Omega, Int. J. Mgmt Sci. Vol. 26, No. 5, pp.

611±621, 1998
# 1998 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0305-0483(98)00007-3 0305-0483/98 $19.00 + 0.00

An Empirical Study of the Performance

Impact of Computerization in the Retail
Industry
THOMPSON SH TEO
POH KAM WONG
National University of Singapore, Singapore 119260, Singapore

(Received January 1997; accepted after revision December 1997)

This paper examines the relationships between the intensity of IT investment and four performance
impact measures relating to computerization. Data were collected via a questionnaire survey of the
retail industry in Singapore. Results suggested that the intensity of IT investment has negligible re-
lationships with information quality and improvement in work environment. Information quality is
positively related to improvement in work environment, managerial satisfaction and organizational
impact. Improvement in work environment is positively related to organizational impact but not to
managerial satisfaction. Implications of the ®ndings are discussed. # 1998 Elsevier Science Ltd.
All rights reserved

Key wordsÐcomputerization, information technology, performance impact, investment intensity,

information quality, satisfaction, retail industry

1. INTRODUCTION been studied include quality of work-life

[2, 11], information quality [10, 17], new organ-
INVESTMENTS IN INFORMATION technology (IT) izational forms [12], improvements in work
have been increasing over the years. As more environment [18, 20], electronic coordination
resources are being spent on the acquisition of [8, 16], organizational ¯exibility [13], control
the latest IT systems and applications in order systems [22], satisfaction [24, 27], and ®nancial
to compete more e€ectively in the increasingly performance [15].
competitive global marketplace, many Evidence of such impacts have ranged from
businesses are concerned about whether these speculative and anecdotal data to longitudinal
IT investments are justi®able. In particular, empirical data on the impacts of IT on organ-
since the retail sector in Singapore has been izational performance. The empirical results
experiencing poor growth in the past few years have generally been mixed. Some researchers
due to increased local and global competition, have found positive relationships between the
many retail companies are investing in IT in intensity of IT investment and performance
order to reduce labor costs as well as to impacts [3, 9] whereas others have not [26].
increase productivity. These retail companies The elusive nature of this relationship has led
are concerned over the tangible and intangible some researchers to refer to it as the ``pro-
payo€s from these IT investments. ductivity paradox'' of IT investment [5, 7].
Numerous research has been conducted to One reason for this paradox is that many
assess the payo€s and performance impacts of other factors (e.g., leadership and manage-
IT investments. Impacts of IT which have ment) besides IT investments may a€ect
611
612 Teo, WongÐPerformance Impact of Computerization in the Retail Industry

organizational performance. Another reason is antecedents to impact on individual perform-

that there is often a time lag between IT ance. In turn, individual impact should even-
investment and performance impacts. tually in¯uence organizational impact.
In a ComputerWorld CEO-CFO survey on In this study, we adapt and modify DeLone
IT, Maglitta [14] commented that many com- and McLean's model to incorporate the fol-
panies have shifted their focus from emphasis lowing variables: intensity of IT investment,
on deriving competitive advantage from IT to and four separate dimensions of IT impact: in-
emphasis on using IT to improve internal formation quality, managerial satisfaction,
quality and eciency. Such shift is not surpris- improvements in work environment and or-
ing since research attempting to link IT invest- ganizational impact. In other words, we recog-
ment with ®nancial performance measures nize the multifaceted nature of IT impacts by
have generally been mixed. One possible examining them at various levels. In addition,
reason is that ®nancial performance measures we will also examine the relationships among
may ignore important elements of perform- the four non-®nancial measures to better
ance that are critical to customers e.g., service understand their relationships with each other.
quality, speed and convenience. It is therefore
not surprising that research attempting to link 2. RESEARCH MODEL AND HYPOTHESES
IT investments to non-®nancial performance
have generally been more successful [23] as Figure 1 presents the research model as well
few can argue against how IT has transformed as the hypothesized relationships among the
organizations by improving eciency and research variables in this study.
e€ectiveness. Note that the model basically postulates
Mahmood and Mann [15] reviewed research possible relationships between the various con-
on IT impacts and organizational performance structs and is not intended demonstrate causal
for the past ten years and found that a wide e€ects. The reason is that the cross-sectional
variety of measures have been used to assess nature of our study makes such causal infer-
IT impacts. Also, past research often di€er in ences dicult. We ®rst postulate relationships
their theoretical background which can range between intensity of IT investment and the
from exploratory studies to research based on four performance variables.
innovation theory or microeconomics.
Ang and Pavri [1] reviewed and classi®ed IT 2.1. Intensity of IT investment and performance
impacts into three main levels: societal, organ- impact measures
izational and individual levels. At the societal Greater intensity of IT investment will
level, IT impacts on infrastructure, employ- enable more ®nancial resources to be allocated
ment, national competitiveness, etc. while at to computerization. These ®nancial resources
the organizational level, IT leads to new or- can take the form of new hardware, software,
ganizational forms as well as facilitates com- manpower and consultancy costs.
petitive advantage. At the individual level, IT Consequently, with more organizational
impacts on user satisfaction, job satisfaction, resources, the computerization e€ort would
work attitudes and user expectations, etc. Ang
and Pavri concluded by emphasizing that a
plurality of perspectives in IT impacts research
is needed since impacts are often multi-faceted
and complex.
Delone and McLean [6] evaluated past
research on IS success and found that a wide
variety of measures are commonly used as sur-
rogates for IS success. They proposed an IS
success model that postulated that system
quality and information quality singularly and
jointly in¯uences both use and user satisfac-
tion. In addition, both use and user satisfac-
tion can in¯uence each other as well as act as Fig. 1. Research model and hypotheses
 Omega, Vol. 26, No. 5 613

have greater chances of success [25] in terms and decision-making, information quality can
of improved information quality, work en- also be positively related to the level of man-
vironment, organizational impact, and man- agerial satisfaction with computerization. It
agerial satisfaction. It follows that: follows that:
H1: The intensity of IT investment is posi- H7: Information quality is positively related
tively related to the information quality. to managerial satisfaction.
H2: The intensity of IT investment is posi-
tively related to improvement in work environ- 2.4. Work environment, organization impact
ment. and managerial satisfaction
H3: The intensity of IT investment is posi- Improvements in work environment can
tively related to organizational impact. motivate employees to work harder and smar-
H4: The intensity of IT investment is posi- ter, thereby resulting in increased eciency
tively related to managerial satisfaction. and e€ectiveness. As well, sta€ morale can be
increased since computerization can automate
2.2. Information quality and work environment routine, repetitive and mundane work pro-
In a delphi study of IS executives, Saunders cesses, thereby making work easier and leaving
and Jones [28] found that information quality more time for other challenging and value-
is often used as a criteria to assess IS function added tasks. This can in turn lead to general
performance. One reason is that many compa- improvement in competitiveness and pro-
nies embarked on computerization programs ductivity. If we consider that improvements in
in the quest for better information for decision work environment is a surrogate measure for
making. Information quality is improved individual impact in DeLone and McLean's
because data can be easily updated, manipu- [6] model, it follows that improvement in work
lated and processed in a timely manner to pro- environment can be positively related to or-
vide relevant information for decision-making. ganizational impact. Hence:
Better information and improved decision- H8: Improvement in work environment is
making can lead to a general improvement in positively related to organizational impact.
the work environment in terms of increased IT can help to improve the work environ-
sta€ morale and making work more interest- ment be expanding job content due to the
ing. Consequently, information quality can be wide range of applications available. In ad-
positively related to improvement in work en- dition, work can be made more interesting
vironment. It follows that: since routine and mundane tasks can be auto-
H5: Information quality is positively related mated. This can lead to an overall increase in
to improvement in work environment. managerial satisfaction with computerization.
It follows that:
2.3. Information quality, organizational impact H9: Improvements in work environment is
and managerial satisfaction positively related to managerial satisfaction.
Information quality in terms of whether it is
valuable, accurate, timely and relevant can 2.5. Organizational impact and managerial sat-
determine its usefulness. If the information de- isfaction
rived from IT is ¯awed, inaccurate, inaccess- User satisfaction has often been used as a
ible in a timely manner or irrelevant, it can measure of IS success or e€ectiveness. We can
have a detrimental e€ect on management con- extend this concept of user satisfaction to
trol, decision making as well as cost savings or encompass the notion of managerial satisfac-
productivity improvements [20]. In other tion with the computerization e€ort. Since, in
words, if the information quality is improved, this study, satisfaction is measured from the
it is more likely that the desired organizational managers' perspective rather than the general
impact will be obtained. It follows that: end users' perspective, we can treat satisfac-
H6: Information quality is positively related tion as an overall measure of IS success rather
to organizational impact. than as an intermediate success measure as
In addition, since one of the major objec- proposed by DeLone and McLean. In other
tives of computerization e€orts is to provide words, if computerization results in perceived
better information for management control organizational impact such as productivity
614 Teo, WongÐPerformance Impact of Computerization in the Retail Industry

improvements, it is likely to increase the level scales that operationalized constructs were
of managerial satisfaction with the computeri- adapted from past research where psycho-
zation e€ort. It follows that: metric properties have been established.
H10: Organizational impact is positively re- 3.2.1. Firm size. This construct is measured
lated to managerial satisfaction. by asking respondents to state their number of
employees [25].
3.2.2. Intensity of IT investment. This con-
3. METHODS struct was measured by dividing total expendi-
ture on computerization by business turnover
3.1. Sample and procedures [15, 30]. This ratio provides an indication of
This study utilized data collected from a the relative size of IT investment compared to
survey of IT usage in the retail industry con- the size (in terms of business turnover) of the
ducted by the National Computer Board, a ®rm. This will enable us to determine the pro-
government agency responsible for the pro- portion of business turnover that are invested
motion and di€usion of IT in Singapore. The in computerization. As well, this ratio gives an
retail industry was chosen as the focus of the indication of how much an organization is
study because IT has often been used to spending on IT relative to other companies in
improve customer service, reduce costs as well the same industry. Note that total IT expendi-
compete more e€ectively [19]. ture and business turnover are with reference
Since retail companies in Singapore may be to the previous ®nancial year since ®gures for
owned by individuals who are predominantly the current year will only be available the fol-
pro®cient in Mandarin, a bilingual question-
lowing year.
naire incorporating both English and
3.2.3. Information quality. This construct
Mandarin languages was used in the study.
was measured using 5-point Likert-type scales
The translation to Mandarin from English was
ranging from (1) strongly agree to (5) strongly
made by a professional translator. As a val-
disagree, that assessed respondents' agreement
idity check, the translated Mandarin version
or disagreement with regards to whether the
was retranslated back to English in order to
information obtained after computerization is
ensure consistency in interpretation of the
valuable, accurate, timely and relevant [10, 17].
items in the questionnaire. After some modi®-
3.2.4. Improvement in work environment.
cations, the ®nal questionnaire was sent to
This construct was measured using 5-point
companies in the retail industry in Singapore.
A sample of 2641 companies was selected Likert-type scales ranging from (1) strongly
from an estimated population of 9000 compa- agree to (5) strongly disagree, that assessed the
nies in the retail industry in Singapore. work-related impacts of computerization in
Strati®ed sampling was used in order to ensure terms of whether sta€ morale has increased,
that all the various sectors (e.g., textiles, whether job content and activities have
books, supermarkets, etc.) of the retail indus- expanded, and whether work has become
try were covered. The questionnaires were sent more interesting [11, 18, 20].
by mail directed at the respective chief infor- 3.2.5. Organizational impact. This construct
mation ocers or senior managers of the com- was measured using 5-point Likert-type scales
panies. Telephone interviews were also ranging from (1) strongly agree to (5) strongly
conducted to improve the response rate. disagree, that assessed the organizational
impact of computerization in terms of whether
3.2. Instrument the ®rm has become more competitive,
Items measuring the variables in this study whether management of the ®rm has
were derived from an extensive review of past improved, and whether productivity has
research on the organizational impact of IT. increased [23, 28].
We de®ned information technology as compu- 3.2.6. Satisfaction. This construct was
ter hardware and software that perform data measured using 5-point Likert-type scales ran-
processing tasks such as capturing, transmit- ging from (1) very satis®ed to (5) very dissatis-
ting, storing, retrieving, manipulating or dis- ®ed, that assessed satisfaction with
playing data. Wherever possible, multiple item computerization in terms of hardware, soft-
 Omega, Vol. 26, No. 5 615

ware, consultant and software vendor support ance measures (information quality, improve-
[24, 27]. ment in work environment, organizational
impact and satisfaction) as these ®rms have
not yet computerized.
4. RESULTS

4.1. Response rate 4.2. Validity and reliability

A total of 1455 companies responded to our For analysis purposes, in order to be con-
survey, thereby giving a response rate of 55 sistent with the direction of our hypotheses,
percent. Of the responding ®rms, 917 reported reverse coding was carried out for information
no usage of IT while 538 reported usage of IT. quality, improvement in work environment,
A comparison between IT and non-IT user organizational impact and managerial satisfac-
groups revealed that the IT-user group is sig- tion. Factor analysis with varimax rotation
ni®cantly bigger in terms of business turnover was used to determine whether the items that
(t = 4.30, p < 0.001) and number of employ- measure each construct actually loaded
ees (t = 7.09, p < 0.001). This is perhaps together. The results showed four distinct con-
expected since larger companies have greater structs (see Table 2). Firm size and the inten-
resources to invest in IT. In addition, IT usage sity of IT investment were not included in the
is often a necessity in larger ®rms in order to factor analysis since they are single item con-
more e€ectively manage the growing complex- structs.
ity of operations. The detailed breakdown by Cronbach alpha was used to assess the re-
di€erent sectors in terms of the two groups liability of the multiple items for each con-
(IT user and non-IT user) is shown in Table 1. struct. A low value of Cronbach alpha would
The largest group of respondents (IT indicate that the items measuring the construct
user + non-IT user) are from the textile and do not belong together. The values of
wearing apparel sector, followed by supermar- Cronbach alpha range from 0.79 to 0.89,
kets, departmental stores and provision shops, thereby indicating relatively high levels of in-
and personal e€ects. This is expected as these ternal consistency among the multi-item scales
three sectors form the bulk of the retail indus- [21]. For statistical analysis, the mean of the
try in Singapore [29]. However, if we compare scale items measuring each construct was used
IT usage in various sectors, we note that as is commonly done in past research.
supermarkets, department stores and provision Since the intensity of IT investment was
shops sector has a relatively low proportion of measured by dividing total IT expenditure by
IT usage. One plausible reason is that many business turnover, conventional computations
shops in this sector are small family-owned of validity and reliability are not possible. As
businesses which had not envision the need to a con®rmation check on the total IT expendi-
use IT. ture, we asked respondents to give a break-
Note that non-IT users were excluded from down of the proportion of total IT expenses
further analysis of our results since we did not spent for hardware, software, manpower, etc.
obtain data pertaining to the various perform- as well as indicate the projected IT expendi-

Table 1. Response rate

Sector IT user (1) Non-IT user (2) Percent (1)/[(1) + (2)]
Textiles and wearing apparel 129 258 33.3
Personal e€ects 105 149 41.3
Athletics, sports and recreational goods 58 42 58.0
Souvenirs, gifts and fancy goods 41 59 41.0
Pharmaceutical and medicinal products 44 76 36.7
Books, magazines and paper products 89 74 54.6
Optical and photographic goods 23 41 35.9
Supermarkets, dept stores and provision shops 49 218 18.4
Total 538 917 37.0
Note: the various retail sectors include both wholesale and retail companies.
616 Teo, WongÐPerformance Impact of Computerization in the Retail Industry

Table 2. Factor analysis and reliability assessment An examination of the correlation matrix
Factors Loadings showed that there are no signi®cant relation-
ships between the intensity of IT investment
F1: Information quality (eigenvalue = 5.56,
variance = 39.7%, alpha = 0.89) and information quality, improvement in work
Information we get now is more environment, and organizational impact.
valuable 0.82 Surprisingly, the relationship between IT
accurate 0.84
timely 0.82 investment and ®rm size is signi®cant in the
relevant 0.83 opposite direction. It appears that as a pro-
F2: Managerial Satisfaction (eigenvalue = 2.47, portion of business turnover, smaller ®rms
variance = 17.7%, alpha = 0.87) actually invest more in IT compared to larger
Satisfaction with overall
hardware 0.68 ®rms. One likely reason is that the turnover
software 0.82 for smaller ®rms may be quite low whereas
consultant 0.89 investment in IT usually require a minimum
software vendor support 0.89
level of commitment which may constitute a
F3: Organizational impact (eigenvalue = 1.45,
variance = 10.4, alpha = 0.81)
higher proportion of business turnover for
We have become more competitive 0.76 smaller ®rms.
Management of the company has improved 0.84 The correlation between investment in IT
Our productivity has increased 0.82
and managerial satisfaction is also negative,
F4: Improvement in work environment thereby implying that more IT investment
(eigenvalue = 1.02, variance = 7.3, alpha = 0.79)
Sta€ morale has increased 0.74 actually lead to lower managerial satisfaction.
Job content and activities have expanded 0.75 One explanation is that payo€s from IT
Work has become more interesting 0.87 may take some time to manifest and are
often elusive. The intercorrelations between
information quality, managerial satisfaction,
tures for the next three years. Most ®rms indi- improvement in work environment and
cated that total IT expenditure will remain organizational impact are all positive and
stable or increase with time rather than signi®cant. This is perhaps not surprising
decrease with time. as these variables are often used to measure
IS success and consequently, may be expected
4.3. Intercorrelations to be correlated with each other. However,
all the correlations are below 0.5, thereby
Table 3 shows the means, standard devi-
indicating that multicollinearity in hierarchical
ations and intercorrelations of the scales used
regression analysis may not be a serious
to assess the variables in this study. Note that
problem.
the logarithms of ®rm size and intensity of IT
investment are used in the analysis as is often
done in past research (e.g., econometric studies 4.4. Hierarchical regression analysis
on organizational performance). This will Hierarchical regression analysis was used to
enable us to improve on the linearity of the re- test the various hypotheses since it allows us
lationships between independent and depen- to control for variables that might confound
dent variables. the results. For hypotheses 1 to 4, the size of

Table 3. Intercorrelations
Variables Mean (S.D.) 1 2 3 4 5

1. Log (®rm size) 0.84 (0.55) 1.00

2. Log (intensity of IT investment) ÿ2.21 (0.82) ÿ0.16** 1.00
3. Information quality 3.85 (0.52) 0.10 0.04 1.00
4. Improvement in work environment 3.42 (0.54) 0.01 0.04 0.38** 1.00
5. Organizational impact 3.54 (0.54) 0.06 0.10 0.49** 0.45** 1.00
6. Managerial satisfaction 3.68 (0.62) 0.02 ÿ0.17** 0.33** 0.16** 0.21**
* p < 0.05.
** p < 0.01.
Note: items 3 to 6 are reverse coded, i.e. 1 (strongly disagree) and 5 (strongly agree).
 Omega, Vol. 26, No. 5 617

the company was controlled because it is formance measures as indicated by the low
plausible to expect that larger ®rms have values of beta coecients and low adjusted
greater human, technological and ®nancial R2. This implies that larger ®rms are not
resources that may allow them to be more suc- necessarily more successful than smaller ®rms
cessful in their use of information systems [25]. in deriving bene®ts from IT. One possible
The results are shown in Table 4 (A1 or reason is that the retail sector in Singapore
Analysis 1). has not been performing well over the past
It is interesting to note that the size of the few years due to increased local and global
®rm has negligible relationships with these per- competition.

Table 4. Hierarchical regression analysis

Beta coecients (A1) Beta coecients (A2)
Variables step 1 step 2 step 1 step 2
H1: DV=INFOQLTY
SIZE 0.01 0.01 0.02 0.03
INTENSITY 0.04 0.08
Adjusted R2 ÿ0.00 ÿ0.01 ÿ0.00 0.00
H2: DV=WORKENV
SIZE ÿ0.03 ÿ0.02 ÿ0.01 ÿ0.00
INTENSITY 0.03 0.06
Adjusted R2 ÿ0.00 ÿ0.00 ÿ0.00 ÿ0.00
H3: DV=ORGIMPACT
SIZE ÿ0.01 0.00 ÿ0.02 ÿ0.01
INTENSITY 0.11 0.13*
Adjusted R2 0.00 0.00 ÿ0.00 0.01
H4: DV=SATISFN
SIZE ÿ0.06 ÿ0.10 0.09 0.08
INTENSITY ÿ0.20** ÿ0.06
Adjusted R2 0.00 0.03 0.00 0.01
H5: DV=WORKENV
SIZE ÿ0.02 ÿ0.03 ÿ0.00 ÿ0.01
INTENSITY 0.03 0.02 0.06 0.03
INFOQLTY 0.39** 0.39**
Adjusted R2 ÿ0.00 0.15 ÿ0.00 0.14
H6, H8: DV=ORGIMPACT
SIZE 0.00 0.00 ÿ0.01 ÿ0.02
INTENSITY 0.11 0.08 0.13* 0.09
INFOQLTY 0.31** 0.31**
WORKENV 0.32** 0.31**
Adjusted R2 0.00 0.28 0.01 0.28
H7, H9: DV=SATISFN
SIZE ÿ0.10 ÿ0.08 0.08 0.09
INTENSITY ÿ0.20** ÿ0.23** ÿ0.06 ÿ0.12*
INFOQLTY 0.32** 0.32**
WORKENV 0.07 0.08
Adjusted R2 0.03 0.16 0.01 0.13
H10: DV=SATISFN
SIZE ÿ0.08 ÿ0.08 0.09 0.08
INTENSITY ÿ0.23** ÿ0.24** ÿ0.12* ÿ0.14*
INFOQLTY 0.32** 0.29** 0.32** 0.29**
WORKENV 0.07 0.02 0.08 0.02
ORGIMPACT 0.15* 0.14*
Adjusted R2 0.16 0.18 0.13 0.15
* p < 0.05.
** p < 0.01.
DV=dependent variables, SIZE=size of ®rm, INTENSITY=intensity of IT investment, INFOQLTY=information
quality, WORKENV=improvement in work environment, ORGIMPACT=organizational impact, SATISFN=
managerial satisfaction.
Note: A1 (Analysis 1): size=number of employees; and intensity=total IT costs/business turnover, A2 (Analysis 2):
size=business turnover; and intensity=total IT costs/number of employees.
618 Teo, WongÐPerformance Impact of Computerization in the Retail Industry

Hypotheses 1, 2 and 3 were not supported support for hypotheses 7 but not hypothesis 9.
thereby indicating that there are no signi®cant One possible reason for the lack of support
relationships between the intensity of IT for hypothesis 9 and support for hypothesis 8
investment and the following variables: infor- is that improvement in work environment was
mation quality, improvement in work environ- evaluated based on IT impact on the general
ment and organizational impact. From the end user work environment whereas satisfac-
low values of the beta coecients and the neg- tion was evaluated based on managerial satis-
ligible values of adjusted R2, it is apparent faction with the computerization e€ort. It is
that higher intensity of IT investment does not plausible that satisfaction of managers may be
necessarily translate into better information linked more to information quality rather than
quality, improvement in work environment improvement in work environment as the lat-
nor organizational impact. One plausible ex- ter a€ects end users to a greater extent than
planation is that there are many other factors the former.
(e.g., management support and leadership, en- For hypothesis 10, we ®rst controlled for
vironmental changes) that can a€ect these per- size, intensity of IT investment, information
formance impact measures. quality and improvement in work environment
Hypothesis 4 was also not supported. before entering organizational impact in step 2
Interestingly, the result is signi®cant in the of the hierarchical regression procedure. The
opposite direction in that greater IT intensity results showed that organizational impact
is signi®cantly related to lower levels of man- explained an additional 2 percent of the vari-
agerial satisfaction. One possible explanation ations in satisfaction, thereby lending support
is the productivity paradox of IT in which the to hypothesis 10.
impacts of IT investment are often elusive or On the advice of a reviewer, we carried out
controversial [5]. Firms with greater IT invest- another set of hierarchical regression analysis
ment are likely to have greater expectations of (A2 or Analysis 2 in Table 3) by de®ning ®rm
its bene®ts. Consequently, when the impacts size in terms of the number of business turn-
of IT are elusive despite high levels of IT over (rather than number of employees in A1
investment, it is likely to lead to lower man- or Analysis 1) and the intensity of IT invest-
agerial satisfaction with the overall computeri- ment in terms of IT costs/number of employ-
zation e€ort. Another explanation is that ®rms ees (rather than total IT costs/business
who invested heavily in IT are often impatient turnover in A1 or Analysis 1). The results are
to reap bene®ts from IT. When these bene®ts similar to Analysis 1 with the exception of hy-
take a longer time to manifest than expected, potheses H3 (which is now supported) and H4
lower satisfaction may result. (which is now not supported in the opposite
For hypothesis 5, after controlling for direction). This implies that the intensity of IT
size and intensity of IT investment, infor- investment (de®ned in terms of IT costs/num-
mation quality has a signi®cant positive ber of employees) has a signi®cant e€ect on
relationship with improvement in work en- organizational impact and an insigni®cant
vironment. Hence, hypothesis 5 was sup- e€ect on managerial satisfaction. One plaus-
ported. For hypotheses 6 to 9, both ®rm size ible reason as suggested by a reviewer is that
and intensity of IT investments are controlled the impact of IT investment is likely to be
by entering it in step 1 of the hierarchical re- more salient on costs rather than revenue.
gression procedure. Information quality and Consequently, IT may help to reduce costs by
improvement in work environment are then maintaining or reducing the size of workforce
entered in step 2. as business grows, thereby resulting in
Hypotheses 6 and 8 were supported in that improved productivity and competitiveness.
information quality and improvement in work However, any reduction in the number of
environment have signi®cant positive relation- employees may decrease the morale of the
ships with organizational impact. However, remaining employees which increases the bur-
the results also showed that information qual- den of managers. In other words, increased
ity, but not improvement in work environ- productivity through IT may not necessarily
ment, has a signi®cant positive relationship produce higher levels of satisfaction as it may
with managerial satisfaction. Hence, there was make it more dicult to manage employees
 Omega, Vol. 26, No. 5 619

who are a€ected by changes brought about by negative returns on investment are likely. A
IT. third possible reason is the time lag in the
impact of IT may be longer than expected.
4.5. Limitations However, on de®ning the intensity of IT
There are three main limitations of this investment in terms of costs (the number of
study. First, the data were collected from a employees) rather than revenue (business turn-
single respondent per company which might over) as shown in Analysis 2 of Table 4, IT
lead to common source bias. We have tried to intensity and managerial satisfaction are no
minimize this limitation by conducting tele- longer negatively associated. However, the re-
phone interviews with respondents to clarify lationship between IT intensity and organiz-
their responses or inquire about uncompleted ational impact is now signi®cant. Taken
portions of the questionnaire. Future research together these results indicate that although IT
can use more than one respondent per com- investment may lead to decrease costs and
pany to reduce common source bias. increase productivity, satisfaction may not
Second, our research model did not come necessarily be increased.
from a single theory of IT and performance. The results also showed that IS success
Instead, we tried to adapt and modify DeLone measures tend to be positively related to each
and McLean's [6] model, not only to examine other. Interestingly, information quality
the relationship between IT investment and appeared to have signi®cant relationships with
performance measures, but also to examine managerial satisfaction, improvement in work
the relationships between performance environments and organizational impact. This
measures themselves. Future research can use result is consistent with Northrop et al.'s [20]
theories from reference disciplines to further ®ndings that computerization generally leads
examine and explain the relationship between to better availability of information for man-
IT and performance. agement control. However, improvements in
Third, the study is cross-sectional in nature work environment appears to be related only
which makes it dicult to draw de®nite con- to organizational impact but not to manage-
clusions about cause and e€ect. A longitudinal rial satisfaction.
methodology to tease out the causal direction The positive relationship between improve-
of the relationships found in this study may be ments in work environment and organizational
conducted in future studies. In particular, impact is instructive in that it demonstrates a
since a number of measures used in this study distinct link between work environment and
is perceptual in nature, future research can use overall organizational impact. In other words,
more objective measures perhaps through research attempting to link IT investment to
detailed case studies of several organizations. organizational impacts should adopt a mul-
tiple impact measures approach, ranging from
5. DISCUSSION AND CONCLUSION impact on information quality to work en-
vironment to organizational impact. In ad-
The study suggests that it is very dicult to dition, the signi®cant positive relationship
®nd evidence linking the intensity of IT invest- between organizational impact and satisfaction
ments to IS success measures such as infor- revealed that managerial satisfaction with
mation quality, work environment, and computerization are not only determined by
organizational impact. Contrary to expec- improved information quality but also by
tation, the results showed that the intensity of overall organizational performance impact.
IT investment is negatively related to satisfac- In summary, this study contributes to exist-
tion. One reason is that with greater IT spend- ing literature in four ways. First, this study
ing, the expectations of the tangible impacts of adapts and modi®es DeLone and McLean's
IT may be greater. Consequently, when the model of IS success to incorporate a multiple
contributions of IT on the ®rm's bottom-line impact approach with four main performance
cannot be clearly distinguished, it may actually measures. Basically, we ®rst postulated that IT
lead to lower satisfaction. Another reason is investment may have direct relationships with
that there may be an optimal level of IT performance measures rather than indirectly
investment [4], beyond which negligible or through intermediate variables. The results in-
620 Teo, WongÐPerformance Impact of Computerization in the Retail Industry

dicate that general linkage between IT invest- REFERENCES

ment intensity and performance measures are 1. Ang, J. and Pavri, F., A survey and critique of the
dicult to establish without considering other impacts of information technology. International
extraneous variables that could in¯uence per- Journal of Information Management, 1994, 14, 122±
133.
formance. Hence, this study helps to reinforce
2. Attewell, P. and Rule, J., Computing and organiz-
previous ®ndings that the relationship between ations: what we know and what we don't know.
IT investment intensity and performance Communications of the ACM, 27(12), 1184±1192.
measures are likely to be complex. 3. Barua, A., Kriebel, C. H. and Mukhopadhyay, T.,
Information technologies and business value: an ana-
Second, by controlling for ®rm size in the lytic and empirical investigation. Information Systems
hierarchical regression analyses, we showed Research, 1995, 6(1), 3±23.
that size has negligible relationship with each 4. Bender, D. H., Financial impact of information pro-
cessing. Journal of Management Information Systems,
of the performance measures. This means that 1986, 3(2), 22±32.
regardless of ®rm size, it is still dicult to ®nd 5. Brynjolfsson, E., The productivity paradox of infor-
a link between IT investment intensity and mation technology. Communications of the ACM,
1993, 36(12), 67±77.
performance measures. Moreover, results 6. DeLone, W. H. and McLean, E. R., Information sys-
showed that ®rm size is actually negatively tems success: the quest for the dependent variable.
correlated with IT investment intensity as il- Information Systems Research, 1992, 3(1), 60±95.
lustrated in Table 2. Future research is needed 7. Due, R. T., The productivity paradox. Information
Systems Management, 1993, 10(1), 68±71.
to examine in greater detail the role of ®rm 8. Gurbaxani, V. and Whang, S., The impacts of infor-
size in IT investment and performance. mation systems on organizations and markets.
Third, this study also examines the relation- Communications of the ACM, 1991, 34(1), 59±73.
9. Harris, S. E. and Katz, J. L., Organizational perform-
ships between various performance measures. ance and information technology investment intensity
While it may be expected that these variables in the insurance industry. Organization Science, 1991,
will be correlated with each other, what is 2(3), 263±295.
10. King, W. R. and Epstein, B. J., Assessing information
interesting is whether these relationships system value: an experimental study. Decision
remained signi®cant when other control vari- Sciences, 1983, 14, 34±45.
ables such as size and IT investment intensity 11. Kraut, R., Dumais, S. and Koch, S., Computerization,
are introduced. productivity, and quality of work-life. Communications
of the ACM, 1989, 32(2), 220±238.
Fourth, instead of using user satisfaction as 12. Lambert, R. and Peppard, J., Information technology
an intermediate variable between information and new organizational forms: destination but no
quality and organizational impact as proposed road map? Journal of Strategic Information Systems,
1993, 2(3), 180±205.
by DeLone and McLean, we have used man- 13. Lucas, H. C. and Olson, M., The impact of infor-
agers' satisfaction as an overall indication of mation technology on organizational ¯exibility.
success of the computerization e€orts. Journal of Organizational Computing, 1994, 4(2), 155±
176.
Although user satisfaction may also be an im- 14. Maglitta, J., It's reality time. ComputerWorld, 25(17),
portant measure of IS success, we feel that April 29, 81±83.
measuring satisfaction from the managers' per- 15. Mahmood, M. A. and Mann, G. J., Measuring the or-
ganizational impact of information technology invest-
spective is important in that managers rather
ment: an exploratory study. Journal of Management
than end users will ultimately decide on Information Systems, 1993, 10(1), 97±122.
whether further investment in IT is necessary 16. Malone, T. W., Yates, J. and Benjamin, R. I.,
and bene®cial. Hence, managerial satisfaction Electronic markets and electronic hierarchies.
Communications of the ACM, 1987, 30(6), 484±497.
with computerization may be used as an im- 17. Miller, J. and Doyle, B. A., Measuring e€ectiveness of
portant IS success measure in future research. computer-based information systems in the ®nancial
It is also of interest to examine whether the services sector. MIS Quarterly, 1987, 12(2), 223±236.
18. Millman, Z. and Hartwick, J., The impact of auto-
research ®ndings can be extended to other mated oce systems on middle managers and their
business sectors. work. MIS Quarterly, 1987, 11(4), 479±491.
19. Moriarty, R. T. and Swartz, G. S., Automation to
boost sales and marketing. Harvard Business Review,
ACKNOWLEDGEMENTS 1989, 67(1), 100±108.
20. Northrop, A., Kraemer, K. L., Dunkle, D. and King,
The authors would like to thank the National Computer J. L., Payo€s from computerization: lessons over time.
Board, Singapore, for permission to use their data on IT Public Administration Review September/October 1990,
usage survey in the retail industry. In addition, the authors 505±514.
would also like to thank the editor and three anonymous 21. Nunnally, J. C., Psychometric Theory, 2nd edition.
reviewers for their insightful comments and suggestions. McGraw-Hill, New York, 1978.
 Omega, Vol. 26, No. 5 621

22. Orlikowski, W. J., Integrated information environment 28. Saunders, C. S. and Jones, J. W., Measuring
or matrix of control? The contradictory implications performance of the information systems function.
of information technology Accounting, Management Journal of Management Information Systems, 1992,
and Information Technology, 1991, 1(1), 9±42. 8(4), 63±82.
23. Quinn, J. B. and Baily, M. N., Information technol- 29. Singapore Department of Statistics, Report on the sur-
ogy: increasing productivity in services. Academy of vey of wholesale and retail trade 1993. Singapore
Management Executive, 1994, 8(3), 28±51. Government Printers, 1995.
24. Raymond, L., Validating and applying user satisfac- 30. Weill, P., The relationship between investment in in-
tion as a measure of MIS success in small organiz- formation technology and ®rm performance: a study
ations. Information and Management, 1987, 12(4), of the valve manufacturing sector. Information
173±179. Systems Research, 1992, 3(4), 307±333.
25. Raymond, L., Organizational context and information
systems success. Journal of Management Information
Systems, 1990, 6(4), 5±20.
26. Roach, S. S., Technology and the service sector: the ADDRESS FOR CORRESPONDENCE: Dr Thompson S. H. Teo,
hidden competitive challenge. Technological Department of Decision Sciences, Faculty of
Forecasting and Social Change, 1988, 34(4), 387±403. Business Administration, National University of
27. Rushinek, A. and Rushinek, S. F., What makes users Singapore, 10 Kent Ridge Crescent, Singapore
happy. Communications of the ACM, 1986, 29(7), 594± 119260. Tel.: +65-772-3036; fax: +65-779-2621;
598. e-mail: [email protected].