM workers to assess
their beliefs and self-reported past usage of an email software and a text editor for which
they had a six month usage experience. One hundred nine workers were users of the
email software, and 75 were users of the text editor. Results showed that perceived
usefulness and ease of use were correlated with self-reported usage in both samples.
However, only perceived usefulness accounted for unique variance of self-reported past
usage. Together they accounted for 31% of variance in email usage and 46% of variance
in text editor usage. In another study, Davis (1989) also found similar results. In this

research, 40 evening MBA students were trained to use two different graphic software

17

applications, C hart-Master and Pendraw, for an hour. After the training, the students
answered questions regarding their perceptions of the perceived usefulness and ease of
use of the applications. They were then asked to predict their future use of the two
software packages. Results showed that both perceived usefulness and ease of use were
correlated with reported usage of either software, but only perceived usefulness
accounted for unique variance.

In a review, Legris et al. (2003) reported that out of 28 possible relations, the
relations between perceived usefulness and self-reported usage were reported 13 times.
Perceived usefulness had a positive relation with self-reported usage eight times and a
non-signiﬁcant relationship ﬁve times. Out of nine reported relations, ease of use was not
related to self-reported usage ﬁve times and positively related four times. The overall
effect sizes could not be estimated as most studies in IT adoption literature did not
provide correlation matrix.

Attempting to extend TAM, a number of studies have examined contingency
factors that moderate the effects of TAM constructs. Most studies reported the
moderating effect of age, gender, and experience on relationships between TAM
constructs and behavioral intention (e.g., Venkatesh, Morris, Davis, & Davis, 2003;
Morris, Venkatesh, & Ackerrnan, 2005). The effect of perceived usefulness on
behavioral intention was found to be stronger among men and young workers, whereas
the effect of ease of use was stronger among women, older workers, and those research
participants with little experience (e. g, Venkatesh, Morris, Davis, & Davis, 2003; Morris,
Venkatesh, & Ackerrnan, 2005). Most studies found non-signiﬁcant moderating effects

when IT usage was used as the criterion (e. g., Venkatesh, Morris, Davis, & Davis, 2003;

18

Morris, Venkatesh, & Ackerman, 2005). Only one study by Venkatesh, Brown,
Maruping, and Bala (2008) reported that experience moderated the effect of behavioral
intention on IT usage such that the effect was stronger with increased level of experience.

In sum, TAM predicts behavioral intention to adopt well. Ease of use and
perceived usefulness were independently related to behavioral intention to use IT
innovations, and together they accounted for a large portion of its variance. However,
TAM were less successful in predicting IT innovation usage. The effect of perceived
usefulness on self-reported usage was consistent, whereas the inﬂuence of ease of use
was not (Legris et al., 2003). As most studies have examined IT usage using subjective
measures, the validity of TAM in explaining actual usage is largely unknown.
Social Cognitive Theory

At the present time, the SCT (Bandura, 1986) is a popular model that is used to
explain individual behavior. It is a broad theory that includes many factors to explain the
dynamic interactions between individuals and their environment. The theory holds that
individuals’ internal processes such as cognition and affect interact reciprocally and
dynamically with overt behaviors and the external environment. Individuals choose the
environment in which to reside and are inﬂuenced by their environment. Behaviors in a
given situation are affected by situational characteristics, which are, in turn, inﬂuenced
by behavior. Behaviors are also inﬂuenced by cognition and affect, which, in turn, are
also affected by them. The detailed processes of the dynamic, reciprocal interactions
between individuals and their environment are complex. Instead of articulating and
applying the SCT in its entirety, researchers in IT adoption literature (e.g., Compeau &

Higgins, 1995; Compeau, Higgins, & Huff, 1999; Hill et al., 1987) took a narrow

l9

approach by focusing on key cognitive constructs in the SCT to explain adoption
behaxiors.

According to the SCT, two key beliefs drive human behaviors. The ﬁrst, outcome
expectation, is expectations regarding outcomes associated with behaviors. Individuals
are more likely to undertake behaviors they believe will result in valued or favorable
outcomes and refrain from those that they think will lead to unfavorable consequences.
Outcome expectation is similar to the construct of pereived usefulness in the TAM
(Davis, 1989; Davis et al., 1989) and is highly related to the concept of attitude in the
TPB framework (Taylor & Todd, 1995).

The second belief, self-efficacy, refers to peoples’ judgment of their capabilities to
organize and execute courses of action required to attain designated types of performance
(Bandura, 1986). It is domain speciﬁc and strongly tied to the behavior of interest. Self-
ef/icaev is concerned not with the skills one has but also with the conﬁdence with which
one uses the skills. Self-efficacy is thought to inﬂuence behavioral choices, effort and
persistence, and task strategy that people employ.

As a construct, self-efﬁcacy closely resembles other constructs encountered in the
TPB and the TAM. It is similar to perceived behavioral control in the TPB (Ajzen,
1991), as Aj zen incorporated perceived behavioral control into the TPB on the basis of
Bandura’s work on self-efficacy. Self-efficacy is also highly related to ease of use in the
TAM (Davis et al., 1989). Davis et al.(1989) assumed that case of use has a direct effect
on attitude toward behaviors on the rationale that it inﬂuences self-efficacy.

Although self-efficacy is related to ease of use, Compeau and colleagues,

(Compeau & Higgins, 1995; Compeau et al.,1999) argued that conceptually it is a distinct

20

construct. Self-efﬁcacy places a stronger focus on individuals as users of technology. It is
thought to be largely driven by individual differences and their experience of technology
usage. In contrast, ease of use is more technology focused and largely determined by
objective technology characteristics. Ease of use represents a psychological cost of using
new technology (Compeau & Higgins; Compeau et al.), and thus it could be considered
as a subset of outcomes associated with technology use.

In addition to the distinctive nature of self-efﬁcacy, Compeau and Higgins (1995)
argued for its role in explaining IT innovation usage on theoretical grounds. If one doubts
one’s capability to successfully use IT innovations, expectations about positive outcomes
of IT usage will be meaningless. The inclusion of self-efficacy should, therefore, explain
additional variance in usage behavior. This theoretical position has important practical
implications. It suggests that IT innovation adoption is not just about convincing people
of the beneﬁts to be harnessed from a technology or improving its quality to make it very
easy to use. It is important to coach, train, and motivate users so that they have requisite
skills and conﬁdence in their ability to use IT innovations successfully.

In IT adoption research, only a few studies have used the SCT belief constructs to
explain 1T innovation usage. Hill et al. (1987) studied four samples of students totaling
157 males and 147 females who learned to use computers. They assessed self-eﬂicacy
regarding learning to use a computer by a four-item measure and outcome expectation by
another four-item measure. They then asked students to indicate their intentions to use a
computer in the future by answering two questions. In the following semester, 12 weeks
after the initial survey, the students were contacted and asked to report whether they

actually enrolled in computer or computer-related courses. In all four samples, structural

21

equation modeling analyses showed that behavioral intention was independently related
to outcome expectation and self-efficacy. In turn, behavioral intention predicted
participants’ actual enrollment in computer or computer-related courses.

In a cross-sectional survey of 1020 Canadian workers, Compeau and Higgins
(1995) developed and validated a 10-item measure of self-efﬁcacy. They found that self-
reported frequency of computer usage was related to self-efficacy (r = .45) and outcome
expectation (r = .41). Path analysis indicated that outcome expectation inﬂuenced self—
reported usage directly and indirectly through satisfaction with computer use, whereas
self-efﬁcaev affected usage directly and indirectly through computer satisfaction and
anxiety. Because the survey was cross sectional and thus does not allow inference about
causality, Compeau et a1. (1999) re-measured self-reported usage, computer satisfaction
and anxiety among people in the same sample a year later in a follow-up study. They then
used self-efﬁcacy and outcome expectation assessed the preceding year as predictors and
found a similar set of relationships between self-efﬁcacy, outcome expectation,
satisfaction, anxiety, and self-reported usage in a sample of 394 respondents. The results
are consistent with a causal interpretation but in a necessary, not sufﬁcient sense.

In sum, although a smaller number of studies have relied self-efﬁcacy and
outcome expectation constructs in social cognitive theory to explain IT innovation
adoption, results appear promising. Results showed outcome expectation and self-efﬁcacy
were independently related to behavioral intention and self-reported usage. Yet, these
few studies have exclusively relied on self reports of IT usage. Given the small number of

studies and the common method variance due to self-report measures, more research with

22

more rigorous measurements and research design is needed to ascertain the validity of
self-efﬁcacy and outcome expectation in predicting IT usage behaviors.
Strengths of Extant Research

The existing research has relied on well-supported theories of human behaviors,
namely the theory of planned behavior and social cognitive theory to study IT innovation
adoption. Studies based on the TPB and its derivative, the TAM, have shown that
perceived usefulness is consistently related to behavioral intention to use IT innovations
and self-reported usage frequency (e.g., Davis, 1989; Davis et al., 1989; Taylor & Todd,
1995). Similarly, a smaller number of studies that have relied on the SCT have found that
outcome expectation, the same construct as perceived usefulness but with a different
label, was related to behavioral intention (Hill et al., 1987) and self—reported IT usage
(C ompeau & Higgins, 1995; Compeau et al.,1999). Thus, research ﬁndings have
converged on the idea that instrumental beliefs, namely perceived usefulness or outcome
expectation, matter in people’s adoption and use of IT innovations.

Researchers also agree that controlling beliefs such as self-efﬁcacy, perceived
behavioral control, or ease of use also plays an important role. Studies that relied on SCT
have found that self-efﬁcacy had a positive relation with behavioral intention (Hill et al.,
198 7) and self-reported IT usage (Compeau & Higgins, 1995; Compeau et al., 1999).
Research that relied on TPB framework, used a similar construct, labeled perceived
he’llaI-‘lOl‘lll control, and similarly reported that perceived behavioral control had positive
effects on behavioral intention and self-reported usage behavior (Taylor & Todd, 1995).
Work that followed TAM showed ease of use, a related construct, may have an

independent effect on behavioral intention and IT usage. However, its effect has been

23

rather inconsistent (Legris et al., 2003). Conceptually, self—efﬁcacy and ease of use,
although related, are distinct constructs (Compeau & Higgins; Compeau et al.; Davis et
al ., 1989). Selfcﬂlcacy is more rooted in individual differences and user experience with

IT innovation usage, whereas ease of use is more driven by the characteristics of IT

innovations (Compeau & Higgins; Compeau et al.).

L inzitations of Extant Research

Although existing studies ofIT adoption place emphasis on different sets of
be] icfs that are responsible for IT usage, they generally follow the model outlined in
F i gure 3. Users’ beliefs regarding usage ofIT innovations are key psychological factors
th a t explain adoption and usage ofIT innovations. They are viewed as the bases on which
u s ers form their behavioral intention, which, in turn, determines frequency and duration

in vvhich they use IT innovations.

With such a model, extant research reveals a number of limitations. The ﬁrst

r Cga rds the weak criteria that they have investigated. Many studies (e. g., Harrison, 1997;
V e 11 katesh, 2000) in the literature have used behavioral intention as a key criterion to
eva I Liate their research models (Legris et al., 2003), on the grounds that behavioral

"’ 3' I 6” IMO]! is a strong correlate of usage behaviors. Yet, existing evidence shows that
h (’11 czvioral intention typically explains only around 30% of variance in self-reported

u -_ . . . . . . . ,
S :13 c behavrors. Much of variance in behavrors remains unexplained, maklng behavioral

tn 1* q); ztion a poor substitute for IT usage. Of those studies that directly examined usage
b ~ . .
Q h :avrors, almost all have used self-reported measures to assess IT usage (Legns et al.).

T ‘1Q employed self-report measures typically contained one to four items asking users to

\nd i cate how frequent and how long they used IT innovations. They asked research

24

aohaomom 5:325 3556::— H— E .252 3.55th ”m 95”:—

 

 

 

xocosvoi owme

 

 

 

55:35
ESSEom

 

 

 

 

 

mconﬁoomxm
Eozom

 

 

 

wEoBoBé
msocowoxm

 

 

 

 

 

25

participants to think retrospectively on how much time they have spent using IT
innovations (e.g., Compeau et al., 1999; Davis, 1989; Davis et al., 1989) and were often
administered at the same time as measures of beliefs (e. g., Compeau & Higgins, 1995;
Hu et al., 1995). This practice of using self-report measures has a number of limitations.
Self-report measures often do not capture actual behaviors accurately (Legris et al.;
Straub, Limayem, & Karahanna-Evaristo, 1995). When measured together with beliefs,
they produce only correlational data and tend to inﬂate the nature of relations due to close
temporal proximity and common methods variance.

The second limitation of existing research regards the assumption that behavioral
intention is the only proximal predictor of IT usage. In the TRA and the TAM and the
studies that were built on them, behavioral intention was treated as the only proximal
determinant of IT usage. The effects of antecedents of IT usage were viewed through
their inﬂuences solely on behavioral intention (e.g., Davis et al., 1989; Venkatesh, 2000).
However, the assumption that behavioral intention is the sole proximal determinant of
behaviors misses the fact that strong intention is not always expressed in behaviors.
Research has shown that the road between intention and performance is rather long
(Gollwitzer, 1999). Typically, behavioral intention is not automatically translated into
actions. People need implementation plans to enact their intentions. They often face
distractions and competing thoughts and at times cannot focus their attention on carrying
out their intentions. It is thus not surprising that the explained variance was modest when
behavioral intention was used to predict self-reported time usage, despite the inﬂated

nature of their relations due to measurement problems.

26

Third, perhaps most importantly, extant research has largely ignored the temporal
dimension of IT usage. Although researchers acknowledged that beliefs, intentions, and
usage behaviors change over time, most studies (e.g., Compeau & Higgins, 1995;
Harrison, Mykytyn, & Riemenschneider, 1997) have used cross-sectional designs to
study the relationships among those variables. Although a number of studies (e. g., Davis
et al., 1989; Venkatesh & Davis, 2000) have attempted to use longitudinal designs, they
also had limitations and have not addressed the dynamic usage of IT innovations. For
example, Davis et al.(1989) and Venkatesh and Davis (2000) measured beliefs and
behaviors at several points in time. Yet their analyses have focused on how beliefs and
behaviors were related at each time point. Such analyses served to validate and replicate
the hypothesized relationships. However, they did not address how beliefs and IT usage
change and inﬂuence one another dynamically over time.

Beliefs and expectations are not ﬁxed constructs. They change over time
(Bandura, 1997). Beliefs and expectations regarding the use of IT innovations are even
more susceptible to change (MeLone, 1990), as they are attached to usage of IT
innovations which are new to and for which users may not have prior expectations and
experience. IT usage behaviors also change over time. Users test out new IT innovations,
use it to carry out tasks, routinize their behaviors, and at some point discontinue using
them (Rogers, 1995). In other words, IT usage behaviors wax and wane.

Despite the reality of dynamic change in beliefs and usage behaviors, existing
research has not addressed this important phenomenon. The extant research has largely

relied on a feed-forward loop framework, as seen in the model in Figure 3, and most

27

studies sampled data at one or two points in time, making inferences about dynamic
change and reciprocal relationships between beliefs and usage behaviors impossible.

To address dynamic change of IT usage and its underlying beliefs and
expectations, we need to conceptualize and investigate change as occurring within
persons over time. The questions we should ask are: (a) how a person’s key beliefs
regarding IT innovations (self-eﬂicacy and outcome expectation) change over time, (b)
what happens to a person's usage behavior as their key beliefs change, (c) what accounts
for the changes in the key beliefs, and (d) whether these change processes are uniform
across persons. To answer those questions empirically, a longitudinal study to sample key
beliefs and usage behaviors within persons frequently over a long period of time is
required. Data analyses should focus on both within- and between-person level so that
they allow us to understand how dynamic process operates within each person as well as
the variation between people.

In order to capture users’ dynamic experience of IT innovation, the present study
attempted to move beyond current models by considering the effects of time and the
dynamic interplay between beliefs and behaviors. It presented a dynamic model in which
innovation self-efficacy and outcome expectation inﬂuence and are inﬂuenced by IT
usage dynamically over time. The study sampled users’ subjective beliefs and objective
behaviors frequently over an extended duration. Data were analyzed within persons as
well as between persons. A model of dynamic user experience of IT innovations was

elaborated next.

28

Self—Regulated User Experience of IT Innovations: A Dynamic Perspective

To address the dynamic changes of self—efficacy and outcome expectation and
their inﬂuence, I drew on self-regulation theory from a social cognitive perspective
(Bandura 1997; Zimmerman, 2000). Self-regulation theory (Zimmerman) can be viewed
as a subset of the broader SCT. It focuses on the relationships between covert internal
processes of cognitions and affect and overt behaviors. The central assumption is that
behaviors affect and are affected by cognitions and affect in a cyclical and reciprocal
manner. According to the theory, individuals regulate behaviors in a cycle of three
phases: forethoughts, performance control, and self-reﬂection. During forethoughts,
individuals engage in cognitive processes that precede efforts to act and set the stage for
action, in our case IT usage. They form deliberate intentions and set strategic plans to
achieve intended goals. Intention formation and strategic planning are inﬂuenced by
beliefs regarding one’s conﬁdence in his or her ability to execute the behaviors, which
Bandura labeled self-efficacy, as well as the expected outcomes from doing so. In
perfonnance control, individuals execute the strategic plans to enact their behavioral
intention. They attempt to control their motoric efforts by focusing attention and
engaging in different task strategies. The effectiveness of their actions is inﬂuenced by
the ability to control attention and the use of verbal self-instruction, imagery, and task
strategies. In self-reﬂection individuals react to executed behaviors. They evaluate their
actions, make causal attributions, experience satisfaction, and adapt their behaviors in a
systematic fashion. These self-reﬂections, in turn, inﬂuence forethoughts regarding

subsequent motoric efforts—thus completing a self-regulatory cycle.

29

With the self-regulation framework, I have developed a model, presented in
Figure 5, to explain users’ dynamic experience ofIT innovations. Consistent with the
self-regulation framework and previous IT innovation research, users’ behavioral
intention (Ajzen, 1991; Davis et al., 1989; Taylor & Todd, 1995) and implementation
intention (Gollwitzer, 1999) are proposed to be the most proximal determinants of IT
usage behaviors. Yet these conscious thoughts reﬂect more distal beliefs regarding the IT
innovation. namely innovation self-efficacji’ (e.g., Compeau et al., 1999; Hill et al., 1987)
and outcome expectation (Compeau et al.; Davis, 1989; Davis et al.; Hill et al.; Venkatesh
& Davis, 2000). Under the inﬂuence of behavioral intention and implementation plans,
individuals carry out actions in interacting with IT innovations. Their behaviors are
manifested in the frequency and duration with which they use the innovations (Brancheau
& Brown, 1993; Igbaria, Pauri, & Huff, 1989; Lee, 1986). According to the model, once
users carry out their intentions and go through (or don’t go through) their interactions
with IT innovations, they reﬂect on their experience. They evaluate their experience of
the IT innovations to determine whether it meets their previous beliefs. Speciﬁcally,
evaluations that one’s ability to use the innovations surpasses his or her self-efficacy
(positive efficacy disconﬁrmation) encourage the upward revision of self-eﬂicacy
(Bandura, 1997). In contrast, judgments that one’s ability to use the innovations falls
short of his or her self—efficacy (negative eﬂicacy disconﬁrmation) leads to a downward
revision of self-efficacy (Bandura). Similarly, evaluations that one’s experience of usage

outcomes exceeds his or her outcome expectation (positive outcome disconﬁrmation)

30

33:95::— .: .8 3:3..oaxm Sm: 2:855 .3 .252 bows—=wom..=om 93588 "m 25w:—

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

coumgccooﬂm t
homo—mum i enmwoowmm
598::
59358032:
omme
8:305:
{E
:otcBE
_82>§om
cough—.3085 5:88me
@8850 it 08830

 

 

 

 

 

31

encourage the upward revision of outcome expectation, whereas judgments that actual
usage outcome is below his or her outcome expectation (negative outcome
(liscon/irmation) drive future outcome expectation downward (Bhattacherj ee &
Premkumar, 2004).

The model described in Figure 5 is consistent with the growth notion that Bandura
(1997) argued for in addressing the dynamic interaction between self-efficacy and
performance. The idea is that self-efficacy and performance feed to one another in a
positive feedback loop. When self-efﬁcacy is high, it is likely to lead to higher
performance. Successful performance (or the discrepancy between perceived
performance and prior performance expectation) tends to raise subsequent self-eﬁicacv,
which in turn tends to enhance future performance. Alternatively, when self-eﬂicacy is
low, it is likely to induce performance. Subpar performance is likely to decrease self-
efj'icaev, which in turn tends to lead to lower future performance. This type of model
assumes that one has ample room to grow and change until one reaches his or her upper
limit, which may happen when one hits his or her physical limits, has to deal with
conﬂicting goals of equal or higher priority, or face other constraining factors. In the
context of technology adoption, it is assumed that early in the adoption process when one
is familiar to the innovations and still has much to learn, one has ample room to change
and grow until his or her behaviors and beliefs stabilize. With this assumption, the model
in Figure 5 does not include constraining factors that could impose ceilings on behavioral
growth in the later stages of adoption process. The early stage of technology adoption

process is thus viewed as a boundary condition for this model.

32

The nature of the constructs and their relationships in the model are elaborated
further below and with the consideration ofthe nature ofthe relationships over time.
Behavioral Intention

According to the TPB (Ajzen, 1991) and the TAM (Davis et al., 1989),
behavioral intention is a central factor that drives human behaviors. It represents the
internal desire to reach certain end points such as performing valued behaviors or
obtaining desirable outcomes. It inﬂuences how much effort people exert and focuses
their attention on the behaviors to which the effort will be directed. Strong behavioral
intention toward displaying a behavior is predicted to lead to a higher likelihood of
engaging in the behavior (Ajzen; Davis et al.).

Research has clearly shown that behavioral intention has a positive effect on self-
report IT usage at between—person level of analysis (Legris et al., 2003). For example,
Davis et al. (1989) found that behavioral intention explained 12% and 40% of variance of
self-report usage of a word processor at two points of measurements in their study.
Venkatesh and Davis (2000) reported that the correlations between behavioral intention
and self-report usage ranged from .44 to .57 in four samples of workers using new
software applications. In a review, Legris et al. found that out of 10 relationships between
behavioral intention and self—report IT usage, nine were signiﬁcant. Although no studies
have examined the relationship between behavioral intention and usage behavior at the
within-person level, I expect to ﬁnd similar relationships within person. An individual is
more likely to engage in a behavior if the individual has strong behavioral intention for

that behavior than when the same person has a weaker behavioral intention.

33

 

Htpothesis I: Stronger behavioral intention will lead to a higher level of I T

usage.

Implementation Intention

Forming a strong behavioral intention is understood as committing oneself to
reaching desired outcomes or to perfomring desired behaviors. However, the time
between forming behavioral intention to behavioral enactment may be long, even with
strong intentions. Behavioral intention is frequently not enacted. The relationship
between behavioral intention and action is not robust. Empirical evidence shows that at
the between-person level of analysis behavioral intention typically accounts for about
25% of behavior variance (Ajzen, 1991) and about 30% of variance in self—reported usage
frequency (Venkatesh, 2000).

The modest link between behavior and behavioral intention is, in part, the result
of people failing to act on their otherwise good intentions (Gollwitzer, 1999). People may
simply forget their previous intentions. They may fail to recognize relevant situational
cues that trigger thoughts and actions. They may not be able to recall an action plan. They
may not have the physical, cognitive, and/or other abilities required to carry out actions.
Or they fail to allocate sufﬁcient cognitive resources because they are distracted by other
competing thoughts and actions.

Conscious implementation planning, expressed in implementation intention, may
play a central role in the enactment of behavioral intention and explain frequent failure to
carry out desirable behaviors. Implementation intention refers to pre-action decision of

when, where, and how behavioral intention or goal is implemented (self-set goal and

34

behavioral intention are similar constructs and are used interchangeably by Gollwitzer,
1999). It is conceptually distinct from behavioral intention in a number of aspects.
behavioral intention refers to a desire to reach certain end points such as perfomring a
particular behavior or obtaining a desirable outcome. It has the structure of“I intend to
reach A” where A is either a behavior or an outcome (Gollwitzer). In expressing their
behavioral intention, people reveal their desire to carry out the behavior of interest. The
consequence of having a strong behavioral intention is a sense of commitment that drives
individuals to carry out the desirable behaviors.

Whereas behavioral intention refers the commitment or desire to carry out a
behavior, implementation intention concerns the way the intended behavior is enacted. It
is subordinate to behavioral intention and instrumental in the execution of the behavior.
Imp/ementation intention speciﬁes when, where, and how intended behavior is enacted. It
has the structure of “When situation X arises, I will respond Y” (Gollwitzer, 1999) and
serves to link anticipated opportunities with appropriate responses that lead to desirable
behaviors or outcomes. With implementation intention, the person commits himself or
herself to respond to a certain situation in a speciﬁc manner in order to carry the desirable
behavior. Thus, it serves the purpose of promoting the enactment of the behavior
expressed in behavioral intention and thus is likely to explain behavioral variance
independent of behavioral intention.

Oollwitzer (1999) suggests implementation intention helps enactment of behavior
because they make individuals more sensitive to situational cues. They automatically
activate goal-directed behaviors, when individuals encounter critical situational cues. As

a result, individuals with strong implementation intention are quick to select the

35

appropriate situations and seize critical opportunities that are short-lived to enact critical
responses that lead to desirable behaviors. Their actions become swift, require little
cognitive resources, and are protected against competing goals, distractions, and habitual
responses.

Gollwitzer (1999) further suggests that implementation intention is instrumental
in the enactment of behaviors for two main reasons. First, implementation intention
causes the representation of anticipated situations to become highly activated and thus
easily accessible. Second, it helps automatize the situation-behavior response so that
action initiation becomes immediate and efﬁcient, and does not require conscious intent.
With these functions, implementation intention allows individuals to quickly identify
relevant situations and carry out appropriate responses automatically that allow
behavioral intention to be enacted.

Research suggests that at the between-person level of analysis, implementation
intention helps individuals carry out their intended behaviors (Gollwitzer, 1999). In a
series of studies, Gollwitzer and Brandstatter (1997) showed that behavioral intention
that is coupled with implementation intention is more likely to be enacted than mere
behavioral intention alone. In one study, student participants were asked prior to
Christmas break to name a difﬁcult-to-implement project they intended to achieve during
the upcoming vacation. Participants indicated intended behaviors such as writing a
seminar paper, settling an ongoing family conﬂict, or engaging in sports activities. When
asked, two-thirds of the participants indicated that they had formed intentions on when
and where to get started (i.e., implementation intention) for the behaviors. After

Christmas, the participants were asked whether they executed their intended behaviors.

36

Two-thirds of the participants with implementation intention had carried them out,
whereas only one-fourth of the participants without implementation intention had done
so. In another study (Gollwitzer & Brandstatter), research participants were asked within
two days after Christmas to write a report on how they spent Christmas Eve and send it to
the experimenters. Half of the participants were told to form implementation intention by
indicating on a paper exactly when and where they intended to write the report during the
two days after Christmas. When the experimenters received the reports, three—fourths of
the participants with implementation intention had written the reports in the requested
time period, whereas only a third of the participants without implementation intention had
done so.

In other studies, Gollwitzer and colleagues have also found similar effects of
implementation intention on other types of behaviors. Implementation intention has a
positive effect on the execution of unpleasant behavior such as self-examination for
evidence of breast cancer (Gollwitzer & Oettingen, 1998; Sheeran & Orbell, 2000) and
taking vitamin supplements (Sheeran & Orbell, 1999). It helps people who have problems
with action control such as addicts and schizophrenic patients (Gollwitzer, 1999) ward
off unwanted habitual responses and enact desired behaviors (Gollwitzer).

The work by Gollwitzer and colleagues suggests that implementation intention
plays an important role in bringing about desirable actions. In light of the evidence, Ajzen
(2001) acknowledged that it may inﬂuence behavioral expression above and beyond
behavioral intention and recommended more research to study its effects. To my
knowledge, no study has attempted to examine the role of implementation intention on

the adoption and usage of IT innovations. Although no studies to date have examined the

37

effect of implenrentation intention at the within-person level of analysis, I expect that
implententation intention will inﬂuence IT usage over-and-above the effect of behavioral

intention.

I It'pothesis 2: Stronger implementation intention will lead to higher levels of I T

usage over-and-above the effect of behavioral intention.

Sel/lefficacv and Eﬂicac'v Disconﬁrmation

While behavioral intention and implementation intention represent the conscious
thoughts of behavioral end points and the means to obtain them, they also reﬂect other
types of beliefs that ultimately drive behaviors. A key belief is self-efﬁcacy, which
Bandura (1997) deﬁnes as beliefs in one's capabilities to organize and execute the courses
of action required to produce given attainments. Self-efﬁcacy is critical to motivation
because people have little incentive to act unless they believe they can produce desired
outcomes by their actions (Bandura). It is domain speciﬁc and strongly tied to behavior in
question. In the context of IT innovation adoption, innovation self-efﬁcacy (e.g.,
computer self-efficacy, Palm Pilot self-efﬁcacy) is deﬁned as beliefs in one’s capability to
use a particular IT innovation (e. g., Compeau & Higgins, 1995; Compeau et al., 1999). It
is similar to the notion of perceived behavioral control in the TPB (Ajzen, 1991) for it
refers to one’s perception of one’s ability to perform a given behavior. Self-eﬂicacy is
also related to the construct of ease of use in the TAM (Davis et al., 1989). Yet,
conceptually it is a different construct (Compeau & Higgins; Compeau et al.; Davis et

al.). Innovation self-efficacy is more rooted in individual differences and experience with

38

IT innovation usage, whereas ease of use is more driven by the characteristics of IT
innovations (Compeau & Higgins; Compeau et al.; Davis et al.). Given this distinction
and the ﬁndings that case of use was inconsistently related to self-report usage (Legris et
al., 2003) and its effect on IT usage was fully mediated by outcome expectation (Davis et
al.), innorraion self-efﬁcacy instead to ease of use is included in the model to represent
controlling beliefs.

Bandura (I997) asserts that self-efﬁcacy affects behavioral intention and actual
behavior because people are motivated to carry out behaviors that they believe they could
successfully enact in order to yield favorable outcomes. Speciﬁcally, self-efﬁcacy
inﬂuences (a) the courses of action that people choose, (b) the level of effort and
persistence they put forward in a task, and (c) the task strategy they follow. The notion
that self-efficacy inﬂuences behavioral choice or intention has received considerable
empirical support at the between-person level of analysis. For example, research has
shown that self-efﬁcacy was positively related to behavioral intention to enroll in a
computer course (Hill et al., 1987) and difﬁculty level of goals people adopt (Bandura,
1997; Locke & Latham, 1990). Although no studies have examined the effect of self-

efﬁcacy on innovation behavioral intention within person, I expect that:

Hypothesis 3: All else equal, higher innovation self-efficacy will lead to stronger

behavioral intention to use IT innovations.

Besides affecting behavioral intention, self-efﬁcacy is expected to inﬂuence

implementation intention. Research has shown this linkage directly and indirectly at the

39

between-person level of analysis. For example, previous studies have found that higher
self-efﬁcacy is associated with more effective task strategy (Bandura, 1997; Seijts &
Latham, 2001 ), a construct similar to implementation intention. In studies where self-
efficaev and implementation intention were not examined together, self-efficacy (Ajzen,
1991; Bandura) and implementation intention (Gollwitzer, 1999; Gollwitzer &
Brandstatter, 1997) were found to have positive effects on behaviors, over and above the
inﬂuence of belu'tvioral intention. This suggests that self-efficacy and implementation
intention may be related. Although no study has examined the relationship between self-
ef/icacv and implementation intention at the within-personal level of analysis, I expected

the same relationship within persons.

IIrpothesis 4 .' Higher innovation self-efficacy will be associated with stronger

implementation intention.

Because self-eﬂicacy is likely to inﬂuence behavioral intention and
implementation intention, it is expected to affect IT usage behaviors. A number of
empirical studies supported this linkage at the between-person level. Compeau and
Higgins (1995) reported that computer self-efﬁcacy was positively related to self-reported
frequency of computer usage in a sample of 1020 Canadian workers in a cross-sectional
survey study (r = .45). In a follow-up survey administered a year later, Compeau et al.
(1999) found a similar relationship between self-reported computer usage and computer
self-ejfﬂ’caev measured the preceding year (r = .43). The positive relationship between

self-efﬁcacy and IT usage was also found in a number of other studies using student

40

samples (e.g., Hill et al., 1987). In addition, research that relied on the TPB framework
similarly reported that perceived behavioral control, a construct similar to self-efficacy,
was positively related to behavioral intention and self-reported IT usage at the between-
person level (e.g., Taylor & Todd, 1995).

Although no study has looked at the effect of self-efficacy on IT usage at the
within-person level, research in Industrial-Organizational Psychology has examined a
similar relationship—the inﬂuence of self-efﬁcacy on task performance‘and reported
some interesting results. Generally, researchers (e. g., Bandura, 1997; Vancouver et al.,
2001, 2002; Vancouver & Kendal, 2006) consider two key effects of self-efficacy on
performance. The ﬁrst regards the inﬂuence of self-efﬁcacy on performance through its
effect on goal level. More efﬁcacious individuals tend to set higher goals, which in turn
lead to more task effort and ultimately higher performance (Locke & Latham, 1990). The
second effect is a direct one on behavior. When goal level is controlled or held constant,
control theorists argued (e.g., Power, 1973, 1978) that self-efﬁcacy has a negative
relationship with performance because individuals are more motivated to reduce the
discrepancy between their perceived capability and their desired level of performance.
Social cognitive theorists (Banduara, 1997; Bandura & Locke, 2003) argued that self-
efﬁcaev effect is largely a positive one because individuals are unlikely to act unless they
believe their action will produce desirable outcomes.

Researchers (e. g., Bandura, 1997; Bandura & Locke, 2002; Vancouver et al.,
2006) tend to agree on the ﬁrst effect of self-eﬁicacy. The second effect, the impact of
self-e/j'icaev on performance when goal level is controlled, is still an open question.

Bandura argued that this effect is positive and reported numerous studies in many

41

different contexts to support his views. However, these studies were conducted to look at
between-person level. Vancouver and colleagues (2001, 2002, 2006) relied on control
theory to argue that self-efﬁcacy would have a negative relationship with performance in
limited context such as a Ieaming, preparatory stage. In a series ofthree studies in
classroom Ieaming settings, they reported that self-efficacy had a weak negative effect on
performance (2001, 2002, 2006) at the within-person level of analysis, but the effect
became insigniﬁcant when goal level was not controlled (Vancouver et al., 2006). These
results suggest that in the Ieaming context self-efﬁcacy may have a negative effect on
performance when goal level is controlled. When goal level is free to vary, the overall
effect of selflefﬁcaev may not be signiﬁcant from zero, as in Vancouver et al.’s (2006)
study where there was not much variance in goal levels among student participants, or be
a positive one, as Bandura argued.

The domain of IT innovation usage over an extended period of time is different. It
is largely a performing context because an IT innovation serves as one means among
several to help users accomplish one or more tasks. Within this context, consistent with
Bandura (1997), innovation self-efﬁcacy is expected to have a positive effect on IT usage.
The reason is that a user is likely to use an IT innovation when he/she believes that he/she
could operate it successfully. In addition, innovation self-efficacy reﬂects users’ actual
ability to use an IT innovation. When a user has high innovation self-efficacy, he/she is

likely to use it more efﬁciently and thus could use the innovation more frequently within

the same time duration. Thus, I expect that:

Ilvpothesis 5 .' Higher innovation self—efficacy will lead higher level of I T usage.

42

An important question is whether the effect of innovation self-efficacy on usage
behavior is uniform over time within person. According to resource allocation theory
(Kahneman, 1973; Kanfer & Ackerman, 1989) people have limited cognitive resources
and have to allocate them to many different tasks in order to achieve desired goal states.
When an individual has a higher level of cognitive resources, he/she is likely to be able to
accomplish more tasks or reach higher goal levels. Yet the effect of cognitive resources
on a given task tends to depend on the individual’s skill level for the task and thus change
over time. Typically, individuals develop their skills sequentially through three stages:
sequential, associative, and autonomous stages (Kanfer & Ackerman). In the sequential
stage, individuals rely heavily on their cognitive resources to learn the correct sequence
of steps in order to perform the task. Their behavior is deliberate, sequential, and still
awkward. Over time, individuals learn the correct behavioral sequence and their actions
require less effort. When individuals reach the autonomous stage, their performance
becomes more or less automatic and is often triggered by external cues instead of
deliberate internal thoughts. It becomes more ﬂuent and requires much less cognitive
resource. In other words, over time, individual performance of a given task depends less
and less on effort and cognitive resources, and the locus of control of their actions moves
from internal, deliberate thoughts to external, situational cues. Given self-efﬁcacy is a
subjective estimate of one’s cognitive resources, it is likely that self-efﬁcacy’s inﬂuence
on performance is moderated by time such that self—efficacy strongly inﬂuences

perfomrance initially, but the effect weakens over time.

43

Although the idea that time moderates the relationship between self-efﬁcacy and
IT usage is a reasonable and important one, little research has attempted to examine the
issue. To my knowledge, only three studies ofinnovation use (e.g., Davis, 1989;
Venkatesh & Davis, 2000, Venkatesh et al. , 2003) have studied the effect of ease of use,
a construct similar to self-efﬁcacy, over time at the between-person level of analysis.
These studies collected usage data at two time points (Davis et al., 1989; Venkatesh et al.,
2003) and three time points (Venkatesh & Davis, 2000). They found that early on (time
I), ease ofuse had an independent effect on IT usage, over-and-above the effect of
outcome expectation. However, the independent effect became insigniﬁcant when users
gained more experience (time 2 or 3). Although no research has examined the role of
time as a moderator at the within-person level of analysis, consistent with resource

allocation theory, I predict that:

llvpothesis 6: The effect of innovation self-efﬁcacy IT usage will weaken over

time.

According to Bandura (1997), self-efﬁcacy for a given task is a malleable
construct. It changes over time and is inﬂuenced by a number of factors. Initially,
individuals develop a sense of self-efﬁcacy regarding a new task by relating it to their
relevant experience, observing others, and/or being verbally persuaded. Yet initial
perception of self-efficacy is often inaccurate as the individuals may not know the task
well, may not anticipate how their ability and constraints interact with the demands of the

tasks, and how their skills develop. Over time, they have opportunities to perform the

44

task, gain ﬁrst-hand experience, and develop their performance skills. Such ﬁrst-hand
experience should become the most important determinant of one’s self-efﬁcacy.
Experience of success or high levels of performance raises one’s self—efficacy, whereas
failure or low levels of performance tends to lower it. The updated self—efficacy then
affects future experience of the task, thus completing the cyclical process.

Numerous studies (e.g., Gist & Mitchell, 1992; Lindsley, Brass, & Thomas, 1995;
Bandura & Locke, 2003; Latham & Locke, 1991; Mitchell, Hopper, Daniels, George-
F alvy, & James, 1994) have examined the effect of past performance on self-efficacy.
They generally found that past performance of a task is a key antecedent of self-efficacy
regarding the task. Successful performance of a task heightens self-efﬁcacy and raises
one’s conﬁdence that, in turn, facilitates further successful performance of the behavior.
Conversely, failing to perfonn a behavior decreases self-efﬁcacy and inhibits future
performance of the behavior, leading to a downward spiral. Because performance level
tends to grow over time as individuals develop skills and because it is strongly

determined by the amount of effort individuals exert, I expect that:

I I vpothesis 7: Innovation self-efﬁcacy will increase over time.

IItpothesis 8: IT usage will have a positive effect on subsequent innovation self-

efficacy.

Bandura (1997) suggests that self-efﬁcacy changes as a result of a cognitive

evaluation mechanism. Once individuals carry out their intended behaviors, they reﬂect

on their experience and evaluate whether their experience matches their prior

45

expectations. Evaluations that one’s current performance of a task such as using an IT
innovation surpasses his or her self—efﬁcacy encourage the upward revision of self-
eff'icacv (Bandura, 1997). In contrast, judgments that one’s performance falls short of his
or her self—efficacy motivate downward revision of self-efficacy (Bandura). It is the
discrepancy between prior self-efficacy and currently perceived performance that compels
users to adjust their future self-efﬁcacy. Cognitive discrepancy is a strong motivating
force because individuals are compelled to avoid cognitive dissonance and keep their
beliefs and behaviors in alignment. In the model in Figure 5, the discrepancy between
self-efficacv and perceived performance reﬂects in the construct of efficacy
disconfirmation, an internal measure of personal performance. Positive efficacy
disconfirmation occurs when users’ prior self—efficacy falls short their perceived
perfomrance and is likely to lead to upward revision of future efﬁcacy expectation.
Conversely, negative efficacy disconﬁrmation happens when prior efﬁcacy expectation
exceeds experienced performance and is likely to result in downward adjustment of
subsequent efﬁcacy expectation. The process of evaluating one’s performance and
adjusting self-eﬂicacy expectation repeats until users reach a state of equilibrium over
time when self-eﬂicacy and perceived performance are in alignment.

Although the idea that efﬁcacy disconﬁrmation mediates the effect of behaviors
on selflefﬁcacj' is reasonable, to my knowledge no study has investigated the issue.

Consistent with Bandura (1997), I expect that:

Hypothesis 9: Efficacy disconfirmation will mediate the effect of I T usage on

subsequent innovation self-efficacy.

46

Outcome Expectation and Outcome Disconfirmation

In addition to self-efficacy, outcome expectation is another key belief that
underlies behavioral intention and propels individuals to action (Bandura, 1997;
Compeau et al.,l999; Davis, 1989; Davis et al., 1989; Zimmerman, 2000). Outcome
expectation regarding an IT innovation refers to the anticipated outcomes that are
believed to be the results of using the innovation (Davis, 1989; Davis et al., 1989;

C ornpeau & Higgins, 1995; Compeau et al., 1999). It is related but conceptually distinct
from selflefficaev. Outcome expectation concerns beliefs about the ultimate ends of using
IT innovations, whereas self-efficacy refers to personal beliefs about having the means to
use them. People expect many psychological and practical outcomes associated with
using IT innovations. These outcomes include the joy of using IT innovations,
development of IT skills, heightened social status, improved performance, and extrinsic
rewards such as promotions or pay raises. Outcome expectation is also similar to
perceived usefulness in the TAM, as they refer to the subjective outcomes of using IT
innovaﬁons.

Outcome expectation inﬂuences behaviors because behaviors are largely regulated
by the actors‘ subjective experience of their consequences through associative Ieaming
and operant conditioning mechanisms. People are more likely to engage in behaviors they
expect will be rewarded, but show little interest in activities that they do not value
(Ajzen, 1991; Bandura, 1997; Davis, 1989; Davis et al., 1989; Vroom, 1964;

Zimmerman, 2000). Individuals engage in activities and extend effort when they

47

anticipate that their behaviors will bring desirable outcomes (Ajzen, 1991; Bandura,
1997; Stajkovic & Luthans, 1997; Vroom, 1964).

In IT adoption research, research has investigated the effect of outcome
expectation using the same construct with a different label, perceived usefulness.
Perceived usefulness was found to have a positive inﬂuence on behavioral intention to
use IT innovations at the between-person level of analysis (Davis et al., 1989; Legris et
al., 2003; Taylor & Todd, 1995; Venkatesh & Davis, 2000). Its effect was stronger than
ease ofuse in many studies (Legris et al.; Vanketesh et al., 2000). In addition, the
inﬂuence of outcome expectation was found to be consistent across different studies. For
example, Legris et al. examined 22 studies that yielded 28 possible relations between
each of the TAM constructs and behavioral intention. Out of 19 reported relations,
perceived usefulness had a positive relation on behavioral intention 16 times and a non-
signiﬁcant relation three times. Although no studies have examined the effect of outcome
expectation on behavioral intention at the within-person level of analysis, I expected the
similar relationship because an individual is likely to have stronger behavioral intention

to use IT innovations when he/she has more positive outcome expectation.

I-Irpothesis l 0: All else equal, higher outcome expectation will be associated with

stronger behavioral intention.

IT adoption research also shows that higher outcome expectation is associated

with higher usage of IT innovations at the between-person level of analysis. Compeau

and Higgins (1995) reported that the correlation between outcome expectation and self-

48

reported time usage ofa computer was .41 in a survey of 1020 workers. When self-report
usage was measured again in a follow-up survey a year later (Compeau et al., 1999), it
was predicted outcome expectation was measured the year before (r = .40). The positive
relationship between outcome expectation and IT usage was also found in other studies
(e.g., Hill et al., 1987). In addition, existing studies have also demonstrated that perceived
usefulness, the same construct but with a different label, predicts behavioral intention and
self-reported usage (e.g., Davis, 1989; Davis et al., 1989; Jeyaraj, Rottman, & Lacity,
2006; Legris et al., 2003; Taylor & Todd, 1995). In their review, Legris et al. reported
that perceived usefulness is the most consistent predictor of self-report usage among all
the TAM and the TPB constructs.

Although no research has studied the inﬂuence of outcome expectation at the
within-person level of analysis, I predicted outcome expectation will have a positive
inﬂuence on IT usage within person because an individual is likely to use IT innovations

when heishe see more positive outcomes resulted from using them.

Hrpothesis II .' Higher outcome expectation will lead to higher levels of I T usage.

An important question is whether the effect of outcome expectation on usage
behavior is consistent over time within persons. As argued previously from a resource
allocation perspective, over time individual performance of a given task is likely to
depend less and less on effort and cognitive resource, and the controls of their actions are
moved from internal thoughts to situational cues. Consequently, the effect of one’s

motivational force on behavior tends to weaken over time. Because outcome expectation

49

is a key construct that captures the motivational force, it is likely that outcome
expectation’s inﬂuence on performance is moderated by time such that outcome
expectation affects IT usage strongly initially, but inﬂuence becomes weaker over time.
Although the idea that time moderates the relationship between self-efficacy and
IT usage is a reasonable and important one, little research has attempted to examine the

issue. Consistent with the resource allocation theory, I predicted that:

Htpothesis 12: The effect of outcome expectation on IT usage will weaken over

time.

Similar to self-efficacy, outcome expectation is expected to have bidirectional
relationships with IT usage. That is, outcome expectation inﬂuence IT usage, which in
tum shapes and adjusts future outcome expectation. This process is driven by users’
desire to avoid cognitive dissonance and keep their beliefs and behaviors in alignment. In
IT usage contexts, users initially form their outcome expectation by evaluating second-
hand information, such as peers’ testimonials, vendors’ claims, or experts’ opinions. Such
communicated information serves as the baseline from which users initially judge the
merits of IT innovation. As users interact with IT innovations, they gain ﬁrst-hand
experience, which allows them to evaluate whether their initial judgment matches with
their actual experience. Because people are motivated to keep their beliefs and behaviors
in alignment, users’ evaluation of discrepancy between prior outcome expectation and
actual experience often lead to revisions of outcome expectation. The updated outcome

expectation then shapes future IT usage. The process reiterates until behavior and belief

50

reach a steady state of equilibrium where they are in alignment. Thus, this process
suggests that the level of positive outcomes experienced by users has a positive inﬂuence
on their subsequent outcome expectation. Because IT innovation usage typically is a
strong correlate of actual usage-related outcomes, on the assumption that the innovation
represents true improvement over past technologies, higher IT usage is likely to lead to
perceptions ofrnore positive outcomes and thus leads to higher levels of outcome
expectation.

Although it is important to understand the impact of IT usage on outcome
expectation, research has not investigated the issue because researchers generally follow
traditional experimental and correlational designs where data were collected at one or few
points in time. Yet, two survey studies by Compeau & Higgins (1995) and Compeau et
al. (1999) found that self-report IT usage and outcome expectation are positively related.
These results provide limited empirical support that IT usage may inﬂuence outcome
expectation at the between-person level of analysis. Although no research has examined
the relationship at the within-person level of analysis, consistent with the rationale

described earlier, I expect that:

Hrpothesis I 3: IT usage will have a positive effect on outcome expectation.

The precise mechanism for change in outcome expectation is thought to occur

through a cognitive evaluation process. Once users go through the experience of using IT

innovations, they are likely to reﬂect on their experience and evaluate whether their

outcome expectation match their perceived outcomes. The discrepancy between prior

51

outcome expectation and perceived outcomes often leads users to adjust their outcome
expectation. The discrepancy of outcome expectation and perceived outcomes reﬂects in
the construct of outcome disconﬁrmation, an internal measure of product performance, in
the model in Figure 5. Positive outcome disconfirmation occurs when users’ prior
outcome expectation falls short their actual experience of positive outcomes and is likely
to lead to upward revision of future outcome expectation. Conversely, negative outcome
tlisconfirnuition happens when prior outcome expectation exceeds of experienced
outcomes and is likely to result in downward adjustment of subsequent outcome
expectation. The process of evaluating one’s experience and adjusting outcome
expectation repeats itself until users reach a state of equilibrium when outcome
expectation and perceived experience of actual outcome are in alignment.

Although no prior research has attempted to investigate the mechanism through
which IT usage updates one’s outcome expectation, consistent with cognitive dissonance
principal I expect that outcome discom’irmation is likely to mediate the inﬂuence of IT

usage on outcome expectation within person.

Hypothesis I4: Outcome disconfirmation will mediate the effect of IT usage on

outcome expectation.

52

METHODS

Overview ofthe Research

In this study, college students, who volunteered for the research, were given Palm
Pilots loaded with several software programs, and their behavioral and attitudinal data
were recorded over an eight-week time period between October and December of 2006.
None ofthe participants had previous experience with Palm Pilots or with the software
programs that they learned to use. The study began with training the participants on each
of three software programs, which were selected because they are seen as useful and
interesting to college students. The Palm Pilots were set up to record the amount of time
that the students used the Palm Pilot software packages. They were also programmed to
sample attitudinal and behavioral data daily over the time period. This research method
enabled the collection of objective adoption data in terms of the amount of time the
participants spent using the IT innovation device as well as attitudinal and perceptual data
over time that were related to adoption behaviors.
S‘_dnlpl_€

Participants were 63 undergraduate students enrolled in Psychology 101 at
Michigan State University. Undergraduate students in Psychology 101 were required to
participate in psychology experiments as part of their class activities, and their
participation was facilitated by the subject pool set up by the psychology department. The
student participants signed up for participation on a ﬁrst—come ﬁrst-serve basis with the
constraints that half were female. They were selected for the study if they met two
conditions. First, they had not used any PDA device prior to the experiment. Second, they

had not used software applications similar to the ones they were to use in study on their

53

personal computers. These conditions were set up to ensure that the Palm Pilot was
perceived as new to the users and thus could be considered as an IT innovation. To screen
participants, only students who indicated they had not used PDA before were invited to
participate in the study. At the beginning of the study, selected participants were asked
whether they have any experience to use PDAs, cell phone, or computers to schedule
appointments, organize ideas, or track to-do lists (see Appendix 1). Two study
participants indicated that they had some experience, and their data were not included in
the analyses. Additionally, students were asked whether Palm Pilots were something new
to them personally (see Appendix 2). Results showed Palm Pilots were viewed as new by
all participants in the ﬁnal sample.

During the course of the study, data from three participants were lost and could
not be recovered due to technical reasons. The ﬁnal sample contained 58 participants, 31
females and 27 males, between the ages of 18 to 27 years (M = 19.7, SD = 1.7). The
ethnic makeup of the sample was: 76% Caucasian American, 10% African American, 5%
Hispanic American, 2% Indian American, and 7% Asian American.
IT Innovation

In this study, Palm Pilots were chosen as an IT innovation to investigate for a
number of reasons. Because Palm Pilots were a relatively new product at the time of
study (1996) and especially to the research participants (which is no longer the case at the
time ofthe publication of this research), who were psychology students at Michigan State
University, it was perceived as a new technology, as suggested by participants' answers to
screening questions discussed at a later point. It was at the time and has been

increasingly becoming a more integrated device that can perform many functions (email,

54

phone, voice-commanded computing, etc...) and thus could be adopted by the mass in
longer temrs. Additionally, Palm Pilots could be programmed to record user behavior
objectively and software applications for collecting cognitive and affective measures used
in this study already existed from previous behavioral sampling research.

The Palm Pilots, in this study, were preinstalled with the following
applications/features.

1. Calendar with which users can record and keep track of their

daily/weekly/monthly schedules and appointments.

2. Contacts which allows users to store and organize their contacts

DJ

131$ enables users to record things to do and keep track of them.

4. Note Pad with which users can draw sketches and write their notes.

5. ﬁrgggg lets users outline their thoughts and rearrange them.

6. Hi Money which enables users to record and track incomes, expenses,

ﬁnancial transactions, and report the overall balance.

7. Diet Helper with which users record and track their daily diet.

Given these features, the Palm Pilot used in the present study can be viewed as an
IT innovation that performs seven different functions implemented by the seven
apphcaﬁons.
Training

The training was about one and a half hours long and was conducted at the ﬁrst
meeting with the participants. The goal was to provide the participants with a basic
understanding of Palm Pilot features: navigation, data input and synchronization

processes. Furthermore, the training was intended to show, in detail, how participants

55

could use the basic features of three main applications in Palm Pilot, namely Calendar,
Contact, and Task. In the training, participants were given an instructional booklet each
(see Appendix 3). The instructor then discussed the contents of each application via a
PowerPoint presentation. Participants were asked to carry out hands-on exercises
frequently at the end of each topic so that they learned to operate Palm Pilots through
hands-on practice. The instructions were adapted from the Getting Started Guide that was
provided with the purchase of the Palm Pilot Zire 31 model used in the experiment. As
can be seen in Appendix 3, the instructions explained the operations of the main icons,
controlling buttons (e. g., Home, Keyboards) and the workspace in Palm Pilot (e. g.,
Grafﬁti writing area, keyboards, and main screen) where data were entered and responses
were viewed. The trainer then brieﬂy described the main purpose of all seven preinstalled
software applications. This was meant to inform users of the main functions of all
preinstalled applications and to explain to the users what the Palm Pilot was capable of
accomplishing. Participants were then shown how to open and close all the preinstalled
applications using the main icons and controlling buttons, but they would be trained only
on the detailed features of the three applications previously mentioned. This pattern of
training is typical in technology training where new users are taught the basic features of
the technology. It often leads to two ways in which innovation adaption could be
manifested. First, users spend time working on the software they are trained on and
practice basic operations. Second, users spend time to learn new applications on their
own, for which they are informed of the applications’ general purpose, if they want to use

them.

56

Instructions on how to access and operate Calendar, Contact, and Task
applications were then explained. These instructions focused on the detailed operations of
the basic features of these applications without discussing the more advanced features.
For example, participants were showed how to schedule an appointment, edit, and review
them by opening different views of the calendar (e.g., daily, weekly, and monthly views).
However, more advanced operations such as how to set up alarms for appointments, how
to repeat the same event over time, how to assign priority to event, how to color code
events for visual effects, and how to attach notes to events were not discussed. The
instructional strategy was intended to teach participants on the fundamentals and help
them get started without being overwhelmed, an approach similar to the actual training
process that typically takes place after the introduction of IT innovations in organizations.
It should be noted that this type of training approach left the burden of Ieaming more
advanced features and mastering IT innovations on the users.

During the training, participants were frequently given time to carry out the
discussed procedures during and at the end of the discussion of each application. The
instructor would stop frequently to ask participants whether they had questions and had
completed the hands-on exercises in order to make sure that they were on the same page
and could operate the Palm Pilots. The instructions ended with a discussion on how Palm
Pilots could be synchronized with computers so that the data could be updated and
backed up. At the end of the presentation, participants were asked to explore Palm Pilot
on their own for 20 minutes. They were instructed to again practice the steps presented in
the presentation by using different buttons and icons to access different applications as

well as performing detailed operations in Calendar, Contact, and Task that had been

57

discussed in the presentation. The investigator also showed the participants where they
could access additional how-to instructions and obtain user manuals for all preinstalled
applications. The additional instructions and manuals were posted on a web page so that
participants could download them ifthey desired.

Procedure

After signing voluntary participation agreements to take part in an eight-week
study using Palm Pilots (see Appendix 4), participants completed a paper-and-pencil
information sheet asking for their gender, age, and ethnicity. They then ﬁlled out a survey
measuring individual difference dimensions (see Appendix 5), which would be used for
additional analyses that are not part of the present study.

After the survey, each of the participants was given a Palm Pilot. The Palm Pilots
were all permanently tagged with unique numbers to identify the users without revealing
their personal infomration. Participants learned how to operate the Palm Pilots with the
aid of the trainer as described in the preceding section.

Participants were then shown how they could respond to a survey administered
and prompted by Palm Pilots. The survey contained questions asking participants about
their expectations, beliefs, intentions, and cognitive evaluations as described in the
“Measure” section. PMAT, a Palm survey application designed by Weiss, Beal, Lucy,
and McDermid (2004), was used to administer the survey and collect participants’ daily
self-reported responses. Once the instructions on how to respond to the PMAT survey
were completed, participants practiced answering a ﬁve-question survey. They were then
asked to respond to the ﬁrst full survey, which took less than ﬁve minutes to complete.

The same survey was administered automatically by PMAT about a day later at the time

58

chosen by participants and was repeated at the same chosen time daily for eight weeks. In
all cases, the administration of the survey was prompted by beeping signals. Survey items
were presented in a random order at each time in order to minimize order effects. A log
system in the Palm Pilot automatically recorded participants’ usage data (namely, the
usage frequency and duration) for the whole bundle as well as each of the applications on
a daily basis.

After participants completed the training survey, they were asked to schedule a
follow-up 30-minute meeting and four subsequent two—minute meetings. The 30-minute
meeting occurred in the same location one week later to help users troubleshoot problems
that arose during their ﬁrst week using Palm Pilots. The short two-minute meetings were
scheduled every two weeks after the training. In these meetings, the experimenter met the
participants a few minutes before the beginning of their Psychology 101 lecture or other
mutually-agreed times and locations and downloaded their survey responses onto his
laptop and usage data so that the data would be secured.

After scheduling the follow-up meetings, participants were reminded that each
would have an opportunity to earn $70. Speciﬁcally, a participant could earn $15 during
the ﬁrst two weeks if he/she (a) met the experimenter to have his/her data downloaded
AND (b) both answered at least 90% of survey items and used the Palm Pilot at least
once a day (with usage duration at his/her discretion) for 12 days out of 14 days.
Similarly, each participant could earn $15 during the second two weeks, $15 during the
third two weeks, and $25 during the ﬁnal two weeks, contingent on BOTH conditions (a)
and (b) described above. Payment was given by check once the whole study was

complete. Participants were then thanked for coming to the ﬁrst meeting and dismissed.

59

At the follow-up meeting a week later, the experimenter asked the participants
whether they ran into any problems using the Palm Pilots. A few, non-signiﬁcant issues
surfaced and were quickly addressed. At subsequent meetings, participants’ survey
responses and usage data were downloaded to the investigator’s computer. The
downloading process took about 30 seconds each time. At the ﬁnal meeting at the end of
the study, participants retumed the Palm Pilot, got paid for their participation, and were
given a short debrieﬁng (see Appendix 6) regarding the purpose ofthe study.
llleasures

The following measures were used in the study].

Time. Time is measured in cumulative days since the start of the study. Time is 1
for the ﬁrst day, 2 for the second, and so forth.

Outcome Expectation. Four items were adapted from Vanketesh et al. (2003) to
assess outcome expectation (see Appendix 7). Each item was rated using a ﬁve-point
scale that ranged from 1 (extremely unlikely) to 5 (extremely likely). A sample item was,
“lfl use Palm Pilot, I would ﬁnd Palm Pilot useful in my work.” The alpha coefﬁcient
for the scale was .92.

Innovation Self-Efficacy. The seven-item measure of self—efficacy (see Appendix
8) followed the procedure recommended by Bandura (1997) and used by Zimmerman and
Bandura (1994). The scales assessed users’ perceived capability to use various features of
Palm Pilots. Users were asked to provide ratings to their perceived capability to use each

of the features of Palm Pilots. For example, “How well can you use the Calendar

 

I Personality, computer self-efﬁcacy, and other individual difference factors were also
measured for a separate study and their scales were described in Appendix 5. These
factors are not needed as controlled variables for within-person design study such as the
present one as they are constant for each person.

60

Application?” Each item was rated using a ﬁve-point scale that ranged from 1 (not well at
all) to 5 (very well). The alpha coefﬁcient for the scale was .89.

Behavioral Intention. Three items were adapted from Vanketesh et al. (2003) to
assess be/uivioral intention to use Palm Pilots (see Appendix 9). Each item was rated
using a ﬁve-point scale that ranged from 1 (extremely unlikely) to 5 (extremely likely). A
sample item was, “I intend to use Palm Pilot during the next 24 hours.” The alpha
coefﬁcient for the scale was .91.

Implementation Intention. Five items assessed implementation intention (see
Appendix 10). They were adapted from Miller’s (2003) study. Each item was rated using
a ﬁve-point scale that ranged from 1 (strongly disagree) to 5 (strongly agree). A sample
item was, “I have planned how I will use Palm Pilot when I see relevant opportunities.”
The alpha coefﬁcient was .94.

Outcome Disconfirmation. Four items were adapted from Bhattacherjee and
Prernkumar (2004) to assess users’ outcome disconfirmation (see Appendix 11). Each
item was rated using a ﬁve-point scale that ranged from 1 (much worse than expected) to
5 (much better than expected). A sample item was, “Compared to my expectation
yesterday, the ability of Palm Pilot to improve my performance was (much worse than
expected much better than expected)” The alpha coefﬁcient for the scale was .91.

Efﬁcacy Disconfirmation. Four items were adapted from Bhattacherj cc and
Premkurnar (2004) to assess users’ efficacy disconﬁrmation (see Appendix 12). Each
item was rated using a ﬁve-point scale that ranged from 1 (much worse than expected) to

5 (much better than expected). A sample item was, “Compared to my expectation

61

yesterday, my ability to use Palm Pilot was (much worse than expected much better
than expected)” The alpha coefﬁcient for the scale was .92 (Bhattacherj ee, 2004).

IT Usage. Objective usage of Palm Pilot was automatically recorded by a log
system on a daily basis. The log system monitored and recorded how often and how long
each application in the Palm Pilots was used daily. I used two different measures of usage
in the present study: usage frequency and usage duration. Usage frequency refers to the
number of times participants used the Palm Pilots each day regardless of the nature of
applications they used. It is not cumulative. For example, ifa participant used the Palm
Pilot three times the ﬁrst day, and ﬁve times the second day, usage frequency was
recorded as three and ﬁve, respectively. Usage duration is the number of minutes
participants spend using the Palm Pilots each day. It is also not cumulative. For example,
ifa participant spent 20 minutes the ﬁrst day and 15 minutes the second day, usage

duration was recorded as 20 and 15 minutes, respectively.

62

DATA ANALYSIS

As the purpose of the present study was to examine the dynamic relationships of
key beliefs and usage behaviors-mot to differentiate user experience of multiple
innovations in a comparative sensezul combined user experience data with the seven
applications in analyzing data. In this way, the data may be viewed as user experience of
the Palm Pilot with the preloaded set of applications or features. The combined data at the
Palm Pilot level would provide more robust measures of user experience as they are
measures ofthe means, minimizing measurement errors and other biases associated each
individual application.

To analyze repeated measures data, a number of analytic techniques exist (e. g.,
WABA, time-series regression, hierarchical linear modeling). Many of these techniques
capture the relationship patterns among the data across persons. Yet, in doing so, the
analyses fail to model relationships within each individual and, thus, may miss out on
important individual differences. To provide a more comprehensive picture, the present
study used both hierarchical linear modeling (HLM) and multiple regressions to model
the data. HLM (Bryk & Raudenbush, 1997; Raudenbush & Bryk, 2002) was used to
analyze data from all individuals in the sample. It works in two steps. First, it models data
for each individual separately in order to estimate within-person effects. Then, it
aggregates estimates of within-person effects across individuals. The resulting output
provides estimates of within-person effects for an average individual as well as the

variance of the effects between individuals.

 

3 For readers who are interested in users' behaviors at the application level, please see
Appendix 13 for an averaged users' usage frequency over time broken out by Palm Pilot
applications.

63

In this study, HLM was considered the primary method to evaluate hypotheses, as
it allows us to evaluate whether the hypotheses hold for an average person. Yet this
method has its limitations as it does not directly reveal whether hypotheses are supported
for speciﬁc individuals. To work around this limitation, when HLM results show
signiﬁcant effect variance between individuals, multiple regressions were run to analyze
data for each individual separately. These supplemental analyses help us determine
whether hypotheses hold for speciﬁc individuals and estimate the proportion of sample
participants for whom hypotheses are supported.

In all analyses, outcome expectation and self-efﬁcacy at time t were used to
predict behavioral intention and implementation intention at time t and IT usage at time t
+1. Behavioral and implementation intention at time t were entered as predictors of IT
usage at time t + 1. Usage at time t was used to predict outcome disconfirmation, efficacy
disconfirmation, outcome expectation, and innovation self-efficacy at time t + 1. Finally,
outcome disconfirmation and self-efficacy disconﬁrmation at time t were entered as

predictors of outcome expectation and innovation self-efﬁcacy at time t + 1, respectively.

64

RESULTS

In Table l, descriptive statistics of studied variables are presented. With the
exceptions of the means and SDs, the numbers below the diagonal space represents zero-
ordercd correlations that were calculated using data at all time points from all sample
participants. These correlations were calculated with data at the most granular level
(person by time) and thus ignored the dependency of data within person. It is noted that
zero-ordered correlation between time and usage duration was signiﬁcantly negative.
However, the correlation was very small (r = -.05, p < .01) and could be signiﬁcant due
the large number of data points and their dependency--which tend to create deﬂated
standard errors. An HLM analysis of this relationship, which addresses data dependency
and biased standard error, showed usage duration did not signiﬁcantly decrease over time
(7 = -0.018,p > .10).

Aside from the negative correlation between time and usage duration, other zero-
ordercd correlations at the most granular level generally supported the hypothesized
relationships summarized in Figure 5. Both behavioral intention and implementation
intention were positively related to usage frequency and usage duration. Innovation self-
efficacv and outcome expectation had positive correlations with behavioral intention,
usage frequency, and usage duration. Usage frequency and usage duration were
positively correlated with outcome disconfirmation and efficacy disconfirmation.
Outcome disconﬁrmation and efficacy disconfirmation, in turn, had positive correlations
with outcome expectation and innovation self-efficacy, respectively.

The numbers above the diagonal space in Table 1 represent person-level zero-

correlations. Variables for each participants were averaged across time.

65

mm H 2 5:3 macaw—2.50 337533 2m 83m :25ch 2: 95% 32:52
2 mm M Z 53» 3.392th 29:3 :m Soc .3an 2:: EN 8 9% maﬁa mecca—oboe 3:829: 85% :28me 2.: 3205 338:2

 

 

 

 

 

2. v a <

_0. V Q “I.

8. v a ..

*Qr. 3&0. **0_. 1.9% “tam. **VN. giwm. tumor 6M0 NW: cozy—5A— owNmD .0

“IKE. 3.50. **w0. flaw. .Ibv. from. 1an. .100. mm; VNM .mozozvohh owumD .0
v0. w0. “I; _. 330. 1.00. simm. **m0. ~0.- wvd N06 EQEQELQEQQMNQ Auuumkaw .5
ON. a. v0. .330. .12. 140. simm. m0.- 590 00.m zemuaatmxrauﬂQ 3.333% .0
#10. 10¢. 00.- V0. 3.3”. **Nm. :3... view... wwd mm.m ﬂsxtmgu: RemnautmESEENﬁ
*wm. :ov. 50. *ON. .1wm. 30¢. *iuv. .32. w.0 vwﬁ =¢-=3=~ Essa—33% .v
ram. 3:: No. *2. (E. *2. 1mm. :9: were Em earsxmweem =§§§£ .m
1.3». “1.0m. v0: *NN. *om. ”Luv. ... _ m. N00 5&0 w0.m :SBUVQRN 3.393% .N
gaw— _0.Nm 25,—. .—

0 w h c m w. M N _ Om 53—2 Snafu.)

 

 

 

833:3» 353m he muzmzﬁm 953.83: ._ 03.;

66

38:32:: .5 .3 3:2..oaxm .85 “Eggn— «o 7:52 regains—2% 95853»: "m 9.sz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cough—magma + r ~83— - o
+ .
1+1
+
5:225
:osﬂcoﬁo—aEH +
owmmD
cosm>occ~
,5
20:52:
+ _§o_>m:om
IT
+ cogccccoaa cosﬁooaxm
2:850 + v 05850

 

 

 

 

 

67

Correlations were then calculated using the resulting averages across persons.
The pattern of con‘elational relationships is similar to those ofthe other types of
correlations described earlier. It should be noted that the magnitudes of correlations of
outcome discrepancy to usage frequency and usage duration are relatively high (.20 and
.21 ) but are not signiﬁcant due to the small sample size of 58 persons.

From a theoretical standpoint, behavioral intention and implementation intention
are two related but distinct constructs. Behavioral intention refers the commitment or
desire to carry out a behavior, whereas implementation intention concerns the way the
intended behavior is enacted. It speciﬁes when, where, and how intended behavior is
enacted. To evaluate their relationship from a measurement standpoint, a principal factor
analysis with Promax (oblique) was performed. Results of the rotated factor pattern
matrix (See Appendix 14) showed that two factors were extracted from the common
variance of eight scale items that were used to measure these two constructs. The eigen
value for third factor was 0.36. The two extracted factors explained 83% of the common
variance, and their correlation was .36. Five items with implementation intention
content had loading factors of more than .78 on the ﬁrst factor and less than .13 on the
second factor. Three items with behavioral intention content had loading factors of more
than .87 on the second factor and less than .01 on the ﬁrst factor. Given these values, it
was concluded that behavioral intention and implementation intention are two related but

distinct constructs as measured by their relevant scale items used in this study.

68

To further analyze the data, HLM and multiple regression analyses were
conducted to evaluate each hypothesis. It should be noted that two measures of IT usage,
usage frequency and usage duration, were assessed in this study. Results showed a
similar pattern of relationships when either of these two measures was used as criterion,
which was not surprising given their intercorrelation of .77 (see Table 1). For simplicity,
only results for usage frequency were presented.

To test Hypotheses 1 (Stronger behavioral intention will lead to a higher level of
IT usage) and 2 (Stronger implementation intention will lead to a higher level of IT usage
over and above the effect of behavioral intention), I ran three different models in HLM.
The results are presented in Table 2. In Model 1, usage frequency was entered without
any predictor so that portions of between- and within-person variances of usage
frequency could be determined. Results showed that the model deviance ratio was
12068.390. ICC was .44. Between-person variance was 1.658 (44% of total variance),
and within-person variance was 2.082 (56% of total variance).

To evaluate Hypothesis 1, behavioral intention was then entered as a level-l
predictor of usage frequency with random slopes. As can be seen in Table 2, the model
deviance ratio was 1 1275.272 or 793.118 lower than Model 1’s, indicating that Model 2
ﬁt the data better than the null model. The slopes of behavioral intention effects were
positive on average, (7 = l.047**, p < .01), but they varied signiﬁcantly from person to
person (0 = 0.048**, p < 01). The addition of behavioral intention reduced within-person
variance to 1.635, explaining 22% of total within-person variance. It lowered between-

person variance to 1.192, accounting for 28% of total between-person variance.

69

O~.VQ <

 

 

 

#0. V D. “.1.
m0. V A *
mmoén <o_o.o Nmod **w$.o 29:28“: aoteaemsmﬁé
Swdo 20.0 Rod iwowd «5:225 NEQSSEQ
390an m3; 898.: Iwnog m .252
091.3 **wvo.o 30.0 $550 $0385 Nexoieemm
mnmdnm: mm“: £5.me imi; N .252
:32
03.302 mwod mmwwvcm gimme.— _ .352
8:525» 225m 353...; 2.35m 3:525,
3:559 EVEom EU :83; EU 2.2m Mm > 833...;
-355» -5233—

 

 

 

 

,mogzcogm @933 .5 .5532: .3.:S§E&QE~ 6:“ aetzﬁs 33.55%.5 .3 Buotm .3 mug—«:6, 21—: .N 93.;

7O

Regression analyses of Hypothesis 1 for each individual showed that behavioral
intention had a signiﬁcant positive effect on usage frequency for 96% of participants (see
Appendix 15). As behavioral intention had a positive effect on usage frequency on
average, Hypothesis 1 was supported. Stronger behavioral intention to use IT innovations
led to higher level of usage frequency within individuals. Furthermore, this effect was
found in almost all ofindividuals.

To test Hypothesis 2, in HLM analyses implementation intention was added as a
level-1 predictor of usage frequency with random slopes in Model 3. As shown in Table
2, the model deviance ratio was 10870641 or 404.632 lower than Model 2, indicating
that Model 3 described the data better. The slopes of the implementation intention effect
were positive on average, (y = 0.808**, p < .01), and they marginally varied from person
to person (0 = 0.010“,p = .053).

The addition of implementation intention reduced within-person variance to
1.443, explaining an additional 10% of total within-person variance. It also lowered
between-person variance to 1.078, accounting for an additional 7% of total between-
person variance.

Regression analyses of Hypothesis 2 for each individual showed that
implementation intention had an incremental positive effect on usage frequency over-and-
above the effect of behavioral intention for 82% of participants (see Appendix 15).

Because implementation intention had an independent positive effect on usage
frequency on average, Hypothesis 2 was supported. Stronger implementation intention
resulted in a higher level of usage frequency within persons. This effect was present in a

large majority of the sample participants.

71

To evaluate Hypotheses 3 (Higher innovation self-efﬁcacy will be related to
stronger behavioral intention to use I T innovations) and 10 (Higher outcome expectation
will be associated with stronger behavioral intention to use IT innovations), two models
were run in HLM, and the results are presented in Table 3. In Model 1, behavioral
intention to use IT innovation without any predictor was entered so that the portions of
between- and within-person variance could be determined. Resulting deviance ratio for
the model was 6613.779. Between-person variance was 0.247 (38% of total variance),
and within-person variance was 0.404 (62% of total variance).

ln Model 2, innovation self-efﬁcacy and outcome expectation were added as level-
1 predictors of behavioral intention. As shown in Table 3, the model deviance ratio was
5509.942 or 1 103.837 lower than Model 1’s, indicating that Model 2 ﬁt the data better.
The average slopes of innovation self-eﬂicacy (y = 0.362**, p < .01) and outcome
expectation (y = 0.912**, p < .01) were both signiﬁcantly positive, but they varied
signiﬁcantly from person to person (0 = 0.057**, p < 01 and o = 0.079**, p < 01,
respectively).

Regression analyses of Hypothesis 3 for each individual showed that outcome
expectation had a signiﬁcant independent effect on behavioral intention for 93% of
participants, whereas innovation self-eﬂicacy effect was signiﬁcant for 53% of
participants (see Appendix 17).

Given that innovation self-efﬁcacy and outcome expectation had independent
positive effects with behavioral intention. Hypotheses 3 and 10 were supported. Higher

levels of innovation self-efﬁcacy and outcome expectation led to a stronger intention to

72

_0. Va **

 

AuuoRxNKNmm 2023625

tbzﬁumeRm NEOUwBO

N 6—52

:32

w .252

 

 

 

 

23...;

 

=o.:§~=~ REFSSEM =o Auaugmubmh .SBSSEQ 6:.“ :etuﬁmakm 2.593% .3 882m he mombw=< 5:: .m «Zak.

73

 

use IT innovations within individuals. The positive effect ofinnovation self-efﬁcacy was
observed in about half of the sample, whereas the impact of outcome expectation was
found in almost all ofthe sample participants.

To examine Hypothesis 4 (Higher innovation self-efﬁcacy will be associated with
stronger implementation intention), two models were analyzed in HLM, and the results
are presented in Table 4. In Model 1, implementation intention was analyzed as a
criterion without any predictor. The model deviance ratio was 7670.179. Between-person
variance was 0.224 (29% of total variance), whereas within-person variance was 0.559
(71% of total variance).

ln Model 2, innovation self-efﬁcaqt and outcome expectation were entered as
level-l predictors of implementation intention with random slopes. As shown in Table 4,
the model’s deviance ratio was 7091.909 or 578.270 lower than Model 1’s, indicating
that Model 2 ﬁt the data better. The average slopes of innovation self-efficacy (y =
0.619***, p < .01) and outcome expectation (y = 0.388**,p < .01) were both
signiﬁcantly positive, and only innovation self-efﬁcacy slopes varied signiﬁcantly from
person to person (0 = 0.029**,p < 01).

Regression analyses of Hypothesis 4 showed that innovation self-efﬁcacy had a
signiﬁcant independent effect on implementation intention for 91% of participants (see
Appendix 18). As innovation self-efﬁcacy had an independent positive effect on
implementation intention on average, Hypothesis 4 was supported. Higher levels of

innovatitm self-efﬁcacy resulted in stronger implementation intention with persons.

74

—O.VQ**

 

 

 

 

mmmdm ioNod wmod .3856 Agnewﬁmkﬁw 223.825
2.9% 20.0 mvod iwwmd 29.....3mxm 3.8.3.5
moo. 50.. Need cow. _ mo **ONN.o N .252
:32
3.6%.. ammd wnvsom. VNNd . .252
3....25» Egam 3....25» Exam 3525»
3.53.5 :83.— EU .595.— EU 2.2m mm > 2925»
-555» $3.53:

 

 

.5532: 29.32.353.55 .5 .Noeogm3394 {$332.5 6.... =e......3mxm~ 2.38.5 he Euchum me 39:23. 24: .v 335..

75

The effect was observed in almost all of the sample participants. It is also noted that
outcome expectation had a signiﬁcant independent effect on implementation intention,
although not predicted, and its effect did not vary between persons.

Two models were run in HLM to evaluate Hypotheses 5 (Higher innovation self-
ef/icacr will lead a higher level of I T usage) and 11 (Higher outcome expectation will
lead to a higher level of 1 T usage), and the results are presented in Table 5. In Model 1,
usage frequency was entered as a criterion without any predictor. The model deviance
ratio was 12068.390. ICC was .44. Between-person variance was 1.658 (44% oftotal
variance in usage frequency), whereas within-person variance was 2.082 (56% of total
variance).

ln Model 2, innovation selfefj’icacy and outcome expectation were added as level-
] predictors of usage frequency with random slopes. The results of the analysis for this
model is presented in Table 5. The model’s deviance ratio was 11415.858 or 652.532
lower than Model 1’s, indicating that Model 2 ﬁt the data better. The average slopes of
innovation self-efﬁcacy (y = 0.607“, p < .01) and outcome expectation (y = 1.675**, p <
.01) were both signiﬁcantly positive, and only innovation self-efﬁcacy slopes varied
significantly between persons (0 = 0.167**, p < 01).

Regression analyses of Hypothesis 5 showed the independent effect of innovation
self-(jiicacv on usage frequency varies: positive for 54%, insigniﬁcant for 40%, and
negative for 6% of participants (see Appendix 19).

Because innovation self-eﬂicacy and outcome expectation had independent

positive effects on usage frequency on average, Hypotheses 5 and 11 were supported.

76

Sun:

 

 

 

 

77

 

 

 

#0. V a as."
.I- . mo. V Q *
voméaz 3.5020 owed "3300.0 AbUQwRQKNmW .52....025
vvmﬁo mood mwod “InmhoA 20...».ummwm 3.59.39
www.2ﬂ2 hwc; :tmwo: **mmm.~ NEUc—Z
=32
oamwoom— Nwo.N mmwwvom **wm©._ — Etc—Z
3525» v.5..am 3.525» ouasdm 3525» I
3.532. :83.— EU 595.. EU 2.2m Hm > 3.525»
-5555 u=oo>50m

 

 

.3533...— 53: .8 .Ggthuhmu. =e.....ae....~ v.5 .32233N 3.32.5 .3 832m 2. woman—5.3. 5:: .m 2.5..

 

Higher levels of im'zovation self-eﬂicacy resulted in higher levels of usage frequency
within individuals. The positive effect of innovation self-efﬁcacy was found for about
half of the sample, whereas the outcome expectation effect was consistent from
individual to individual.

To evaluate Hypotheses 6 (The effect of innovation self—eﬁcacy on IT usage will
weaken over time) and 12 (The effect of outcome expectation on IT usage will weaken
over time), three models were run in HLM. The results are presented in Table 6. In
Model 1. outcome expectation, innovation self-eﬂicacy, and time (number of days) were
entered as predictors of usage frequency so that it could be used as a base model to
evaluate interaction effects. Resulting deviance ratio for Model 1 was 11356.372.

In Model 2, the interaction term for outcome expectation and time was added. As
shown in Table 6, the model deviance ratio was 11361.621 or 5.249 higher than Model
1’s, indicating that Model 2 did not ﬁt the data better than Model 1. The average slope of
the interaction terms (7 = 0.003, p > .05) was not signiﬁcantly different from zero. These
results suggested the interaction effect of outcome expectation by time was not present.
Hypothesis 12 was thus not supported.

In Model 3, the interaction term for innovation self-efﬁcacy and time was added to
Model 1. As shown in Table 6, the model’s deviance ratio was 11358.463 or 7.909 lower
than the Model 1’s, suggesting Model 3 ﬁt the data better. The average slopes of the
interaction tenns (y = -0.008**, p < .01) were signiﬁcantly negative and varied from
individual to individual (0 = 0.163**,p < 01). Regression analyses showed that
innovation self-efﬁcacy X time interaction term was signiﬁcantly negative for only 12%

of participants (see Appendix 20).

78

2.1:

 

 

 

 

 

Eva:
mova ._.
3.3 1.88... So... tweed- 2.5 3.38.5.5... 53.65..
5.8. 3.88... So... So... as...
.38 18.... as... :58... 38%....33 .2385:
$25: 33 32.8 :3. £20 :3 Sod to... 223.83.338.30
22...:
832: £283 go... So... 0523223333.... 35836
83.2 1.88... 8o... 80.... as...
$32 1.2... .8... 39%... 58.33.53 223.65..
5.33: .83 333 1.382 .8... 1.2... 5o... 1%... 23.33333335830
2252
832 :88... 89o Sod- as...
2...; .18.... as... 3.83.0 38.33.35 22335..
.28.: 23 $352128... $2. *2; £0... ta... 22.38%... 358:6
:25:
3.525» 35....m 3.525» 35:5m 3.52.3
3533a :35.— EU :85.— EU 2.2m mm > 3352.3
-555» -505...

 

 

 

3E; .52. 3:353...— 352. .8 A333m§m~$3h .523325 6.5 =3......33S..m~ 353325 .3 83.3— ..o weak—5.2. 21—: .c 335,—.

79

Given that the interaction term of innovation self-efﬁcacy by time was
si gnificantl y negative on average, Hypothesis 6 was supported. This suggests that
innovation soif-cf/iatev effect on usage frequency weakened over time within persons.
However. this effect was not frequently observed, as it was found in a very small percent
of sample participants.

Two models were run in HLM to evaluate Hypothesis 7 (Innovation self—efﬁcacy
will grow overtime), and the results are presented in Table 7. In Model 1, innovation self-
eﬂicaev was used as a criterion without any predictor. Resulting deviance ratio for the
model was 3862.666. ICC was .48. Between-person variance was 0.160 (48% of total
variance in behavioral intention), whereas within-person variance was 0.175 (52% of
total variance).

ln Model 2, time (number of days) was added as a level-l predictor of innovation
sci/lef/it'ac')‘ with random slopes. As shown in Table 7, the model deviance ratio was
2580.782 or 1281.884 lower than Model 1’s, indicating that Model 2 fit the data better.
The slopes were positive on average, (7 = 0.012**, p < .01), and they varied significantly
from person to person (0 = 0.00003**, 1) < 01).

Regression analyses showed that time had a significant positive effect on
innovation self-eﬂicaQI for all participants (see Appendix 21).

As time had a positive effect on innovation self-efﬁcacy on average, Hypothesis 7
was supported. Innovation self-efﬁcacy increased within individuals over time. This

effect was observed in all sample participants.

80

—O.VQ**

 

 

 

.3de 38256 nowoood .12 86 on...»
$56me 3 .d 203? .1. 6.6 N .252
:22
www.mowm mid mmnw 5m 1.03.0 — .252
3.325» 3535 3.525» 3.25m 3.525»
353.69 :83.— EU :83.— EU 2.2m Mm > 2925»
-553. $850..

 

 

 

3...; .95 53.20 Auuumwmxmubmm. .3..§8==~ .3 392.22. 5:: .5 «35—.

81

To test Hypotheses 8 (IT usage will have a positive effect on innovation self-
ef/icacv) and 9 (Emcacy disconﬁrmation will mediate the effect of I T usage on innovation
sci/lclficacv), three models were run in HLM, and the results are shown in Table 8. In
Model 1, innovation self-efﬁcacv was used as criterion without any predictor. Resulting
deviance ratio for the model was 3862.666. ICC was .48. Between-person variance was
0.160 (48% of total variance in innovation self-efficacy), whereas within-person variance
was 0.175 (52% oftotal variance).

In Model 2, usage frequency was added as a level-1 predictor of innovation self-
efﬁaicv with random slopes. As shown in Table 8, the model deviance ratio was
3682997 or 179.669 lower than Model 1’s, indicating that Model 2 fit the data better.
The slopes of usage frequency effect were positive on average, (7 = 0.047”, p < .01),
and they significantly varied from person to person (0 = 0.006**,p < .01).

Regression analyses showed that usage frequency had a positive effect on
innovation self-efﬁcacy for 30% of participants, a negative effect for 6% of the sample,
and non-significant effect for the rest (see Appendix 22).

In Model 3 efficacy disconﬁrmation was added as a level-1 predictor with random
slopes in HLM analyses. As shown in Table 8, the resulting deviance ratio for the model
was 3302.435 or 380.562 lower than Model 2’s, indicating that Model 3 fit the data
better. The slopes of efﬁcacv disconﬁrmation and usage frequency effects were positive
on average, (7 = 0.285**, p < .01; and y = 0.056**, p < .01, respectively), and
only usage frequency slopes significantly varied from person to person (0 = 0.006**, p <
.01 ). This suggests efﬁcacy disconﬁrmation and usage frequency had independent

positive effects on innovation self—efficacy.

82

~O.VQ**

 

 

 

52.3 .000 30.0 **mwm.0 zones-.mxzooﬁnNAoeoRxm
8:: 1.805 :35 5.585 585.525. .83:
3.80m N30 mm..mmm 1.32.0 m .252
Roam 5.8.3 :90 $23.0 .8885”. 83:
$3 3.0 R._m0m 3.20 N .0632
:22
$.mowm 3.0 mm.m.0m 300.0 . .352
35.25» 3.51% 3.525» 35:0m 3.325»
3.5.69 :83.— EU :83.— EU one—m mm > 03525»
-555. -5353

 

 

 

A35§m~$mh 3.335: .8 3.35.0?35-5 33.05% a...“ 53:03,..— owamD .3 330%.. .3 mama—«.3. 532 .0 oz:-

83

To further evaluate Hypothesis 9, two other models were run in HLM to evaluate
the inﬂuence of IT usage on efficacy disconfirmation. Results are showed in Table 9. In
Model 1, efficacy disconﬁrmation was used as criterion without any predictor. Resulting
deviance ratio for the model was 4520.854. ICC was close to 1.0. Between-person
variance was close to .001 (0% of total variance in efﬁcacy disconﬁrmation), whereas
within-person variance was 0.228 (100% of total variance).

ln Model 2, usage frequency was added as a level-1 predictor of efﬁcacy
iliscon/irination with random slopes. As shown in Table 9, the model deviance ratio was
4510.345 or 9.491 lower than Model 1’s, indicating that Model 2 ﬁt the data better. The
slopes of usage frequency effect were positive on average, (y = 0.018**, p > .01), and
they did not significantly vary from person to person (0 = 0.004, p > .10). This shows that
usage frequency had a positive effect on efﬁcacy disconﬁrmation. Given that usage
frequency had a positive effect on innovation self-efficacy on average, Hypothesis 8 was
supported. Higher levels of usage frequency led to higher levels of innovation self-
efﬁcacv within persons. However, this effect was observed in a small portion of sample
participants (30%).

Because (a) usage frequency and innovation disconﬁrmation had independent
positive effects on innovation self—efficacy on average and (b) usage frequency had
positive influence on efficacy disconfirmation on average, Hypothesis 9 was supported.
These outcomes suggest that efficacy disconﬁrmation partially mediated the effect of

usage frequency on innovation self-efficacy.

84

~O.VQ**

 

 

 

5.38 3.5.0 wood .55 So 55:55”. 555
$2 _ 3 2.8 $5.? .8... N .282
:52
5.3.82 53 mag .85 . 5.52
3:25» 3.3—6 3:..25» 35:0m 3:525»
3:25: :83.— EU :83.— EU one—m Hm > 20525»

5.55.3 -5255:

 

 

 

=:..3=...N\=3M5 Aoeomxtxm :0 55:03,: «mam: a: 330:— 0: mama—5:4,. $3: .0 205,—-

85

To evaluate Hypotheses 13 (IT usage will have a positive effect on outcome
expectation) and 14 (Outcome disconﬁrmation will mediate the effect of I T usage on
outcome expectation), three models were run in HLM, and the results are presented in
Table 10. In Model 1, outcome expectation was entered as a criterion without any
predictor. Resulting deviance ratio for the model was 1685.8. ICC was .40. Between-
person variance was 0.134 (60% of total variance), whereas within-person variance was
0.090 (4091- oftotal variance).

In Model 2, usage frequency was added as a level-1 predictor of outcome
expectation with random slopes. As shown in Table 10, the model’s deviance ratio was
1463.44 or 222.36 lower than Model 1’s, indicating that Model 2 fit the data better. The
slopes of usage frequency effect were positive on average, (7 = 0.055**,p < .01), and
they varied from person to person (0 = 0.00067”, p < .01).

Regression analyses indicated that usage frequency had a signiﬁcant positive
effect on outcome expectation for 72% of participants (see Appendix 23). In Model 3
outcome-disconfimiation was added as a level-1 predictor with random slopes in HLM
analyses. As shown in Table 10, the model’s deviance ratio was 822.84 or 640.6 lower
than Model 25, indicating that Model 3 fit the data better. The slopes of outcome
(lisr'on/irmation and usage frequency effects were positive on average (7 = 0.212**, p <
.01; and y = 0.065**, p < .01), and only usage frequency slopes significantly varied from
person to person (0 = 0.00065**,p < .01).

To further evaluate Hypothesis 14, two additional models were run in HLM to

anal yzc the influence of usage frequency on outcome disconﬁrmation. Results are

86

_O.VQ**

 

 

 

 

 

www.mm 0000.0 000.0 1.2.60 noccﬁxﬁtooﬁQmEooSQ
0vw.mmw @000 08.08? 5.020 308$ 1.30000 m00.0 1.30.0 55500500.me
2352
03.83 30.0 mmwsgm 3.02.0 $0.0: 1.500000 m00.0 *._.mm0.0 homozcoemowmmD
:25:
000.32 000.0 wowsmov 102.0 :32
20032
3.525» «.335 353...; Pas—um 8.325»
3:559 5200 EU ace—om EU 2.2m Hm > 033...;
-555» -5938:

 

 

 

.S.:E.§5.m 2:335 :c greatmﬁsamQ East‘s 6:5 55:09.... «mam: ho £85m 00 mom.m_a:< 2.:— .0_ «BF—t

87

shown in Table l 1. In Model 1, outcome disconfirmation was used as criterion without
any predictor. Resulting deviance ratio for the model was 5646.069. ICC was close to
1.0. Between-person variance was .001 (0% of total variance in efﬁcacy disconﬁrmation),
whereas within-person variance was 0.321 (100% of total variance).

In Model 2, usage frequency was added as a level—1 predictor of outcome
tliscon/irtnation with random slopes. As shown in Table 11, the model deviance ratio was
563 I .38‘) or 14.680 lower than Model 1’s, indicating that Model 2 fit the data better. The
slopes of usage frequency effect were positive on average, (7 = 0.024**, p < .01), and
they did not significantly vary from person to person (0 = 0.000, p > .10).

Given that usage frequency had a positive effect on outcome disconﬁrmation on
average. Hypothesis 13 was supported. Higher levels of usage frequency led to higher
levels ofoutcome expectation within persons. This effect was observed in a majority of
sample participants (72%).

Because (a) usage frequency and outcome disconﬁrmation had independent
positive effects on outcome expectation on average and (b) usage frequency had positive
influence on outcome disconﬁrmation on average, Hypothesis 14 was supported. These
results suggest that outcome disconﬁrmation partially mediated the effect of usage

frequency on outcome expectation.

88

_O.VQ**

 

 

 

53% o «85 15.85 3:83.: 05$:
3.20m 05.0 Km.mm 000.0 N 3:52
:52
50.05% _mm.0 Seam _00.0 :25:
3:525» 25:0m 3:525» 2535 3:525»
3:532— :83: EU :83: EU 2.2m mm > 53525»
55:?» -5553—

 

 

 

:SEEhPSMS 55825 :: 5:232...— ow5mD no 38tm a: 85355.. SE: .2 5.05,:

89

DISCUSSION

.S'ununatjv of Findings

The objective ofthe present study was to evaluate a dynamic model in which
innovation self—efficacy and outcome expectation change, inﬂuence, and are influenced by
IT usage over time within persons. Results, with a few omissions noted below, were
summarized in Figure 6. The discussions below focused on usage frequency as a measure
of usage frequency, it should be noted that a similar pattern of relationships was found
when usage duration was used.

At a broad level, results showed that innovation self-eﬂicacy—which increased
over time~and outcome expectation affected and were affected by IT usage dynamically.
Speciﬁcally, both innovation self-efficacy and outcome expectation had independent
positive effects on usage frequency. The effect of innovation self-efficacy varied from
person to person and was signiﬁcantly positive for 54% of sample participants, whereas
the effect of outcome expectation was consistent and did not change between persons.
The effect of innovation self-efﬁcacy on usage frequency weakened over time (not noted
in Figure 6), but this effect pattern was observed in only 12% of the sample. The
relationship of innovation self-efﬁcacy and outcome expectation with IT frequency was
also evidenced in the reverse direction. Usage frequency had positive effects on
innovation self-efficacy and outcome expectation. The effect on innovation self-eﬂicacy
varied between persons and was positive for 30% of sample participants, whereas the
effect on outcome expectation also changed from person to person and was positive for

72% ofthe sample.

90

2555.5:5 r: .8 3:50.505”..— ..8: 855:»: a: .232 Eo~5_:wom¢_om 03.35% 50 950E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

+
5558.: :82 +
. c .o v 58:55.28
3550*” 855.550
+ .
+
[T
50:85
:25558295 +
ow5mD
500535:
{E 50:85 +
+ .5535nom
+ 8058:0585 + :ocﬁoomxm
580850 + v 2:850

 

 

 

 

 

 

 

 

91

At a more detailed level, the present study found that the effects of innovation
s'el/letiicacj‘ and outcome expectation on usage frequency operated through behavioral
intention and implementation intention. Innovation self-eﬂicacy and outcome expectation
were found to have independent positive effects on behavioral intention. The effect of
innovation self—efficacv varied between persons and was positive for 71% of the
participants. whereas the effect of outcome expectation also varied between persons and
was positive for 93% of the sample. Innovation self-efficacy and outcome expectation
were also found to have independent influence on implementation intention. The effect of
innovation self-efﬁcacy on implementation intention varied between persons and was
positive for 91" 0 of the sample, where as the effect of outcome expectation did not
change from person to person. Behavioral intention and implementation intention in turn
had positive independent effects on usage frequency. The effect of behavioral intention
varied between persons and was positive for 96% of participants, whereas the effect of
imp/ententation intention also varied between persons and was positive for 82% of the
sample.

Results also showed that the effects of usage frequency on innovation self-efﬁcacy
and outcome expectation on usage frequency were partially mediated by efﬁcacy
disconfirntation and outcome disconﬁrmation, respectively. Usage frequency had a
positive effect on efﬁcacy discomiirmation, and its effect did not vary between persons.
Efficacy disconﬁrmation in turn had a positive inﬂuence on innovation self-efﬁcacy, and
its effect also did not vary between persons. Similarly, usage frequency had a positive

effect on outcome disconfirmation, and its effect did not vary between persons. Eﬂicacy

92

iliscon/irtnation in turn had a positive influence on innovation self-efficacv, and its effect
did not vary between persons.

In summary, the present study found that innovation self-efﬁcacy and outcome
expectation affected and were affected by IT usage dynamically. The effect of innovation
s‘c>l/1cf[fi('tic;i‘ and outcome expectation on usage frequency operated through behavioral
intention and implementation intention, whereas the influence of usage frequency on
innovation .sel/lefﬁcacv and outcome expectation were partially channeled through
efficaev tlisconﬁrnuition and outcome disconﬁrmation, respectively. Yet, there were
significant individual differences in most of these relationships. The relationships
between outcome expectation and usage frequency were more consistent or frequently
observed in individuals than the relationships between innovation self-efﬁcacy and usage
frequency.

Research Contribution and Implications

The present study is the first to evaluate a within-person process model of how
key beliefs influence and are inﬂuenced by IT innovation usage behaviors. It contributes
to the current knowledge base in four key areas. lFirst, this study supported existing
research showing that innovation self-efﬁcacy, outcome expectation, and behavioral
intention are key beliefs that drive IT usage behaviors. More importantly, it showed that
these key beliefs change and explain varying IT usage within persons over time and thus
accounted for behavioral changes within persons. This idea is not new and is assumed in
dominant models such as the TAM (e.g., Davis et al., 1989) and the TPB (Ajzen, 1991).
Yet, existing studies, relying on between-person analyses of cross-sectional data or

datasets collected at a few time points, have not been able to provide evidence for such

93

intra-person processes. Rather, they showed that a person with high outcome expectation
(or innovation self-efﬁcacy) is likely to have a stronger intention to use IT innovation,
which leads him to more IT innovation usage than another person who has low outcome
expectation (or innovation self—efficacy) does. Such a between-person analysis does not
reveal the dynamic processes that occur within persons. By focusing on intra-person
processes, the present study showed that when a person has higher outcome expectation
(or innovation .S'el/lefficaq’), he tends have a stronger intention to use IT innovation (or
higher levels of IT usage) than he does when he has lower outcome expectation (or
innovation self-efficacy). The difference, although it may seem minor on the surface, is
important. Knowledge of within-person dynamic processes that account for behavioral
change over time is the foundation of intervening efforts to train and/or persuade people

to use IT innovations.

Second, the present study showed that implementation intention is a proximal
determinant of IT usage in addition to behavioral intention. This ﬁnding expands the
current view of key psychological factors that are important in predicting and explaining
IT usage. Current IT adoption models such as the TAM (e.g., Davis et al., 1989) and the
TPB (Ajzen, 1991) assume that behavioral intention is the only proximal determinant of
IT usage. To predict IT adoption, researchers typically backtrack and try to identify
antecedents of behavioral intention. This approach is most useful if behavioral intention
is the only proximal determinant of IT usage. The present research does not dispute the
important role of behavioral intention. In fact, it found that behavioral intention
explained 22% of within-person variance in IT usage and 28% of between-person

variance. Yet, it argued that behavioral intention is not the only proximal determinant.

94

Implementation intention was found to have an incremental effect on IT usage,
explaining an additional 10% of within-person variance and 7% of between-person
variance above-and-beyond behavioral intention. As a result, implementation intention is

viewed as another proximal determinant of IT innovation usage.

The ﬁnding of implementation intention effect is consistent with Gollwitzer’s
(1999) work that shows implementation intention increases the probability of behaviors
to be enacted when individuals have general behavioral intention to carry out such
behaviors. Its effect stems from the fact that the path from behavioral intention to
behavior is not automatic. People need cognitive resource to execute their behavioral
intention, but they often face many distractions and competing thoughts. Implementation
intention helps people focus, connect situational cues to behavioral responses, and shield
themselves from distracting alternatives (Gollwitzer). It, thus, gives people a better

chance to carry out their intended behaviors.

Given the role of implementation intention, future studies perhaps should
incorporate this construct into their models and evaluate its effect. Additionally, effort to
examine potential antecedents of implementation intention—other than innovation self-
e[/icac_v and outcome expectation—could be useful, as it could lead to additional

psychological factors responsible for IT usage.

The third contribution of the current research regards the ﬁndings that IT usage
influenced innovation self-efﬁcacy and outcome expectation within persons and the
processes through which the effects occur. These ﬁndings extend our knowledge base in
a signiﬁcant way. Current models of IT innovation adoption such as the TAM (e. g.,

Davis et al., 1989) and the TPB (Ajzen, 1991) propose that equivalents of innovation self-

95

efficacv and outcome expectation (ease of use and perceived useﬁilness) influence IT
usage. Although they assume that these two psychological drivers are changed as results
of IT experience, they have not speciﬁed speciﬁc linkages, and no studies, to my best
knowledge, have examined this issue. The present study showed that IT usage had
positive effects on innovation self-efficacy and outcome expectation. F urthermore,
efficacv tliscon/irmation—the discrepancy between prior innovation self-efﬁcacy and
perceived ability, an internal measure of personal performance—partially mediated the
effect of IT usage on innovation self-efﬁcacy within person, whereas outcome
disc'on/irmation—the discrepancy between prior outcome expectation and perceived
outcomes, an internal measure of product performance—partially mediated the effect of
IT usage on outcome expectation. In other words, IT usage directly and indirectly through
e/j’icacv tlixt'on/irmation and outcome disconﬁrmation affects innovation self-efficacy and
outcome expectation within persons. With these ﬁndings, user experience is viewed as a
dynamic process in which key beliefs—innovation self-eﬂicacy and outcome
expectation-—determine and are determined by IT usage within persons. It is a cyclical

process in which beliefs and behaviors mutually reinforce each other.

Fourth, though neither last nor least, the present study analyzed whether
individuals differ in their intra-process and quantiﬁed the proportion of individuals to
whom a speciﬁc process is applied (Proportions of the sample for which speciﬁc
processes are applied are reported in this study. However, for simplicity I did not infer
conﬁdence intervals for those proportions, which could be easily estimated). This type of
analysis is tedious and time consuming. It may also seem to clutter the study’s results and

obfuscate key gleaned insights. Yet, it presents a more precise knowledge of intra-person

96

processes among individuals. Researchers often ask whether a process exists (e.g., Does
innomtion self-efficacv influence IT usage within person?) and typically detemiine the
answer at an aggregate, “average” level (e. g., Innovation self-efﬁcacy has a positive

in II ucnce on IT usage for an average person.) What does not get asked is if the process
exists at an aggregate level, what is the proportion of people for whom the process is at
work? The latter question is important, if not more, to answer from research and practical
standpoints. For example, this study found that outcome expectation influenced IT usage
within person for an average person, and this effect did not signiﬁcantly vary between
persons. Similarly, innovation self—eﬁcacv was also found to have a positive impact on
IT usage on average, but its effect changed between persons and was signiﬁcantly
positive for 52% of the sample. From a research standpoint, these ﬁndings raise the
question of why the effect of innovation self-efficacy was present in half of the sample
but not the other. Perhaps, individual factors could be at work, an issue that future
research could examine. From a practical standpoint, if practitioners have to choose
whether to alter innovation self-efﬁcacy or outcome expectation in order to encourage
Palm Pilot usage, these ﬁndings suggest that the better bet is outcome expectation, as its
effect is consistent across individuals. In another example, the current research found that
the effect of innovation self-efﬁcacy on usage frequency weakens over time for an
averaged person. However, this effect varied from person to person and was signiﬁcant
for only 12% of the sample. These ﬁndings raise the question why the weakening effect
exists in some individuals but not others, an issue that future studies could address. They
also suggest that the weakening effect of innovation self-efficacy may not be of

importance and perhaps could be ignored .from a practical standpoint, as it is evidenced in

97

a very small group of individuals. These examples illustrate the signiﬁcance of knowing
proportions of individuals for which a speciﬁc process or relationship is relevant. They
present a new approach to look at data—one that the present study, to my best
knowledge, is the ﬁrst to conduct.

Practical Implications

As this research is based on observational data, causation inferences may not be
warranted. If the ﬁndings in this study are to be replicated in controlled experiments
where drivers of the change processes are manipulated and systematically altered, a
number of implications may be relevant to practitioners. The ﬁrst set of implications
concerns the issue of product design. Because outcome expectation may be a consistent
predictor of IT usage over time, it may be an important design factor. Products that
generate hiin outcome expectation among users may be more likely to be adopted.
Product designers may beneﬁt signiﬁcantly by viewing the product from users’
perspective and providing product features that users think are useful. Useful features and
hi gh levels of beneﬁts may enhance users’ outcome expectation and thus likely encourage
adoption and usage. Besides outcome expectation, users’ perception of their ability to use
the innovation (innovation self-eficacy) may be another important factor to consider.
Perceived ease of use may enhance adoption rates because it may inﬂuence IT innovation
usage above-and-beyond outcome expectation. When design engineers focus on products’
functionalities while neglecting users’ abilities to use them, they may end up with great
features that users may have troubles Ieaming and adopting. It may be greatly beneﬁcial
to users when products are built with user friendly, easy-to-use features from their

perspective. Useful and easy to use products from users' standpoint may be more likely

98

to be adopted and used, as indicated by the positive independent effects of outcome

expectation and innovation self-efﬁcacy on usage behaviors in this study.

The second set of implications regards how IT innovations, once designed, should
be introduced to users. Because outcome expectation may be an important driver of usage
behavior, innovations’ beneﬁts may need to be communicated to users early, in order to
help shape their opinions. Communication attempts may address what beneﬁts the
innovations provide and how they might fulﬁll users’ needs. In addition to “selling”
innovative products, the ﬁndings in this study suggest that training and coaching may be
crucial, especially during the initial phase of product adoption. Training and coaching
may help the adoption process as they may enable users to use the products well and
boost their sch-efﬁcacy. Beyond “selling” and training, developing an implementation
mindset may also facilitate users' adoption. As this research and other studies suggested,
having an implementation mindset may help users carry out intended behaviors. It
enables users to develop strong associations of when and where they should use the

innovations and to follow through with their intention to use the innovations.

The third set of implications concerns what may be done to enhance the quality of
user experience. The present study found that when users’ expectations were
unrealistically high, they were likely to adjust their expectations downwards after going
through actual experience. Failing to meet expectations were assumed to generate
negative affect and lead to a higher likelihood of downward adjustment of expectations.

Conversely, when users’ expectations were unrealistically low, they were likely to adjust

99

their expectations upwards. Exceeding expectations were assumed to create positive

feelings and a higher likelihood of upward revision of expectations3.

These ﬁndings thus suggest that practitioners may need to avoid setting
unrealistically high expectations or oversell the products. Doing so may create negative
user experience, which may discourage future usage. Instead, communicating the beneﬁts
of IT innovations in an incremental way and then providing training to help a user meet
or exceed their expectation at every turn may be more beneﬁcial. This process could be
done in iterations. Initially, realistic beneﬁts of IT innovations that are easier to obtain
may be explained to users. Users may need to be convinced that the innovations are
useful to them and they could successfully use the innovations. Expectations that are high
enough to convince users of the beneﬁts of the innovations may be more helpful, but they
should not be set too high to create a disconﬁrming experience. Training and coaching
that enable users to realize their expectations may provide additional beneﬁts. The
process of communicating realistic beneﬁts, training, and coaching users materialize their
expectations may be repeated until users reach advanced skills that allow them to fully
exploit potential beneﬁts of the innovations. Although this continuous “realistic-promise
and over-deliver” strategy may need frequent assessments of users’ expectations and
skills and interventions along the way, it have the potential of providing a positive user

experience and fostering commitment to long-term usage of new innovations.

 

" Although one may attribute these adjustments as the results of regressions to the means and the resulting
fluctuations as artifacts, it is unlikely to be the case. Results showed that IT innovation self-efficacy
increased with time, and thus suggested that changes in innovation self-efficacy were real and meaningful.
Although outcome expectation did not signiﬁcantly increase or decrease over time, its variance was
predicted by outcome disconfirmation, a construct that captures the discrepancy between experience and
expectation. This suggests that random regression to the means was not at play as the ﬂuctuations in
outcome expectation were explained and thus not random.

100

Bountlatjv Conditions and Future Research

The results of the present study have a number of boundary conditions. First,
because the sample contained only undergraduate students who were young and had
limited work experience. The ﬁndings are applicable to this population. Questions remain
whether results would be generalizable to adults in work settings. In some work settings,
the effect of confomting to social norms and authority may be stronger. In addition,
workers may be obligated to carry out certain behaviors regardless of their perceptions of
benefits other than fulﬁlling ones’ obligations. The effects of social inﬂuence and social
contract may be at a preconscious level but nevertheless impact behaviors. As a result,

much research is needed to replicate the ﬁndings in work settings with workers’ samples.

The second boundary condition regards the present study’s focus on the early
stage of IT adoption. This focus allows us to capture users experience at a critical stage
when their beliefs and behaviors form and change. Examinations of user experience at the
early stage of adoption are critical because users make their decision to continue usage of
the innovations at this stage. However, the ﬁndings at the early stage may not be
applicable to user experience late in the adoption process. In later stages, users’ beliefs
and behaviors may stabilize and may not exhibit much variance. The intra-person model
in the present study may not be applicable. Clearly, research is needed to examine user
experience at a later stage in the adoption process.

Another boundary condition regards the nature of the IT innovation. In this study,
we examined the adoption of Palm Pilots with a speciﬁc set of applications. This type of
technology could be operated independently by individuals. Usage by one individual does

not affect others. Although this type of technology reﬂects the nature of many

101

innovations, it is different from the other innovations that require collaborative usage of
multiple individuals (e.g., Enterprise Resource Planning system). Thus, future research
could evaluate whether the ﬁndings in the present study generalize user-interdependent

IT innovations.

Besides testing additional boundary conditions, future research could beneﬁt by
extending the current model by identifying and testing exogenous factors that inﬂuence
outcome expectation and innovation self-efficacy, key drivers of IT innovation usage. In
past between-person studies, numerous factors have been found to inﬂuence self-efﬁcacy
and outcome expectation. For example, persuasion and behavioral modeling alter self-
etiicacv and outcome expectation (Bandura, 1997). IT innovations’ qualities such as job
relevance, output quality, and result demonstrability have signiﬁcant impacts on outcome
expectation (Venkatesh & Davis, 2000), whereas qualities such as complexity, reliability,
ﬂexibility. integration, accessibility, and timeliness inﬂuence innovation self-efficacy
(Wixom & Todd, 2005). Future research could examine the effects of those factors with
respect to time using within-person design. Such research would be useful in determining
what factors to change and when to implement the changes in order to encourage IT
innovation usage.

As signiﬁcant between-person variances were found for many relationships
described in this research, a fruitful area would be to investigate and identify individual
difference factors that may be responsible for the observed variances across people. As
technology innovations become more integrated and sophisticated, ability to cope with
cognitive overloading can be a factor. As results, cognitive ability may be an important

factor affecting adoption behaviors. Individuals with high levels of cognitive ability may

102

be more likely to adopt and adapt to new technology innovations as they have the
capacity to deal and cope with cognitive overloading associated with new innovations.
Personality factors such as conscientiousness and openness to experience may be
signiﬁcant factors. Research in personality domain has showed that individuals with high
levels ofconscientiousness tend to be organized, disciplined, and efﬁcient. They may be
likely to be drawn into certain types of innovations such as integrated device that would
help them organize tasks and activities and keep things in orderly, efﬁcient ways.
Individuals who are more open to new experience may tend to adopt new innovations.
Yet they may be willing to abandon older innovations when newer ones are available and
thus may adopt more innovations over time.
Conclusions

The present study is the ﬁrst to evaluate a within-person process model of how
key beliefs influence and are inﬂuenced by IT innovation usage behaviors. Results
showed that innovation self-eﬁcacy and outcome expectation affected and were affected
by IT usage dynamically within persons. The effect of innovation self-efficacy and
outcome expectation on usage frequency operated through behavioral intention and
imp/ementation intention, whereas the inﬂuence of usage frequency on innovation self-
etiicacv and outcome expectation were partially channeled by efficacy disconfirmation
and outcome disconﬁrmation, respectively. Signiﬁcant individual differences were
observed in many of these relationships.

With these ﬁndings, the present study contributes to existing knowledge in four
main ways. First, it showed that innovation self-efficacy, outcome expectation, and

behavioral intention changed and accounted for varying IT usage within persons over

103

time and thus explained behavioral changes within persons. Second, the present study
argued and provided evidence that implementation intention is a proximal predictor of IT
usage in addition to behavioral intention, expanding our knowledge of key proximal
factors that are important in understanding users’ adoption behaviors. Third, this study
showed that IT usage inﬂuenced innovation selfefﬁcacy and outcome expectation within
persons and the processes through which the effects occur. With these ﬁndings, user
experience is viewed as a dynamic process in which key beliefs—innovation self-efﬁcacy
and outcome expectation—determine and are determined by IT usage within persons.
Fourth. the present study is the ﬁrst to quantify the proportion of individuals for whom
speciﬁc within-person processes are applied. It thus presents a more precise knowledge
of intra-person processes among individuals. Future research should examine which
individual factors account for variances in within-person processes among individuals. It
is also important to extend the current work by identifying and testing external factors
that inﬂuence outcome expectation and innovation self-efficacy with respect to time using
within—person design. Such research would be useful in determining what factors should
be changed and when changes should be implemented in order to encourage IT

innovation usage for different groups of individuals.

104

APPENDICES

105

Appendix 1: Previous PDA Experience
How familiar are you with using PDAs (e. g., Palm Pilots or Blackberry) to track your
daily and weekly appointments?
How familiar are you with using PDAs (e. g., Palm Pilots or Blackberry) to organize your
to-do lists?
How familiar are you with using PDAs (e.g., Palm Pilots or Blackberry) to systematically
outline and organize your ideas?
How familiar are you with using cell phones to track your daily and weekly
appointments?
How familiar are you with using cell phones to organize their to-do lists?
How familiar are you with using cell phones to organize and outline your ideas?
How familiar are you with using other computer device (e.g. personal computers) to track
your daily and weekly appointments?
How familiar are you with using other computer device (e. g. personal computers) to
organize your to-do lists?
How familiar are you with using other computer device (e. g. personal computers) to
systematically organize and outline your ideas?

1. None at all.

IQ

Little.
3. Some.

4. Average

Ur

Above average

6. Way above average

106

Appendix 2: Innovativeness of Palm Pilots
Is using a Palm Pilot to track your daily and weekly appointments something new to you
personally"?
Is using a Palm Pilot to organize your to-do lists something new to you personally?
Is using a Palm Pilot to outline and organize your ideas something new to you
personally?

1. Totally new

Ix.)

Somewhat new

 

3. Average
4. Somewhat familiar

5. Very familiar

107

Appendix 3: Palm Pilot Training Booklet
Purpose
. Course Description:
a This training is designed for the new owner of a Palm Pilot. The goal is to provide
an overview of Palm Pilot and its variety of uses and applications.
. Training Objectives:
0 Upon completion ofthe course, you will:
0 Have a basic understanding of Palm features, navigation, data input and
synchronization.
0 Develop the ability to use the main applications: calendar, contact, and task,
Why use a Palm Pilot?
Advantages over a cell phone
°Handwritten notes
°Longer Battery Life
°Bigger Screen
Advantages over a PC
°Portable
°Accessibility
°Cheaper

Locating controls on your Palm Pilot

' Screen Displays the applications and information on your Palm Pilot. The screen is

touch sensitive.

Clock icon Displays the current time and date.

108

 

Display icon Lets you adjust the brightness and contrast of your Palm Pilot’s display.
Input area Lets you enter info with Graffiti® writing or open the onscreen
keyboard.

Locating the controls on your Palm Pilot (Cont.)
Power button Turns your Palm Pilot on or off
Application Opens the Calendar and Contacts applications.

5-way navigator Helps you move around and select info to display on the screen.
What’s on my Palm Pilot?

Your Palm Pilot comes with many applications preinstalled and ready to use. Open
these applications by selecting the icons on the Home screen.
Calendar Manage your schedule, from lunch with a friend, to weekly meetings, to
annual events like holidays, to extended events like conferences and vacations. Even
color—code your schedule by category.
Contacts Store names and addresses, phone numbers, e-mail and web site addresses—
cven photos and birthdays. Organize your contacts into categories.

What’s on my Palm Pilot? (Cont)
Tasks Stay on top of your to—do list. Enter things you need to do, prioritize them, set
alarms, and then monitor your deadlines.
Note Pad Write on the screen in your own handwriting or draw a quick sketch.
Progect. Let you outline your thoughts on the move, rearrange them, and export the
outlines to computers.

Diet Helpers: Track your daily diet and exercise to help you stay in shape.

109

° Hi Money Track your personal ﬁnance and provide reports after you synchronize with
your computer.
Using the 5-way navigator
- The navigator lets you access your information quickly with one hand and without the
stylus. The navigator does various things based on which type of screen you’re on. To
use the navigator, press Up, Down, Right, or Left; or press Select in the center.
Moving around the Home screen
- In the Home screen, use the navigator to open an application.
- Right or Left Scrolls to the next or previous application category.
- Select Inserts the selection highlight. When the selection highlight is present:
0 Up, Down, Right, or Left scrolls to the next icon in the corresponding direction.
0 Select Opens the selected application.
Using Menus
Menus allow you to Access hidden options or preferences and locate of many key
features in programs.
° Open an application-
0 Tap Menu to open the menus.
- Tap a menu, and then tap a menu item.
Calendar
° Staying on top of your schedule is an important part of being productive both at
work and at home. Calendar can help you remember important events
(appointments, birthdays, homework deadline, school holidays) and spot schedule

conﬂicts.

110

Calendar (Cont.)
° You can view your calendar by day, week, or month, or as an agenda list that
combines your list of tasks with your appointments.

Scheduling an appointment
° Press the Calendar application button.
- Tap the Day View icon
° Select the date ofthe appointment:

. Tap Go To.

- Tap the arrows to select the year

- Tap the month.

. Tap the date.

Tap the line next to the time the appointment begins and enter a description.
Data inputs
Use Grafﬁti or pop-up keyboard!

Grafﬁti gives you the ability to use a form of shorthand to add text to various Palm
applications. Yet it is quite slow and prone to errors. Keyboards are recommended. Tests
have shown entering information through keyboard is quicker and more accurate.
~Prcss abc for a letter keyboard

Press 123 the number keyboard

111

Checking your schedule
Sometimes you want to look at your schedule for a particular date, while other
times you want to see an overview of a week or month.
' Press the Calendar application button.
° Tap the icons in the lower-left comer to see four different Calendar views:
Contacts
Say good-bye to a paper address book that you need to update manually every time I
someone moves, changes their e-mail address, or gets a new work extension. With l
Contacts, not only is it easy to enter information such as names, addresses, and phone
numbers, but it is just as quick to view, update, and organize contact information. You
can back up contact information to your computer, easily share info with other
Adding a contact

Press the Contacts application button.
Add your contact infomiation:

Select New.

Tap each ﬁeld where you want to enter information, and enter it.

Tap the scroll arrows to move to the next page.

Locating a contact on your list

Press the Contacts application button.
Search for the contact:

Tap the Look Up line at the bottom of the screen and

enter the ﬁrst letter of the name you want to ﬁnd.

112

Enter the second letter of the name, and so on, until you can easily scroll to the contact
you want.
Select the contact to open it.

Tasks
Some of the most successful people in the world are also the busiest. When asked how
they manage to do it all, busy people usually say, “I make lists.” The Tasks application
on your Palm Pilot is the perfect place to make a list of the things you need to do and
create subtasks. You can set priority, alarms, and attach other information about the tasks.
Tasks allows you set priority, track deadlines, and stay focused.
Create a task
Select Tasks from Home screen
Tap New.
Enter a description ofthe task.
Organizing your tasks
Sometimes you want to look at all the things you need to do, while other times you want
to see only certain types oftasks.
Go to the Home screen and select Tasks .
In the Tasks list, select one of these options:
All Displays all your tasks.
Date Displays tasks that are due in a speciﬁc time frame. Tap the pick list in the

upper-right to select Due Today, Last 7 Days, Next 7 Days, or Past Due.

113

Marking a task complete
You can check off a task to indicate that you’ve completed it.
Go to the Home screen and select Tasks.
Select the check box on the left side ofthe task.
Synchronizing Your Palm Pilot
with Your Computer
Maybe you only think to use your Palm Pilot on its own to look up phone numbers, enter
appointments, and so on. But you can do much more with your Palm Pilot if you
synchronize it with your computer. Synchronizing simply means that information that has
been entered or updated in one place (your Palm Pilot or your computer) is automatically
updated in the other. No need to enter information twice.
Synchronizing Information

Prepare your Palm Pilot:

Connect the HotSync® cable to the USB port on your computer, and then insert the
other end into the mini-USB connector on your Palm Pilot.

Make sure your Palm Pilot is on.
Synchronize your Palm Pilot with your computer:

Tap HotSync .

When synchronization is complete, a message appears at the top of your Palm. Pilot
screen, and you can disconnect your Palm Pilot from the cable. Be patient;

synchronization may take some time.

114

Practice on your own

- In the next 15 minutes, perform the following operations on your own.

° Tap the Home Icon or use the 5-point navigator to access various software applications.

° Create a contact, schedule an appointment, and set up a task that you may want to
complete in a near future.
Palm Pilot do’s and don’ts
Do
° Use only the stylus to tap the screen—no pens, pencils, paper clips, or other sharp
objects.
° Charge the battery daily.
° Use so ft reset to recover from freezing problems.
Don’t
- Do not carry your Palm Pilot in your back pocket; you might sit on it by mistake.
0 Do not expose your Palm Pilot to very hot or cold temperatures.
- Never let your Palm run out of battery.
° Do not install any other applications to your Palm during the entire 8 weeks of the
study. Installation of addition software will compromise the study and disqualify you

from further participation and any cash rewards.

115

Additional Help!
° You can download how-to manuals for the installed applications in your palm at:
0 www.msu.edu/~dangchi/palm.htm
° Contact Chi Dang if you have any questions or problems with your palm
° Phone: 337-7873

' Email: dangchi@msu.edu

116

 

Appendix 4: Palm Pilot Study Consent Form

This study is designed to investigate the dynamic usage of Palm Pilots. If you choose to
participate in this study, you will be asked to complete a series of survey items and learn
how to use a Palm Pilot. You will be then provided a Palm Pilot to take home and will be
asked to use it daily at your discretion and answer survey items administered by the Palm
Pilot twice a day for eight weeks. In addition, you will be asked to meet the investigator
biweekly for a few minutes at pre-scheduled times and locations (a few minutes before
the beginning of your Psychology 101 lecture or mutually-agreed times and locations) so
that your Palm usage data can be downloaded and secured. At the end of the study, you
will return the Palm Pilot to the experimenter and listen to a short debrieﬁng.

The ﬁrst meeting and Palm Pilot training should take approximately 1.5 hours, and the
subsequent four meetings and survey responses (about 5 minutes daily) should take a
total of 5.5 hours. For your participation, you will receive research credits for subject
pool participation. In addition, you will have an opportunity to earn $70. Speciﬁcally,
you will earn $15 during the ﬁrst two weeks if (1) you meet the experimenter to have
your two-week usage data downloaded AND (2) you answer or provide non-applicable
responses for at least 90% of survey items and use the provided Palm Pilot daily at least
12 days out of 14 days with usage duration at your discretion. Similarly, you will earn
315 during the second two weeks, $15 during the third two weeks, and $25 during the
ﬁnal two weeks, contingent on BOTH conditions (1) and (2) described above.
Approximately within a few weeks after you ﬁnish the study, you will be informed via
email whether you meet the conditions for the cash payment and a check will be mailed
to you if you qualiﬁed.

Your participation in this research is completely voluntary. You are free to decline to
answer any questions or to terminate your participation at any time without penalty. Your
participation in this study will be kept conﬁdential to the maximum extent allowable by
11w. Your data will be included in a summary report with the data from others. The report
will not include any information that will allow anyone to identify any of your individual
responses.

If you have any questions about this study, please contact the responsible investigator,
Daniel Ilgen, by phone: (517) 355-7503, email: ilgen@msu.edu, or by regular mail: 340A
Psychology Building, East Lansing, MI 48824. You may also contact the secondary
investigator, Chi Dang, by phone: (517) 337-7873, email: dangchi@msu.edu , or regular
mail: 348 Psychology Building, East Lansing, MI 48824.

 

If you have any questions or concerns regarding your rights as a study participant, or are
dissatisﬁed at any time with any aspect of this study, you may contact - anonymously, if
you wish - Peter Vasilenko, Ph.D., Director of the Human Subject Protection Programs at
Michigan State University, by phone: (517) 355-2180, fax: (517) 432-4503, email:
irb@‘nrsrr.edu, or regular mail: 202 Olds Hall, East Lansing, MI 48824.

117

PARTICIPANT’S STATEMENT

 

I agree to participate in the Palm Pilot Study. I have been informed that I will learn how
to use a Palm Pilot, use it daily at my discretion, answer survey questions twice daily for
eight weeks, and return the Palm Pilot to the investigator at the completion of the
experiment. My materials will be kept conﬁdential and will not be seen by anyone other
than the research team. I consent to having these materials used for research purposes. I
was infomied that I will earn $70 if I answer at least 90% of the survey items, use the
provided Palm Pilot daily at least 12 days every two weeks, and meet the investigator at
mutually-agreed times to have my usage data downloaded biweekly.

My participation is voluntary. I may discontinue participation and return the Palm Pilot at
any time without penalty. I was informed that all of my individual responses will be kept

strictly conﬁdential, and that I will not be identiﬁed in any report of this study.

Signature: Date:

 

 

Printed name: Email:

 

 

Address to which the check should be sent if I were to meet conditions for the cash
payment:

 

118

Agreement of Responsibility

1 agree to participate in the “Adoption of Palm Pilots” study conducted by Dr. Daniel
Ilgen and Chi Dang in return for research credits. As part of the study, I received a Palm
Pilot and will use it for 8 weeks. I understand that the Palm Pilot is a property of the
Management Department at MSU and will use it with care until I return it to the
experimenters no later than December 5, 2006. Furthermore, I understand that I will be
ﬁnancially responsible for the market value of the Palm ($100) if it is lost or damaged
due to non-accidental causes.

Full Name (Print):

 

 

Signature:

Date:

119

Appendix 5: Demographic Information
and Individual Differences

Demographic Information

What is your age? Gender? GPA?

 

SAT score? Or ACT score?

 

What is your ethnicity? (Check all that apply):

[:I White/Caucasian
E] African America

E] Hispanic

[:1 Native American
[1 Asian

What is the estimated total family yearly income?
D $20,000 or less
[:I $20,001 -40,000
[:| $40,001 ~60,000
E] $60,001 -80,000
E] $80,001-100,000
[:1 $ 1 00,001 - 120,000
C] $120,000 or more

NEO Five-Factor Inventory

 

l. I am not a worrier.

2. I like to have a lot of people around me.

3. I don’t like to waste my time daydreaming.
4. I try to be courteous to everyone I meet.

5. I keep my belongings clean and neat.

6. I often feel inferior to others.

7. I laugh easily.

120

 

 

9.

10.

11.

Once I ﬁnd the right way to do something, I stick to it.
I often get into arguments with my family and co-workers.
I’m pretty good about pacing myself so as to get things done on time.

When I’m under a great deal of stress, sometimes I feel like I’m going to pieces.

. I don’t consider myself especially “light-hearted.”

. I am intrigued by the patterns I ﬁnd in art and nature.
. Some people think l’rn selﬁsh and egotistical.

. I am not a very methodical person.

. I rarely feel lonely or blue.

. I really enjoy talking to people.

. I believe letting students hear controversial speakers can only confuse and

mislead them.

. I would rather cooperate with others than compete with them.
. I try to perform all the tasks assigned to me conscientiously.

. I often feel tense and jittery.

. I like to be where the action is.

. Poetry has little or no effect on me.

. I tend to be cynical and skeptical of others’ intentions.

. I have a clear set of goals and work toward them in an orderly fashion.

. Sometimes I feel completely worthless.
. I usually prefer to do things alone.
. I often try new and foreign foods.

. I believe that most people will take advantage of you if you let them.

121

 

 

L’J
A
V

. I waste a lot of time before settling down to work.

3 I. I rarely feel fearful or anxious.

32. I often feel as if I’m bursting with energy.

33. I seldom notice the moods or feelings that different environments produce.

34. Most people I know like me.

35. I work hard to accomplish my goals.

36. I often get angry at the way people treat me. ﬁx. 1'

37. I am a cheerful, high-spirited person.

38. I believe we should look to our religious authorities for decision on moral issues.

39. Some people think of me as cold and calculating.

40. When I make a commitment, I can always be counted on to follow through.

41. Too often, when things go wrong, I get discouraged and feel like giving up.

42. I am not a cheerful optimist.

43. Sometimes when I am reading poetry or looking at a work of art, I feel a chill or
wave of excitement.

44. I’m hard-headed and tough-minded in my attitudes.

45. Sometimes I’m not as dependable or reliable as I should be.

46. I am seldom sad or depressed.

47. My life is fast—paced.

48. I have little interest in speculating on the nature of the universe or the human
condition.

49. I generally try to be thoughtful and considerate.

50. I am a productive person who always gets the class done.

122

51.

52.

50.

57.

58.

59.

60.

I often feel helpless and want someone else to solve my problems.

I am a very active person.

. I have a lot of intellectual curiosity.
. Ifl don’t like people, I let them know it.

. I never seem to be able to get organized.

At times I have been so ashamed I just wanted to hide.

I would rather go my own way than be a leader of others.

I often enjoy playing with theories or abstract ideas.

If necessary, I am willing to manipulate people to get what I want.

I strive for excellence in everything I do.

Computer Self-Efﬁcacy

 

I could use computer to do my work. . .

. if there was no one around to tell me what to do as I go.

. ifl had only the software manuals for reference.

. if I had seen someone else using it before trying it myself.

. ifl could call someone for help ifl got stuck.

. if someone else had helped me get started.

. . ifl had just the built-in help facility for assistance.

. . . ifsomeone showed me how to do it ﬁrst.

Computer Anxiety

Computers do not scare me at all.

Working with a computer makes me nervous.

I do not feel threatened when others talk about computers.

123

 

 

It wouldn’t bother me to take computer courses.

Computers make me feel uncomfortable.

I feel at ease in a computer class.

I get a sinking feeling when I think oftrying to use a computer.
I feel comfortable working with a computer.

Computers make me feel uneasy.

Social In fluence

 

How many people that you know use PDAS (e.g., Palm Pilots or Blackberry) to track
their daily and weekly appointments?

How many people that you know use PDAs (e. g., Palm Pilots or Blackberry) to organize
their to-do lists?

How many people that you know use PDAs (e.g., Palm Pilots or Blackberry) to outline
and organize their ideas?

How many people that you know use cell phones to track their daily and weekly
appointments?

How many people that you know use cell phones to organize their to-do lists?

How many people that you know use cell phones to organize and outline their ideas?
How many people that you know use computers to track their daily and weekly
appointments?

How many people that you know use computers to organize their to-do lists?

How many people that you know use computers to systematically organize and outline

their ideas?

124

 

‘!"J V I Kl)-
.' L .

n

I

Computer skills

 

I frequently read computer magazines or other sources
of information that describe new computer technology.
I know how to recover deleted or “lost data” on a computer.
I know what a LAN (Local Area Network) is.
I know what an operating system is.
I know how to write computer programs.
I know how to install software on a computer.
I know what a database is.
I frequently use a computer for getting information on the Internet.
I regularly use a computer for word processing

Internet skills

 

I know how to navigate the www using hyperlinks or entering its address (URL) in the
browser.

I know how to search for information using search engine (e. g., google, yahoo) on the
Internet.

I know how to send and receive email using the Internet.

I know how to upload ﬁles from my computer to the Internet.

I know how to download ﬁles from the Internet to my computer.

I know how to install applications downloaded from the Internet to my computer.

I know how to program a web page using simple or advanced computer language.

Technology Use

 

How often do you use a computer? (circle a number)

125

 

 

l. I don’t use a computer 5. everyday, for less than 1 hour.

[\J

. about once a month 6. everyday, for 1 to 3 hours

'0)

. a few times a month 7. everyday, for more than 3 hours
4. a few times a week

How many computers do you have at home? (write 0 if you have no

 

C omputer).

How often do you use the Internet? (circle a number)

I. I don’t use the Internet 5. everyday, but for less than 1 hour
2. about once a month 6. everyday, for 1 to 3 hours
3. a few times a month 7. everyday, for more than 3 hours

4. a few times a week
Do you have Internet at home? (circle one) Yes No
If Yes, how do you connect to the Internet? (circle one)
1. Telephone dial up 4. Wireless connection
2. Cable modem 5. T-l ﬁber optic
3. DSL-enabled phone line 6. Don’t know
Do you ever go online using a wireless devise, like a PDA or cell phone? Yes No
PDQ you ever go online using a wireless computer or laptop? Yes No
Overall, how would you rate your computer skills?
1. No skills 4. Good skills
2. Poor skills 5. Very good skills

3. Average skills

126

Technology Activities

There are many things people can do when they use a computer or other technology.
Please indicate how often you do each of the following things.

 

1 2 3 4
Never Some Often Very
times often

 

1) Create a Web page

 

2) Surf the Web

 

3) Talk on a cell phone

 

4) Text message on a cell phone

 

5) Lise a photo editing program

 

6) Play videogames on a computer

 

7) Play videogames on a console

 

8) Play video games in an arcade

 

9) Instant message with friends

 

10) Instant message with strangers

 

l 1) Download music ﬁles

 

12) Buy something online

 

13) Create documents for school

 

14) Save images/graphics

 

15) ) Take a course online

 

16) Take a survey online

 

17) Take a test online

 

18) Visit a chat room

 

 

 

19) Post to a mailing list

 

20) Use a laptop

 

21) Read mailing list messages

 

22) Use an MP3 player

 

23) Use a DVD player

 

24) Use email

 

25) Use a PDA (like blackberry)

 

26)Search the Web for information about

 

27) health, diet, ﬁtness

 

28) news. current events

 

 

 

 

J9) religion, spirituality

 

gonobs

 

  

31) colleges, universities, education

 

  

32) entertainment (movies, pop stars)

 

  

33) politics, political leaders

 

  

34) depression, mood, mental illness

 

  

3 5) government services

 

  

3 6) something I need for a school report

 

 
 

 

 

 

 

ﬂh—li—dI—ID—DI—lh—lI—Ir—‘DsﬁU—ih-dr—Ih—dhaﬁ—Iﬁ—‘rsh—iHD—‘h—iI—‘ﬂh—ﬁ—‘r—‘I—ir—lr—dp—‘p—I—Ir—i—OH
NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN
we.)wmumwt»wwwwwwwwwuwwwwwwwwwwwwwwwL»moat»
h$bhhhbh$bh¥hhbbhhhhh#hh-bhhAhAbA-§hhhh

3 7) an interest or hobby of mine

 

 

127

Appendix 6: Debriefing Form

The study that you just participated examines how beliefs/perceptions about Palm Pilots
influence users’ reactions and usage of Palm Pilots. Speciﬁcally, the study investigates
how beliefs about the usefulness of technology and one’s conﬁdence his/her capability to
use Palm Pilot change and influence intentions to use, satisfaction with, and usage of
Palm Pilots dynamically over time. It also seeks to explain how individual difference
factors such as computer anxiety and computer efﬁcacy affect the patterns of usage
behaviors. If you are interested in the detailed results of the study, please contact the
second investigator in May 2007 when the results are expected to be available.

Vv’ithin a few weeks when we have the chance to look at your responses, we will inform
you via email whether you meet the conditions for the cash payment and a check will be
l'llilllCd to you if you qualiﬁed.

It is very important that you not discuss your experiences with this study with your
friends and classmates for 2 reasons:

1. Talking to your classmates may bias their reactions during the experiment.
2. They may participate in a different study altogether.

'Thank you very much for your participation today. Research 1ike this cannot go forward
\vithout your help.

128

l\)

(1)

Appendix 7: Outcome Expectation
. Using Palm Pilot enables me to accomplish tasks more quickly.
. 1 would ﬁnd Palm Pilot useful in my work.
. Using Palm Pilot would enhance my productivity.

. lfl use Palm Pilot, I would improve my personal performance.

129

Appendix 8: Innovation Self-Efﬁcacy

How well can you use the following applications in Palm Pilot?

1. Calendar

2. Address

3. Task

4. Note Pad
5. Progect

o. HiMoney

7. Diet Helper
Response scale

1. Not well at all

3. Neither

U1

. Very well

130

[0

Appendix 9: Behavioral Intention

. I intend to use Palm Pilot during the next 24 hours.
. I predict 1 would use Palm Pilot during the next 24 hours.

. 1 plan to use Pam Pilot during the next 24 hours

131

IQ

Appendix 10: Implementation Intention
l have thought about particular situations in which I will use the Palm Pilot for the
next 24 hours.
I have thought about when I will use the Palm Pilot for the next 24 hours.
I already know the situations in which I will use the Palm Pilot for the next 24
hours.
I have planned how I will use the Palm Pilot for the next 24 hours.

I already know how I will use the Palm Pilot for the next 24 hours.

132

Appendix 11: Outcome Disconﬁrmation

1. Compared to my expectation expressed in the last survey, the ability of Palm Pilot to
improve my performance was (much worse than expected much better than expected).
2. Compared to my expectation expressed in the last survey, the ability of Palm Pilot to
increase my productivity was (much worse than expected much better than expected).
3. Compared to my expectation expressed in the last survey, the ability of Palm Pilot to

be useful was (much worse than expected much better than expected).

133

Appendix 12: Efﬁcacy Disconﬂrmation
1. Compared to my expectation expressed in the last survey, my ability to use Palm Pilot
is (much worse than expected much better than expected).
2. Compared to my expectation expressed in the last survey, my ability to execute various
features in Palm Pilot is (much worse than expected much better than expected).
3. Compared to my expectation expressed in the last survey, my ability to use different

applications in Palm Pilot is (much worse than expected much better than expected).

134

 

Appendix 13: Usage Frequency by Applications over Time

 

 

 

Week Contacts Calendar Tasks Note Progect Hglgter Mrriey Total
1 4 8 4 4 1 O 0 22
2 1 5 4 5 3 2 1 21
3 1 5 5 5 3 3 1 23
4 1 5 5 4 3 2 1 21
5 1 5 5 5 3 3 1 24
6 1 6 5 6 4 3 1 26
7 1 5 4 5 3 2 1 21
8 1 5 5 5 3 2 1 23

 

 

 

 

Notes: As data were missing for some days of the week, daily averages was ﬁrst
calculated. They were then multiplied by 7 to estimate numbers of weekly usage

frequency

135

 

 

Appendix 14: Oblique Rotated Factor Pattern for Behavioral Intention
and Implementation Intention

 

 

 

Scale Item Factor 1 Factor 2
Implementation intention 1 0911* -0.002
Implementation intention 2 0885* 0.002
Implementation intention 3 0780* -0.005
Implementation intention 4 0904* 0.013
Implementation intention 5 0922* —0004
Behavioral intention 1 0.002 0879*
Behavioral intention 2 0.003 0884*
Behavioral intention 3 -0003 0879*
Explained Variance 58% 25%
Inter-Factor Correlation 0.36

 

 

 

Factor loadings > .32 are marked with an asterisk

136

t .

_ Al
all! I

 

 

Appendix 15: Regression Analyses of Effect of
Behavioral Intention on Usage Frequency within Persons

 

 

 

 

Person Behavioral Intention
1 1.09
2 0.69
3 0.98
4 0.92
5 0.26
6 1.39
7 1.11
8 1.2
9 0.34
10 0.86
1 1 0.87
12 1.01
13 1.34
14 0.79
15 1.05
16 0.91
17 1.16
18 0.85
19 1.66
20 0.74
21 0.79
22 0.82
23 0.82
24 1.21
25 1.17
26 0.93
27 0.98
28 0.65
29 1.14
30 1.32
31 1.35
32 1.16
33 0.93
34 1.1

 

137

 

Appendix 15 (Continued): Regression Analyses of Effect of
Behavioral Intention on Usage Frequency within Persons

 

 

 

 

 

Person Behavioral Intention
35 1.11
36 1.05
37 0.98
38 1.25
39 1.05
40 1.2
41 1.25
42 0.93
43 2.27
44 1.14
45 0.88
46 1.06
47 1.49
48 2.31
49 1.9
50 1.58
51 1.46
53 0.82
54 0.71
55 1.33
56 1
57 0.91
58 0.85

 

 

 

Note: Numbers in bold were insigniﬁcant Innovation self-eﬂicacy
slopes

138

 

Appendix 16: Regression Analyses of Effects of Behavioral Intention and
Implementation Intention on Usage Frequency within Persons

 

 

 

 

 

 

 

 

 

Person Behavioral Intention Implementation Intention
l 0.63 0.79
2 0.58 0.58
3 0.8 0.7
4 0.66 0.67
5 0.25 0.15
6 0.96 0.8
7 0.83 0.69
8 0.98 0.38
9 0.19 0.39
10 0.64 0.38
1 1 0.8 0.35
12 1.03 1.04
13 1.14 0.36
14 0.8 -0.04
15 0.85 0.55
16 0.77 0.46
17 1.06 0.2
18 0.56 0.87
19 1.57 0.15
20 0.66 0.59
21 0.67 0.45
22 0.65 0.51
23 0.82 -0.02
24 1.1 0.49
25 1.13 1.01
26 0.71 0.69
27 0.82 0.54
28 0.6 0.67
29 0.79 0.57
30 1 0.65
31 1.06 0.78
32 0.89 0.52
33 0.68 0.99
34 0.88 0.57
35 0.71 0.75

139

 

Appendix 16 (Continued): Regression Analyses of Effects of Behavioral Intention
and Implementation Intention on Usage Frequency within Persons

 

 

 

 

 

 

 

 

 

 

 

Person Behavioral Intention Implementation Intention
36 0.93 0.51
37 0.84 0.32
38 1.06 0.36
39 1.04 1.3
40 0.84 0.66
41 1.02 0.45
42 0.95 1.57
43 1.65 0.77
44 0.66 0.75
45 0.64 0.82
46 0.91 0.47
47 1.25 0.43
48 1.7 0.87
49 1.72 0.78
50 0.9 0.71
51 1.07 0.8
53 -0.27 0.91
54 0.64 0.29
55 0.89 0.65
56 0.75 0.49
57 ‘ 0.52 0.65
58 JL 0.69 0.73

Note: Numbers in bold were insigniﬁcant implementation intention slopes

140

Appendix 17: Regression Analyses of Effects of Outcome Expectation an Innovation
Self-Efﬁcacy on Behavioral Intention within Persons

 

 

 

 

 

 

Person Outcome Expectation Innovation SelﬁEf/icacy
1 1.77 0.468
2 1.24 -0.102
3 0.63 0.369
4 2.11 0.727
5 1.41 ~0.017
6 1.31 0.561
7 1.06 0.601
8 1.5 0.647
9 0.23 0.053
10 1.49 0.403
1 1 1.53 0.674
12 0.25 0.24
13 0.96 0.387
14 0.9 2.123
15 1.09 0.892
16 0.76 0.879
17 1.08 0.755
18 0.91 0.841
19 1.01 0.388
20 1.97 0.399
21 1.29 0.673
22 g 0.39 0.402
23 0.6 -0.04
24 0.68 0.186
25 1.27 0.273
26 0.77 0.359
27 0.96 0.446
28 1.02 -0.196
29 1.69 0.265
30 0.75 0.757
31 0.52 0.358
32 1.01 0.505
33 0.61 0.316
34 1.07 0.133

 

 

141

Appendix 17 (Continued): Regression Analyses of Effects of Outcome Expectation
and Innovation Self-Efﬁcacy on Behavioral Intention within Persons

 

 

 

 

 

Person Outcome Expectation Innovation Self-Efﬁcacy
35 1.46 0.568
36 0.97 -0.043
37 0.87 0.865
38 0.53 0.464
39 1.2 0.069
40 0.93 0.286
41 0.91 0.68
42 0.71 -0.322
43 0.58 0.354
44 1.18 0.376
45 1.12 0.07
46 0.89 0.322
47 0.87 0.483
48 0.85 0.191
49 0.85 -0.124
50 0.69 0.275
51 0.62 0.793
52 0.01 1.832
53 0.33 0.154
54 1.01 -0.049
55 0.87 0.239
56 1 0.36
57 0.93 0.636
58 0.42 0.383

 

 

 

 

Note: Numbers in bold were insigniﬁcant innovation self-efficacy slopes

142

15:

 

Appendix 18: Regression Analyses of Effects of Outcome Expectation and
Innovation Self-Efficacy on Implementation Intention within Persons

 

 

 

Person Outcome Expectation Innovation Self-Efﬁcacy
1 0.212 0.848
2 0.401 0.394
3 0.033 0.504
4 0.703 1.258
5 0.538 1.896
6 1.139 0.448
7 0.899 0.733
8 0.998 0.821
9 0.133 0.349
10 0.257 0.852
1 1 -0.573 0.949
12 -0.133 0.41
13 0.672 0.93
14 0.51 1.307
15 0.316 1.041
16 0.365 0.537
17 0.203 1.184
18 0.44 0.869
19 0.515 0.926

20 0.376 1.287
21 0.744 0.452
22 0.234 0.426
23 0.29 0.299
24 0.037 0.805
25 0.473 1.439
26 0.487 0.378
27 0.671 0.618
28 0.087 0.127
29 -0.225 0.707
30 0.469 0.835
31 -0.057 1.197
32 0.539 0.61

33 0.169 0.417
34 0.395 0.81

 

 

143

 

 

 

Appendix ]8 (Continued): Regression Analyses of Effects of Outcome Expectation
and Innovation Self-Efﬁcacy on Implementation Intention within Persons

 

 

 

 

 

Person Outcome Expectation Innovation Self-Efﬁcacy
35 -0.107 1.037
36 0.268 0.336
37 0.592 0.855
38 0.526 0.304
39 0.143 0.629
40 0.543 0.836
41 0.619 0.69
42 0.154 1.385
43 0.274 0.749
44 -0.251 0.83
45 0.76 0.439

. 46 0.314 0.441
47 0.71 0.814
48 1.084 0.112
49 0.531 0.145
50 1.174 0.622
51 1.007 0.617
52 0.44 8.124
53 -0.027 0.894
54 0.365 0.407
55 0.116 0.517
56 0.736 0.551
57 -0.609 0.867
58 -0.136 0.472

 

Note: Numbers in bold were insigniﬁcant innovation self-efﬁcacy slopes

144

 

 

Appendix 19: Regression Analyses of Effects of Outcome Expectation and
Innovation Self-Efﬁcacy on Usage Frequency within Persons

 

 

 

 

 

Outcome Expectation Innovation Self-Efﬁcacy
Person Slope Slope
1 2.18 0.69
2 1.64 ~0.41
3 1.37 0.74
4 1.47 1.51
5 3.18 1.64
6 3.3 1.33
7 1.43 1.16
8 3.51 0.53
9 0.73 -0.17
10 2.16 0.21
I 1 2.36 0.92
12 0.32 0.8
13 1.85 1.52
14 2.09 2.56
15 1.94 1.7
16 1.18 0.96
17 2.44 0.47
18 1.33 0.93
19 2.92 0.98
20 2.51 0.76
21 1.2 1.12
22 1.14 0.32
23 0.99 -0.06
24 1.3 0.27
25 2.62 0.09
26 1.17 0.7
27 0.9 1.19
28 0.76 —0.5
29 2.42 0.3
30 0.96 1.28
31 1.78 0.49
32 1.96 0.81
33 1.18 0.43
34 0.9 0.92

 

 

 

 

145

Appendix 19 (Continued): Regression Analyses of Effects of Outcome Expectation
and Innovation Self-Efﬁcacy on Usage Frequency within Persons

 

 

 

Innovation Self-Eﬂicacy
Person IOutcome Expectation Slope Slope
35 2.19 0.86
36 1.91 -0. 7
37 1.31 0.89
38 0.95 0.87
39 2.54 -0.31
40 0.85 2.25
41 1.94 0.95
42 1.98 0.24
43 2.64 1.28
44 2.02 0.86
45 1.59 0.32
46 1.5 0.55
47 1.56 1.49
48 2.52 0.81
49 2.07 -0.06
50 2.41 0.86
51 1.41 1.2
52 1.08 7.34
53 1.94 0.26
54 1.85 -0.4
55 1.57 0.76
56 1.49 0.48
57 0.49 0.79
58 0.69 1.21

 

 

 

Note: Numbers in bold were insigniﬁcant innovation self-efﬁcacy slopes
Numbers in italics were signiﬁcantly negative innovation self-eﬁ’icacy slopes

146

Appendix 20: Regression Analyses of Effects of Innovation Self-Efficacy on
Usage Frequency within Persons over Time

 

 

 

Outcome Innovation Time x Innovation
Person Time Expectation Self-Efﬁcacy Self-Efﬁcacy

l 0.01 2.01 1.01 —0.03
2 —0.04 1.05 0.45 0

3 0.03 1.16 0.37 -0.01
4 0.02 1.37 2.43 -0.05
5 0.04 2.69 -0.59 0.15
6 -0.02 3 .3 0.1 0.05
7 -0.03 1.46 2.65 -0.03
8 -0.02 3.26 1.1 0

9 —0.02 0.49 0.37 0
10 -0.01 1.87 0.54 -0.02
1 1 -0.03 2.16 2.65 -0.05
12 0.03 0.22 0.05 0.01
13 0.01 1.79 0.9 0.03
14 -0.01 1.64 8.19 -0.3
15 -0.02 1.89 2.2 -0.01
16 -0.03 1.14 2.94 -0.05
17 -0.02 2.16 1.21 -0.01
18 -0.01 1.14 2.56 -0.06
19 0.01 2.88 0.04 0.03
20 0.03 2.1 -0.1 0.05
21 0.03 0.64 0.35 0.01
22 0.03 0.74 0.06 -0.01
23 -0.03 0.67 0.27 0.01
24 -0.02 0.99 0.92 -0.01
25 0.02 2.36 -0.54 0.05
26 0.03 0.86 0.35 -0.01
27 0 0.45 2.63 -0.05
28 -0.04 0.32 0.35 -0.01
29 0 2.07 1.01 -0.04
30 -0.05 1.12 3.61 -0.06
31 -0.03 1.57 -0.37 0.07
32 0.01 1.87 1.32 -0.02
33 0.04 0.84 -1.04 0.01
34 0.02 0.91 0.85 -0.02

 

 

147

 

 

Appendix 20 (Continued): Regression Analyses of Effects of Innovation
Self-Efﬁcacy on Usage Frequency within Persons over Time

 

 

 

Outcome Innovation Time x Innovation
Person Time Expectation Self-Efﬁcacy Self-Efﬁcacy

35 0.02 1.5 2.04 -0.07
36 -0.03 1.37 -0.04 -0.01
37 -0.03 1.4 3.17 -0.06
38 -0.02 0.85 1.53 -0.01
39 -0.04 1.76 2.57 -0.03
40 0 0.93 2.32 0

41 0.04 1.67 1.6 -0.04
42 -0.03 1.4 0.88 0.04
43 0.04 2.56 3.68 —0.07
44 0.02 2.09 1.02 -0.02
45 -0.03 0.97 0.15 0.03
46 0.03 1.11 0.13 -0.01
47 -0.02 1.69 2.29 -0.01
48 -0.03 2.05 4.24 -0.11
49 0.01 1.65 0.95 -0.07
50 -0.01 2.45 1.29 0.00
51 -0.03 1.42 2.6 -0.04
53 0.02 3 -0.94 0.02
54 -0.03 1.18 -0.17 0.01
55 0.04 1.49 -0.6 0.02
56 0.05 1.67 0.7 -0.04
57 0 0.49 1.31 -0.04
58 0.03 1.38 -0.95 0.04

 

 

 

Note: Numbers in bold were insigniﬁcant interaction effects

148

 

 

Appendix 21: Regression Analyses of Innovation Self-Efﬁcacy
Growth over Time

 

 

 

 

Person Time
1 0.023
2 0.025
3 0.017
4 0.012
5 0.004
6 0.006
7 0.009
8 0.008
9 0.021
10 0.019
1 1 0.013
12 0.018
13 0.005
14 0.003
15 0.004
16 0.009
17 0.007
18 0.01
19 0.005

20 0.004
21 0.011
22 0.016
23 0.014
24 0.017
25 0.003
26 0.018
27 0.013
28 0.02
29 0.023
30 0.01
31 0.007 0
32 0.007
33 0.019
34 0.015

 

 

 

 

149

Appendix 21 (Continued): Regression Analyses of Innovation Self-Efficacy
Growth over Time

 

 

 

 

 

 

Person Time
35 0.019
36 0.014
37 0.009
38 0.009
39 0.007
40 0.006
41 0.009
42 0.005
43 0.007
44 0.017
45 0.018
46 0.018
47 0.009
48 0.005
49 0.004
50 0.006
51 0.009
52 0.001
53 ' 0.017
54 0.022
55 0.015
56 0.014
57 0.019
58 1 0.014

 

150

Appendix 22: Regression Analyses of Effect of Usage Frequency on
Innovation Self-Efﬁcacy within Persons

 

 

 

 

 

 

Person Usage Frequency
1 0.198
2 -0.188
3 0.169
4 0.087
5 -0.019
6 0.02
7 0.064
8 -0.001
9 -0.089
10 0.123
1 1 0.086
12 0.083
13 0.051
14 0.008
1 5 -0.006
16 0.041
17 -0.005
1 8 0.054
19 0.039

20 -0.037
21 0.155
22 0.094
23 -0.099
24 0.14
25 0.007
26 0.159
27 0.1
28 -0. 23
29 0.176
30 0.047
3 1 0.019
32 0.029
33 0.085
34 0.1 15

 

151

Appendix 22 (Continued): Regression Analyses of Effect of Usage
Frequency on Innovation Self-Efﬁcacy within Persons

 

 

 

Person Usag Frequency
35 0.14
36 -0.079
3 7 0.05
3 8 -0.003
39 -0.002
40 0.026
41 0.045
42 0.017
43 0.03
44 0. 175
45 -0.007
46 0.217
47 0.03
48 -0.007
49 -0.001
50 0.015
5 I 0.062
52 -0.002
53 0.063
54 -0.134
55 0.153
56 0.175
57 0.208
58 0.1 16

 

 

Note: Numbers in bold were insigniﬁcant slopes

Numbers in italics were signiﬁcantly negative slopes

152

 

Appendix 23: Regression Analyses of Effect of Usage
Frequency on Outcome Expectation within Persons

 

 

 

Person Usage Frequency
1 0.055
2 0.128
3 0.055
4 0.009
5 0.017
6 0.058
7 0.12
8 0.026
9 0.068
10 0.096
1 1 0.036
12 -0.005
13 0.047
14 -0.001
15 0.012
16 0.124
17 0.039
18 0.143
19 0.065

20 0.048
21 0.073
22 - 0.055
23 0.217
24 0.081
25 0.055
26 0.06

27 0.018
28 0.164
29 0.089
30 0.077
31 0.062
32 0.029
33 0.043
34 0.011

 

 

153

 

Appendix 23 (Continued): Regression Analyses of Effect of Usage
Frequency on Outcome Expectation within Persons

 

 

 

 

 

Person Usage Frequency
35 0.042
36 0.107
37 0.094
38 0.006
39 0.034
40 0.034
41 0.027
42 0.06
43 0.045
44 0.057
45 0.082
46 0.05
47 0.061
48 0.013
49 0.024
50 0.046
51 0.076
52 0.034
53 0.011
54 0.12
55 0.066
56 0.01
57 ‘ -0.017
58 0.038

 

 

Note: Numbers in bold were insigniﬁcant slopes

154

 

BIBLIOGRAPHY

155

BIBLIOGRAPHY

Ackerman, P. L., & Kanfer, R. (1993). Integrating laboratory and ﬁeld-study for
improving selection: Development of a battery for predicting air-trafﬁc—controller
success. Journal of Applied Psychology, 78, 413-432.

Adam, F., 8: O'Doherty, B. (2003). ERP projects: Good or bad for SME. Cambridge,
UK: Cambridge University Press.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human
Decision Processes, 50, 179-211.

Ajzen. I. (2001 ). Nature and operation of attitudes. Annual Review ofPsyChologv, 52, 27-
58.

Ajzen. I. (2002). Perceived behavioral control, self-efﬁcacy, locus of control, and the
theory of planned behavior. Journal of Applied Social Psychology, 32, 665-683.

Ajzen, 1., & F ishbein, M. (1977). Attitude—behavior relations: Theoretical-analysis and
review of empirical research. Psychological Bulletin, 84, 888-918.

Bandura, A. (1997). Theoretical perspectives. In A. Bandura (Ed.), Self-efﬁcacy: The
exercise of control (pp. 1-35). New York: Freeman.

Bandura. A., & Locke, E. A. (2003). Negative self-efficacy and goal effects revisited.
Journal of/lpplied Psychology, 88, 87-99.

Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive
Theorv. En glewood Cliffs, NJ: Prentice Hall.

Bandura. A., & Locke, E. A. (2003). Negative self-efﬁcacy and goal effects revisited.
Journal o/‘Applied Psychology, 88, 87-99.

Bhattacherjee, A. (2001). Understanding information systems continuance: An
expectation-conﬁrmation model. MIS Quarterly, 25, 351.

Bhattacherjee, A., & Premkumar, G. (2004). Understanding changes in belief and attitude
toward information technology usage: A theoretical model and longitudinal test. MIS
Quarterly 28, 229-254.

Brancheau, J. C ., & Brown, C. V. (1993). The management of end-user computing:
Status and directions. Computing Surveys, 25, 437-482.

Bryan, .I. F., & Locke, E. A. (1967). Goal setting as a means of increasing motivation.
Journal oprplied Psychology, 51, 274-277.

156

Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and
data analysis methods. Thousand Oaks, CA: Sage Publications, Inc.

C hau, P. Y. K., & Hu, P. J. H. (2001). Information technology acceptance by individual
professionals: A model comparison approach. Decision Sciences, 32, 699-719.

Chin. W. W., & Gopal, A. (1995). Adoption intention in Gss: Relative importance of
beliefs. Data Base for Advances in Information Systems, 26, 42-64.

Compeau, D. R., & Higgins, C. A. (1995). Computer self-efﬁcacy: Development ofa
measure and initial test. MIS Quarterly, 19, 189-211.

Compeau, D. R., Higgins, C. A., & Huff, S. (1999). Social cognitive theory and
individual reactions to computing technology: A longitudinal study. MIS Quarterly,
23, 145-158.

C urley, K. F., & Pyburn, P. J. (1982). Intellectual technologies: The key to improving
white-collar productivity. Sloan Management Review, 24, 31-39.

Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Quarterly, 13, 319-340.

Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of computer-
technology: A comparison of two theoretical models. Management Science, 35, 982—
1003.

de Vries, P., Midden, C., & Bouwhuis, D. (2003). The effects of errors on system trust,
self—conﬁdence, and the allocation of control in route planning. International Journal
ofI-Iunum—Computer Studies, 58, 719-735.

Dewett, T. & Jones G. R., (2001). The role of information technology in the
organization: a review, model, and assessment. Journal of Management, 3, 313-346.

Erez. M., & Zidon, I. (1984). Effect of goal acceptance on the relationship of goal
difﬁculty to performance. Journal of Applied Psychology, 69, 69-78.

Eveland, J. D., Hetzner, W. A., & Tomatzky, L. G. (1983). Innovation processes research
implications for public-policy. IEEE Transactions on Engineering Management, 30,
37-39.

Gist, M. 13., & Mitchell, T. R. (1992). Self-efﬁcacy: A theoretical-analysis of its
determinants and malleability. Academy of Management Review, 1 7, 183-211.

Gollwitzer, P. M. (1999). Implementation intention: Strong effects of simple plans.
American Psychologist, 54, 493-503.

157

Gollwitzer, P. M., & Brandstatter, V. (1997). Implementation intention and effective goal
pursuit. Journal of Personality and Social Psychology, 73, 186-199.

Gollwitzer, P. M., & Oettingen, G. (1998). The emergence and implementation of health
goals. Psychology & Health, 13, 687-715.

Harrison, D. A., Mykytyn, P. P., & Riemenschneider, C. K. (1997). Executive decisions
about adoption of information technology in small business: Theory and empirical
tests. Information Systems Research, 8, 171-195.

Hill, T., Smith, N. D., & Mann, M. F. (1987). Role of efﬁcacy expectations in predicting
the decision to use advanced technologies: The case of computers. Journal of Applied
Psychology, 72, 307-313.

Hu. P. 1., Chau, P. Y. K., Sheng, O. R. L., & Tam, K. Y. (1999). Examining the
technology acceptance model using physician acceptance of telemedicine technology.
Journal of Management Information Systems, 16, 91-112.

Igbaria, M., Pavri, F. N., & Huff, S. L. (1989). Microcomputer applications: An empirical
look at usage. Information & Management, 16, 187-196.

Ives, B., & Learmonth, G. P. (1984). The information system as a competitive weapon.
Communications ofthe ACM, 27, 1193-1201.

.lames. D., 8.: Wood, M. L. (2000). A second wind for ERP. McKinsey Quarterly, 2, 100-
107.

.laspcrson, J. 8., Carter, P. E., & Zmud, R. W. (2005). A comprehensive
conceptualization of post-adoptive behaviors associated with information technology
enabled work systems. MIS Quarterly, 29, 525-557.

.leyaraj, A., Rottman, J. W., & Lacity, M. C. (2006). A review of the predictors, linkages,
and biases in IT innovation adoption research. Journal of Information Technology,
2/(1),l-23.

Johansen, R., & Swigart, R. (1996). Upsizing the individual in the downsized
organization: Managing in the wake of reengineering, globalization, and
overwhelming technological change. Reading, MA: Addison-Wesley.

Kalmeman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.

Kanfer, R., & Ackerman, P. L. (1989). Motivation and cognitive abilities: An integrative

aptitude treatment interaction approach to skill acquisition. Journal of Applied
Psychology, 74, 657-690.

158

Kanfer, R., Ackerman, P. L., Murtha, T. C., Dugdale, B., & Nelson, L. (1994). Goal-
setting, conditions of practice, and task-performance: A resource allocation
perspective. Journal of Applied Psychology, 79, 826-835.

Landauer, T. K. (Ed.). (1995). The trouble with computers: Usefulness, usability, and
prmluctivity. Cambridge, MA: MIT Press.

Latham, G. P., & Locke, E. A. (1991). Self-regulation through goal-setting.
Organizational Behavior and Human Decision Processes, 50, 212-247.

Lee, D. M. S. (1986). Usage pattern and sources of assistance for personal-computer
users. MIS Quarterly, 10, 313-325.

Legris, P., Ingham, J ., & Collerette, P. (2003). Why do people use information
technology? A critical review of the technology acceptance model. Information &
Management, 40, 191-204. ‘

Lindsley, D. H., Brass, D. J ., & Thomas, J. B. (1995). Efﬁcacy-performance spirals: A
multilevel perspective. Academy of Management Review, 20, 645-678.

Locke, E. A., & Bryan, J. F. (1969). Knowledge of score and goal level as determinants
of work rate. Journal of Applied Psychology, 53, 59-65.

Locke, E. A., & Latham, G. P. (Eds). (1990). A theory ofgoal setting and task
performance. Englewood Cliffs, NJ: Prentice Hall.

Martocchio, J. J ., & Webster, J. (1992). Effects of feedback and cognitive playfulness on
performance in microcomputer software training. Personnel Psychology, 45, 553-578.

Mathieson, K. (1991). Predicting user intentions: Comparing the technology acceptance
model with the theory of planned behavior. Information Systems Research, 2, 173-
I91 .

Melone, N. P. (1990). A theoretical assessment of the user satisfaction construct.
Management Science, 36, 76-92.

Milner, K. (2002). Understanding and improving training transfer motivation: an
application of recent advances in motivational theory. Unpublished Dissertation.
Michigan State University.

Mitchell, T. R., Hopper, H, Daniels, D., George-Falvy, J ., & James, L. R. (1994).
Predicting self-eﬂicacy and performance during skill acquisition. Journal of Applied
Psvchologv, 7.9, 506-517.

Moore, G. A. (1991). Crossing the chasm: Marketing and selling high-tech products to
mainstream customers. New York: Harper-Collins.

159

Morris, M. G., Venkatesh, V., & Ackerman, P. L. (2005). Gender and age differences in
employee decisions about new technology: An extension to the theory of planned
behavior. leee Transactions on Engineering Management, 52, 69-84.

Muir, B. M. (1994). Trust in automation .1. Theoretical issues in the study of trust and
human intervention in automated systems. Ergonomics, 37, 1905-1922.

Norman, D. A. (1993). Things that make as smart: Defending human attributes in the age
ofthe machine. Reading, MA: Addison-Wesley.

Olson, J. C ., & Dover, P. A. (1979). Disconﬁrmation of consumer expectations through
product trial. Journal of Applied Psychology, 64, 179-191.

Powers, W. T. (1973). Behavior: The Control of Perception. Chicago: Aldine.

Powers, W. T. (1978). Quantitative analysis of purporsive systems: Some spadework at
the foundations of scientiﬁc psychology. Psychological Review, 85, 417-435.

Rockart, J. F., Earl, M. J ., & Ross, J. W. (1996). Eight imperatives for the new IT
organization. Sloan Management Review, 38, 43-47.

Rogers, E. M. (1995). Diffusion of innovations (4th ed.). New York: The Free Press.

Ross, .1. W., & Weill, P. (2002). Six IT decisions your IT people shouldn't make. Harvard
Business Review, 80, 84-92.

Seijts, G. H., & Latham, G. P. (2001). The effect of distal Ieaming, outcome, and
proximal goals on a moderately complex task. Journal of Organizational Behavior,
22, 291-307.

Sheeran, P., & Orbell, S. (1999). Implementation intention and repeated behaviour:
Augmenting the predictive validity of the theory of planned behaviour. European
Journal of Social Psychology, 29, 349-369.

Sheeran, P., & Orbell, S. (2000). Using implementation intention to increase attendance
for cervical cancer screening. Health Psychology, 19, 283-289.

Sichel, D. E. (1997). The computer revolution: An economic perspective. Washington,
DC: The Brookings Institution.

Stajkovic, A. D., & Luthans, F. (1997). A meta-analysis of the effects of organizational
behavior modiﬁcation on task performance, 1975-95. Academy of Management

Journal, 40, 1122-1149.

Stajkovic, A. D., & Luthans, F. (1998). Self-eﬂicacy and work-related performance: A

160

meta—analysis. Psychological Bulletin, I 24, 240-261.

Stratopoulos, T., & Dehning, B. (2000). Does successful investment in information
technology solve the productivity paradox? Information & Management, 38, 103-1 17.

Straub, D., Limayem, M., & Karahanna-Evaristo, E. (1995). Measuring system usage —
Implications for IT theory testing. Management Science, 41, 1328-1342.

Taylor, S., & Todd, P. (1995). Assessing IT usage: The role of prior experience. MIS
Quarterly, 19, 561-570

Vancouver, J. 8., Thompson, C. M., & Williams, A. A. (2001). The changing signs in the
relationships among self-efficacy, personal goals, and performance. Journal of
Applied Psychology, 86, 605-620.

Vancouver, J. 8., Thompson, C. M., Tischner, E. C., & Putka, D. J. (2002). Two studies -1
examining the negative effect of self-efﬁcacy on performance. Journal of Applied 1
Psychology, 87, 506-516.

\r’ancouver, J. B., & Kendal, L. N. (2006). When self efﬁcacy negatively relates to
motivation and performance in a Ieaming context. Journal of Applied Psychology, 91,
1 146-1 153.

Venkatesh, V. (2000). Determinants of perceived ease of use: Integrating control,
intrinsic motivation, and emotion into the technology acceptance model. Information
Svstems Research, I l , 342-365.

Venkatesh, V., Brown, S. A., Maruping, L. M., & Bala, H. (2008). Predicting different
conceptualizations of system use: The competing roles of behavioral intention,
facilitating conditions, and behavioral expectation. Mis Quarterly, 32, 483-502.

Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the Technology
Acceptance Model: Four longitudinal ﬁeld studies. Management Science, 46, 186-
204.

Venkatesh, V., Davis, F. D., & Morris, M. G. (2007). Dead or alive? The development,
trajectory and future of technology adoption research. Journal of the Association for

In/brnuttion Systems, 8, 267-286.

Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of
infomiation technology: Toward a uniﬁed view. MIS Quarterly, 27, 425-478.

Vroom, V. H. (1964). Work and Motivation. New York, NY: John Wiley and Sons.

Weiner, L. R. (1993). Digital woes: Why we should not depend on software. Reading,
MA: Addison-Wesley.

161

Weiss. H. M., Beal, D. J., Lucy, S. L., & MacDermid, S. M. (2004). Constructing EMA
studies with PMAT: The Purdue Momentary Assessment Tool user's manual.
Software manual available at http://www.mfri.purdue.edu/pmat.

Will, R. P. (1991). True and false dependence on technology: Evaluation with an expert
system. Computers in Human Behavior, 7, 171-183.

Wixom, B. H., & Todd, P. A. (2005). A theoretical integration of user satisfaction and
technology acceptance. Information Systems Research, 16, 85-102.

Zimmemtan, B. J. (2000). Attaining self-regulation: A social cognitive perspective. In M.
Bockaerts, P. R. Pintrich & M. Zeidner (Eds), Handbook of self-regulation (pp. 13-

39). San Diego: Academic Press.

Zimmerman, B. J., & Bandura, A. (1994). Impact of self-regulatory inﬂuences on writing
course attainment. American Educational Research Journal, 31, 845-862.

162