Uncovering Consumer Behaviour in
E-commerce: the Role of Language and Payment
Methods
University of Amsterdam
Faculty of Economics and Business
Master of Science in Business Administration
Marketing track
Under supervision of: dhr. drs. Frank Slisser
By:
Student: Karima Bouchtaoui
Student number: 10071733
Statement of originality
This document is written by Student Karima Bouchtaoui who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents Table of Figures ... 4 Abstract ... 6 1. Introduction ... 7 2. Theoretical Background ... 11 2.1. Purchase Intentions ... 11 2.2. Language ... 12 2.3. Payment Methods ... 18 2.4. Research Gap ... 26 2.5. Conceptual Framework ... 27 3. Methodology ... 28 3.1. Pre-test ... 28 3.2. Participants ... 29 3.3. Measurement of Variables ... 31 3.4. Statistical Procedure ... 34 4. Results ... 37 4.1. Correlation Analysis ... 37 4.2. Direct Effects ... 42 4.2. Moderating Effects ... 48 5. Discussion ... 56
5.1. Theoretical and Practical Implications ... 56
5.2. Strengths and Limitations ... 64
6. References ... 66
Table of Figures
Figure 1. Preferred Online Payment Methods of Dutch consumers……….. 25
Figure 2. Conceptual Framework ... 27
Table 1. Descriptive statistics for non-Dutch (=,00) and Dutch (=1,00) consumers ... 31
Table 2. EM Means: Testing for Missing Data ... 34
Table 3. Means, Standard Deviations, Correlations and Reliabilities Non-Dutch ... 39
Table 4. Means, Standard Deviations, Correlations and Reliabilities Dutch ... 39
Table 5. Exploratory Factor Analysis for Purchase Intentions (1), Language Similarity Importance (2) and Payment Method Familiarity Importance (3); rotated component matrix for non-Dutch (,00) and Dutch (1,00) ... 40
Table 6. Multicollinearity Coefficients (non-Dutch = .00, Dutch=1.00) ... 41
Table 7. Bootstrap results for the direct relationships between the control variables age, gender and education, the predictor variables Language and Payment Methods and the outcome variable Purchase Intentions for non-Dutch consumers ... 45
Table 8. Bootstrap results for the direct relationships between the control variables age, gender and education, the predictor variables Language and Payment Methods and the outcome variable Purchase Intentions for Dutch consumers ... 46
Table 9. Bootstrap results for testing significance difference of Beta coefficients Language variable ... 47
Table 10. Bootstrap results for testing significance difference of Beta coefficients Payment Methods variable ... 47
Table 11. Bootstrap results for the moderating effect of Language Similarity Importance on the relationship between Language and Purchase Intentions for non-Dutch consumers ... 52
Figure 3. Moderating effect of Language Similarity Importance for non-Dutch consumers (Low Language = non-native, High Language = native) ... 52
Table 12. Bootstrap results for the moderating effect of Payment Method Familiarity Importance on the relationship between Payment Methods and Purchase Intentions for non-Dutch consumers ... 53
Figure 4. Moderating effect Payment Method Familiarity Importance for non-Dutch consumers (Low Payment Method=non-preferred, High Payment Method=preferred) ... 53
Table 13. Bootstrap results for the moderating effect of Payment Method Familiarity Importance on the relationship between Payment Methods and Purchase Intentions for Dutch consumers ... 54
Figure 5. Moderating effect Payment Method Familiarity Importance for Dutch consumers
(Low Payment Method=non-preferred, High Payment Method=preferred) ... 54
Table 14. Bootstrap results for testing significance of difference of interaction coefficients
between non-Dutch and Dutch consumers ... 55
Abstract
Online shopping has largely increased over the past decade; however, there is a missing potential of e-commerce due a home bias of consumers. Research has found that although distance related costs have decreased, language and payment methods related costs have greatly increased in the online environment, compared to the usual brick and mortar shopping. Studies have indicated that consumers are reluctant to purchase from e-stores that are offered in a foreign language and e-stores that fail to offer the preferred payment method of the consumers. This thesis has empirically studied consumer behaviour in the process of online shopping, by specifically testing how these two identified barriers to cross-border trade affect online purchase intentions of Dutch versus non-Dutch EU consumers, by the means of an online survey. Results indicate that e-stores that do not offer the preferred payment method of the consumer negatively affect purchase intentions for both consumer groups, whereas
language differences have a negative effect on purchase intentions only for non-Dutch EU consumers.
1. Introduction
The financial and economic crises seem to have had less negative implications for e-commerce than for many other industries (Silicon Republic, 2009). The crisis of 2008 has given e-commerce a boost, as consumers were seeking the price advantages of online
shopping (OECD Background Report, 2009). E-commerce can be defined as the production, distribution, marketing, sales or delivery of goods and services by electronic means
(Leinbach, 2001). E-commerce has three basic forms: business-to-business transactions (B2B), business-to-consumer transactions (B2C), and consumer-to-consumer transactions (C2C). This study will focus on the consumer-oriented form of B2C market, a steadily growing market worldwide (Hande, Gosh and Govil, 2015), because it is worthwhile to examine factors that eventually shape consumer buying-behaviour in this growing market.
Although the benefits of e-commerce are well known, the Civic Consulting (2011) report pointed out that the true potential of e-commerce has yet to be reached. In 2011, from the on average 40% of the European Union (EU) citizens that shops online, only 9% indicated they had purchased across borders within the EU (European Commission, 2012). It was thought that the rise of the Internet has led to a ‘death of distance’ and hence would increase cross-border trade, however there are apparently more hurdles than distance to overcome. Language and payment system costs were negligible in the mostly B2B offline international trade (Wei, 1996), but have proven to be a great new source of trade costs in the online international trade.
Gomez-Herrera, Martens and Turlea (2014) empirically studied and compared offline and online trade flows between EU countries. They found that although distance related costs have reduced in the online environment, language-related trade costs have increased and moreover, online trade introduced a new source of trade costs: online payment systems. According to the authors, the wide variety of payment systems impedes cross-border
e-commerce, because consumers are reluctant to purchase from websites that do not offer a payment system that consumers are familiar with and trust. The Civic Consulting (2011) report on e-commerce in the EU B2C market confirmed the consumers’ concerns regarding online payment systems, causing a barrier to online shopping across borders that causes the missing potential of e-commerce. In addition, the report Bringing e-commerce benefits to
consumers by the European Commission (2012) conclude similar results on the importance of
language similarity and payment method familiarity for the consumer online decision-making process. They concluded that both language barriers and payment systems concerns are a major reason for consumers to not purchase online across borders.
The importance of the two variables language and payment methods for the online decision-making processes is confirmed by several e-commerce studies on website features that result in website success. Chen, Hsu and Lin (2010) empirically studied website features that affect consumer purchase intentions. They found that consumers positively value
websites that provide a variety of payment options and offer native language versions of the website, which in turn increases their purchase intentions. Liang and Lai (2002) also studied website success by exploring features that positively influence Taiwanese consumer purchase intentions. They found that language similarity and offering credit card payment methods (i.e. the preferred payment method of the Taiwanese consumers) positively influences purchase intentions.
Above studies have identified store languages and the type of payment methods e-stores offer as important predictors for consumer purchase intentions, that eventually influences consumers’ willingness to shop online and across borders. For this reason, this study has chosen to confine itself to studying the role of these two variables in the online decision-making process. Theory suggests that language similarity and familiarity with the payment method positively influence consumer purchase intentions. However, little research
has been done to study if and how these variables influence the online purchase intentions of consumers from different nations. There are reasons to believe that these two main barriers to cross-border trade may have different roles in the consumer decision-making process for consumers from different countries, depending e.g. on the level of multilingualism and the level of preference for a given payment method (i.e. strong, moderate, weak preference) in a country. The Eurobarometer (Eurobarometer 386, 212) reported that 91% of the Dutch population claims to be bilingual, compared to 54% of the EU27 average. This suggests that language may not impede cross-border online shopping for Dutch consumers to the same extent as for non-Dutch EU consumers. Furthermore, a study on preferred online payment systems conducted by Gesellschaft für Konsumforschung ([GfK], 2014) revealed that the Dutch consumers strongly prefer a Dutch payment system iDeal, whereas most EU consumers indicate moderate preferences for several online payment methods (CyberSource: Global Payment Options, 2011). Thus, considering a strong versus moderate preference for a given payment method, it can be expected that the role of payment method in the online decision-making process may be greater for Dutch compared to non-Dutch consumers.
Although sufficient research has been done on the e-commerce decision-making processes and some research exists in the field of cross-border e-commerce, more study is needed to advance our knowledge in the field of cross-border online trade, to help overcome the missing potential of e-commerce. There are to my knowledge as yet no official statistics or data available on online cross-border transactions within the EU (also according to Gomez-Herrera et al., 2014), which complicates research in the field of cross-border e-commerce. Therefore, this thesis will aim to contribute to the field by studying consumer behaviour in the process of online shopping. Specifically, this study will empirically test how the predictors language and payment methods affect the purchase intentions of Dutch versus non-Dutch consumers. It is valuable to test the effects and importance that consumers from different
nations place on the two variables that obstruct cross-border trade, because (a) it advances the knowledge of if and how these predictors affect purchase intentions for different nations and (b) it provides e-retailers with valuable information for developing nation-targeted online marketing strategies. This master thesis addresses the following research question:
What is the role of ‘language’ and ‘payment methods’ in the online purchase decision-making process for non-Dutch EU consumers compared to Dutch consumers?
1.1. How do language differences in e-stores affect purchase intentions for
Dutch consumers versus non-Dutch consumers?
1.2. How do differences in payment method offered in e-stores affect purchase
2. Theoretical Background
Present study investigates the roles language and payment methods in the online shopping activities for Dutch consumers compared to non-Dutch EU consumers. This section will develop a testable framework that includes the two important variables that influence
consumer purchase intentions. These variables are based on cross-border trade studies online (Gomez-Herrera et al., 2014; European Commission staff working document, 2012; Flash Barometer No. 313, 2011; Civic Consulting, 2011); and on e-commerce studies regarding website features that influence purchase intentions (Maity & Dass, 2014; Chen, Hsu & Lin, 2010; Kim et al., 2008 & 2009; Pavlou & Fygenson, 2006; Lee & Lin, 2005; Chang, Cheung & Lai 2005; Liang & Lai, 2002). This chapter is structured as follows. The first sub-chapter will explain the notion of purchase intentions. The second two sub-chapters will elaborate on the two identified variables. This chapter will conclude with describing the research gap and presenting the conceptual model.
2.1. Purchase Intentions
Online purchase intention refers to an outcome of criteria assessment of consumers, regarding the quality of a website, quality of the products and information search and reflects the desire of consumers to make a purchase through a website. (Chen, Hsu & Lin, 2008). The theory of planned behaviour (TPB) is an influential theory for explaining and predicting various types of human behaviour, such as purchase behaviour (Ajzen, 2002). According to TPB, human behaviour is guided by three kinds of considerations: belief about the likely consequences or attributes of the behaviour (behavioural beliefs), beliefs about the normative expectations of other people (normative beliefs) and beliefs about the presence of factors that may further
obstruct performance of the behaviour (control beliefs). Combined, these believes will form the behavioural intention of human beings (Ajzen, 2002).
Chang et al. (2005) have identified more than 80 antecedents of purchase intentions that they grouped into three categories: website characteristics; product characteristics; and consumer characteristics. Building on studies regarding online decision-making processes and cross-border e-commerce that have identified the website characteristics of language and payment systems as important predictors of purchase intentions, this study confines itself to studying the effect of website characteristics on consumer purchase intentions. Product characteristics that may influence purchase intentions have to do with quality and price of the product. Consumer characteristics that may influence purchase intentions are largely studied in the realm of cross-cultural marketing. These latter studies explore the role of national culture and individual consumer attitudes on (cross-border) online purchase intentions.
Finally, drawing on the TPB, many e-commerce studies have proven that consumer intentions to engage in online transactions are a significant predictor of consumers’ actual behaviour (Pavlou & Fygenson, 2006; Kim et al., 2009), which is a justification for inferring consumers actual buying behaviour based on their intentions.
2.2. Language
Language similarity/dissimilarity is an antecedent of purchase intentions that is related to the category of website characteristics as described by Chang et al. (2005). Studies on border e-commerce have identified language dissimilarity as an important barrier to cross-border e-commerce and studies regarding website success have confirmed that language dissimilarity will adversely affect consumers’ online purchase intention (Gomez-Herrera et al., 2014; Chen et al., 2010). However, this may be less the case for nations that are highly
multilingual. Such is the case for The Netherlands, with a higher than average EU-average level of multilingualism (Eurobarometer 386, 2012). This implies that for Dutch consumers language dissimilarity may not adversely affect purchase intentions to the same extent as theory would predict. This sub-chapter will aim to answer the first sub-question: How do
language differences in e-stores affect purchase intentions for Dutch consumers versus non-Dutch consumers? This sub-chapter is structured as follows. The first part will explain the
role of language in the online purchase decision-making process, continue with elaborating on multilingualism in The Netherlands and other EU nations and conclude with formulating the hypotheses.
Gomez-Herrera et al. (2014) have studied the online trade flows within the EU Digital Market (the online equivalent of the EU Single Market) and found that language differences cause a significant barrier to international trade in the online context. They found that consumers are reluctant to purchase from e-stores that are offered in languages other than their native language. Language differences complicate the information search about the quality of products and/or services, which adversely affects the consumer’s willingness to purchase from online stores that do not offer a version in their native language (Gomez-Herrera et al., 2014). A report by the European Commission Bringing e-commerce benefits to consumers (2012) concludes similar results on the importance of language similarity for the online shopping activities of EU consumers. It is not only where the e-merchant is geographically based, i.e. the country of origin of an e-store, but also the language in which an e-store is offered, that induces or discourages consumers to purchase from a given e-store (European Commission, 2012). The report by the European Commission asked respondents to indicate reasons for not buying products from foreign e-retailers. Respondents were presented with 16 possible answers. The main answers of the respondents were: ‘There is enough choice in their
own country (42%)’, ‘I do not speak the language of the foreign websites’ (11%), and 21% indicated they had concerns regarding the payments in foreign e-stores. Thus, from cross-border e-commerce studies it can be concluded that consumers are reluctant to purchase from foreign based e-stores if they do not offer a native language version of the website.
Several e-commerce studies on website features that influence purchase intentions have also emphasized the importance of language similarity. Because of the specific
characteristics of the online environment, in which the transactions are blind (i.e. consumers cannot rely on face-to-face service), borderless, occur 24 hours a day and 7 days a week and are non-instantaneous (payments may occur before delivery/consumption), consumers
perceive online shopping as risky and uncertain. Websites that offer a native language version will alleviate some of the risk, because the processes of information search is simplified, resulting in an increase of consumer purchase intentions (Liang & Lai, 2002; Chen et al., 2010; Maity & Dass, 2014).
Chen et al. (2010) confirm that a website offered in the native language will increase purchase intentions. According to the authors, product uncertainties (e.g. product quality) are a major issue of online shopping. In addition, consumers have a growing desire for (product) information sharing. Offering native language versions of e-stores will (a) reduce consumers’ perceived product uncertainties, because it enables information search and consumption; and (b) it facilitates and simplifies information sharing. This will in turn positively affect
consumer purchase intentions (Chen et al., 2010).
Language preferences indeed play an important role in a consumer’s choice for a website. A survey report requested by the Directorate-General Information Society and Media of the European Commission, examined the Internet users’ attitudes and opinions towards the use of different languages on the Internet (Flash Eurobarometer 313, 2011). The report confirms above findings regarding consumers strongly preferring native language websites
over non-native language websites. According to the report, 90% of the EU27 respondents revealed that when given a choice of languages, they always visited a website in their own language for different Internet activities, such as reading or watching content. In addition, the report indicated that only 18% of the EU27 respondents admit to frequently searching or buying products from a website in a foreign language, whereas nearly a majority of 42% admits to never shop on websites offered in a foreign language (Flash Eurobarometer 313, 2011).
Above studies indicate that language similarity will positively affect consumers’ purchase intentions and consequently it will also increase consumers’ purchase intentions from foreign based e-retailers if they offer a native language version. Although Gomez-Herrera et al. (2014) found that language differences reduced online trade flows within the EU, they did not study the role of language in the consumer decision-making process for consumers from different nations, and neither did they empirically test the effects of language differences on purchase intentions. Rather, they proposed differences in language as a
universal barrier to cross-border online trade. However, there are reasons to expect that this so called language effect depends on the level of multilingualism in a country, and hence
language differences may play a smaller role in the online decision-making process for countries with high levels of multilingualism. According to a 2012 European Commission report on Europeans and their Languages (Eurobarometer 386, 2012), 91% of the
respondents in The Netherlands claims to be bilingual, compared to 54% of the average EU27 population. Furthermore, 77% of the Dutch population claims to speak at least two foreign languages well enough to have a conversation, compared to 25% of the EU average.
Specifically, only the population in Luxembourg (84%) is able to do so more than the Dutch. Furthermore, 56% of the Dutch population claims to understand English well enough to be able to use it in the online environment, compared to 26% of the EU average, with only Malta
(64%) and Denmark (58%) scoring higher.
The high level of multilingualism in The Netherlands is most likely due to the educational system in The Netherlands, which educates students in at least one foreign language during their high school years. All students in The Netherlands take final exams in English and 45% and 35% undertake exams in German and French respectively (Centraal Bureau voor de Statistiek, 2015). The Netherlands additionally score very high on the English Proficiency Index (EPI) (EF English Proficiency Index, 2014). EPI calculates a country’s average adult English skill level using data from English tests, including grammar and vocabulary (EF EPI Country fact sheet: Nederland, 2014). Above findings imply that the Dutch are particularly comfortable with the English language, which suggests that e-stores in this language may not reduce purchase intentions for Dutch consumers to the same extent as other foreign languages would. However, this thesis distinguishes between languages based on being ‘native’ or ‘non-native’ and therefore suggests that future research will study the effects of e-stores offered in different languages (e.g., French, English, German) on consumer’s purchase intentions.
Based on theories of international trade in the online context; e-commerce studies regarding website features that influence purchase intentions; and a report on consumer language preferences, it can be expected that language plays an important role in the consumer’s online shopping decision-making process. Specifically, theory explains that language similarity will positively influence consumer online purchase intentions, thus e-stores offered in the native language will increase consumer purchase intentions. However, considering the higher level of multilingualism in The Netherlands compared to other EU countries, it can be expected that there is only a positive effect of language similarity on purchase intentions for the non-Dutch EU consumers.
Therefore this study hypothesizes:
H1a: ‘Language’ (0=non-native, 1=native) will have a positive relationship with purchase intentions for non-Dutch EU consumers.
H1b: ‘Language’ (0=non-native, 1=native) will not have a positive relationship with purchase intentions for Dutch consumers.
Besides language, other factors such as ‘product price’, ‘website design’ and ‘mood of the consumer’ will also influence the purchase intentions of consumers. Consumers may for example indicate high purchase intentions when shopping in an e-store that was coincidently offered the native language, where in fact the high purchase intentions were due to these other factors. This complicates the process of testing the direct effect of language on purchase intentions, and issues of content validity may arise. Content validity is the extent to which a data collection tool provides enough data to answer the research question and meet all the objectives (Field, 2013). To overcome the problem of content validity, a moderator that measures the importance that consumers place on e-stores being offered in their native language is included in the analysis. This variable moderates the relationship between language and purchase intentions in such a way, that the positive relationship of language with purchase intentions for non-Dutch consumers will be stronger for those consumers that place high importance on language similarity compared to those that place a lower
importance. The interaction effect will be able to show evidence for the direct language effect, if indeed the purchase intentions are higher for those consumers who score high on the
language importance variable, in e-stores that indeed were offered in the native language. Although the use of this moderator is unorthodox, it will serve as extra support to test for the
existence of the language effect on purchase intentions for non-Dutch consumers. It will furthermore provide a better picture of whether or not consumers do indeed consider the language of e-stores in their purchase decision-making process. Considering that this thesis expects the language effect only for the non-Dutch consumers, the moderating effect will also only be tested for this group. This leads to the following hypothesis:
H2: ‘Language Similarity Importance’ moderates the relationship between ‘language’ (0 is non-native, 1 is native) and ‘purchase intentions’ in such a way that the higher (lower) the importance the non-Dutch consumers place, the stronger (weaker) the positive relationship.
2.3. Payment Methods
The variable payment methods can be considered as a second antecedent of purchase intentions, that is related to the category of website characteristics as described by Chang et al. (2005). This variable is identified based on studies in cross-border e-commerce and on several studies regarding the e-commerce decision-making process. The study by Gomez-Herrera et al. (2014) suggests that the Internet has introduced payment systems as a new source of trade costs. According to the authors, consumers are overwhelmed with the number of online payment systems and solutions that exist and are reluctant to purchase from e-stores that offer payment options that they do not trust and are unfamiliar with, i.e. the non-preferred payment methods. Furthermore, several studies on the e-commerce decision-making
processes found a significant positive effect of offering the preferred payment solution of consumers and offering a wide variety of payment solutions on consumer purchase intentions.
The development of online payment methods and the type of preferred payment methods is driven by country specific factors (Online payments: Moving beyond the web,
2012). Furthermore, some countries indicate very clear and strong preferences for methods, other nations within the EU are more indifferent when it comes to online payment solutions (Online payments: Moving beyond the web, 2012). Therefore, it can be expected that the role of payment methods in the online decision-making process differs for consumers from different nations, depending on e.g. the level of preference for a certain payment method that consumer indicate (e.g. weak, moderate, strong preference for a given payment option). To my knowledge, little research exists that empirically tested the role of payment methods in the online decision-making process. Specifically, little research exists that studied how payment methods (i.e. e-stores that offer or fail to offer the preferred payment method of the consumer) affect consumer purchase intentions for consumers from different nations.
This sub-chapter will aim to answer the second sub-question: How do differences in
payment method offered in e-stores affect purchase intentions for Dutch consumers versus non-Dutch consumers? This sub-chapter is structured as follows. First, this sub-chapter will
explain the role of payment methods in cross-border e-commerce and continues with its role in the online decision-making process. The second part elaborates on studies regarding payment systems in The Netherlands and other EU nations and concludes with formulating the hypotheses.
Gomez-Herrera et al. (2014) studied trade flows within the EU Digital market by applying the Gravity Model of trade and comparing offline to online international trade flows. They report that the Internet has introduced a new source of trade costs: payment systems. The authors found that online payment systems have a significant effect on cross-border trade. Their results indicate that online trade flows within the EU Digital Market decreased significantly relative to offline trade, when the payment system variable was introduced. According to the authors, this was due to differences in payment systems across EU countries. They state that
the concerns regarding payment systems are generally higher in the B2C market than the B2B market, considering the higher perceptions of risk of consumers relative to businesses.
Consumers are reluctant to purchase from e-merchants that offer a payment method that they do not prefer and is not commonly used in their country. The authors state that in order to increase cross-border trade, e-retailers require flexible online payment systems in order to cater to the demands and wishes of all EU consumers. They predict that within the EU Digital Market, a 1% increase in the use of flexible and harmonized payment systems could increase cross-border e-commerce by as much as 17% (Gomez-Herrera et al., 2014).
There have been two important publicly led initiatives to harmonize the fragmented online payments landscape in Europe: the Payment Services Directive (PSD) and the Single Euro Payments Area (SEPA). The aim of the PSD is to ensure the user-friendliness, safety and efficiency of payments throughout the EU Digital Market and it furthermore introduced a system of licenses for payment service providers to help achieve these aims (Payment
Solutions and Payment Processing: PSP License, 2015). There are no exact numbers of licenses issued, however just over half of the e-stores in the United Kingdom (UK) have obtained the license. Considering that the UK is the clear frontrunner in Europe, the
harmonizing effect in the EU of the registered license system can be questioned. The second initiative SEPA’s aim is to abandon the concept of cross-border payments and to treat Europe as one domestic region for payments, by harmonizing and simplifying credit card transfers and direct debits across 32 European countries (SEPA Introductie, 2015). However, the harmonizing effect is as stated limited to the European region, and thus excluded non-EU e-stores from upcoming economies such as Brazil, Russia, India and China (BRIC-countries), with growing B2C e-commerce markets (MarketWatch, 2015). Second, it does not tackle the payment method problem from the perspective of the consumers, regarding the wide variety of payment options offered in e-stores. Rather, it solved the problem behind the scene, the
problem of online payment inefficiency. Despite the attempts to harmonize the EU online payment market, the variety of payment methods that e-stores offer (inside and outside the EU Digital zone) remains a barrier for consumers to purchase online (and across-borders).
The European Commission consumer survey report Bringing e-commerce benefits to
consumers (2012) have also identified payment related issues as an important barrier for
cross-border online shopping. When asked for the reasons for not buying products/services from a foreign e-retailer, 21% of the respondents indicated that they had concerns regarding online payment methods. Other reasons to not shop across borders were that there is enough choice in their own country (42%) (European Commission, 2012). Thus, in the realm of cross-border e-commerce, payment systems are an important predictor of consumer purchase intentions. Familiarity with and trust of the payment systems offered in (foreign) e-stores will reduce payment related concerns, which positively influence consumers’ purchase intentions to shop online (Civic Consulting, 2011).
Several e-commerce studies on website features that influence purchase intentions have also emphasized the importance of payment systems. Liang and Lai (2002) conducted an empirical study to evaluate how e-store features affect consumer purchase intentions. From a consumer-oriented perspective, they derived functional requirements of websites that they grouped into three categories: motivators, media richness and hygiene. These requirements will in turn influence consumer purchase intentions. ‘Media Richness’ refers to channel’s ability to communicate richness of information (Daft & Lengel, 1984). ‘Hygiene’ refers to the website’s ability to offer delivery and return options, and product tracking options. Finally, ‘Motivators’ refer to e.g., having good search engines and creating convenience for the consumer during the transaction process (Liang & Lai, 2002). The results of their
empirical study among 9000 Taiwanese shoppers show that hygiene factors and motivational factors play key roles in the consumer decision-making process. An important ‘Motivator’
related to creating consumer convenience is the type of payment methods a website provides. Specifically, the authors found that offering ‘credit card’ payment options is desirable among the Taiwanese consumers, and offering ‘multiple payment solutions’ is even more desirable among the consumers. According to the authors, offering a payment method that consumers from a given nation prefer and are familiar with, reduces levels of consumers’ perceived risk, resulting in higher purchase intentions (Liang and Lai, 2002).
Chen et al. (2010) identify similar criteria based on which consumers assess websites and e-stores: website quality (technology factors), product evaluation (product factors) information search (shopping factors). ‘Technology factors’ comprise the quality of the website that facilitates the technical aspects of online activities. ‘Product factors’ refer to the evaluation of the quality/price of the products. Finally, ‘shopping factors’ refer to individual website attributes that relate to consumer convenience. Consumer convenience refers to the services of e-stores that reduce consumer time and effort during the Internet activity. Services such as offering a variety of payment methods will reduce the efforts and time consumers spend to shop online, because they can choose a payment option that they trust and are familiar with. This service will increase the levels of consumer convenience, which in turn will increase consumer online purchase intentions (Chen et al, 2010). Other services naturally also increase consumer convenience; however, considering the scope of this thesis, this study has chosen to focus solely on the two aspects of language and payment methods.
Above studies on website features that influence consumer purchase intentions have established the importance of payment methods in the online decision-making process for consumers. Specifically, the studies have indicated that e-stores that offer the preferred
payment methods of consumers (e.g. by offering flexible payment options and offering a wide variety of payment options) will positively influence consumer’s online purchase intentions.
However, over 200 online payment methods and solutions exist (About Payments, 2015) and consumers worldwide have different preferences for payment systems. E-commerce Europe, a national e-commerce association representing over 25,000 online companies, published an online payment report that provided an overview of e-commerce payment methods and habits across the globe. In Europe, the most common used and offered online payment methods are credit card payments, dominated by major players MasterCard and Visa, with an exception for Germany and the Eastern European region. In Hungary, Poland, Romania and Slovakia consumers still largely prefer cash-on-delivery payments, mainly due to a lack of
card-payment infrastructure. In Germany consumers prefer online card-payment services such as PayPal and direct debit transfers (58%) and credit card payments are preferred slightly less (43%) (Online payments: Moving beyond the web, 2012). The consumers from most EU countries indicate a moderate preference for credit cards payments, with other online payment options such as direct debit transfers and PayPal being slightly less popular (Belgium, France, Greece, Italy, Luxembourg, Norway, Sweden, UK) except for the Danes that indicates a strong
preference for credit card payments (Online payments: Moving beyond the web, 2012). The Dutch have a strong preference for iDeal (GFK, 2014), an e-commerce payment method developed in 2005 by the Dutch company Currence. It conducts payments via direct bank transfers between nine major Dutch banks (iDeal, 2015). Although iDeal is promoting their payment system among e-retailers worldwide, 80% of their payments still occur on Dutch websites and 15% on Belgian Flemish websites (Thuiswinkel Waarborg, 2015). Figure 1 depicts the results of the study by GfK and reveals that 53% of consumers exclusively use the iDeal payment system, which is greatly preferred over the second most used method credit card payments (12%).
Based on the theory about payment methods and its role in the consumer decision-making process, combined with above facts about consumers’ (weak, moderate or strong)
preference for a payment method, we can expect that not offering the preferred payment method will have a negative effect on purchase intentions. We can also expect that this payment method effect differs for consumers from different nations, depending on how strongly they have a preference for a given method. This study has hypothesized the following:
H3a: Payment Methods (0=non-preferred, 1=preferred) will have a positive relationship with purchase intention for non-Dutch consumers.
H3b: Payment Methods (0=non-preferred, 1=preferred) will have a positive relationship with purchase intention for Dutch consumers.
The same issue of content validity arises when trying to measure the direct effect of payment methods on purchase intentions, as with measuring the language effect on purchase intentions. Consumers may for example indicate high purchase intentions when shopping in an e-store that coincidently offered the preferred payment method, where in fact the high purchase intentions were due to other factors such as the design of the website or the price of the product. This complicates the process of testing the direct effect of payment methods on purchase intentions. To overcome the problem of content validity, a moderator will be included in the analysis. The moderating variable ‘payment method familiarity’ will measure the importance that the consumers place on e-stores offering a payment method that the consumers prefer, and moderates the relationship between payment methods and purchase intentions. Results of this interaction effect will depict whether the purchase intentions are indeed higher (lower) on websites that offered (failed to offer) the preferred payment method, for those consumers that score high on the moderating variable. If this is indeed the case,
additional evidence is found for the payment method effect on purchase intentions and thus for the fact that consumers have considered the payment method offerings of the e-store during their decision-making process. Although the use of this moderator is unorthodox, it will serve as extra support to test for the existence of the language effect on purchase intentions for non-Dutch consumers.
To conclude, it can be expected that for consumers that place high (low) importance on e-stores offering their preferred payment method, the positive relationship of payment methods with purchase intentions is stronger (weaker). This leads to the following hypotheses:
H4a: ‘Payment Method Familiarity Importance’ moderates the relationship between ‘payment methods’ and ‘purchase intentions’ in such a way that the higher (lower) the moderating variable, the stronger (weaker) the positive relationship for non-Dutch consumers.
H4b: ‘Payment Method Familiarity Importance’ moderates the relationship between ‘payment methods’ and ‘purchase intentions’ in such a way that the higher (lower) the moderating variable, the stronger (weaker) the positive relationship for Dutch consumers.
2.4. Research Gap
According to a report by Civic Consulting (2011) and The European Commission (2012), there is a missing potential in commerce. Consumers do not reap all the benefits that e-commerce platforms offer them, because of a home bias in their online shopping activities. Although the Internet mitigated distance related costs, an important barrier to cross border trade in the offline context, it simultaneously introduced new sources of trade costs that are related to language and payment methods, causing the home bias. Several studies that
uncovered the barriers to cross-border e-commerce have pointed out that language differences and payment issues (i.e. payment methods that consumers are unfamiliar with and hence distrust) impede cross-border e-commerce and result in the missing potential online shopping. Besides presenting these two predictors of purchase intentions as important impediments to cross-border trade, little research has been done to study if and how these variables influence the online purchase intentions of consumers from different nations.
This study will aim to fill this gap by studying consumer behaviour in the process of online shopping. Specifically, this study will empirically test how the two variables language and payment methods affect the purchase intentions of consumers from The Netherlands compared to other EU-countries. Are these barriers for cross-border trade the same for e.g. consumers from nations that are largely multilingual and have a high, moderate or low preference for a payment method system (that e.g. is largely offered in national e-stores)? This study builds on the findings by Gomez-Herrera et al. (2014) and others studies, but dives deeper into roles of these variables by studying the effects of these variables on consumer purchase intentions. By doing so, it will add to the (limited) knowledge in the field of consumer behaviour in cross-border e-commerce.
2.5. Conceptual Framework
To test the proposed hypotheses, below conceptual framework is developed.
Figure 3. Conceptual Framework Figure 2. Conceptual Framework
3. Methodology
The aim of this study is to empirically test the role of language and payment methods in the online decision-making process for Dutch versus non-Dutch consumers. Considering the research question and the sub-questions of this thesis, an explanatory research design is chosen. The study will be conducted with the means of an online survey, distributed via several online platforms such as Facebook and Instagram and Intranet channels to both Dutch and other EU consumers. The survey will include questions on demographic details of the potential respondents including gender, age, and highest level of education. The second part of the questionnaire will include items covering the independent -and the dependent variables of this study. All items will be presented as five-point Likert-scale response questions with 1 being ‘strongly disagree’ to 5 ‘strongly agree’. The multi-scaled items used to measure the constructs are adopted and/or adapted from different English research studies.
This chapter furthermore represents the start of the empirical part of this study. First the pre-test and its results will be discussed, then the most evident characteristics of the participants will be outlined. Afterwards, the variables included in the questionnaire and corresponding reliabilities are discussed. Finally, a description will be given of the statistical approach that was taken in order to test for the expected relationships as discussed in the previous chapter. The appendix will include the complete questionnaire in English.
3.1. Pre-test
Two pre-test have been conducted on the 9th and the 11th of July, to ensure validity and that the questionnaire worked well according to the guidelines of Hunt, Sparkman & Wilcox (1982). The pre-test of the questionnaire paid extra attention and care to the questions that
measured the purchase intentions, considering the difficulty to capture the purchase intention of consumers by the means of an online survey. It is important to accurately formulate a sentence that instructs consumers to go back to their latest online shopping experience, prior to actual purchase behaviour, i.e. prior to the moment the consumers made an actual purchase or not. Several versions of sentences have been presented to a panel of 5 respondents (fellow students and laymen) and asked them to explain what moment they pictured when reading the instruction. With their input, the formulation of the sentence has been adjusted several times which resulted in the final formulation: ‘Please think of the last website you have visited when you were looking to shop online, prior to your decision to purchase or not. It is not relevant whether or not you have made an actual purchase from that Website.’ A second pre-test with 10 respondents (5 additional respondents) confirmed that this formulation was indeed a sufficient instruction to get consumers to return to their latest online shopping experience prior to their purchase decision. Two of the ten panel members indicated a low purchase intention for their last shopping experience, because of payment method concerns (it was not possible to pay with iDeal), which was the reason behind their low purchase
intention.
3.2. Participants
The sample consisted of 49.3% Dutch consumers, and 50.7% non-Dutch consumers (largest groups: 9.7% Germany, 7.2 % Belgium and 5.3 % Denmark). Six respondents indicated they were from a country not listed. Respondents were approached via online platforms Facebook (other Facebook users have shared the survey with their friends etc.) and Instagram, face-to-face at the University of Amsterdam’s library (Koningsplein), university mail, and in addition e-mails were send to all employees working at the insurance company Nationale Nederlanden
(Amsterdam office) and at the news channel RTV Rijnmond. This resulted in 284
questionnaires, of which 217 have been fully completed and from which eventually 207 were used for the data analysis. The six respondents from ‘other’ nations and four respondents with no previous online shopping experience were excluded from further analysis. There is no known explanation for the large amount of respondents that have started the survey, but never finished it.
Taking all 207 respondents together (Meanage= 27.39, SDage= 4.95, age-range is 19-51)
a majority of 61.8% is female. The sample covered mainly higher educated respondents, 30.4% has a 4-year College and 48.8% has a Master’s Degree. The remaining 20.8% completed High School (2.9%), Some College (9.2%) and 2-year College Degree (6.3%). Detailed descriptive statistics can be found in table 1. For both consumer groups (i.e. non-Dutch and non-Dutch consumers), the importance of familiarity with the payment method that websites offer is more important than websites offering a native language version (non-Dutch: Meanpayment importance= 3.39, Meanlanguage importance=2.74, Dutch: Meanpayment importance= 2.97,
Meanlanguage importance= 2.53). Although not including this variable in the conceptual framework,
the survey included a question on the willingness of respondents to accept English language e-stores as an alternative to e-stores in their native language (native English speakers were excluded from this analysis). Dutch consumers (MeanEnglish language= 4.17) more often than
non-Dutch consumers (MeanEnglish language= 4.07) indicated that websites offered in English
would be a satisfying substitute of a native language website (the native English speaking respondents were excluded). Furthermore, the Dutch consumers have more often been browsing in e-stores offered in the native language: 24.5% of the Dutch consumers indicate that the last e-store they visited was not offered in the native language, compared to 45.7% of the non-Dutch consumers. Finally, most consumers recall that the e-store they last visited did offer their preferred payment method (non-Dutch: 82% and Dutch: 73.5%).
Table 1. Descriptive statistics for non-Dutch (=,00) and Dutch (=1,00) consumers Nationality N Minimu m Maximu m Mean Std. Deviation ,00 Education 105 1 7 4,99 1,282 Age (years) 105 19 51 28,09 5,476 Gender 105 1 2 1,61 ,490 Language 105 ,00 1,00 ,5429 ,50055 Payment_Method 105 ,00 1,00 ,8857 ,31968 Purchase_Intentions 105 1,50 5,00 4,0682 ,72873 Language_Importance 105 1,00 5,00 2,7381 ,91946 Payment_Importance 105 1,00 5,00 3,3905 ,97890 English_Language 102 2,00 5,00 4,0762 ,82852 Valid N (listwise) 105 1,00 Education 102 2 7 5,36 ,952 Age (years) 102 20 43 26,68 4,245 Gender 102 1 2 1,63 ,486 Language 102 ,00 1,00 ,7549 ,43227 Payment_Method 102 ,00 1,00 ,8235 ,38310 Purchase_Intentions 102 1,83 5,00 3,9165 ,75518 Language_Importance 102 1,00 4,25 2,5319 ,72241 Payment_Importance 102 1,00 5,00 2,9691 1,06315 English_Language 97 2,00 5,00 4,1718 ,58157 Valid N (listwise) 97
Note: Gender (1=male; 2=female) Language (0= non-native, 1=native); Payment Method (0=non-preferred, 1=preferred).
3.3. Measurement of variables
All items used in the questionnaire were derived from English studies. The questionnaire included questions on demographic details of the respondents, such as age, gender and
educational background. The respondents have been clearly instructed to remember their most recent online shopping experiences on a website, prior to their purchase decision. This
instruction was repeated above all the questions that measured their purchase intentions and the two items that measured the two independent variables, to ensure they were recalling their last shopping experience when answering the questions. Furthermore, the questionnaire
included statements covering the moderating variables ‘language similarity importance’ and ‘payment method familiarity importance’.
The dependent variable ‘purchase intention’ was based on six items, measured on a five point Likert-scale. The items used were adopted from previous research questionnaires. Examples of items are: ‘I would make another purchase from the website’ and ‘I would recommend this site to a friend’ (Thamizhvanan & Xavier, 2013; Gefen, Karahanna & Straub, 2003). A reversed item is included, meaning that a relatively low score indicates relatively high levels of purchase intentions. Reverse coding helps to reduce the bias caused by response styles, especially acquiescence bias (Field, 2013).
The independent variables ‘language’ and ‘payment methods’ are categorical variables that can be ranked. ‘Language’ (non-native = 0 and native=1, because theory indicates that native languages are positively related to purchase intentions) and ‘payment methods’ (non-preferred=0 and preferred=1, because theory indicates that preferred payment methods are positively related to purchase intentions). The respondents have been asked to indicate the language on the website that they last visited when they were looking to shop online (i.e. ‘Was the language on this website offered in your native language?’) and the payment method on that website (i.e. ‘Did the website offer your preferred payment method?’).
To assess the potential moderating effect of language similarity importance on the relationship between language and purchase intention, a measurement scale was used to capture the importance consumers place on language similarity. The moderator ‘language similarity importance’ was based on four items, measured on a five point Likert-scale. The items have been adopted from the Flash Eurobarometer report (Flash Barometer 313, 2011). Examples of items are: ‘When there is a choice of languages, I will always visit a website in my native language’ and ‘I would also purchase from websites that are not offered in my
native language’ (reversed). A reversed item is included to reduce the bias caused by response styles.
The moderator ‘payment methods familiarity importance’ measures the importance consumers place on a website offering their preferred payment system. The variable was based on three items, measured on a five point Likert-scale. The items have been adapted from previous research (Civic Consulting, 2011). Examples of items are: ‘I prefer online shopping on websites that offer my preferred payment’ and ‘I would shop on websites, even if the website does not offer my preferred payment option’ (reversed). This thesis has adjusted the items by changing the wordings ‘credit card’ into ‘my preferred payment’. In addition, the survey includes items that measure consumers’ willingness to purchase from e-stores that are offered in English. Although this study will not further analyse its role, measuring this variable will provide more insight regarding the language effect on purchase intentions.
To conclude, the two independent variables are ranked, and the moderators and dependent variable are on the ordinal data level. Considering the conceptual model and the variables, this study will test the hypotheses by running (moderation) regression analyses and a second set of regression analyses to test whether the Beta coefficients of the two groups of Dutch and non-Dutch EU consumers significantly differ from each other. Results of the current study control for three variables: age, gender, and educational background. These items were included in the last section of the questionnaire. This study is interested in the effect of the predictor variables on purchase intentions, and not in the possible effects of the control variables. Limited research has been conducted on the effects of socioeconomic characteristics on online shopping, but there is reason to believe that these characteristics can influence online shopping behaviour (Hernández, Jiménez & Martín, 2011). Based on the results of the regression analyses, this study can draw conclusion on the effects of language and payment methods on purchase intentions, for both non-Dutch and Dutch consumers.
3.4. Statistical Procedure
Data were collected by means of an online survey. The survey administration started on July 13th 2015, and was closed on July the 16th 2015. For the data analysis the Statistical software Package for Social Sciences (SPSS) was used. The data has been cleaned and checked for missing values before the analysis. Specifically, the data has been checked for patterns in the missing values by conducting a MCAR-Little’s test (Missing Completely At Random). The MCAR-Little’s test is testing whether or not the missing data are missing completely at random. The H0 for this test is that there is no systematic pattern in the missing data. The
Expectation Maximization (EM) statistics indicate a significance level of .196, which means that the null hypothesis cannot be rejected (See table 2). Therefore, the data are probably missing in a random way and data may be imputed. The missing data has been imputed using the mean values of the variables.
Table 2. EM Means- Testing for Missing Data
Q34 Q11 Q18 Q19 Q41_6 Q41_7 Q13_1 Q13_3 Q13_4 Q13_5 Q15_1 Q15_2 Q15_3 Q15_4 Q17_1 Q17_2 Q17_3 Q39_1 Q39_2 Q39_3 Q7 2.02 1.00 1.34 11.15 12.80 13.04 4.05 3.91 2.07 3.98 3.32 3.10 2.04 3.91 3.35 3.18 2.98 9.26 9.24 8.84 1.00 a. Little's MCAR test: Chi-Square = 182.450, DF = 167, Sig. = .196
After recoding the reverse coded items, scale reliabilities, descriptive statistics, skewness, kurtosis and normality tests were computed. From the in total eight variables, the variables language similarity importance and payment method familiarity importance are normally distributed. Educational background is not normally distributed and is negatively skewed in both groups, indicating a relatively high-educated group. This may have
implications for the results, which will be discussed in the fifth chapter. Age is not normally distributed due to the sample consisting of relative young respondents, with a few older respondents. These outliers are few in number, and in order to keep the valuable data, the outliers were dealt with, using the Winsorizing technique. This technique assigns the outliers
the same value as the next highest or lowest value found in the sample that is not an outlier (Field, 2013). The outliers of the non-Dutch and Dutch consumers were given the age 30 and 28 respectively.
The variables language, payment methods and purchase intentions are non-normally distributed data, according to the Levene’s test (p < 0.05). However, when sample sizes are relatively large, small differences in group variances can produce a Levene’s test that is significant (Field, 2013). Therefore a second look has been taken at the data, which has been checked for normality by looking at the kurtosis and skewness levels. The kurtosis and skewness levels indicate whether the data is normally distributed. The Zscores of skewness and
kurtosis of the variables were calculated for both groups using the following formula:
Zskewness= 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆S−0 and Zkurtosis= 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆K−0
When |S| and |K| > 1. 96 the skewness is significantly different from zero (alpha=5%) (Field, 2013).
After calculating the |S| and the |K| values of the independent variables for both non-Dutch and non-Dutch consumers, it became apparent that the values exceed 1.96 and the data is significantly different from normally distributed data. A third step has been undertaken to transform the data, using the log transformation, square root transformation and the reciprocal transformation. However, the Levene’s test and calculating the |S| and the |K| values indicated that the data remained non-normally distributed.
The absence of a normal distribution in the dependent variable can be explained by the fact that purchase intentions can be considered as a binary variable. Respondents will most likely indicate an either high or low purchase intention, and few will be indifferent when it comes to purchase intentions. This is not surprising, considering the consumer’s perceived
risk of online shopping (Kim et al, 2008): consumers have either high purchase intentions when they were satisfied or low purchase intentions when they were dissatisfied about the e-retailer and e-store. The non-normally distributed data of the predictor variables is neither surprising, indicating that for both Dutch and non-Dutch consumers the last shopping experience have been on a native language website where their preferred payment method was offered, rather than a foreign language e-store that failed to offer the preferred payment method.
Based on the model (i.e. two categorical predictor variables, one continuous outcome variable and two moderating variables) and above findings and conclusions about the normally distributed data, the bootstrapping technique was applied. Bootstrapping is a non-parametric resampling procedure, which randomly resamples a set of data with replacement (i.e. when an item is sampled it is immediately replaced) multiple times and statistical conclusions are drawn from this data collection (Henderson, 2005). This technique reduces the assumptions required to validate analyses, and therefore bootstrapping will allow for regression analyses to test the proposed hypotheses, despite of the non-normally distributed data. As recommended by Henderson (2005), this study has used the Bias Corrected
accelerated (BCa) confidence intervals as a basis for statistical conclusions, because it adjusts for bias and skewness of the dataset. Furthermore, the recommendation to resample 5,000 times was followed.
Direct relationships were examined by the use of hierarchical regression. In step 1, the control variables age, gender and education level were entered. In order to prevent
multicollinearity among the independent variables for testing the interaction effect, the variables have been mean-centered (the mean-score is subtracted from each independent variable prior to computing the interaction terms) (Field, 2013). Finally, an alpha level of .05 was used in all statistical procedures.
4. Results
This chapter will first discuss the correlation matrices (See table 3 and 4). Subsequently the results from the regression analyses will be outlined. This chapter will start with presenting the results of the direct relationships between language and payment methods and purchase intention, and continue with presenting the results of the moderation effects.
4.1. Correlation analysis
The tables with the descriptive statistics, correlations and scale reliabilities are presented in table 3 (non-Dutch consumers) and table 4 (Dutch consumers) and are derived using the bootstrapping technique. To be able to distinguish between the constructs of purchase intentions, language importance and payment method importance, the Cronbach alphas (reliability indicators) are calculated and displayed in the table between brackets. In addition, exploratory factor analysis was conducted to test whether the items indeed measure separate constructs. The results are stated in table 5 and indicate that all questionnaire items belonging to purchase intention, language similarity importance or payment method familiarity
importance, report higher factor loadings on that given component, compared to the other components. This provides support for the assumption that the different variables are indeed three separate constructs. The tables 3 and 4 indicate that for both Dutch and non-Dutch respondents the two independent variables of language and payment methods are correlated with each other. A test that had been undertaken to check the severity of multicollinearity resulted in low levels of variance inflation factor (VIF) for both groups, indicating that there is no problem of multicollinearity (See table 6). VIF indicates whether a predictor has a strong linear relationship with the other predictor (Field, 2013).
The number of significant correlations between purchase intentions and the two predictors imply direct relationships. Payment methods and purchase intentions are strongly positively correlated for both the non-Dutch (r= .425, p < .01 ) and Dutch (r=.388, p < .01) consumers. Language and purchase intentions are also positively correlated for the non-Dutch consumers (r=.345, p < .01), implying a second direct relationship. Furthermore, the
independent variable language is positively related with the importance consumers place on language similarity for both groups (r= .427 and .509 for non-Dutch and Dutch respectively,
p < .01), which is not surprising. It is likely that consumers that claim to place high
importance on e-stores offering a native language version, will indeed shop on a native language website. Surprisingly, no significant correlation is found between payment methods and the importance consumers claim to place on payment methods familiarity for either the non-Dutch or the Dutch respondents. Furthermore, for the non-Dutch groups the importance consumers place on language similarity is significantly positively related with the importance consumers place on payment method familiarity (r=.421, p < .01). This implies that both language and payment methods are important predictors for online decision-making processes for this group of consumers. This is however not the case for the Dutch consumers, indicating that they place different, i.e. less importance on language similarity compared to payment method familiarity. Finally, there is a significant positive relationship between age and the importance the non-Dutch consumers place on language similarity, suggesting that the older generation of non-Dutch consumers generally have a greater preference for native language e-stores than the younger generation.
Table 3. Means, Standard Deviations, Correlations and Reliabilities - Non-Dutch
*. Correlation is significant at the 0.05 level (2-tailed) **. Correlation is significant at the 0.01 level (2-tailed)
Note: Language Importance refers to Language similarity importance, Payment Importance refers to Payment Method Familiarity Importance Table 4. Means, Standard Deviations, Correlations and Reliabilities - Dutch
*. Correlation is significant at the 0.05 level (2-tailed) **. Correlation is significant at the 0.01 level (2-tailed)
Note: Language Importance refers to Language similarity importance, Payment Importance refers to Payment Method Familiarity Importance
Variables Number
of items
M SD 1 2 3 4 5 6 7 8
1. Age 105 27.43 4.125 -
2. Gender (0=male, 1=female) 105 .6095 .4902 -.325** -
3. Education 105 4.99 1.282 -.050 .071 -
4. Language (0=non-native, 1=native) 105 .5429 .5006 .180 .049 -.321** -
5. Payment Methods (0=non-preferred, 1=preferred) 105 .8939 .30756 .030 .142 -.050 .211* - 6. Purchase Intentions 105 4.0682 .7287 .193* -.039 .053 .345** .425** (.87) 7. Language Importance 105 2.7381 .9789 .301** -.154 -.157 .427** .273** .104 (.85) 8. Payment Importance 105 3.3905 1.282 .163 -.240* -.280** .073 .134 -.091 .421** (.85) Variables Number of items M SD 1 2 3 4 5 6 7 8 1. Age 102 26.11 3.008 -
2. Gender (0=male, 1=female) 102 .6275 .4859 .136 -
3. Education 102 5.36 .952 -.014 .167 -
4. Language (0=non-native, 1=native) 102 .7549 .4322 -.018 .079 .098 -
5. Payment Methods (0=non-preferred, 1=preferred)
102 .7353 .4434 .017 -.144 .096 .454** -
6. Purchase Intentions 102 3.9165 .7552 .049 .058 -.072 .154 .388** (.89)
7. Language Importance 102 2.5319 .7224 -.005 -.058 -.046 .509** .226* .146 (.78)
Table 5. Exploratory Factor Analysis for Purchase Intentions (1), Language Similarity
Importance (2) and Payment Method Familiarity Importance (3); rotated component matrix for non-Dutch (,00) and Dutch (1,00)
,00 Component 1 2 3 Q41_6 (Recoded) .802 Q41_7 (Recoded) .735 Q13_1 .899 Q13_2 .868 Q13_3 .649 Q13_4 (Recoded) .767 Q15_1 .842 Q15_3 .881 Q15_5 .785 Q15_4 (Recoded) .716 Q17_1 .867 Q17_2 .881 Q17_3 (Recoded) .789
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 5 iterations.
1,00 Component 1 2 3 Q41_6 (Recoded) ,838 Q41_7 (Recoded) ,815 Q13_1 ,805 Q13_2 ,843 Q13_3 ,715 Q13_4reverse ,751 Q15_1 ,801 Q15_3 ,800 Q15_5 ,738 Q15_4 (Recoded) ,737 Q17_1 ,902 Q17_2 ,901 Q17_3 (Recoded) ,894
Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. a. Rotation converged in 5 iterations.
Table 6. Multicollinearity Coefficients (non-Dutch = .00, Dutch=1.00)
Nationality Model Collinearity Statistics
Tolerance VIF
.00 1 Payment_Method .957 1.045
Language .957 1.045
1.00 1 Payment_Method .930 1.076
Language .930 1.076
4.2. Direct effects
The first set of hypotheses (H1a,b) proposed that there is a direct positive relationship between language similarity and purchase intentions for non-Dutch consumers, which is absent for the Dutch consumers. The second set of hypotheses (H3a,b) proposed a direct positive relationship between payment methods and purchase intentions for both non-Dutch and Dutch consumers. Tables 7 and 8 present the results of the regression analyses using the bootstrapping techniques for both the non-Dutch and the Dutch respondents. A hierarchical multiple regression analysis was undertaken, to allow to control for the possible effects of age, educational background and gender on purchase intentions. In the first step the control variables were entered, in the second step the two predictor variables were additionally entered. The overall model fit increased for both consumer groups from the first model to the second model, as indicated by an increase of the adjusted R2 value.
The R2 value is a useful measure of how well the model fits the data. It tells us how much of the variance in the outcome variable is accounted for by the regression model from our sample. But an even better measure to analyse whether the model fits the data is the adjusted R2 . This value attempts to adjust for statistical shrinkage and indicates how much variance in the outcome variable would be accounted for if the model had been derived from the population from which the sample was taken (Field, 2013). Therefore, this study has chosen to present and rely on adjusted R2 values to assess the fit of the models. The hierarchical regression analyses results indicate that the adjusted R2 values have increased from .024 to .255 ( p < .01) and from -.001 to .127 ( p < .01) for non-Dutch and Dutch consumers respectively, which indicates that predictive power was added to the model by the addition of the two predictor variables in step 2.
