Int. Journal of Business Science and Applied Management, Volume 5, Issue 1, 2010

Driving online shopping: Spending and behavioral

differences among women in Saudi Arabia

Talal Al-maghrabi

Brunel Business School, Brunel University

West London, UB8 3PH, United Kingdom

Tel: +44 (0) 1895 267171

Email: talal.almaghrabi@brunel.ac.uk

Charles Dennis

Brunel Business School, Brunel University

West London, UB8 3PH, United Kingdom

Tel: +44 (0) 1895 265242

Email: charles.dennis@brunel.ac.uk

Abstract

This study proposes a revised technology acceptance model that integrates expectation confirmation

theory to measure gender differences with regard to continuance online shopping intentions in Saudi

Arabia. The sample consists of 650 female respondents. A structural equation model confirms model

fit. Perceived enjoyment, usefulness, and subjective norms are determinants of online shopping

continuance in Saudi Arabia. High and low online spenders among women in Saudi Arabia are

equivalent. The structural weights are also largely equivalent, but the regression paths from perceived

site quality to perceived usefulness is not invariant between high and low e-shoppers in Saudi Arabia.

This research moves beyond online shopping intentions and includes factors affecting online shopping

continuance. The research model explains 60% of the female respondents’ intention to continue

shopping online. Online strategies cannot ignore either the direct and indirect spending differences on

continuance intentions, and the model can be generalized across Saudi Arabia.

Keywords: internet shopping, e-shopping, technology acceptance, male and female examination,

continuance online shopping, Saudi Arabia

Acknowledgements: The authors thank the respondents, the editors, and the anonymous reviewers for

their many helpful suggestions. Special thanks for their families for their continued support.

Talal Al-maghrabi and Charles Dennis

1 STUDY MOTIVATION

Globalization continues to drive the rapid growth of international trade, global corporations, and

non-local consumption alternatives (Alden et al. 2006; Holt et al. 2004), and advances of the Internet

and e-commerce have diminished trade boundaries. E-commerce and e-shopping create opportunities

for businesses to reach to consumers globally and directly, and in turn, business and social science

research now focuses specifically on cross-national and cross-cultural Internet marketing (Griffith et al.

2006).

The Internet had changed how businesses and customers customize, distribute, and consume

products. Its low cost gives both businesses and consumers a new and powerful channel for information

and communication. In 1991, the Internet had less than 3 million users worldwide and no e-commerce

applications; by 1999, about 250 million users appeared online, and 63 million of them engaged in

online transactions, which produced a total value of $110 billion (Coppel 2000). Business-to-consumer

online sales in the United States grew by 120% between 1998 and 1999 (Shop.org and Boston

Consulting Group, 2000). According to a U.K. payment association, the number of consumers who

shop online has increased by more than 157%, from 11 million in 2001 to more than 28 million in 2006

(cited in Alsajjan and Dennis, 2009). E-commerce transactions also are growing in the Middle East

(19.5 million Internet users) and in the Gulf States. In Saudi Arabia, online transactions have increased

by 100%, from $278 million in 2002 to $556 million in 2005 (Al Riyadh 2006). In 2007, Internet sales

increased to more than $1.2 billion worldwide and are expected to continue to rise (World Internet

Users and Population Stats 2007).

An unpublished study by the Centre for Customer Driven Quality also highlights some potential

savings: For one retailer, the cost of an in-store customer contact was estimated to be $10, the cost of a

phone contact $5, and the cost of a Web contact $0.01 (Feinberg, et al. 2002). In the airline industry,

the savings are similar. According to the International Air Transport Association, airlines currently

issue approximately 300 million paper tickets per year at a cost of $10 per ticket to process (Arab News

Newspaper, 2007). One e-ticket process costs only $1(Arab News Newspaper, 2007).

Despite the impressive online purchasing growth rates though, compelling evidence indicates that

many consumers who search different online retail sites abandon their purchases. This trend and the

proliferation of business-to-consumer e-shopping activities require that online businesses understand

which factors encourage consumers to complete their purchases. Acquiring new customers also can

cost as much as five times more than retaining existing ones (Bhattacherjee 2001b; Crego and Schiffrin

1995; Petrissans 1999). For example, a 5% increase in customer retention in the insurance industry

typically translates into an 18% reduction in operating costs (Bhattacherjee, 2001a; Crego et al, 1995).

Online customer retention is particularly difficult. Modern customers demand that their needs be

met immediately, perfectly, and for free, and they are empowered with more information to make

decisions (Bhattacherjee 2001b; Crego and Schiffrin 1995). They also have various online and offline

options from which to choose, and without a compelling reason to choose one retailer over another,

they experiment or rotate purchases among multiple firms (Bhattacherjee 2001b; Crego and Schiffrin

1995).

To employ the savings derived from e-businesses, companies might engage in tactics to increase

switching costs and thereby retain more customers. E-retailers might recall details about the customer

that reduce the customer effort demanded in future transactions; they could also learn more about the

customer to tailor those future interactions to the customer’s needs (Straub and Watson 2001). Better

product quality, lower prices, better services, and increased outcome value should help companies build

sustainable relationships with their customers.

Theoretical explanations of online shopping intentions suggest several important factors. For

example, Rogers (1995) suggests that consumers reevaluate their acceptance decisions during a final

confirmation stage and decide to continue or discontinue. Continuance may be an extension of

acceptance behavior that covaries with acceptance (e.g., Bhattercherjee 2001a; Davis et al. 1989;

Karahanna et al. 1999). We adopt the extended expectation confirmation theory (ECT; Bhattacherjee

2001b) and the technology acceptance model (TAM; Davis et al. 1989) as a theoretical basis,

integrating ECT from consumer behavior literature to propose a model of e-shopping continuance

intentions, similar to the way in which the TAM adapts the theory of reasoned action (TRA) from

social psychology to postulate a model of technology acceptance.

The TAM, as expanded by Davis and colleagues (1992) and Gefen (2003), and the ECT

(Bhattacherjee 2001a; Oliver 1980) have been used widely in research in the industrialized world, but

they are less commonly applied to developing countries. Moreover, the TAM stops at intention and

does not investigate continuance intentions or behavior.

Int. Journal of Business Science and Applied Management / Business-and-Management.org

As another issue in prior research, no widely acceptable definition for e-commerce exists. Coppel

(2000) calls it doing business over the Internet, including both business-to-business and business-to-

consumer markets. For the purpose of this research, we adopt the following definition: E-shopping,

electronic shopping, online shopping, and Internet shopping are the same. All these activities include

the activity of searching, buying, and selling products and services through the Internet. In recent years,

the Internet has grown to include a wider range of potential commercial activities and information

exchanges, such as the transaction and exchange of information between government agencies,

governments and businesses, businesses and consumers, and among consumers. We focus mainly on

the business-to-consumer (B2C) arena, which has been the source of most online progress and

development.

Previous research also finds that gender differences significantly affect new technology decision-

making processes (Van Slyke et al. 2002; Venkatesh et al. 2000). Venkatesh and colleagues (2000)

report that women tend to accept information technology when others have high opinions of it and are

more influenced by ease of use. Men rely more on their evaluations of the usefulness of the technology.

However, in many cultures, women represent the primary decision makers in families and households’

main shoppers. Greater e-commerce exposure and decision-making power may imply that women can

attain greater satisfaction from online shopping (Alreck and Settle 2002).

Finally, no previous research considers Internet shopping in Saudi Arabia or, specifically,

continuance intentions for online shopping in Saudi Arabia, nor do studies address gender-based

differences in shopping behavior online in Saudi Arabia. This research attempts to provide a validated

conceptual model that integrates different factors, including gender, and clarifies the theoretical

problems of continuance intentions in the unique context of Saudi Arabia.

The remainder of this article proceeds as follows: We offer a review of existing literature, and then

detail our proposed model, hypotheses, and methodology. After describing the structural equation

model and analysis, we provide our results. We conclude with some limitations and recommendations

for further research.

2 THEORETICAL BACKGROUND

The TAM (Davis 1989) represents an adaptation of the TRA, tailored to users’ acceptance of

information systems. It helps explain determinants of computer acceptance and can explicate user

behaviors across a broad range of computing technologies and populations; it also is parsimonious and

theoretically justified (Davis et al. 1989). The major determinants are perceived usefulness and ease of

use. Perceived usefulness significantly influences attitude formation (Agarwal and Prasad 1999; Davis

1989; Dishaw and Strong 1999; Gefen and Keil 1998; Igbaria et al. 1996; Moon and Kim 2001; Taylor

and Todd 1995; Venkatesh 2000; Venkatesh and Davis 2000), but evidence regarding perceived ease of

use remains inconsistent. Many studies simplify the original TAM by dropping attitude and studying

just the effect of perceived usefulness and ease of use on intention to use (Gefen and Straub 2000;

Leader et al. 2000; Teo et al. 1999).

Updates to the TAM add antecedents of perceived usefulness and ease of use (Venkatesh and

Davis 2000), such as subjective norms, experience, trust, and output quality. Ample evidence confirms

that both usefulness (i.e., external motivation) and intrinsic enjoyment (i.e., internal motivation) offer

direct determinants of user acceptance online (Davis et al. 1992; Leader et al. 2000; Moon and Kim

2001; Teo et al. 1999; Venkatesh 1999).

Expectation confirmation theory (ECT) in turn helps predict consumer behavior before, during,

and after a purchase in various contexts, in terms of both product and service repurchases (Anderson

and Sullivan 1993; Dabholkar et al., 2000; Oliver, 1980, 1993; Patterson et al. 1997; Spreng et al.

1996; Swan and Trawick 1981; Tse and Wilton 1988). According to ECT, consumers define their

repurchase intentions by determining whether the product or service meets their initial expectations.

Their comparison of perceived usefulness versus their original expectation of usefulness influences

their continuance intentions (Bhattacherjee 2001a; Oliver 1980). Their repurchase intentions depend on

their satisfaction with the product or service (Anderson and Sullivan 1993; Oliver 1980).

However, the ECT ignores potential changes in initial expectations following the consumption

experience and the effect of these expectation changes on subsequent cognitive processes

(Bhattacherjee 2001a). Pre-purchase expectations typically are based on others’ opinions or

information from mass media, whereas post-purchase expectations derive from first-hand experience,

which appears more realistic (Fazio and Zanna 1981). After such first-hand experience, expectations

may increase if consumers believe the product or service is useful or contains new benefits and features

that were not part their initial expectation.

Talal Al-maghrabi and Charles Dennis

Venkatesh and colleagues (2003) suggest that usage and intentions to continue usage may depend

on cognitive beliefs about perceived usefulness. Gefen (2003) also indicates that perceived usefulness

reinforces an online shopper’s intention to continue using a Web site, such that when a person accepts a

new information system, he or she is more willing to alter practices and expend time and effort to use it

(Succi and Walter 1999). However, consumers may continue using an e-commerce service if they

consider it useful, even if they are dissatisfied with its prior use (Bhattacherjee 2001a).

Site quality and good interface design enhance the formation of consumer trust (McKnight et al.

2002a), and if a consumer perceives a vendor’s Web site to be of high quality, he or she should trust

that vendor’s competence, integrity, and benevolence (McKnight et al. 2002a). Gefen and colleagues

(2003) integrate trust into the TAM in a B2C e-shopping context and find trust positively affects

consumers’ intention to use a Web site. Building trust with consumers is an essential mission for e-

retailers, because purchasing decisions represent trust-related behaviors (Jarvenpaa et al. 2000;

McKnight et al. 2002b; Urban et al. 2000).

A person’s beliefs about what important others think about the behavior also should directly

influence subjective norms. Therefore, if e-shopping is a socially desirable behavior, a person is more

likely to e-shop (George 2002).

Childers and colleagues (2001) also find that enjoyment can predict attitude towards e-shopping,

just as much as usefulness can. However, usefulness was the better predictor for grocery items, whereas

enjoyment offered better results for hedonic purchases. With regard to e-shopping, the hedonic

enjoyment constructs in the TAM may reflect the pleasure users obtain from shopping online, which

reinforces continuance intentions.

3 PROPOSED MODEL AND HYPOTHESES

3.1 Site Quality

Initial trust forms quickly on the basis of available information (Meyerson et al. 1996). If

consumers perceive a Web site as high quality, they trust it and will depend on that vendor (McKnight

et al. 2002a). Site information quality and a good interface design enhance consumer trust (Fung and

Lee, 1999). Web site quality helps predict behavior (Business Wire 1999; Carl 1995; Meltzer 1999).

Perceptions of Web site quality affect trust and perceptions of usefulness. In addition, e-shoppers

should perceive a Web site as more trustworthy if it appears more attractive because of its contents,

layout, and colors, which represent site quality. On the basis of previous research, we therefore predict:

H1a. Perceived site quality relates positively to perceived usefulness.

H1b. Perceived site quality relates positively to customer trust to use online shopping.

3.2 Trust

Trust refers to an expectation that others will not behave opportunistically (Gefen 2003). Trust

therefore implies a belief that the vendor will provide what has been promised (Ganesan 1994). In turn,

perceived usefulness should occur only for an e-vendor that can be trusted (Festinger 1975). Thus:

H2. Perceived trust relates positively to perceived usefulness.

3.3 Perceived Usefulness

According to Burke (1997), perceived usefulness is the primary prerequisite for mass market

technology acceptance, which depends on consumers’ expectations about how technology can improve

and simplify their lives (Peterson et al. 1997). A Web site is useful if it delivers services to a customer

but not if the customers’ delivery expectations are not met (Barnes and Vidgen 2000). The usefulness

and accuracy of the site also influence customer attitudes. Users may continue using an e-commerce

service if they consider it useful, even if they may be dissatisfied with their prior use (Bhattacherjee

2001a). Consumers likely evaluate and consider product-related information prior to purchase, and

perceived usefulness thus may be more important than the hedonic aspect of the shopping experience

(Babin et al. 1994). In a robust TAM, perceived usefulness predicts IT use and intention to use (e.g.,

Adams et al. 1992; Agarwal and Prasad, 1999; Gefen and Keil 1998; Gefen and Straub 1997;

Hendrickson et al. 1993; Igabria et al. 1995; Subramanian 1994), including e-commerce adoption

(Gefen and Straub 2000). Therefore:

H3a. Perceived usefulness relates positively to increasing customer subjective norms.

Int. Journal of Business Science and Applied Management / Business-and-Management.org

H3b. Perceived usefulness relates positively to increasing customer enjoyment.

H3c. Perceived usefulness relates positively to increasing customer continuance intentions.

3.4 Subjective Norms

According to Venkatesh and colleagues (2003), social influences result from subject norms, which

relate to consumers’ perceptions of the beliefs of other consumers. Shim and colleagues (2001)

consider subjective norms only marginally significant on e-shopping intentions, whereas Foucault and

Scheufele (2005) confirm a significant link between talking about e-shopping with friends and

intention to e-shop. Enjoyment also is relevant to social norms, because involving Web sites facilitate

e-friendship and enforce e-shopping as a subjective norm. Thus,

H4a. Perceived subjective norms relate positively to increasing customer enjoyment.

H4b. Perceived subjective norms relate positively to increasing customer continuance intentions.

3.5 Enjoyment

Enjoyment in using a Web site significantly affects intentions to use (Davis et al. 1992; Igbaria et

al. 1995; Teo et al. 1999; Venkatesh et al. 2002). Shopping enjoyment (Koufaris 2002), perceived

entertainment value of the Web site (O’Keefe et al. 1998), and perceived visual attractiveness have

positive impacts on perceived enjoyment and continuance intentions (van der Heijden 2003). Thus:

H5. Perceived enjoyment relates positively to increasing customer continuance intentions.

4 METHODOLOGY

To validate the conceptual model and the proposed research hypotheses, we developed an online

survey, which is suitable for collecting data from large geographical areas. In addition, compared with

traditional surveys, online surveys offer lower costs, faster responses, and less data entry effort.

4.1 Measures

The measures of the various constructs come from previous literature, adapted to the context of

online shopping if necessary. All online survey items use 1–7 Likert scales, on which 1 indicates

strongly disagree and 7 is strongly agree. The site quality and trust items come from McKnight and

colleagues (2002a, 2002b). The perceived usefulness items derive from Gefen (2003). Perceived

enjoyment is a measure from Childers (2001). Shih and Fang (2004) provide the subjective norm items.

The continuance intention items were adapted from Yang (2004).

The pilot study suggested some clarifications to the survey. Both Arabic and English language

versions were available. The Arabic questionnaire employed Brislin’s (1986) back-translation method

to ensure that the questionnaires have the same meaning in both languages.

5 DATA ANALYSIS

Survey respondents were people who were actively engaged in Internet and online shopping in

Saudi Arabia, including undergraduate and postgraduate students and professionals. As we show in

Table 1, the sample consists of 650 female participants in Saudi Arabia. This somewhat surprising

participation level illustrates the high rate of Internet use among women in Saudi Arabia. Most

respondents are in their late 30s (2.5% younger than 18 years of age, 26.6% between 18 and 25, 42.8%

are 26–35, 22% are 36–45, and 6.2% are older than 46 years). Similarly, 60% of the Saudi population

is younger than 30 years of age. The vast majority (92.6%) of participants came from the three main

regions in Saudi Arabia: 24.6% from the east, 27.8% from the central region, and 40.2% from the

western region. The education levels indicate 1.5% of respondents earned less than a high school

degree, 10.9% attended high school, 12.9% had diplomas, 52.9% had bachelor’s degrees, and 21.7%

were postgraduates. Most respondents thus are well-educated. Moreover, 36% of them work in the

public sector (government employee), 35.4% in the private sector, 6.5% were businesspeople, and

22.22% were students.

As we show in Table 2, 52.2% of the respondents visited at least five different online sites to

purchase each month, and 66.9% used the Internet for actual shopping. The western region reveals the

highest percentages in most categories, such that 31.1% spend £100–£500 per year online, and 51.3%

spend more than £501 per year. Furthermore, 49.7% of the respondents used the Internet in the prior six

months to make flight booking or purchase airline tickets, 37.5% made hotel reservations, 35.2%

purchased clothing, 58.6% bought books, and 37.8% purchased CD-DVDs or video tapes. To indicate

Talal Al-maghrabi and Charles Dennis

why they used the Internet, as we summarize in Table 3, 82% referred to information search, 56.8% to

social communication, 52.5% to banking, 64.8% to entertainment, 51% to work-related tasks, and 69%

used it for study-related tasks.

Table 1: Demographic Items

Question

Count

Percentage

Gender

Total Female Participants

650

100

Age

Less than 18

2.5

Between 18-25

173

26.6

Between 26-35

278

42.8

Between 36-45

143

22.0

Above 46

6.2

Education Level

Less than high school

1.5

High school

10.9

Diploma

12.9

Bachelor

344

52.9

Post-graduate

141

21.7

Occupation

Government employee

234

36.0

Private sector

230

35.4

Business people

6.5

Student

144

22.22

Income Level

<SR4,000 (£1,000)

105

16.2

SR4,000-SR6,000 (£1,000-2,000)

12.0

SR6,001-SR8,000 (£2,001-4,000)

13.7

SR8,001-SR10,000 (£4,001-7,000)

11.8

SR10,001-SR15,000 (£7,001-10,000)

128

19.7

>SR15,001 (>£10,000)

123

18.9

Dependent on others

7.7

Region

East region

160

24.6

West region

261

40.2

Central region

181

27.8

North region

4.5

South Region

2.9

Int. Journal of Business Science and Applied Management / Business-and-Management.org

Table 2: Items Purchased Online and Reasons

Items purchased in the

last six months

Region in Saudi Arabia

East

West

Middle

Buying Books

16.9%

25.5%

16.2%

Music CD, DVD,

Videotape

12.2%

15.5%

10.1%

Cloth

13.3%

14.7%

7.2%

Sports equip

7.9%

6.5%

3.6%

Travel reservation and

ticketing

15.5%

23.0%

11.2%

Hotel booking

11.2%

18.0%

8.3%

Reason for using the Internet

Info. Search

100

25.9%

36.0%

20.1%

Entertainment

21.6%

28.1%

15.1%

Social Communication

16.9%

27.3%

12.6%

Work

14.0%

25.5%

11.5%

Study

15.8%

26.6%

16.2%

Purchasing

19.8%

32.4%

14.7%

Banking

12.9%

25.9%

13.7%

Table 3: Important Issues when Shopping Online

Important issues to e-

shoppers

Region in Saudi Arabia

East

West

Middle

Security

100

27%

36%

21%

Price

104

27%

37%

20%

Service, Delivery

27%

35%

21%

Quality

102

27%

37%

22%

Payment

100

26%

36%

21%

Language Barrier

22%

29%

15%

6 ANALYSIS

The Cronbach’s alphas (Table 4) are all greater than 0.7 (Bagozzi and Yi 1988). The squared

multiple correlation cut-off point is 0.7, and the average variance extracted cut off-point is 0.5 or higher

(Bagozzi 1994; Byrne 2001; Hair et al. 2006) (Table 5). We thus confirm the convergent reliability and

discriminant validity.

Talal Al-maghrabi and Charles Dennis

Table 4: Scale Properties and Correlations

Factor Correlations

Model

Constructs

Mean

Std.

Dev.

Cronbach’s

alpha

Trust

Enj

CIU

26.92

6.38

0.927

1.000

32.97

7-86

0.946

.749

1.000

Trust

21.74

5.03

0.947

.655

.695

1.000

18.73

6.19

0.943

.259

.275

.395

1.000

Enj

28.39

8.61

0.931

.438

.465

.668

.536

1.000

CIU

31.48

7.98

0.961

.397

.421

.606

.533

.745

1.000

Table 5: Measurement Model

Constructs/Indicators

S. Factor

S.E

C.R.

AVE

Squared

Multiple

Correlation

Site Quality (SQ)

0.757

SQ 1

0.922

0.039

26.510

0.85

SQ 2

0.844

0.038

26.414

0.71

SQ 3

0.855

0.035

26.972

0.73

SQ 4

0.857

0.74

Perceived usefulness

0.813

PU 3

0.911

0.039

37.788

0.83

PU 4

0.909

0.027

37.135

0.83

PU 5

0.914

0.84

PU 6

0.871

0.029

33.487

0.76

Trust

0.804

Trusting Beliefs Integrity 1

0.896

0.028

35.069

0.80

Trusting Beliefs Integrity 2

0.886

0.023

42.297

0.79

Trusting Beliefs Integrity 3

0.896

0.027

35.167

0.80

Trusting Beliefs Integrity 4

0.909

0.83

Subjective Norm

0.804

SN 3

0.731

0.53

SN 4

0.973

0.054

25.507

0.95

SN 5

0.955

0.057

24.647

0.91

SN 6

0.908

0.055

23.875

0.82

Enjoyment

0.744

Enj 4

0.705

0.50

Enj 5

0.94

0.055

22.934

0.88

Enj 6

0.925

0.055

22.918

0.86

Enj 8

0.858

0.052

20.672

0.74

Continuance Intention

0.864

CIU 1

0.827

0.024

35.466

0.69

CIU 2

0.928

0.017

55.752

0.86

CIU 3

0.981

0.96

CIU 4

0.974

0.012

78.936

0.95

Int. Journal of Business Science and Applied Management / Business-and-Management.org

6.1 Structural Equation Model

As the first step in testing the proposed model, which operationalizes the hypotheses and the

factors involved in continuance e-shopping intentions in Saudi Arabia, we estimate the goodness-of-fit

indices (Figure 1). Bentler and Bonnett (1980) suggest the Chi-square/Degrees-of-freedom (CMIN/DF)

ratio as an appropriate measure of model fit, which should not exceed 5 (Bentler 1989).

A structural equation model (SEM) with AMOS 5.0 software determines additional goodness-of-

fit indices, including critical ratio (CR), chi-square (CMIN), degrees of freedom (df), chi-

square/degrees of freedom (CMIN/DF), root mean square residual (RMR), root mean square error of

approximate (RMSEA), goodness-of-fit index (GFI), comparative fit index (CFI), normal fit index

(NFI), incremental fit index (NFI), and the relative fit index (RFI). In general, GFI, NFI, RFI, IFI, and

CFI values greater than 0.90 indicate good model fit (Bentler 1989). As we illustrate in Table 6, all the

hypotheses are statistically significant and supported, with CRs ranging from 20.318 to 4.888, which

are greater than 1.96 and thus indicate acceptable results (Hair et al. 2006; Holmes-Smith 2000). As

illustrated in Table 7, the goodness-of-fit indices of the proposed model of continuance intentions fit

the data reasonably well, as confirmed by the chi-square CMIN = 875.370, df = 236, CMIN/DF =

3.709, RMR = 0.231, GFI = 0.905, CFI = 0.964, RMSEA = 0.065, NFI = 0.952, IFI = 0.965, and RFI =

0.944.

Table 6: Regression Weights

Hypotheses

Paths

Standardized

Regression

Weights

(B)

Standard

Error

S.E.

Critical

Ratio

C.R.

Value

Hypotheses

Findings

H1 a

<---

.305

.051

6.262

***

Supported

H1 b

Trust

<---

.749

.036

20.318

***

Supported

<---

Trust

.467

.053

9.324

***

Supported

H3 a

<---

.395

.037

9.580

***

Supported

H3 b

Enj

<---

.541

.032

13.473

***

Supported

H3 c

CIU

<---

.183

.040

4.888

***

Supported

H4 a

Enj

<---

.322

.032

8.876

***

Supported

H4 b

CIU

<---

.178

.039

5.521

***

Supported

CIU

<---

Enj

.527

.062

11.481

***

Supported

*** p < 0.001.

Talal Al-maghrabi and Charles Dennis

Table 7: Goodness-of-Fit Indices

Confirmatory Factor Analysis CFA

(Goodness-of-fit measure)

Acceptable Values

Value

Chi-Square CMIN

875.370

Degree of freedom

236

CMIN/DF

Chi square/ df ≤5

(Bentler and Bonnett, 1989)

3.709

P value

p≤0.05 (Hair et al., 2006)

0.000

Root mean square residual (RMR)

No established thresholds

(the smaller the better)

(Hair et al., 2006)

0.231

Goodness-of-fit (GFI)

≥ 0.90 (the higher the better)

(Hair et al., 2006)

0.905

Comparative fit index (CFI)

≥ 0.90 (Hair et al., 2006)

0.964

Root mean square error of approximate

(RMSEA)

< 0.08 (Hair et al., 2006)

0.065

Normal fit index (NFI)

≥ 0.90 (Hair et al., 2006)

0.952

Incremental fit index (IFI)

≥ 0.90 (Hair et al., 2006)

0.965

Relative fit index (RFI)

≥ 0.90 (Hair et al., 2006)

0.944

Next, we examine the regression weights (path significance) of each relationship in our research

model and the variance explained (R

value) by each path. The AMOS software reports the

standardized regression weights, standard error, and CR for each path, which we provide in Table 6.

The hypothesized associations are strongly significant at p = 0.000. Perceived enjoyment is the

strongest predictor of continuance intention (B = 0.53), followed by perceived usefulness (B = 0.18),

and then subjective norms (B = 0.18). The model explains 60% of the variance in continuance

intentions (Figure 1).

Figure 1: Internet Continuance Intention Shopping Model

Int. Journal of Business Science and Applied Management / Business-and-Management.org

6.2 Invariance Analysis

When comparing cultures or groups, research participants may not recognize the same meaning of

survey items. To minimize the bias in cross-national and cross-cultural research derived from the data

collection, we applied back-translation (Brislin 1986; Yi et al. 2008). In addition, we assessed the

measurement invariance across the groups to consider the constructs’ factorial invariance (Cheung et

al. 1999).

The invariance analysis indicates whether any differences occur between high and low online

spenders among women. If we find that the annual online spending effect on the measurement

invariance of the construct and the score of the group analysis is significant, the construct measurement

differs for the two groups, and they cannot be compared directly.

To compare the annual spending (low and high spending) among female respondents, we use

factorial invariance to assess the extent to which measures from both groups have the same meaning

(Hair et al. 2006). The CMIN = 1404.966, df = 499, CMIN/DF = 2.816, RMR = 0.249, GFI = 0.855,

CFI = 0.951, RMSEA = 0.053, NFI = 0.925, IFI = 0.951, and RFI = 0.918 indicate outstanding

goodness-of-fit indices across the groups (Table 8).

Table 8: Goodness-of-Fit Indices

Confirmatory Factor Analysis CFA

(Goodness-of-fit measure)

Acceptable Values

Value

Chi-Square CMIN

1404.966

Degree of freedom

499

CMIN/DF

Chi square/ df ≤5

(Bentler and Bonnett, 1989)

2.816

P value

p≤0.05 (Hair et al., 2006)

0.000

Root mean square residual (RMR)

No established thresholds

(the smaller the better)

(Hair et al., 2006)

0.249

Goodness-of-fit (GFI)

≥ 0.90 (the higher the better)

(Hair et al., 2006)

0.855

Comparative fit index (CFI)

≥ 0.90 (Hair et al., 2006)

0.951

Root mean square error of approximate

(RMSEA)

< 0.08 (Hair et al., 2006)

0.053

Normal fit index (NFI)

≥ 0.90 (Hair et al., 2006)

0.925

Incremental fit index (IFI)

≥ 0.90 (Hair et al., 2006)

0.951

Relative fit index (RFI)

≥ 0.90 (Hair et al., 2006)

0.918

Assuming the unconstrained model is correct, rather than constraining all factorial paths, the result

across groups indicates changes in df (Δdf) = 18, chi-square (Δχ²) = 21.721, and p = 0.245, which is

greater than Byrne’s (2001) 0.05 cut-off. Tests of this measurement invariance appear in Table 9,

which shows that changes in the chi-square and df are insignificant (p = 0.245). Therefore, the

goodness-of-fit indices are comparable across low and high online spending groups, which justifies the

invariance of the unconstrained and constrained models. Thus, we establish metric equivalence and can

proceed in our analysis to regression paths.

Table 9: Invariance Analysis

Model

Δdf

Δχ²

Measurement weights

21.721

.245

Structural weights

20.273

.016

The coefficient (regression paths) invariance analysis determines if high and low spenders among

female respondents have the same relationships with same variables in the research model. The

findings in Table 9 suggest coefficient non-invariance between low and high online spending among

women across the research model with all regression paths constrained (Δχ² = 20.273, Δdf = 9, p =

0.016). To consider the relationships between model constructs for the source of non-invariance

differences, we conducted an invariance analysis between low and high online spenders.

Talal Al-maghrabi and Charles Dennis

The findings in Table 10 indicate that low and high online spenders are non-invariant for certain

relational paths. Differences in their behavior in the context of online shopping continuance in Saudi

Arabia result from different coefficients of perceived site quality  perceived usefulness (change in

chi-square = 5.033, p = 0.025). For the high spenders, this influence is greater than that for low

spenders.

Table 10: Structural Factorial Analysis: High and Low Online Spending by Women

Hypotheses

Paths

Low Online

Spenders

High Online

Spenders

Invariance

C.R.

value

C.R.

Value

CMI

P Value

H1 a

<---

.185

.070

2.641

.412

.073

5.665

5.033

.025

H1 b

Trust

<---

.790

.052

15.310

.692

.050

13.815

1.853

.173

<---

Trust

.570

.073

7.849

.436

.077

5.638

1.604

.205

H3 a

<---

.393

.054

7.253

.321

.050

6.362

.953

.329

H3 b

Enj

<---

.477

.050

9.515

.402

.042

9.558

1.310

.252

H3 c

CIU

<---

.208

.063

3.292

.196

.053

3.699

.021

.886

H4 a

Enj

<---

.228

.047

4.883

.323

.045

7.253

2.118

.146

H4 b

CIU

<---

.239

.058

4.143

.202

.053

3.807

.223

.637

CIU

<---

Enj

.741

.095

7.772

.689

.082

8.399

.166

.683

The results of the latent mean online spending analysis appear in Table 11. The group analysis

between the low and high spender samples exhibits latent mean invariance for the research constructs.

Table 11: Means: Annual Spending Sample (Low Spender – High Spender)

Latent mean

S.E.

C.R.

P Value

-.113

.102

-1.110

.267

Trust

-.077

.104

-.744

.457

Enj

.065

.105

.618

.537

CIU

-.093

.107

-.868

.386

-.040

.105

-.380

.704

-.004

.088

-.050

.960

6.3 Direct and Indirect Effect Analysis

The direct and indirect effects in Table 12 reveal that the greatest total influences of direct and

indirect (mediated) effects on continuance intentions come from perceived enjoyment for both the low

online spenders (0.741) and high online spenders (0.689) samples. The next greatest influences derive

from perceived usefulness for low (0.721) and high (0.610) online spenders. Trust has a greater

influence on low online spenders’ (0.411) than on high online spenders’ (0.266) continuance intentions.

Thus, site quality, trust, perceived usefulness, and subjective norms all play significant roles for

continuance intentions toward online shopping in Saudi Arabia among women with both low and high

online spending habits.

Int. Journal of Business Science and Applied Management / Business-and-Management.org

Table 12: Direct and Indirect Influences on Continuance Intentions (Low Spender – High

Spender)

Construct

CIU (Low Spender)

CIU (High Spender)

Direct

Indirect

Total

Direct

Indirect

Total

--------

.458

--------

.435

TRUST

--------

.411

--------

.266

.208

.513

.721

.196

.414

.610

.239

.169

.408

.202

.223

.424

ENJ

.741

--------

.741

.689

--------

.689

= 0.60

7 DISCUSSION

This research attempts to provide a validated conceptual model that integrates different factors and

clarifies the theoretical problems of continuance e-shopping intentions and behavioral differences

among women on online annual spending in Saudi Arabia. The online field survey validates the

hypothesized model, and the model findings confirm that perceived enjoyment, perceived usefulness,

and subjective norms are the main determinants of continuance intentions in Saudi Arabia, explaining

60% of continuance e-shopping intentions. However, enjoyment is more influential (see Table 6; srw =

0.527, cr = 11.481), followed by subjective norms (srw = 0.178, cr = 5.521), and then perceived

usefulness (srw = 0.183, cr = 4.888). These findings are consistent with previous research (e.g.,

Bhattacherjee 2001a; Childers 2001; Davis et al. 1989; George 2002; Shih and Fang 2004; Taylor and

Todd 1995; Teo et al. 1999; Venkatesh et al. 2003). Enjoyment, perceived usefulness, and subjective

norms have positive influences (direct and indirect) on consumers’ continuance e-shopping intentions.

The measurement weights of low and high online spending among female shoppers, based on

metric invariance, are invariant. Testing for factorial regression paths invariance, we find that

relationship path between site quality  trust; trust  perceived usefulness; perceived usefulness 

subjective norms; perceived usefulness enjoyment; subjective norms  enjoyment; perceived

usefulness  continuance intentions; subjective norms  continuance intentions; and enjoyment 

continuance intentions are similar for both low and high online spenders among women in Saudi

Arabia. However, the site quality  perceived usefulness relationship path is non-invariant (high

spenders rw = 5.665; low spenders rw = 2.641). That is, higher online spenders tend to accept

technology, because these women perceive high quality content, good design, simple navigation, ease

in finding necessary information, and ease of communication (utilitarian and hedonic experiences),

which in turn increase the level of usefulness they perceive.

The model factorial paths for site quality and trust indicate strong antecedents of perceived

usefulness on the regression weights (site quality srw = 0.361, cr = 5.804; trust srw = 0.430, cr = 6.754)

(see Table 6). Both site quality (east = 0.438; west = 0.413; central = 0.415) and trust (east = 0.266;

west = 0.284; central = 0.280) have great indirect effects on continuance intentions (see Tables 17–19).

These findings match the collectivist culture of Saudi Arabia, where people tend to trust only those

within their in-group (Yamagishi and Yamagishi 1994). They similarly may apply to other collectivist

cultures or other nations in the Gulf States that appear similar to Saudi Arabia in various traits.

Trust and site quality do not have direct effects on continuance intentions toward the online

retailer.