Can the Popularity of Higher Education in the UK be explained by the Country-of-Origin Effect? And which Role will Brexit Play?

(1)

Can the Popularity of Higher Education in the UK be explained

by the Country-of-Origin Effect?

And which Role will Brexit Play?

Master Thesis

M.Sc. Double Degree Advanced International Business Management &

Marketing

Rijksuniversiteit Groningen & Newcastle University Business School

Supervisors:

Prof Dr Sjoerd Beugelsdijk

Faculty of Economics and Business, Rijksuniversiteit Groningen Dr Eleftherios Alamanos

Faculty of Humanities and Social Sciences, Newcastle University Business School

(2)

(3)

I Purpose – Attempting to explain the popularity of British HEIs through the COO in light of close competition from other HE exporting countries and Brexit as a factor with potentially great negative impact on the industry. Therefore, the country-of-origin construct was com-bined with the Theory of Reasoned Action to empirically test the framework.

Design / methodology / approach – A repeated-measures design was employed to collect quantitative data through an online questionnaire, distributed to (future) students throughout Europe. The data were then statistically analyzed by using a Friedman’s ANOVA and several linear regressions.

Findings – The COO-effect can be applied to the service industry and explain the popularity of British HE. Moreover, consumer attitudes are a direct predictor of the students’ intention to enroll while the COO, consisting of a cognitive and affective component – is an indirect pre-dictor. Lastly, Brexit is a negatively perceived issue and diminishes the likelihood of EU stu-dents enrolling at UK HEIs. Age and gender to not influence this decision, whereas prior vis-its to the UK and study programs attended in the UK do.

Research limitations – No deeper insights into the specific reasons for EU students prefer-ring one country over another to attend HE in could be derived. Moreover, the sample size was relatively small and the premises for the statistical tests were violated in some cases. Therefore, non-parametric tests were utilized. In addition to that, the theoretical conception of Brexit was subject to some restrictions.

Practical implications – Brexit might hit the British HE industry severely and the UK should not rely on its superiority when counterbalancing the decline in demand from international students. Furthermore, Switzerland might be an additional close competitor for the UK in terms of exporting international education and should be further examined.

Originality / value – There is little to no research on Brexit’s influence on HE and the con-cept of the COO-effect has not been applied to the Theory of Reasoned Action in thus way before. In addition to that, the COO-effect has not been applied to the service sector of HE before and thus complements the body of literature concerning the COO effect in the service industry.

(4)

II

Acknowledgements

(5)

III

6 Discussion ... 43 7 Conclusion ... 45 7.1 Theoretical Implications ... 45 7.2 Practical Implications ... 46 7.3 Limitations ... 47 Bibliography ... 49 Appendix A ... 53 Appendix B ... 59 Appendix C ... 62 Appendix D ... 70 Appendix E ... 85 Appendix F ... 88 Appendix G ... 89 Appendix H ... 90 Appendix I ... 94 Appendix J ... 100 Appendix K ... 103 Appendix L ... 111 Appendix M ... 116 Appendix N ... 121 Risk Assessment Form

(7)

V

List of Figures

Figure 1.1: Non-UK Domicile Students Studying at HE Institutions in the UK ... 2

Figure 3.3.1: Theory of Reasoned Action ... 15

Figure 3.4.1: Proportion of UK Population Voting For and Against Brexit by Age ... 16

Figure 3.5.1: Conceptual Model ... 18

Figure 4.2.1: Set-up of the Online Questionnaire ... 19

Figure 5.2.1: Study Programs Pursued by Respondents as Percentage of Total Sample N = 222 (rounded off) ... 27

Figure 5.2.2: Students Having Visited the UK and Students Having Studied in the UK as Percentage of Total Sample N = 222 (rounded off) ... 28

Figure 5.2.3: Respondents’ Preferred Destinations to Attend HE in as Percentage of Total Sample N = 222 ... 29

Figure 5.4.1: Distribution of the Brexit Variable by Percentage of Total Sample N = 222 (rounded off) ... 33

(8)

VI

List of Tables

Table 2.3.1 Overview of Definitions of the COO Effect………7

Table 3.2.1 Summary of COO Cues……….12

Table 5.2.1: Profile of Participants (rounded off)……….26

Table 5.3.1: Reliability of Scales………..31

Table 5.4.1: Means by Variable and Scenario………..33

Table 5.6.2.1: Mean Ranks for the Variable Intention to Enroll for each Scenario………….36

Table 5.6.2.2: Wilcoxon Test: Monte Carlo Significance Values (1-tailed) for the Variable Intention to Enroll for each Scenario………37

Table 5.7.1.1: Premises of a Regression Analysis according to Field (2009) and their Applica-tion to this Research………..38

Table 5.7.2.2.1: Mediation Effect of Attitude between Country Image and Intention to Enroll for all Scenarios………40

(9)

VII

List of Abbreviations

ANOVA Analysis of Variance

cf. confer

CI Country Image

COO Country of Origin

EU European Union

e.g. exempli gratia

HE Higher Education

HEI Higher Education Institution

p. Page

SPSS Statistical Package for Social Sciences

THE Times Higher Education

TRA Theory of Reasoned Action

UK United Kingdom

(10)

(11)

1

1 Introduction

In a world which becomes increasingly globalized and connected students seek experiences abroad, not only for leisure purposes, but also to attend higher education (HE). The European Union (EU) poses a special case since students are free to move across borders and follow a study program literally wherever they wish. While completing a semester abroad was recog-nized as a factor of distinction by employers only a few years ago, international experience has now become a requirement for many positions (Lipsett, 2008).

These current developments, including globalization, also lead to fierce competition on the supply side: Especially universities in English-speaking countries, such as the United King-dom (UK), are under pressure to attract international students. These shifts call for proactive international marketing measures by higher education institutions (HEIs) and have lead to a so-called marketization of HE with students increasingly being regarded as consumers (Hemsley-Brown & Oplatka, 2006; Hemsley-Brown & Goonawardana, 2007; Hanover Re-search, 2014; Rogers, 2017).

(12)

2 Figure 1.1: Non-UK Domicile Students Studying at HE Institutions in the UK (source: HESA, 2017)

But what makes British universities that popular that the whole nation projects the image of being known for offering excellent HE? One possible but yet unexplored explanation is the so-called country-of-origin (COO) effect or country image (CI). Traditionally, COO is used for products, usually being labelled with the made in sign (Amine, Chao & Arnold, 2005 as cited in Chattalas, Kramer & Takada, 2007). Surprisingly, only very little research has been conducted on the construct’s applicability to the service industry. Scholars like Javalgi, Cutler & Winans (2001), Ahmed, Johnson, Pei Ling, Wai Fang & Kah Hui (2002) and Bose & Pon-nam (2011), nonetheless, were able to prove its significance for the entertainment and leisure industry (e.g. airlines, ski resorts) as examples for the service sector.

(13)

3 A third reason for examining the UK as focal nation in this research is the recent event of Brexit, which has the potential to heavily disrupt the whole industry and may impact the stu-dents’ – especially from within the EU – preference for the UK as destination to attend HE, due to weakened ties and integrity as well as insecurity (e.g. Morgan, 2015; Kuznar & Men-kes, 2017; European University Association, 2017). Therefore, the perception of students from a selection of European countries is most interesting to investigate in this context in or-der to reveal more about the complex mechanisms behind the decision to study abroad, within the EU. In this study, Brexit is thereby incorporated as a subjectively perceived factor, nega-tively influencing the students’ choice to pursue their studies in the UK.

Having outlined the topicality of the issue and the background to this study, a specific re-search gap is to be addressed. Firstly, the general applicability of the COO effect on services rather than products remains highly unexplored to this point in time. This research aims at narrowing this gap down by extending the few existing papers on the entertainment sector (Javalgi et al., 2001) by results on the sector of HE. Secondly, the concept of COO effects regarding services has not been empirically tested in combination with the TRA by Ajzen & Fishbein (1980) before, which forms the basis for this research. Thirdly, the whole event of Brexit being that recent it was not yet feasible to conduct any thorough research on it. Thus, this paper looks more deeply into the possible consequences of Brexit and the UK as COO of HE for the British HE sector by building on the work of Roth & Diamantopoulos (2009).

The aforementioned research gap results in a set of questions this study aims to answer, taking the perspective of the consumer – the student – rather than that of the UK or the universities:

I. Does the COO effect apply to services, more specifically to HE?

II. Does information on the COO of the HE institution influence the student’s attitude to-wards the offered HE and the intention to enroll at that institution?

III. Do positive associations (affective and cognitive) positively affect the students’ atti-tude towards the UK and their intention to enroll at a UK university?

IV. Does the student’s perception of the prospect of Brexit influence the intention to apply to a UK HE institution?

(14)

fol-4 lowed by a thorough review of existing literature on this topic as well as the development of a conceptual framework and research hypotheses. Secondly, having established a solid basis for this research, the methodological approach to collecting and analyzing data as well as the study design will be explained. Lastly, the results will be presented and discussed and impli-cations as well as limitations for further research will be outlined.

2 Definitions

Having outlined the importance of the topic and set the framework for this study, several key terms are to be defined carefully in order to accurately develop an unambiguous theoretical construct for the aforementioned research problem.

2.1 Students as Consumers

With the marketization of HE, students increasingly resume the role of consumers looking to purchase a service – education. According to Molesworth, Scullion & Nixon “HE became a tradable service, based on demand and supply laws under which students became key con-sumers while universities and their staff were the providers“ (2011, p. 142). Proponents of this view state that it is indeed the student who determines the quality of his or her education, while opponents are concerned about the commodification of HE. In addition to that, the pri-vatization of HE institutions is harshly criticized in terms on consumerism. It is argued that only the students with substantial financial resources are given the opportunity to higher grades and high-quality education, which in turn disempowers the student as a consumer and deprives him or her of protection by the market (Molesworth et al., 2011).

(15)

5 & Steenkamp, 1999). It is thus essential to be aware of the subjective nature of this variable and the comparatively low validity of the results.

Nevertheless, university rankings, being mainly based on student perceptions, still form a highly relevant element when it comes to choosing a HE institution for a study program. Cur-rently, the UK has five1 universities in the top 25 of the Times Higher Education (THE) World University Ranking with the University of Oxford being number one (Times Higher Education, 2017). This is likely to put HEIs in the UK in a favorable position internationally. Derived from individual perceptions of students regarding the quality of HEIs, these rankings also relate to attitudes students develop towards such institutions and thus lead to a closer in-vestigation of the field of consumer attitudes.

2.2 Consumer Attitudes

A fundamental element of this research is the encompassing theory of consumer attitudes which belongs to the general field of consumer behavior. The paramount power of attitudes can be attributed to the attitudes’ influence on human behavior and simultaneously their com-position of a diverse range of external and internal factors surrounding the individual decision maker. The concept goes back to LaPiere (1934) and Katz (1937) and is closely related to the theory of COO since the country image usually evokes a certain attitude in the consumer by forming a perception about the quality of the product / service offered (e.g. Lin & Sternquist, 1994).

Attitudes can be defined as “an individual’s internal evaluation of an object” (Mitchell & Ol-son, 1981, p. 318). One reason why the concept of attitudes finds such great application in the prediction of consumer behavior is that attitudes are generally considered as relatively stable over time (Mitchell & Olson, 1981). Thus, they can be viewed as a highly complex, but en-during and rather psychological notion. This approach is consistent with the reasoning of Fishbein & Ajzen (1975) who have substantially shaped the theoretical construct of consumer attitudes by developing the Theory of Reasoned Action. Representing the basis for the con-ceptual model of this study, the TRA will be examined more closely in section 3.3.

1_{University of Oxford, University of Cambridge, Imperial College London, University College London, London}

(16)

6 Grounded in the need for a conceptually clear definition that can be operationalized into a variable, this research adopts the view of Argyriou & Melewar (2011, p. 444) and regards attitudes as “evaluative judgments measured via categorization on a continuum“. This defini-tion is in line with the aforemendefini-tioned one but more specific regarding its measurement. To customize the definition of consumer attitudes for this research framework, the notion that an attitude evolves with respect to a service offered by a service provider based in a distinct (for-eign) country, is added. Consequently, whenever referring to a student’s attitude, this specific attitude is meant. The body of research is ambiguous about the formation of attitudes but sug-gests that attitudes originate from both cognitions and emotions (LaPiere, 1934; Argyriou & Melewar, 2011). As the consecutive review of literature will demonstrate, these two compo-nents will appear as the common ground between attitudes and the COO effect.

2.3 The Country-of-Origin Effect (COO)

The sheer amount of studies in the field of COO that has not accurately defined the term itself prior to conducting the research is startling (see e.g. Bilkey & Nes, 1982; Laroche, Papado-poulos, Heslop & Mourali, 2003; Koschate-Fischer, Diamantopoulos & Oldenkotte, 2012). Furthermore, it is to be noted that some scholars refer to the COO effect as country image, which is more explicitly centered around the consumers’ perceptions (e.g. Laroche et al., 2003; Roth & Diamontoploulos, 2009; Maher & Carter, 2011) and essentially refers to the same phenomenon.

The whole construct becomes increasingly ambiguous with regard to research such as by Ma-her & Carter (2011), who advocate that the country image and product-country-image are two different concepts since the country image corresponds to an attitude the consumer forms about the country and its inhabitants, as opposed to the product-country-image, which refers to an attitude developed regarding products from that specific country. Since the vast majority of research conducted in this field, however, employs both terms country image and product-country-image under the same definition (as table 2.3.1 summarizes and illustrates) and thus blurs the distinctive line between these to a great extent, the concepts become hard to disen-tangle.

(17)

7 „[…] country of origin is not merely a cognitive

cue for product quality, but also relates to emo-tions, identity, pride and autobiographical mem-ories. Such symbolic and emotional connota-tions transform country of origin into an ‘ex-pressive'’ or ‘image'' attribute’.“

Product Verlegh & Steenkamp (1999, p. 523)

“[...] information pertaining to where a product is made”

Product Zhang (1996, p. 51) “The image of countries as origins of products

is one of many extrinsic cues, such as price and brand name, that may become part of a prod-uct’s total image”

Product Eroglu & Machleit (1989) as cited in Laroche et al. (2003, p. 97)

“Country image is the overall perception con-sumers form of products from a particular try, based on their prior perceptions of the coun-try’s production and marketing strengths and weaknesses.”

„[...] country-image can be viewed as an opera-tional concept: a variable, a holistic network, a complex of beliefs, an attitude construct and a triple-component attitude construct.“

Product

Roth & Romeo (1992, p. 480)

Brijs, Bloemer & Kasper (2011) as cited by Adina, Gabriela & Roxana-Denisa (2015, p. 423)

“[…] the country-of-origin cue triggers a global evaluation of quality, performance, or specific product/service attributes. Consumers infer at-tributes to the product based on country stereo-type and experiences with products from that country.”

Service Bruning (1997, p. 60)

“CO is a multi-dimensional construct that

evokes a wide range of cognitive responses.” Service Ahmed et al. (2002, p. 280) “To improve the applicability of COO to

ser-vices marketing, the treatment of COO not as a single cue, but as one among an assortment of cues is especially important.”

Service Javalgi et al. (2001, p. 568)

Table 2.3.1: Overview of Definitions of the COO Effect (own illustration)

(18)

8 as cited in Adina et al. (2015, p. 424) “[c]ountry-of-origin effects are defined as the differen-tial consumer response to a product, due to the country that is perceived as its source“; a defi-nition, which is slightly more specific but does not specify the mechanisms triggered in the consumer by the COO. Moreover, the scholars do not refer to a possible application of the effect to services either.

Bruning is one of the few scholars who did not adopt this narrow, product-centered vision and thus defines the COO as a cue, which “triggers a global evaluation of quality, performance, or specific product/service attributes“ (1997, p. 60). Furthermore, he states that consumers match a product / service with specific attributes that originate from a stereotype of or a past experi-ence with the country of origin, which relates back to the concept of attitudes. This is also in line with the understanding of the COO effect of Diamantopoulos, Schlegelmilch & Paliha-wadana (2011) despite the fact that they only research the effect on products. With this specif-ic perspective on the COO effect, Bruning includes not only the servspecif-ice industry, but also links the framework to the concept of stereotypes and refers to both affective and cognitive forces in play. This more encompassing and less generic definition is therefore taken as the basis for this research.

The COO itself in this research is seen as the respective nation, in which the HE institution (usually a university) has its physical appearance. This country consequently represents the ground for the COO effect. As previously outlined, the terms COO effect, country image and product-country-image essentially refer to the same phenomenon. For reasons of a clear oper-ationalization, it will be refrained from the utilization of product-country-image, whereas both COO effect and country image will be used instead, interchangeably. Bruning’s (1997) defini-tion acts as the basis for this terminological refinement.

2.4 Stereotypes

(19)

9 case of a whole nation. They usually act as mental shortcuts for human beings and, in particu-lar situations, enable a person to make a decision based on cues that fill the gaps of existing external information (Hilton & von Hippel, 1996). This means that with respect to HE, a country like the UK, known for excellent education, is potentially more likely to be selected as a domicile for academic studies compared to a country to which no such stereotype is at-tributed.

Even though these cultural simplifications often evoke unfavorable clichés, they usually hold a certain level of truth and are therefore a very popular tool in international marketing. For instance, Chattalas et al. (2007) state that consumers tend to evaluate unknown products based on national stereotypes and that the COO establishes ties between these stereotypes and “cog-nitive, affective, and normative connotations” (2007, p. 58). Findings by Binsardi & Ekwulu-go (2003) indicate that although the UK is losing some of its competitive edge over universi-ties in the U.S. and Australia – and current rankings confirm this development (Times Higher Education, 2017) – it is still known for superior education across national borders. Interesting-ly, the authors see HE as a product rather than a service as many others, including this re-search, do. Nevertheless, this seems to be due to a lack of terminological distinction and therefore should not substantially impact the further analysis.

Among the countless models aiming to explain the formation and nature of stereotypes (see Hilton & von Hippel, 1996; Chattalas et al., 2007), most of them investigate the act of stereo-typing regarding a group of people rather than a nation and are therefore less applicable to this research. The view all these researchers share, however, is that stereotypes are evoked by a multitude of factors, which makes it almost impossible to determine a single, generally appli-cable theory to explain the mechanism. To not exceed the scope of this research and instead focus on the core subject of interest, this degree of intangibility of the stereotype construct is accepted. As a consequence, stereotypes are regarded as composing a substantial part of the country image but will not be investigated deeper.

3 Theoretical Background and Hypothesis Development

(20)

10 are three different streams of literature relevant to this study, which will be combined in order to gain a deeper insight into the issue and develop research hypotheses. An overview of the key authors and topics can be found in appendix A.

3.1 The Decision-Making Process of Students

Moogan, Baron & Harris (1999) examined the decision-making behavior of British potential students and found that the traditional steps (1) problem recognition, (2) information search, (3) evaluation of alternatives, (4) purchase and (5) post-purchase evaluation are applied. Most distinctive to the decision-making process of a traditional consumer is the consultation of and reliance on the parents’ and teachers’ opinion and expertise. Ivy (2008, p. 289) sup-ports this assessment and adds that „[as] choices available to students grow, life changing decisions about where to study are becoming more complex, with the decision-making pro-cess becoming longer while prospect students assess alternative offerings of competing busi-ness schools“.

On average, the sample of the study by Moogan et al. (1999) selected six different institutions for their studies, which were then narrowed down within the next ten months. Since - at that point - students usually do not know their final grade point average, the decision is made on the basis of current grades. This whole process, consisting of sequential stages, can be catego-rized as extensive problem solving in the case of HE, which can be explained through the large investment, the high involvement of the potential students, the amount of time needed to make the decision and the extensive search for alternatives (Moogan et al., 1999). The authors consider the geographical location of the HEI as potentially important given distance and fi-nancial resources but do not engage in the topic more deeply. In light of Brexit, this result leaves open whether recent economic or political developments in the geographic area of the HE institution would also impact the evaluation stage of the future student since experts on this topic see the Brexit as a serious threat to British HE (e.g. Mayhew, 2017).

(21)

11 supports the development of research hypotheses in this context.

3.2 The Country-of-Origin Effect

First revealed by Dichter in 1962 and backed up by Schooler in 1965, the COO effect builds on the concept of stereotypes and is often used by companies for international marketing pur-poses. Despite the fact that this theory is comparatively old and a broad base of literature chal-lenging the relevance of this phenomenon has evolved (e.g. Samiee, Shimp & Sharma, 2005; Usunier, 2006), a substantial amount of follow-up studies has been conducted and found the impact of a product’s country of origin on consumer choice still to be significant. For exam-ple, Verlegh & Steenkamp (1999) suggest that the country of origin influences not only the consumers’ emotions but also the attitude towards a product as well as its perceived quality and generally matters to consumers. In line with that, “[f]indings from [product-country im-age] studies can provide valuable strategic information to firms exporting their products, manufacturing abroad, and/or competing in their home markets against foreign companies“ (Laroche et al., 2003, p. 97).

Interestingly, the authors also found that the COO has significant implications on the product evaluation even if the consumer is not familiar with the respective product. Diamantopoulos et al. (2011) furthermore reveal a significant indirect impact of COO on purchase intentions through the country image. In a different line is the finding that consumers are even willing to spend more on a product from a country, which is positively known for the respective product (Koschate-Fischer et al., 2012). In this context it is astonishing that, until now, COO research has almost exclusively been conducted regarding products. This study, however, defines HE as a service that can be exported across national borders to EU students. As mentioned in the introduction, a few studies exist that were able to make a case for the validity of the COO effect in the service industry, particularly in the entertainment sector. Nevertheless, there has not yet been any scientific research analyzing the effectiveness of the COO effect regarding HE. Regardless of the specific composition of the COO effect and based on Binsardi & Ekwulugo’s (2003) findings concerning the popularity of British HEIs as well as the by Ver-legh & Steenkamp (1999) suggested impact of the country image, it is hypothesized that:

(22)

12 Reviewing the prevalent literature on COO or country image, it becomes apparent that it is a highly ambivalent and complex construct, which is subject to various different interpretations and theoretical conceptions as well as types of empirical research. Consequently, there are scholars defining the COO as only an informational cue (e.g. Bruning, 1997; Michaelis, Woi-setschläger, Backhaus & Ahlert, 2008; Kumara & Canhua, 2010), while others postulate in-novativeness, design, prestige and workmanship to be dimensions of the country image (e.g. Roth & Romeo, 1992; Bose & Ponnam, 2011).

Studies analyzing the COO effect as a single (informational) cue have, over time, been harsh-ly criticized since more recent investigations show that the COO is such a complex construct that it consists of more than one cue (e.g. Bilkey & Nes, 1982; Javalgi et al., 2001). During the last two centuries, however, a consensus has been reached again, and most scholars adapted the perspective of the country image involving three distinct mechanisms, namely cognitive, affective and normative cues (e.g. Verlegh & Steenkamp, 1999; Laroche et al., 2005; Chattalas et al., 2008; Koschate-Fischer et al., 2012; Adina et al., 2015) (see table 3.2.1). The following table briefly describes each of these cues and gives an illustrative exam-ple.

Type of Cue Description Example Source

Cognitive COO as an indication for (perceived) product quality

Reliability, reduced risk, safety

Verlegh & Steenkamp (1999); Lobb, Maz-zocchi & Traill (2007)

Affective COO evokes an

emo-tional response Status, self-esteem, pride Adina et al. (2015) Normative COO results into

iden-tification or disidentifi-cation

Affinity/animosity, ethnocentrism

Josiassen, Assaf & Karpen (2011) Table 3.2.1: Summary of COO Cues (own illustration)

(23)

cog-13 nitive and the affective component. Both factors thereby result in a country image and thus shape a distinctive attitude in the consumer’s mind. Other research findings (e.g. Dubé, Cer-vellon & Jingyuan, 2003) are consistent with this view, Maher & Carter (2001) being able to confirm this two-component view again, two years later. Based on these findings, a second research hypothesis is derived:

H2: Country cognition and affection together form the construct of country image. The cognitive aspects of the COO effect are generally seen as inferences concerning the per-ceived quality of the product or service, whereby the image of the product / service is matched with the image of its COO. This process is largely based on national stereotypes (e.g. Verlegh & Steenkamp, 1999). According to Maher & Carter, “[t]he cognitive component captures the beliefs held of another country“ (2011, p. 561). In this case, it is assumed that the UK is known for providing superior HE, which leads students to positively evaluate the UK as COO of HE, project this country image on the service and thus prefer studying in the UK over other destinations.

The affective feature, on the other hand, refers to the emotional connotations, the COO of a product / service evokes in the consumer. As indicated in the review of common definitions of the COO effect in section 2.3, these impressions may be based on the consumer’s past experi-ences with the country, either direct or indirect (e.g. through the media). When considering the purchase of a product / service, consumers link their (emotional) memories concerning the respective COO with the product / service, which consequently results in the formation of an attitude towards the product / service from a specific country (e.g. Verlegh & Steenkamp, 1999). If a student, for example, has been on a holiday in the UK before or has gathered in-formation about the nation in the newspaper, this is likely to substantially affect his / her atti-tude towards HE in the UK. A closer examination of these mechanisms leads to the following hypothesis:

(24)

14

3.3 The Theory of Reasoned Action

The Theory of Reasoned Action derives from the field of social psychology and was devel-oped by Ajzen & Fishbein (1980) and Fishbein & Ajzen (1975). Since then, the model has not only been revised by the authors several times, but has found its application in countless sci-entific papers and numerous distinct streams of literature. Since social psychology in general has become a valuable tool in the field of consumer behavior and, on a more general level marketing, the related concept of the COO effect integrates well into this framework. Moreo-ver, consumer attitudes – a main component of the TRA - have already been identified as a notion, crucial to the theory of country image, and thus become an essential variable for this framework.

Generally, the TRA is utilized to predict a certain human behavior based on the individual’s intention to perform this particular behavior (see figure 3.3.1). The intention, in turn, is influ-enced by two distinct factors: subjective norm and attitude. According to Fishbein & Ajzen (1975), subjective norm and attitude each have two further antecedents. While subjective norm is composed of social, normative beliefs and the motivation to comply, an attitude is developed through the evaluation as well as strength of the respective belief. At first sight, these terms seem fairly abstract; a deeper engagement with the framework, however, reveals their rather intuitive nature. Normative beliefs refer to what the individual believes the social environment thinks of his / her actions. Consequently, the motivation to comply describes the extent to which the individual intends to act according to these beliefs or expectations.

The TRA is highly applicable to the construct of the country image since it is believed that the country image influences the consumer’s attitude towards the product / service that is consid-ered for purchase and plays a considerable role in influencing the ultimate buying behavior (e.g. Koschate-Fischer et al., 2012). Thus, the TRA is employed to conceptualize the impact of a HEI’s COO on the student’s intention to enroll at the respective HEI. For reasons of sim-plicity and context-specificity, this research omits the normative path (indicated in grey in figure 3.3.1) and focuses only on the relation between the consumer’s attitude and the behav-ioral intention. Building on Ajzen & Fishbein’s (1980) work, it is assumed – as students are regarded as consumers and believed to make a decision to purchase the service of HE accord-ing to traditional models (Molesworth et al., 2011; Moogan et al., 1999) – that:

(25)

like-15 lihood to enroll at a HEI in the UK, mediated by the EU students’ attitude towards British HE.

Figure 3.3.1: Theory of Reasoned Action (based on: Fishbein & Ajzen, 1975, p. 16) (own illustration)

3.4 Brexit

(26)

16 Figure 3.4.1: Proportion of UK Population Voting For and Against Brexit by Age (source: BBC, 2016)

If recent predictions of Brexit having a rather negative impact on the HE sector within the UK are to be believed, the effectiveness of the country image in this context is to be questioned as well. Based on different perspectives voiced in several articles, observations and papers (e.g. Morgan, 2015; McGrath, 2016; Cambridge, 2017), this research defines Brexit as an external source of (negatively) perceived insecurity about future developments for EU students, inter-ested in pursuing their studies in the UK. This insecurity may arise from several factors whose impact remains yet unknown. Possible examples are potentially increasing tuition fees, stricter visa regulations as well as obsolete partnerships between EU and UK universities and the perception of not being welcome to the UK any longer (Morgan, 2015; European Univer-sity Association, 2017; McGrath, 2016; Cambridge, 2017). For instance, the ERASMUS pro-gram, which acts as a main fund for EU students to pursue international higher education, is about to end in 2020 but may be terminated for the UK prior to that date due to Brexit (McGrath, 2016).

(27)

17 Sampson & Wadsworth, 2016), a broad tendency for a negative perception of Brexit, espe-cially regarding the educational sector, becomes apparent. For example, McGrath states that “it is difficult to find anyone in the higher education sector supporting a leave vote“ (2016). This tendency supports again the rather negative operationalization of Brexit for this research.

Since the UK fears the impact of Brexit due to declining demand for HE in the UK (BBC Ac-tive, 2016), it can be assumed that, vice versa, EU students, generally interested in pursuing their studies in the UK may refrain from this option and decide for a study program else-where. In fact, the Guardian already reports shrinking numbers of international students en-rolled in the UK (Reidy, 2017). Based on these observations and the set definition of Brexit, it is assumed that:

H5: EU students perceive Brexit as a negative event.

H5a: The prospect of Brexit renders EU students who show a positive bias to-wards HE in the UK less likely to enroll at a university in the UK.

3.5 Conceptual Model

(28)

18 Figure 3.5.1: Conceptual Model (own illustration)

Figure 3.5.1 summarizes the aforementioned relations in a visual manner and thus represents the conceptual model to this research. Based on this, the respective variables will be opera-tionalized into specific items and scales and the research hypotheses tested empirically.

4 Methodology

Building on the conceptual model and the according hypotheses developed in the previous section, a specific methodology is set and followed throughout every stage, ranging from the collection of data to the analysis of the results in order to answer the research questions.

4.1 Research Design

In order to scientifically investigate these questions, a research paradigm following the posi-tivist philosophy and, accordingly, a deductive and objective approach was employed. In line with this procedure, the research is of quantitative nature. An advantage of this type of re-search is that the data derived from the sample tend to be generalizable (Collis & Hussey, 2014). More specifically, a repeated-measures experiment (cf. Field, 2009) was conducted via an online questionnaire distributed to students from the EU. Therefore, respondents were asked to score their attitude as well as the likelihood of them enrolling at the displayed HEI for six different, subsequent scenarios, each manipulated by the COO.

4.2 Procedures and Sample

(29)

19 Collis & Hussey, 2014) of European students or pupils that are going to be students in the near future. The online questionnaire was distributed via e-mail and social media platforms during the time between October 12, 2017 and October 23, 2017 in order to effectively reach the desired group of respondents. The survey link was posted in Erasmus and student ex-change groups in order to reach the desired target group and a broader portfolio of EU nation-alities. A drawback of such an approach can be seen in the fact that the sample is a conven-ience sample (Collis & Hussey, 2014). However, the potentially greater sample is expected to countervail this disadvantage. The prevalent literature and other papers in the same field sug-gest a sample size of approximately 150 to 200 to allow for generalizability of the results (e.g. Ahmed et al., 2002; Bose & Ponnam, 2011; Collis & Hussey, 2014). This method is rather feasible in terms of timely and financial aspects. The students were given the opportunity to enter a prize draw for two € 20 gift cards for Amazon.com in order to incentivize them to par-ticipate in the survey.

The requirements for a respondent to participate in the survey were twofold: Firstly, the re-spondent should either be a student or a pupil planning on studying within the next 12 months. Secondly, he or she should be a resident of the EU in order to narrow the sample down to a feasible size and to achieve the prospect of gaining insights into possible impacts of Brexit on the UK’s higher education exports. These requirements are secured through screen-ing questions. If the respondent didn’t fulfill these two conditions, he or she was directly transferred to the end of the survey.

Figure 4.2.1: Set-up of the Online Questionnaire (own illustration)

(30)

20 is composed of the different scenarios. These each include items regarding the perceived qual-ity of the HE, the respondents’ emotions and cognitions towards the respective HEI as well as their attitude. A third block represents a decision scenario; a set of items regarding the percep-tion of Brexit as well as demographic factors compose the end of the survey.

Participation in the questionnaire was entirely voluntary and anonymous. Prior to the begin-ning of the questionnaire, the respondent was assured that his or her answers cannot be traced back to the individual. Additionally, information on the purpose of the study, the institution, the researcher’s name and e-mail address in case of questions, as well as the time it takes to complete the questionnaire was provided. Each respondent had to confirm that he or she was older than 18, had read the information regarding the storage of the data at qualtrics.com and voluntarily took part in the survey. The survey did not contain any insensitive questions. This procedure was chosen in order to ensure that the data collection does not interfere with any form on ethical concern (cf. Collis & Hussey, 2014).

4.3 Measurement

This study roughly adopted the questionnaire by Ahmed et al. (2002), extended by items op-erationalizing the cognitive and affective component of the students’ attitude as well as the perception of Brexit (see appendix D).

4.3.1 Variables

Measuring both affective and cognitive components (as independent variables) accurately is subject to the individual context of each study (Roth & Diamantopoulos, 2009). Since HE in the UK sets a very specific framework, scale items should be carefully adapted to this subject through the use of pre-tests. In order to measure the cognitive component of the country im-age, the items from Crites, Fabrigar & Petty (1994) were adopted, as suggested by Roth & Diamantopoulos (2009). Even though these scales are comparatively old, they have been vali-dated by several more recent studies (e.g. Huskinson & Haddock, 2004) and thus appear to serve as an appropriate tool to measure the cognitive variable. A 5-point Likert scale was em-ployed with opposite cognitions on each end (cf. Crites et al., 1994).

(31)

21 and product-country image, it is crucial to refer back to the set, operational definition of the HEI’s COO as a cue, which “triggers a global evaluation of quality, performance, or specific product/service attributes“ (Bruning, 1997, p.60). This results in a specific operationalization of the cognitive and affective variables, jointly composing the country image. Consequently, both factors were transformed into items, which refer explicitly to the education offered and the HEI in a specific country rather than solely to the COO. Since this approach poses an ob-stacle to the isolation of the COO effect, the respondents were solely provided with the COO of the respective HEI in order to ensure that the emotions and cognitions causally derive from the country image.

To measure the affective component, a selection of the items developed by Richins (1997) was employed since they more accurately capture potential emotions encountered by students regarding HE relative to other emotional measurements used in the literature (e.g. cf. Mano & Oliver, 1993). Here, the respondents were asked to rate the strength of distinct negative and positive emotions on a 5-point Likert scale, ranging from not at all to very much (e.g. cf. Wat-son, Clark & Tellegen, 1988). Both independent variables were manipulated insofar as the COO of the HE, meaning the country in which the university to be evaluated by the students is located, changed for each of the six different scenarios.

The first scenario was kept neutral, without the COO manipulation, since only the photo of a random university and a classroom was provided without a COO given. In the following sce-narios, the respondent was asked to imagine that the HE institution he or she had just seen was now located in a distinct country. To avoid a potential bias resulting from favorable, pres-tigious brands such as the University of Oxford or Harvard and to isolate the effect of the COO, the student was intentionally and solely provided with the COO information.

(32)

22 pre-test were closely tied to the COO (such as language, culture and reputation) (see appendix B), only price, size and study program were chosen as fixed factors.

Provided with this selected information, the respondents were asked to rate their (1) percep-tion of the quality of educapercep-tion offered by the HEI, (2) their emopercep-tions regarding the educapercep-tion and institution, (3) their related cognitions, as well as (4) their overall attitude towards the HEI for each of the six scenarios. The COOs were chosen with respect to the Times Higher Education ranking, being based on students’ perceptions. In line with that, Switzerland and the U.S. were chosen as high-quality COOs for HEIs, whereas Bulgaria and China were se-lected as low-quality counterparts (Times Higher Education, 2017). Five countries were picked in order to set the UK as a COO into a broader context and evaluate the potential pow-er of the UK’s country image for suppow-erior HE relative to othpow-er countries. The initial question-naire by Ahmed et al. (2002) includes nine scenarios, which was categorized as too long for a voluntary student survey.

As suggested by Ahmed et al. (2002), the perceived quality of education was simply meas-ured on a 5-point Likert scale ranging from very low quality to very high quality while the attitude towards each HEI was captured on a similar scale from dislike very much to like very much. The ultimate intention to enroll at one of the displayed universities - the dependent var-iable - was operationalized as the likelihood of enrolling, measured on a 5-point scale ranging from not likely to very likely (cf. Ahmed et al., 2002). This question was positioned in the questionnaire as a block subsequent to the scenarios, containing every HEI.

(33)

23

4.3.2 Pre-Tests

Prior to the distribution of the final questionnaire, two pre-tests were run in order to ensure that the questions and the design in general were well understood and in order to correct po-tential shortcomings (Remenyi, Williams, Money & Swartz, 1998) (see appendix B, C).

The first pre-test was conducted via telephone interviews with 13 students from the EU on October 5, 2017. Telephone interviews were chosen as a preferred method due to the geo-graphical dispersion of the respondents and the need for potentially more extensive, qualita-tive information (cf. Remenyi et al., 1998). The respondents were first asked to state and rate the five most important aspects they consider when choosing a HEI for their studies. After that, the students were introduced to a list of four decision factors, established by Binsardi & Ekwulugo (2003) and asked to rate these according to their applicability on a 5-point Likert scale from not at all applicable to very applicable.

The respondents then listened to the seven self-developed Brexit items and rated them on the same scale. This first pre-test allowed for a list of aspects that students consider when choos-ing a HEI and simultaneously validated the 2002 survey by Binsardi & Ekwulugo (2003) to some degree. These factors could then be rated. Out of the most important ones, only those aspects that were separable from the COO were set as predetermined conditions for the sce-narios in the questionnaire. As there was a high degree of variance within the group of re-spondents for the Brexit items, all seven items were included in the additional pre-test ques-tionnaire (see appendix B).

(34)

24 refinements to the final survey.

Most importantly, the results showed that 88 % of the respondents were highly influenced by the different pictures used to illustrate each scenario, which diminished the ability to conduct manipulation checks. However, the students’ likelihood to enroll at the respective universities roughly matched the Times Higher Education ranking: Bulgaria was perceived as the least attractive COO, followed by China. On the other hand, Switzerland was rated most favorably, while the U.S. and the UK came on second and third place (see appendix C). According to the THE ranking, the UK resumes the leading position, followed by the U.S. and Switzerland. This result may also be explained by the fact that these three countries are extremely close in the THE ranking (Times Higher Education, 2017). Based on these findings, only the neutral scenario was assigned a visual presentation in the final questionnaire in form of two pictures, while the following scenarios were only described by the COO.

In addition to that, the respondents stated that they would find an order of the scenarios from European COOs to non-European COOs more logical, that the poles of the cognitive items should be switched and that the questionnaire is fairly long (see appendix C). All these re-marks were incorporated in the final questionnaire and the items were adapted accordingly. As for the Brexit items, the three with the lowest rating were erased from the block of Brexit items in order not to further prolong the questionnaire. In addition to that, it became apparent that respondents who indicated that they were slow readers, took up to 30 minutes to complete the questionnaire (see appendix C). Since this time frame is less feasible and runs the risk of leading to respondents’ fatigue, thereby impacting the results (Collis & Hussey, 2014), one item of each the cognition and emotion variable was deleted as well. For this procedure, the two items that were perceived as not applicable by some respondents (see appendix C) were chosen in order not to compromise on the variable’s reliability.

5 Results

(35)

25

5.1 Data Cleansing

Prior to running the actual statistical analysis, the dataset was cleansed in several ways. While 373 responses were recorded in total, 71 of these were screened as unfit to perform the ques-tionnaire. 53 respondents indicated not to be a student and 18 respondents did not fulfill the requirement of being a EU resident. These individuals were reached unintentionally through the snowball sampling procedure. Initially, all questionnaires that were not fully completed were deleted, which amounts to a number of further 78 invalid responses. 50 respondents out of the 78 ended the survey after the first two screening questions. The rest - 28 respondents - discontinued the questionnaire after this stage. Having removed these cases from the dataset, the questionnaires completed by respondents being a national of one of the countries repre-sented in the scenarios were taken out in order to avoid an ethnocentric bias. These were three respondents from Bulgaria. As a result, 222 valid responses were recorded and entered the statistical analysis.

In a next step, the indicated study fields were grouped into 19 distinct categories. Additional-ly, the respondents’ nationalities as well as their preferred country to study in were standard-ized. These items were open-ended questions (cf. Collis & Hussey, 2014) and thus subject to individual spelling mistakes, etc. Furthermore, a boxplot analysis of the different variables indicated a relatively high amount of outliers and extremes (see appendix I). These are likely to substantially impact the means of the variables and could thus distort the results in the case of a relatively small sample (Collis & Hussey, 2014). In order not to compromise further on the sample size but to improve the explanatory power of the variables, these outliers and ex-tremes were set to missing in SPSS. Examining the demographic background (age, gender and nationality) of the respondents responsible for these values does not reveal any pattern and leaves the reason for the outliers and extremes unexplained.

5.2 Descriptive Nature of the Sample

(36)

26 % of the sample are women, whereas the male account for approximately 47 %. As expected, the distribution of age shows a large bias towards respondents in their mid-20s. This is partial-ly a natural consequence of the premise of onpartial-ly students and pupils commencing their studies within the next 12 months being allowed to the sample. The residual variance can be ex-plained by the response collection process being a convenience sample.

Profile of Participants N = 222 Gender Female 52.4% Male 47.2% Age 18 to 20 21.7% 21 to 24 52.1% 25 to 29 24.4% 30 or older 1.8% Nationality Austrian 1.8% Belgian 2.7% Danish 1.3% Dutch 8.4% Estonian 0.4% Finnish 2.2% French 5.3% German 56.9% Greek 0.9% Hungarian 3.1% Italian 5.3% Luxembourgish 0.9% Norwegian 0.9% Polish 1.3% Portuguese 0.9% Romanian 2.7% Spanish 3.1% Swedish 1.7% Degree Postgraduate 60.8% Undergraduate 39.0%

Table 5.2.1: Profile of Participants (rounded off) (own illustration)

(37)

conven-27 ience sample – whereas the second largest nationality is Dutch (8 %), followed by Italian (5 %), Spanish and Hungarian (3 %, respectively). All in all, an adequate representation of the different geographic parts from the EU was reached. Ultimately, the respondents were asked to indicate which type of HE degree they considered when filling in the questionnaire. The result is depicted in the last two rows of the table: Among all students, 61 % consider a post-graduate degree while 39 % refer to an underpost-graduate degree.

Completing the demographic section of the survey, the study program of the students was inquired. As shown in figure 5.2.1 the majority of respondents (39 %) pursues a degree in Business and Economics. The area of Design and Arts composes the second largest propor-tion with 10 %, followed by Medicine (8 %), Engineering and Sports (5 %, respectively). In total, 19 distinct areas of study programs are represented in the sample.

Figure 5.2.1: Study Programs Pursued by Respondents as Percentage of Total Sample N = 222 (rounded off) (own illustration)

Complementary to the demographic section two items were employed to score how many students in the sample have ever been to the UK as well as how many have already stayed in

1% 39% 10% 1% 5% 0% 1% 4% 2% 4% 4% 8% 4% 4% 2% 0% 5% 4% 0%

What is your (future) study field?

(38)

28 the UK in order to pursue their studies. Figure 5.2.2 demonstrates these proportions: While 81 % have visited the UK in the past, only 13 % have studied there. Even though these items do not belong to the core investigation on the popularity of British HE, they may be employed in order to analyze the presence or absence of differences in scores between the two groups.

Figure 5.2.2: Students Having Visited the UK and Students Having Studied in the UK as Percentage of Total Sample N = 222 (rounded off) (own illustration)

Prior to commencing the core questionnaire, the students were asked to indicate their pre-ferred destination for attending HE. Since this item was an open-ended question, the respond-ents were not restricted by any predetermined set of possible choices. The results are highly interesting (see figure 5.2.3). Without informing the students about the UK being the focal country of interest to this study, the majority (20.4 %) refers to the UK as their preferred choice. This figure gains even more explanatory power when comparing it to the second fa-vorite country, Germany (20 %), since, unlike Germany, the UK was not one of the nationali-ties included in the sample. Concluding, the possibility that there are students in the sample simply not desiring to study abroad and thus indicating their home country as preferred op-tion, can be excluded in the case of the UK. Besides that, two additional countries (as high-lighted in yellow in figure 5.2.3) employed as COOs in the scenarios were mentioned as study destinations: The U.S. (9.7 %) as well as Switzerland (1.3 %).

No 19%

Yes 81%

Have you ever visited the

UK?

No 87% Yes

13%

(39)

29 Figure 5.2.3: Respondents’ Preferred Destinations to Attend HE in as Percentage of Total Sample N = 222 (own illustration)

In summary, the frequencies (see appendix J) demonstrate that the sample is equally distribut-ed regarding gender and incorporates an acceptable level of diversity with respect to the re-spondents’ study program, nationality, age and preferred study locations, even though biased due to the convenience sample. Moreover, most students have already visited the UK at least once, which means that they are likely to form an accurate attitude towards HEs in the UK based on their cognitions and emotions.

5.3 Generation of Variables

Based on this modified dataset, distinct variables were generated (cf. Field, 2009). While the

Argentina Australia Austria Botswana Canada Denmark Finland France Germany Iceland Ireland Italy Japan Lebanon Netherlands New Zealand Norway Peru Portugal Singapore Slovenia South Africa Spain Sweden Switzerland Taiwan Thailand UK USA 0% 5% 10% 15% 20% 25%

(40)

30 dependent variable intention to enroll and the mediator variable attitude as well as the extent of knowledge about Brexit, favorite country, visited the UK, and studied in the UK were all kept as single-item variables, four new variables for (1) cognition, (2) affection, (3) country image, and (4) Brexit were generated through the means of the distinct items measuring the respective constructs (cf. Field, 2009). A comprehensive overview of all variables can be found in appendix E. The variable perceived quality was omitted from the further analysis as it was primarily intended as an optional, additional indication for the cognitive aspect of the country image due to its utilization in the original – scientifically tested - questionnaire by Ahmed et al. (2002). For this reason, perceived quality could have replaced the cognition var-iable in the case of absence of reliability. However, this varvar-iable proved to be highly relvar-iable and thus was used for the analysis instead due to its multi-item character.

Prior to the generation of the affective variable, the negative emotion embarrassed was recod-ed in order to make its scores comparable to the residual positive emotions (cf. Field, 2009). Therefore, a new variable was recoded with opposite poles, which means that a high score on this item now reflected the feeling of not being embarrassed. In addition to that, the age of the respondents was grouped into for categories in order to improve the compactness of the data. After the generation of variables the three multi-item scales cognition, affection and Brexit were tested for reliability by evaluating the items’ Cronbach’s alpha (cf. Collis & Hussey, 2014). As table 5.3.1 indicates and as expected - since the items composing the cognition and affection variable have been derived from existing literature - these items appear to be (high-ly) reliable, showing a Cronbach’s alpha greater than 0.7 (cf. Field, 2009). However, the SPSS outputs indicate that the overall reliability of the affective variable may be improved by leaving out the emotion embarrassed (recoded) (see appendix H). Therefore, all variables for affection of the six scenarios were generated again, consisting of only three emotions.

Variable Item Deleted Number of Items Cronbach's Alpha

(41)

31 Cognitive UK No 4 0.758 Cognitive BG No 4 0.919 Cognitive CN No 4 0.849 Cognitive US No 4 0.808 CI No 7 0.794 CI CH No 7 0.822 CI UK No 7 0.854 CI BG No 7 0.944 CI CN No 7 0.898 CI US No 7 0.890 Brexit Yes 3 0.627

Table 5.3.1: Reliability of Scales (own illustration)

Since cognition and affection are expected to compose the country image and thus measure the same construct, both variables were combined and transformed into a country image vari-able for each scenario. Prior to testing their reliability, the construct validity was analyzed through the use of correlations. For all scenarios the correlation coefficients between affection and cognition are positive and significant, thus demonstrating convergent validity. When, however, correlated with an unrelated construct, such as age, the significance disappears which leads to the assumption of discriminant validity (see appendix K) (cf. Sin et al., 2005). Regarding the scale reliability all country image variables show a Cronbach’s alpha greater than 0.7 and are therefore reliable.

(42)

32 (2001), a Cronbach’s alpha of 0.6 can be regarded as acceptable in some cases, for instance if the scale is composed of few items. For this reason, the variable will be used for the further analysis.

The specific nature of the Brexit variable requires to additionally test its validity. Since there is no other variable measuring the same construct, its convergent and discriminative validity, however, cannot be analyzed. The same holds true for criterion validity, justified by the fact that no other comparable dataset to test against is available. However, an Exploratory Factor Analysis (see appendix H) demonstrates that all items composing the Brexit variable load onto the same factor which is a sign for these three items measuring the same underlying construct – Brexit (cf. Field, 2009). Therefore, the Brexit variable will be employed as a moderator var-iable to be able to test the respective research hypothesis.

5.4 Distribution of the Variables

Prior to the actual analysis, an assessment of the variables’ descriptive nature (see appendix J) is in order to develop a preliminary understanding of the data distribution. Table 5.4.1 sum-marizes and compares the means of the dependent and independent variables between the different COOs. Highlighted in yellow is the highest mean for each variable. As outlined in the literature review, the popularity of British HE still exists but is subject to international competition. This is underlined by the value of the means in comparison to the other countries as shown in the table. In terms of the means of the independent variable intention to enroll, the UK performs best, which is an indication for H1.

However, Switzerland was scored higher than the UK on affection and total country image. Consistent with the THE ranking, the UK performs best and Bulgaria worst. However, Swit-zerland shows unexpectedly high scores, outperforming the U.S. and demonstrating an even higher mean for cognition than the UK. China, on the other hand, is rated as expected as se-cond last of the six COOs. Since any nationalities appearing in the scenarios were excluded from the sample and the COO was used as the only information to evaluate each scenario, it is highly likely that the differences in means derive indeed from the distinct COOs.

Scenario Cognitive Affective CI Attitude Enroll

Neutral 3.85 3.53 3.69 3.77 3.61

(43)

33

UK 4.08 3.93 3.93 4.09 4.24

Bulgaria 3.04 2.75 2.97 2.83 1.95

China 3.48 3.21 3.36 3.22 2.27

U.S. 3.68 3.21 3.70 3.77 3.63

Table 5.4.1: Means by Variable and Scenario (own illustration)

The frequencies also indicate a mean of 3.65 (see appendix J) for the Brexit variable, which is above average. Furthermore, it is skewed to the right (- 0.291) and is almost normally distrib-uted with a Kurtosis of – 0.031 (see figure 5.4.1). Therefore, H5 can already be confirmed due to this distribution showing that Brexit is perceived as a rather negative event by EU students. Additional insights can be derived from frequencies of the different items composing the Brexit variable (see appendix J). Generally, the students are comparatively well informed about the issue Brexit; the variable’s mean being 3.77. This leads one to believe that they do not score their perception of Brexit solely based on intuition. The highest mean (3.84) was reached by the item That people voted for Brexit makes me angry, which indicates a substan-tial emotional component in the perception based on external sources. I think that it will be more difficult to get a visa after Brexit was scored on average at 3.66, whereas 3.44 is the mean belonging to the item Brexit will negatively impact my future. Consequently, EU stu-dents show a substantial level of concern when considering Brexit and, more importantly, its implications for themselves as individuals.

Figure 5.4.1: Distribution of the Brexit Variable by Percentage of Total Sample N = 222 (rounded off) (own illustration) 1% 6% 33% 47% 13% 0% 10% 20% 30% 40% 50% 1 2 3 4 5

Overall, Negative Perception of Brexit on a 5-Point Likert Scale

(44)

34

5.5 Analysis of the Conceptual Model

Preceding the ANOVA of the different scenarios, the actual construct as proposed in the con-ceptual model in figure 3.5.1 was tested by running correlations between the different varia-bles present in the model (cf. Field, 2009). According to the construct, affection and cognition encountered by a student regarding a HEI form the country image and positively influence the attitude towards that HEI. This attitude, in turn, is predicted to positively influence the stu-dent’s intention to enroll at the respective HEI.

Figure 5.5.1 shows the strength of the (positive) significant correlation at a 0.01 significance level between (1) affection and attitude (r = 0.69), (2) cognition and attitude (r = 0.56), (3) affection and intention to enroll (r = 0.43), and (4) cognition and intention to enroll (r = 0.28) for the case of the UK. The visualization of these correlations clearly shows that, overall, the effect of affection on the two variables is greater but that both cognition and affection have a smaller effect on intention to enroll that on attitude. A further analysis reveals that the two components of the country image as well are positively and significantly correlated with each other (r = 0.58) at a 0.01 significance level (see appendix K). Together with the high level of reliability of the country image variables this correlation coefficient provides evidence for H2, which is therefore accepted.

Figure 5.5.1: Correlation of Affection and Cognition with Attitude in the UK Scenario (own illustration)

In a second step, the potential relation between attitude and intention to enroll was examined.

0,69 0,56 0,43 0,28 0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 UK_AFFECTIVE UK_COGNITIVE

Can the Popularity of Higher Education in the UK be explained by the Country-of-Origin Effect? And which Role will Brexit Play?