Вы находитесь на странице: 1из 13

PERGAMON Omega, Int. J. Mgmt. Sci.

27 (1999) 327±339

The development of a tool for measuring the perceived


impact of information technology on work
G. Torkzadeh a, *, W.J. Doll b
a
College of Business Administration, The University of Texas at El Paso, El Paso, TX 79968, USA
b
The University of Toledo, Toledo, USA
Received 1 July 1997; accepted 1 July 1998

Abstract

The impact of information technology on work life has been one of the most talked about issues over the recent
years. Chief executive ocers spending millions of dollars on information technology face the critical issue of
assessing the impact of this technology on work. Information system managers are increasingly required to justify
technology investment in terms of its impact on the individual and his/her work. Measures of impact of information
technology have narrowly focused on productivity impacts. This study uses a broader concept that is based on the
impact of technology on the nature of work literature. This literature recognizes the multiple impacts of technology
on work at the level of the individual. A review of the literature enabled us to generate thirty-nine items that were
grouped into four constructs. In a pilot study, these constructs were assessed by observers in structured interviews
with eighty-nine users to provide a criterion measure. Next, the users completed the thirty-nine item questionnaire.
The unidimensionality, internal consistency and criterion-related validity of each construct were assessed. The pilot
results suggest a four factor 12-item instrument that measures how extensively information technology applications
impact task productivity, task innovation, customer satisfaction and management control. In a large scale study, a
sample of 409 respondents was gathered to further explore this 12-item instrument and its relationships with other
constructs (user involvement, user satisfaction, system usage). The results support the four factor model. Evidence
of reliability and construct validity is presented for the hypothesized measurement model and future research is
discussed. # 1999 Elsevier Science Ltd. All rights reserved.

Keywords: Information technology impact; System impact; Technology impact on work; Instrument development; System to value
chain; System success measures

1. Introduction its impact. Organizations that spend millions of dollars


on information technology are primarily concerned
Both academics and practitioners recognize that the about how their investment will in¯uence organiz-
success of information technology can potentially be ational and individual performance. MIS researchers
measured through its impact on work at the level of have quite appropriately increased their attention to
the individual end-user. The widespread use of infor- measurement issues in general and success measures in
mation technology by non-data processing pro- particular.
fessionals and growing line management responsibility To evaluate information systems success, MIS
for managing information technology, have further researchers have measured a variety of constructs such
increased the potential of information technology and as perceived usefulness [10], currency [1], accuracy [13]
and relative advantage [41]. Despite the contributions
made by MIS researchers in the development of suc-
* Corresponding author. Tel.: +1-915-747-5496; fax: +1- cess measures, Melone [39] calls for the development
915-747-5126; e-mail: reza@utep.edu. of e€ectiveness measures that are outcome-oriented.

0305-0483/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 3 0 5 - 0 4 8 3 ( 9 8 ) 0 0 0 4 9 - 8
328 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

Goodhue [19] proposes evaluation constructs that can 2. The impact of information technology on work
usefully link underlying systems to their relevant
impacts. Based on attitude-behavior theory, Doll and
The impact of information technology is broad and Torkzadeh [15] describe a `system to value chain' of
multifaceted providing signi®cant research opportu- system success constructs from beliefs, to attitudes, to
nities and challenges. There has been a growing inter- behavior, to the social and economic impacts of infor-
est on the part of MIS researchers to study mation technology (see Fig. 1). In this taxonomy,
information technology impacts: on competitive impact is a pivotal concept that embodies downstream
advantage [50], on organizational strategy [35], on time e€ects. It is dicult to imagine how information tech-
utilization [53], on middle managers [40, 43] and on nology can be assessed without evaluating the impact
industry level competitive advantage [49]. While the it may have on the individual's work. Impact is said to
research on impact of information technology has been occupy a pivotal position in a chain of system to value
diverse, it has not focused on work at the level of the constructs because it is a direct consequence of usage
individual. and, in turn, it is a major factor determining organiz-
Palmquist [42] suggests that while the societal ational impact.
changes represent a fascinating and complex research
challenge, information scientists must bear some 2.1. Information technology impact
responsibility to anticipate the impact of improved
access to information. Reviewing the relevant litera- Information technology has and continues to exert a
ture, he observed that there seemed few attempts to strong in¯uence on the individual and his/her work. It
examine the impact of information technology empiri- may be central to our future understanding of the
cally. Palmquist's review of the literature found that of structure and function of work organization for
all the research done to examine the new information human productivity, but for the individual it holds the
technologies, the impact on the individuals and their power to enhance a job or render human labor
work is least addressed. He believes it will become unnecessary [41]. Zubo€ [55] contends that technology
more and more important for researchers to examine cannot be considered neutral; it is brimming with
the impacts of information technology on all aspects valence and speci®city in that it both creates and fore-
of an individual's life. closes avenues of experience. Coates [4] asserts that
Explicit objectives of this research were to develop technology intrinsically, not accidentally, forces econ-
an instrument that: (1) identi®es the multidimensional omic, political or social change.
nature of information technology's impact at the level Danziger [9] identi®es the computer as the key tech-
of the individual end-user; (2) is short, easy to use and nological device producing the third great revolution
appropriate for both academic research and practice in human history. He suggests that the impacts of
and (3) can be used with con®dence across a variety of computing are highly contingent upon the context of
applications and contexts (i.e. generalizability as evi- use. That is, the impacts are quite varied in their
denced by reliability and predictive validity). These nature and their levels, not only across di€erent types
goals in¯uenced the methods used to generate items, of settings, but also across di€erent computing con-
the nature of the sample, the procedures used to elim- texts within comparable settings and even across di€er-
inate items to create a shorter and easier to use ques- ent individuals within comparable settings and
tionnaire, the nomological network of constructs computing contexts. In today's end-user computing en-
chosen to assess construct validity, and the methods vironment, the natural concern over how well designed
used to assess the generalizability of the instrument. In information systems are must be augmented with an
the following sections, we ®rst provide the theoretical equally ardent concern about how e€ectively they are
underpinning for the measurement of information used or what impact they have on work.
technology impact. Then, we describe research Literature [2, 30, 33, 37, 42, 46, 51, 54, 55] suggests
methods and surveys followed by practical and theor- that in the industrial model, technology was used to
etical applications of the proposed measures. substitute capital for labor with impacts on labor pro-

Fig. 1. System to value chain.


G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 329

ductivity and enhanced management control as pre- and Zubu€ [55] the researchers developed explicit de®-
dictable machines replaced variable human behavior. nitions for four impact dimensions (see Table 1). These
In the post-industrial model, the focus is still on pro- four impact dimensions, taken together, describe `how'
ductivity and management control but the stochastic an application impacts the individual in an organiz-
nature of the technology makes management control ational context. These impact dimensions are de®ned
more dicult. In addition, the post-industrial model at the application level. An application is de®ned as
includes innovation and customer satisfaction as the use of information technology to accomplish work.
necessary impacts of technology [7, 21, 22, 25, 31, 47, 55]. When applications are used in these ways, workers are
A conception of technology's impact that is limited empowered by information technology.
to productivity and/or management control is rooted in
an out-of-date paradigm that ignores organizationally 2.2.1. Task productivity
relevant impacts essential to the success and survival of Information technology impact on task pro-
modern organizations. In the context of information ductivity is suggested by numerous authors
technology and white-collar productivity, Davis [11] [2, 6, 7, 9, 10, 18, 24, 30, 32, 43, 54, 55]. Braverman [2]
suggests that improved e€ectiveness in serving internal suggests that organizations automate to improve pro-
and external customers is a better basis for judging ductivity and management control. Zubo€ [55] and
most white-collar work and that too much emphasis on Weick [54] discuss the problems of making full use of
lowering costs and increasing output may prove coun- information technology in a post-industrial work en-
terproductive. Traditional concepts of technology's vironment to improve productivity. Zubo€ suggests
impact emphasize task productivity and largely ignore that information technology essentially alters the con-
the role of technology in serving customers. tours of reality, productivity increases by an order of
magnitude and measures of productivity become more
2.2. Identifying and operationalizing impact dimensions abstract and dicult to determine. Weick argues that
new technologies mean many things because they are
The impact of information technology literature lar- simultaneously the source of events that are stochastic,
gely provides ad hoc examples and illustrations of how continuous, and abstract requiring di€erent measures
systems may impact the individual. Systematic atten- of productivity. Li€ [32] reported that most clerical
tion has not focused on conceptualizing the extent and and secretarial employees in her survey believed that
dimensions of this impact. Based on the works of `oce automation' had increased the interest and skill
Braverman [2], Curley and Pyburn [7], Danziger [9], requirements of their jobs resulting in higher pro-
Davis [11], Filiatrault et al. [17], Harvey [20], Harvey, ductivity.
et al. [22], Hirschhorn and Farduhar [24],
Hirschhorn [25], Kraemer and Danziger [30], Larson 2.2.2. Task innovation
and Fielden [31], Li€ [32], Long [33], Palmquist [42], The impact of information technology on task inno-
Sulek and Marucheck [43], Shaiken [51], Weick [54] vation has gained increased recognition in recent

Table 1
De®nitions of information technology's impacts on work

Label De®nition of construct Literature support

Task productivity The extent that an application improves Braverman [2], Curley and Pyburn [7], Hirschheim
the user's output per unit of time and Farduhar [24], Kraemer and Danziger [30],
Li€ [32], Sulek and Marucheck [43], Weick [54]
and Zubo€ [55]

Task innovation The extent that an application helps users Curley and Pyburn [7], Davis [11], Harvey et al. [22],
create and try out new ideas in their work Hirschhorn [25], Larson and Fielden [31], Long [33]
and Zubo€ [55]

Customer The extent that an application helps the user Curley and Pyburn [7], Filiatrault et al. [17], Harvey and
satisfaction create value for the ®rm's internal or external Filiatrault [21]], Harvey et al. [22], Hirschhorn [25, 26],
customers Schlesinger and Haskett [47]

Management The extent that the application helps to Braverman [2], Hirschhorn [26], Kraemer and
control regulate work processes and performance Danziger [30], Shaiken [51] and Zubo€ [55]
330 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

years [7, 11, 17, 22, 25, 31, 33, 54, 55]. Information tech- ted systematic and purposeful agency to the manage-
nology facilitates innovation through learning. In dis- rial use of technology [2, 51]. They argue that
cussing the impact of `industrial humanism managers are interested exclusively in technology as a
management', Long [33] suggests that employees will means of controlling, limiting, and ultimately weaken-
use information technology to enrich and broaden jobs ing their work force. Shaiken [51] anticipates fewer
(i.e. innovate). Larson and Fielden [31] suggest the skills, less worker input and increased managerial con-
need for task innovation or problem solving activity as trol. He suggests that a high level of skill is often
opposed to number of word measures of productivity. embodied in computerized systems which will lay the
Harvey et al. [22] suggest that many innovative compa- basis for a considerable transfer of power to manage-
nies are using the technology and exploring new ways ment. In some cases, sophisticated management, es-
to interface with customers. In the emerging post- pecially in small shops specializing in intricate
industrial setting, worker skill is not the skill of execut- prototype work, have felt that ecient production
ing but the skill of problem solving and the ability to requires skilled workers. Under these circumstances,
learn and innovate [25]. Work becomes heavily depen- the machine is viewed as a tool for the worker rather
dent on information processing, increasingly abstract than as a vehicle to control the worker. Shaiken
and mediated by sensing mechanisms. Thus, in study- suggests that the design of the technology determines
ing the impact of information technology, we cannot the context in which labor and management forces
just study what workers do, but rather how innovative play themselves out. Zubo€ suggests a more compli-
they are in what they have to do. The overall impact cated reality. With the di€usion of management func-
of technology is to di€use the management function in tion, management control has a di€erent meaning.
new ways as work (1) becomes a fusion of working Sometimes management is part of a group that moni-
and learning, (2) is identi®ed with developmental ac- tors his/her own performance. Using a socio-technical
tivity and (3) is organized through team relationships. perspective, Hirschhorn [26] contends that information
technology is used in new ways in an emerging post-
2.2.3. Customer satisfaction industrial labor process and has decisive implications
Information technology empowers employees to pro- for the general texture of life. He suggests that if we
vide better and faster service to both internal and are to make full use of the new technologies, we must
external customers [7, 17, 21, 22, 25, 26, 47]. Describing confront the constraints imposed by management sys-
the role of technology in service quality, Harvey et tems, capitalistic interests and old organizational de-
al. [22] point out the more dicult task of ``balancing signs.
short-term, directly measurable productivity gains
against the more important and more pervasive, but
more dicult to measure gains that come from produ- 3. Methods
cing and delivering more value to the customer''. They
suggest a model that describes how services that pro- A literature review was conducted to ensure that a
vide `value-added partnerships' can be created through comprehensive list of items was generated to measure
information technology. Failiatrault et al. [17] suggest dimensions of application impact. The works of MIS
a close relation between service quality, productivity researchers [18, 27, 33, 34, 36, 38, 43, 44, 48, 52], works
and customer satisfaction. They argue that improving on the social and economic impact of information
productivity can improve customer satisfaction. technology [2, 8, 9, 24, 25, 28, 29, 33, 36, 42, 45, 51, 54, 55]
Harvey and Filiatrault [21] review technology adoption and works on white collar productivity [11, 20±23, 31±
in the banking industry and describe, for example, 33, 47, 53] were reviewed. This literature, especially the
how on-line technology can speed up client queries or descriptions of information technology's impact on the
implement client requests, increasing customer satisfac- nature of work, is a rich source of illustrations or
tion. Improved service quality is perceived through examples of how information systems are used by indi-
close interaction and real-time ¯ow of information. viduals in their work context. Based upon this review,
Information technology applications can be used to the researchers generated thirty-nine items to measure
create customer `empowerment' that will ultimately aspects of impact (see Appendix A). A ®ve point
result in customer satisfaction [47]. Likert-type scale was used where 1 = not at all, 2 = a
little, 3 = moderately, 4 = much and 5 = a great deal.
2.2.4. Management control The researchers developed a structured interview
Many authors suggest that one of the initial intents questionnaire and process for the pilot study to: (1)
for using a new technology is to increase management assess whether the instrument was capturing the
control. [2, 26, 30, 51, 55]. Although technology rede- phenomenon desired by the researchers; (2) verify that
®nes the possible, it cannot determine which choices important aspects of impact dimensions were not
are taken up and to what purpose. Some have attribu- omitted and (3) enable an interviewer to make an
G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 331

assessment (observation) of the dimensions of impact Finally, the reliability (internal consistency) of the
described above. First, personal interviews with the remaining items comprising each dimension was
end-users were conducted, then the self-administered examined using Cronbach's alpha to see if additional
thirty-nine item questionnaire was distributed. The items could be eliminated without substantially low-
interviewers (6) were experienced systems analysts ering reliability. Items were eliminated if the re-
(from two to six years as a professional systems ana- liability of the remaining items would be at least
lyst) who were carefully trained to conduct the inter- 0.90. Where deleting either of two items would have
views and make observations. The interviews took had the same impact on alpha, the item with the
about 2 h each. highest correlation with the observer criterion was
The users were asked to demonstrate their appli- retained.
cation, provide sample output reports, and describe After these deletions, an exploratory factor analy-
how they used the application in their work. For each sis of the remaining items was conducted to deter-
dimension, users provided qualitative information mine whether the entire set of items had a simple
describing the impact of their applications. As each factor structure and to determine the number of
category of impact was discussed, the interviewer made factors. Finally, to assess criterion-related validity,
an observation of the extent of impact using a 5-point the four constructs were correlated with the four
Likert-type scale (1 = not at all to 5 = a great deal). observer criteria: (1) to assess whether their scales
This observer's score was used as one criterion to vali- were more highly correlated with the observer score
date the items on the self-administered questionnaire. for their theoretical construct than with observer
scores for the other constructs and (2) to assess
3.1. Assessment of measurement properties whether the observer scores for each construct were
more highly correlated with the scale for its theor-
The questionnaire responses and observer criterion etical construct than with the scales for other con-
scores were explored with several objectives in mind: structs.
puri®cation, unidimensionality, reliability, brevity and After determining the factor structure, the research-
simplicity of factor structure. First, the researchers ers examined construct validity by assessing the corre-
wanted to purify the items before doing factor analysis lation between impact dimensions and the nomological
(i.e. eliminate `garbage items'). The need to purify the network of constructs. These constructs included a
items that are suggested as measures of a construct is measure of perceived user involvement, user satisfac-
described by Churchill [5]. He contends that when fac- tion, and usage pattern. User involvement was
tor analysis is done before puri®cation, there seems to measured using an eight item scale developed by Doll
be a tendency for factor analysis to produce many and Torkzadeh [14]. User satisfaction was measured
more dimensions than can be conceptually identi®ed, using the 12-item end-user computing satisfaction
confounding the interpretation of the factor analysis. (EUCS) instrument [13]. Both instruments use a ®ve
Two independent criteria were used to eliminate point Likert-type scale: 1 = not at all; 2 = a little;
items. First, items were eliminated if their corrected- 3 = moderately; 4 = much; 5 = a great deal). The 30-
item total correlation (each item's correlation with the item perceived usage pattern scale includes items for
sum of the other items in its category) was less than decision support, work integration and customer
0.70. The domain sampling model provides a rationale service [16]. A ®ve point Likert-type scale is used
for this procedure. The key assumption in the domain (1 = not at all; 2 = a little; 3 = moderate; 4 = much;
sampling model is that all items, if they belong to the 5 = a great deal).
domain of the concept, have an equal amount of com-
mon core. If all the items in a measure are drawn from
the domain of a single construct, responses to those 4. Pilot study
items should be highly intercorrelated. The corrected-
item total correlation provides a measure of this [5]. Eighty-nine usable interviews were obtained using a
Items were also eliminated if their correlation with structured interview procedure followed by the self-
the observer's criterion score, obtained through the administered questionnaire. Qualitative comments
pilot study, was less than 0.70. The users' descriptions from the structured interviews were compared with the
and demonstrations of their applications, the examin- responses to items in each of the four categories. This
ation of output reports, and the qualitative interview enabled the researchers to verify that the respondents
comments recorded for each dimension enabled the knew what the items were asking. To make the results
research assistants to make observations of an applica- more generalizable, the researchers gathered data from
tion's impact on the interviewee. These observer scores 30 di€erent ®rms. We tried to survey a variety of user
(using a 1 to 5 scale) for each category of items were situations including individuals engaged in computer
also used to eliminate `garbage items'. intensive work as well as more casual users.
332 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

4.1. Purifying items structure was easy to interpret and corresponding


with the four categories described above. The range
Twenty four of the thirty-nine items illustrated in for factor loading was 0.90 to 0.92 for task pro-
Appendix A were eliminated because they had either ductivity, 0.87 to 0.94 for task innovation, 0.88 to
a corrected-item total correlation or a correlation 0.94 for customer satisfaction and 0.94 to 0.96 for
with the observer criterion less than 0.70. About management control. The four-factor model
90% of eliminated items failed both criteria and the explained 89.9% of the variance.
remainder failed (about equally) either corrected- Finally, to assess criterion-related validity, the rec-
item correlation or correlation with the criterion. ommended scales for the four constructs were corre-
Reliability analysis resulted in further deletion of lated with the observer criterion scores for each
three items: coecient alpha for the task innovation construct. The correlations for each construct and its
items could be improved if one item was deleted, corresponding observer criterion score were 0.90 for
coecient alpha for management control items task productivity, 0.89 for task innovation, 0.84 for
could be improved if two items were deleted. customer satisfaction and 0.91 for management con-
Table 2 reports the corrected-item total correlations trol. These correlations are signi®cant at p < 0.001.
and correlations with the observer criterion for the Each observer criterion is more correlated with its cor-
remaining 12 items. Coecient alpha for each con- responding scale than it is with the scales measuring
struct is also provided in Table 2. Once the items the other constructs.
were puri®ed and reliability was established, the These pilot study methods enabled the researchers to
construct validity of the four scales was examined. verify that the four dimensions of information techno-
logy's impact on work made sense in terms of the end-
4.2. Construct validity results user's work context. The high correlation between
observer ratings and end-user responses support the
The factor analysis of the items indicated unidi- contention that these impacts of technology on work
mensionality of each construct. There were no mul- can be veri®ed by trained observers as well as user re-
tiple loadings for any category. The ratio of sample sponse data. Thus, pilot study resulted in a 12-item
size to number of items (7.4:1) was lower than instrument that operationalizes four constructs (task
accepted standard (10:1), but considered adequate for productivity, task innovation, customer satisfaction
exploratory analysis at the pilot stage. The factor and management control).

Table 2
Measures of application impact on work (n = 89)

Corrected item total Correlation with


correlation criterion

Task productivity (alpha = 0.90)


P1 this application saves me time 0.83 0.80
P2 this application increases my productivity 0.80 0.84
P3 this application allows me to accomplish more work than 0.76 0.82
would otherwise be possible

Task innovation (alpha = 0.95)


I1 this application helps me create new ideas 0.93 0.85
I2 this application helps me come up with new ideas 0.91 0.82
I3 this application helps me try out innovative ideas 0.82 0.86

Customer satisfaction (alpha = 0.94)


C1 this application improves customer service 0.92 0.80
C2 this application improves customer satisfaction 0.89 0.81
C3 this application helps me meet customer needs 0.84 0.78

Management control (alpha = 0.97)


M1 this application helps management control the work process 0.94 0.86
M2 this application improves management control 0.93 0.88
M3 this application helps management control performance 0.91 0.90
G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 333

5. The survey iety of applications support the generalizability of the


®ndings.
To further explore this 12-item instrument, a sample Respondents were asked to identify their position
of 409 computer end-users was gathered from 18 or- within the organization; they responded as follows: 20
ganizations including 8 manufacturing ®rms, 1 retail top level managers, 80 middle managers, 76 ®rst level
®rm, 2 government agencies, 2 utilities, 2 hospitals, 2 supervisors, 143 professional employees, and 90 oper-
educational institutions and one `other'; this rep- ating personnel. 47% of the respondents stated that
resented half of the ®rms contacted. The response rate their application was a personal computer (micro) ap-
for ®rms willing to participate with the study was over plication. 36% of the computer applications were
70%. This second data set enabled the researchers to developed primarily by end-users but only 12% of the
re-examine factor structure and reliability. To assess respondents personally developed the applications
construct validity, the researchers also gathered data themselves. 50% of the applications provided data
on other constructs (i.e. user involvement, user satis- analysis capabilities (spreadsheet, modeling, simulation,
faction, usage pattern) across the system to value optimization or statistical routines). 46% of the appli-
chain. cations provided a data base with ¯exible inquiry capa-
The sample represents 139 di€erent applications bilities.
including accounts payable, accounts receivable, bud-
geting, CAD, CAD-CAM, customer service, service 5.1. Reliability and validity results
dispatching, engineering analysis, process control,
work order control, general ledger, manpower plan- An exploratory factor analysis was conducted using
ning, ®nancial planning, inventory, order entry, pay- principal components as the means of extraction and
roll, personnel, production planning, purchasing, oblimin as the method of rotation. Without specifying
quality, sales analysis, student data and pro®t plan- the number of factors, there were four factors with
ning. The large number of organizations and the var- eigenvalues greater than 1.0. The four factor solution

Table 3
Factor pattern of application impact items (n = 409)

Factor 1 Factor 2 Factor 3 Factor 4

C1 0.907
C2 0.897
C3 0.876

I1 0.896
I2 0.873
I3 0.867

M2 0.927
M3 0.879
M1 0.876

P2 0.872
P1 0.866
P3 0.827

Factor correlation matrix


Factor 2 0.35
Factor 3 0.36 0.36
Factor 4 0.53 0.31 0.50

Mean 9.52 6.85 9.12 10.72


Standard deviation 3.86 3.46 3.87 3.64

Eigenvalue 6.01 1.79 1.65 1.08

Total variance explained = 87.7%.


334 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

reported in Table 3 represent a simple solution. The tor) than with items of its own theoretical variable.
items are grouped by their highest (primary) factor Campbell and Fiske [3] suggest determining whether
loading and listed in descending order. Loadings this count is higher than one-half the potential com-
greater than or equal to 0.30 are reported. The four parisons. However, in this case, common method var-
factors explained 87.7% of the variation in the 12 iances are present so it is unclear how large a count
items and were interpreted as customer satisfaction, would be acceptable. An examination of the matrix for
task innovation, management control and task pro- the 12-item instrument reveals zero violation (out of 66
ductivity, respectively. Table 3 also provides factor cor- comparisons) of the condition for discriminant validity.
relation matrix, means and standard deviations for the Each of the twelve items are more highly correlated
four factors. Thus, the four factor model identi®ed in with the other items in its group than with any of the
the pilot study was supported by this large sample ex- items measuring other variables.
ploratory analysis.
Using this sample of 409 responses, the corrected- 5.3. General applicability results
item total correlation and reliability (alpha) for each of
the four impact scales were calculated. The corrected- Correlations between the four impact scales being
item total correlations range from 0.82 (P1) to 0.94 proposed and related constructs thought to comprise
(C1). Reliability scores are 0.93, 0.95, 0.96 and 0.93 for the nomological network were examined. In this
task productivity, task innovation, customer satisfac- sample, the 8-item user involvement and 12-item user
tion and management control, respectively. Overall re- satisfaction scales have reliabilities (Cronbach's alpha)
liability for the 12-item scale was 0.92. of 0.96 and 0.92, respectively. The 30-item usage pat-
tern scale has a reliability of 0.96. The correlation
5.2. Convergent and discriminant validity results matrix for the impact scales and the related constructs
in the nomological network is provided in Table 4.
The instrument's correlation matrix was analyzed User involvement has signi®cant correlations with
for convergent and discriminant validity [13]. This all potential success measures. Correlations between
approach to convergent validity tests that the corre- user involvement and usage pattern (r = 0.31), per-
lations between measures of the same theoretical con- ceived impact (r = 0.32) and EUCS (r = 0.30) are
struct are di€erent than zero and large enough to remarkably similar and signi®cant ( p < 0.01). EUCS
warrant further investigation. The smallest within vari- has signi®cant ( p < 0.01) correlation with usage pat-
able (factor) correlations are: task productivity = 0.79; tern (r = 0.29) and the four dimensions of impact scale
task innovation = 0.73; customer satisfaction = 0.81 (task productivity, r = 0.47; task innovation, r = 0.30;
and management control = 0.73. For a sample of 409, customer satisfaction, r = 0.28 and management con-
these are signi®cantly ( p < 0.001) di€erent than zero trol, r = 0.17). EUCS's signi®cant correlations with
and large enough to encourage further investigation. overall as well as subscales of perceived impact suggest
Using this approach, discriminant validity is tested that EUCS may have a direct as well as an indirect
for each item by counting the number of times it corre- (through usage patterns) relationship with perceived
lates more highly with an item of another variable (fac- impact.

Table 4
Correlations among constructs (n = 409)

1 2 3 4 5 6 7 8

(1) Overall impact (12 items)


(2) Task productivity (3 items) 0.80**
(3) Task innovation (3 items) 0.74** 0.52**
(4) Customer satisfaction (3 items) 0.77** 0.55** 0.38**
(5) Management control (3 items) 0.71** 0.37** 0.37** 0.37**
(6) Usage pattern (30 items) 0.70** 0.52** 0.64** 0.46** 0.49**
(7) User involvement (8 items) 0.32** 0.26** 0.40** 0.10 0.21** 0.31**
(8) User satisfaction (12 items) 0.40** 0.47** 0.30** 0.28** 0.17** 0.29** 0.30**

Mean 36.17 10.72 6.85 9.52 9.12 62.9 14.6 47.9


Standard deviation 11.20 3.64 3.46 3.86 3.87 24.1 9.7 8.3

*
Signi®cance less than or equal to 0.05.**Signi®cance less than or equal to 0.01
G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 335

The four dimensions of impact scale are signi®- e€ectiveness and productivity, it should be of great
cantly correlated with each other with the closest as- interest to executive management; (3) as an outcome
sociation existing between task productivity and measure, it might stimulate new research that has prac-
customer satisfaction (r = 0.77). The lowest corre- tical implications for how systems are designed and
lations (r = 0.37) exist between management control developed and (4) it may help us gauge and reduce a
and other three dimensions of the impact scale. Each perceived gap between the potential of information
of the four impact dimensions have signi®cantly technology and its actual impact.
( p < 0.01) higher correlations with perceived usage The results of pilot study were substantive and gave
pattern than other measures (i.e. user satisfaction, the researchers con®dence that the four dimensions
user involvement), suggesting that they are more clo- make sense in term of the end-user work context and
sely associated with usage than user satisfaction or represent real phenomenon that can be observed/veri-
user involvement. Means, standard deviations, and ®ed by a trained professional. The results based on
correlations for the 12-item scale are provided in observer criterion scores also supports the use of end-
Appendix B. user responses as a way of measuring the impacts of
The reliability of the impact scales for the entire information technology.
sample and for di€erent hardware platforms, types of
applications, modes of application development, and
required usage conditions are reported in Table 5. All 6.1. Implications
45 Cronbach alpha scores are high (ranging from
0.89 to 0.97), indicating remarkable stability across Substantial experimentation and learning may be
all contextual variables. These reliability alphas sup- necessary before we can develop ®nal guidelines for
port the generalizability of the impact scales across a using these instruments. In this section, we provide
variety of hardware platforms, application types, some tentative suggestions for the instrument's appli-
modes of development and required or voluntary cation.
usage situations. This 12-item instrument may be used to evaluate the
impact of applications. The Likert-type scales are easy
to administer and the instrument is short enough to be
used in most research or applied settings. It is our
6. Discussion assumption that both performance-related behaviors
and attitudes determine the impact of information
A multidimensional measure that identi®es how in- technology. Thus, user satisfaction measures should be
formation systems impact work may have several ad- augmented, not replaced, by impact measures. These
vantages: (1) it might help us distinguish between measures of impact can be used for a variety of situ-
e€ective and ine€ective systems; (2) as a measure of ations including: (1) where ¯exible technology is pre-

Table 5
Reliability by nature and type of application

Overall Task Task Customer Management


impact productivity innovation satisfaction control

Mode of development
End-user development (n = 129) 0.92 0.93 0.95 0.96 0.93
Traditional development (n = 259) 0.90 0.93 0.90 0.94 0.91

Hardware platform
Personal computer (n = 181) 0.89 0.92 0.93 0.94 0.94
Mini or mainframe (n = 204) 0.93 0.93 0.91 0.95 0.90

Usage required?
Yes (n = 249) 0.92 0.93 0.91 0.94 0.89
No (n = 49) 0.92 0.93 0.94 0.97 0.92

Type of application
Decision support (n = 185) 0.89 0.91 0.89 0.96 0.92
Data base (n = 175) 0.90 0.92 0.91 0.96 0.90
Other (n = 131) 0.91 0.92 0.94 0.93 0.92
336 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

sent and may have di€erent degrees of impact on the customer needs and control the work process within
individual; (2) where there are substantial di€erences and between groups. In longitudinal research designs,
between users in how e€ectively they use a given appli- these impact measures provide a tool for assessing
cation; (3) where user attitudes are likely to be biased whether information technology is, indeed, being used
and (4) where one wishes to implement a behaviorally- in new ways.
based training program and one desires some initial
behavioral benchmarks.
The impact scales might be used to make compari- 7. Conclusions
sons between users of the same software packages (e.g.
to identify di€erences between individuals and assess The ultimate question for use of information tech-
training needs). Also, one might wish to use di€erent nology by individuals and organizations may be
parts of the instrument to evaluate di€erent types of more related to how the technology impacts than
applications. For example, the task productivity scale how technology is designed or regarded. Despite
might be the only scale relevant to evaluating a ®rm's wide acceptance of and call for outcome oriented
task productivity applications. measures of information systems success, the impact
As measures of e€ective or ine€ective application, construct, its conceptualization and measurement, has
these instruments open promising new areas of inquiry. received inadequate research attention. This research
For example, information systems research on learning reports on an initial e€ort to conceptualize the
processes or continuous improvement behaviors has impact construct and develop measures of infor-
been hampered by the lack of an appropriate depen- mation technology impact that have adequate re-
dent variable [12]. Researchers might use these instru- liability and validity. The four impact dimensions
ments to facilitate the identi®cation of situational identi®ed (task productivity, task innovation, custo-
factors and processes that determine both the e€ective- mer satisfaction and management control) measure
ness of application and the rate at which individuals the extent of information technology impact on
learn to apply new technologies which are made avail- work. Further research is needed to con®rm the hy-
able to them. Research on the link between upstream pothesized measurement model, assess the stability of
causal factors such as design features or development the instrument and develop standards for evaluating
methods and their downstream consequences (i.e. how speci®c applications.
e€ective information systems are) can also bene®t from
these instruments. Impact dimensions have an advan-
tage as a construct for integrating MIS research with
studies of information technology usage; a causal link
between use and impact is quite plausible. Appendix A. Measures of application impact on work
(thirty nine items used in pilot study)
6.2. Extensions
Impact on task productivity:
There is a clear need for further research and con®r-
matory methods to examine the second-order nature of (1) This application improves the quality of my
the impact on work construct and its components, work.
®rst-order dimensions. It is intuitively plausible to (2) This application increases my productivity.
expect a second-order model for the impact of technol- (3) This application saves me time.
ogy on work and this study clearly suggests the four (4) This application enables me to accomplish tasks
components. However, only further research can pro- more quickly.
vide a clearer picture for this phenomenon employing (5) This application supports critical aspects of my
con®rmatory factor analysis. Further research should job.
also examine the relationship between the impact of in- (6) This application allows me to accomplish more
formation technology construct and outcome variables work than would otherwise be possible.
such as information technology usage. (7) This application enables me to spend more time
In the emerging post-industrial setting, scholars con- on productive activities.
tend that information technology is used in new ways (8) My e€ectiveness on the job is enhanced by this
that have substantial implications for the nature of application.
work, productivity and economic growth. Information (9) The quality of my work is improved by this ap-
technology has the potential to di€use management plication.
functions throughout the work force. Individuals will (10) This application is useful in my job.
e€ectively use information technology to improve pro- (11) This application assists me in performing my job
ductivity, create new ideas, meet internal and external better.
G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 337

(12) The quality of my work depends upon this appli- (26) This application enables me to respond to chan-
cation. ging customer needs.
(27) This application helps me accommodate individ-
ual customer needs.
Impact on task innovation: (28) This application enables me to deal more strate-
gically with internal and/or external customers.
(13) This application helps me identify innovative
ways of doing my job.
(14) This application helps me come up with new sol- Impact on management control:
utions to job problems.
(15) This application helps me create new ideas. (29) This application improves management control.
(16) This application helps me come up with new (30) This application helps management control the
ideas. work process.
(17) This application helps me solve job problems. (31) This application helps management control per-
(18) This application helps me ®nd new ways to formance.
improve my job performance. (32) This application enables management to com-
(19) This application helps me try out innovative pare work performance to standards.
ideas. (33) This application helps management identify
when corrective action is required.
(34) This application enables management to monitor
Impact on internal/external customer satisfaction: and correct errors.
(35) This application enables management to control
(20) This application improves customer satisfaction. work schedules.
(21) This application improves customer service. (36) This application enables management to monitor
(22) This application makes me more customer work progress.
oriented. (37) This application enables management to ensure
(23) This application helps me create value for custo- a timely completion of tasks.
mers. (38) This application enables management to control
(24) This application helps me meet customer needs. resource allocation.
(25) This application helps me adapt to changing cus- (39) This application enables management to control
tomer needs. quality.

Appendix B. Means standard deviations and correlations for the 12-item impact scale (n=409)
Table 6

P1 P2 P3 I1 I2 I3 C1 C2 C3 M1 M2 M3

P1
P2 0.84
P3 0.79 0.81

I1 0.43 0.45 0.48


I2 0.40 0.45 0.48 0.83
I3 0.41 0.45 0.47 0.80 0.73

C1 0.52 0.47 0.48 0.35 0.32 0.30


C2 0.48 0.43 0.46 0.38 0.33 0.32 0.87
C3 0.53 0.51 0.51 0.34 0.30 0.31 0.88 0.81

M1 0.33 0.35 0.39 0.33 0.35 0.24 0.35 0.30 0.34


M2 0.27 0.31 0.32 0.33 0.33 0.27 0.33 0.26 0.31 0.81
M3 0.29 0.34 0.34 0.36 0.32 0.30 0.37 0.31 0.37 0.73 0.83

Mean 3.7 3.5 3.5 2.3 2.4 2.2 3.2 3.1 3.2 3.2 3.1 2.9
Standard deviation 1.2 1.3 1.3 1.3 1.2 1.3 1.4 1.3 1.4 1.4 1.4 1.4
338 G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339

[21] Harvey J, Filiatrault P. Service delivery processes: new


technology and design. Int J Bank Market
1991;9(1):2531.
[22] Harvey J, Lefebvre E, Lefebvre L. Technology and the
creation of value in services: a conceptual model.
Technovation 1993;13(8):481±95.
[23] Hirschheim R. Oce automation: concepts, technologies
References
and issues. Reading, MA: Addison-Wesley, 1978.
[24] Hirschhorn L, Farduhar K. Productivity, technology and
[1] Bailey JE, Pearson SW. Development of a tool for
the decline of the autonomous professional. Oce
measuring and analyzing computer user satisfaction.
Technol People 1985;2:245±65.
Manage Sci 1983;29(5):530±45.
[25] Hirschhorn L. The post-industrial labor process. N Pol
[2] Braverman H. Labor and monopoly capital. New York,
Sci 1981;0:11±32.
NY: Monthly Review Press, 1974.
[26] Hirschhorn L. Beyond mechanization: work and technol-
[3] Campbell DT, Fiske DW. Convergent and discriminant
ogy in a postindustrial age. Cambridge, MA: The MIT
validation by the multitrait±multimethod matrix. Psychol
Press, 1984.
Bull 1959;56(1):81±105.
[27] King WR, Rodriquez JI. Evaluating MIS. MIS Q
[4] Coates JF. Science, technology and human rights.
1978;2(3):43±52.
Technol Forecast Social Change 1991;40:389±91.
[28] Kling R, Scacchi W. The web of computing: computer
[5] Churchill GA, Jr. A paradigm for developing better
technology as social organization. Adv Comput
measures of marketing constructs. J Market Res
1980;21:1±90.
1979;16(1):64±73.
[29] Kling R. Social analysis of computing: theoretical per-
[6] Cooper RB, Zmud RW. Information technology im-
plementation research: a technology di€usion approach. spectives in recent empirical research. Comput Surv
Manage Sci 1990;36(2):123±39. 1980;12(1):61±110.
[7] Curley KF, Pyburn PJ. `Intellectual' technologies: the [30] Kraemer KL, Danziger JN. The impacts of computer
key to improving white-collar productivity. Sloan technology on the worklife of information workers.
Manage Rev 1982;0:31±9. Social Sci Comput Rev 1990;8(4):592±613.
[8] Danziger JN, Kraemer K. People and computers: the [31] Larson RW, Fielden JS. The managerial clerk syndrome.
impacts of computing on end-users in organizations. Bus Horizon 1985;0:26±34.
New York, NY: Columbia University Press, 1986. [32] Li€ S. Clerical workers and information technology: gen-
[9] Danziger JN. Social science and the social impacts of der relations and occupational change. N Technol Work
computer technology. Social Sci Q 1985;66(1):3±21. Employment 1990;18:4±28.
[10] Davis F. Perceived usefulness, perceived ease of use, and [33] Long RJ. The impact of new oce information technol-
user acceptance of information technology. MIS Q ogy on job quality of female and male employees.
1989;13(3):319±42. Human Relat 1993;46(8):939±61.
[11] Davis TRV. Information technology and white-collar [34] Lucas HC, Jr. Performance and the Use of information
productivity. Acad Manage Executive 1991;5(1):55±67. systems. Manage Sci 1975;21(8):908±19.
[12] DeLone WH, McLean ER. Information systems success: [35] Mahmood MA. A comprehensive model for measuring
the quest for the dependent variable. Inf Syst Res the potential impact of information technology on organ-
1992;3(1):60±95. izational strategic variables. Decision Sci 1991;22(4):869±
[13] Doll WJ, Torkzadeh G. The measurement of end-user 97.
computing satisfaction. MIS Q 1988;12(2):259±74. [36] Markus ML, Robey D. Information technology and or-
[14] Doll WJ, Torkzadeh G. The measurement of end-user ganizational change: causal structure in theory and
software involvement. OMEGA 1990;18(4):399±406. research. Manage Sci 1988;34(5):583±98.
[15] Doll WJ, Torkzadeh G. The measurement of end-user [37] Markus ML. Understanding information systems use in
computing satisfaction: theoretical and methodological organizations. Ph.D. dissertation, Case Western Reserve
issues. MIS Q 1991;15(1):5±10. University, 1979.
[16] Doll WJ, Torkzadeh G. Developing a multidimensional [38] Mason RO. Measuring information output: a communi-
measure of system-use in an organizational context. Inf cation systems approach. Inf Manage 1978;1(5):219±34.
Manage 1998;33:171±85. [39] Melone NP. A theoretical assessment of the user-satisfac-
[17] Filiatrault P, Harvey J, Chebat J. Service quality and ser- tion construct in information systems research. Manage
vice productivity management practices. Ind Market Sci 1990;36(1):76±91.
Manage 1996;25:243±55. [40] Millman Z, Hartwick J. The impact of automated oce
[18] Fuerst WL, Cheney PH. Factors a€ecting the perceived systems on middle managers and their work. MIS Q
utilization of computer-based decision support systems. 1987;11(4):479±91.
Decision Sci 1982;13(4):554±69. [41] Moore GC, Benbasat I. An empirical examination of a
[19] Goodhue DL. Understanding user evaluations of infor- model of the factors a€ecting utilization of information
mation systems. Manage Sci 1995;41(12):1827±44. technology by end users. Working Paper. University of
[20] Harvey J. Operations management in professional service British Columbia, 1992.
organizations: a typology. Int J Oper Product Manage [42] Palmquist RA. The impact of information technology on
1990;10(4):5±15. the individual. In: Williams ME, editor. Annual review
G. Torkzadeh, W.J. Doll / Omega, Int. J. Mgmt. Sci. 27 (1999) 327±339 339

of information science and technology (ARIST), vol. 27. [50] Sethi V, King WR. Development of measures to assess
Medford, NJ: Learned Information, 1992:3±42. the extent to which an information technology appli-
[43] Pinsonneault A, Kraemer KL. The impact of infor- cation provides competitive advantage. Manage Sci
mation technology on middle managers. MIS Q 1994;40(12):1601±27.
1993;17(3):271±92. [51] Shaiken H. Work transformed: automation and labor in
[44] Robey D. User attitudes and management information the computer age. New York: Holt, Rinehart and
system use. Acad Manage J 1979;22(3):527±38. Winston, 1985.
[45] Rule J, Attewell P. What do computers do?. Social Prob [52] Srinivasan A. Alternative measures of system e€ective-
1989;36(3):225±41.
ness: associations and implications. MIS Q
[46] Shannon CE, Weaver W. Work transformed: automation
1985;9(3):243±53.
and labor in the computer age. Lexington, MA:
[53] Sulek JM, Marucheck AS. A study of the impact of an
Lexington Books, 1986.
[47] Schlesinger LA, Haskett L. The service driven service integrated information technology on the time utilization
company. Harvard Bus Rev 1991;0:71±81. of information workers. Decision Sci 1992;23:1174±91.
[48] Schewe CD. The MIS user: an exploratory behavioral [54] Weick KE. Technology as equivoque: sensemaking in
analysis. Acad Manage J 1976;19(4):577±90. new technologies. In: Goodman PS, Sproull LS et al.,
[49] Segars A, Grover V. Strategic group analysis: a meth- editors. Technology and organizations. San Francisco:
odological approach for exploring the industry level Jossey-Bass, 1990:1±44.
impact of information technology. Omega 1994;22(1):13± [55] Zubo€ S. In the age of the smart machine: the future of
34. work and power. New York, NY: Basic Books, 1988.

Вам также может понравиться