Consulting A General Practitioner (GP) Online: A Revolutionary Practice

Consulting A General Practitioner (GP) Online: A Revolutionary Practice

By

Nastassja Tijssen

University of Groningen

Faculty of Economics and Business

MSc BA – Marketing Management & Marketing Research

Master Thesis August 2014

n.tijssen@student.rug.nl

s1632388

Abstract

Healthcare delivery is being transformed by advances in e-Health and by the empowered, computer-literate consumer. With (mobile) technology playing an increasingly important role in consumers’ lives and already being used to manage many aspects of their lives, it is only a matter of time before consumers will (have the ability to) turn to online channels for the delivery of general practitioner (GP) healthcare as well. No longer will it be necessary to make a time-consuming appointment just to find out what to do about that minor, but bothersome, health complaint. Having the ability to make use of such online services can be very beneficial for both patient and doctor, as it not only adds value in terms of time and cost saving, but also in terms of (perceived) convenience in the broadest sense of the word. The question, however, is whether consumers are open to receiving GP healthcare via online channels instead of the traditional face-to-face appointment? And if they are, what should such a service look like?

This study uses a Choice-Based Conjoint analysis to gain more insight into what Dutch consumers would find most important with regard to obtaining online doctor’s consultations. Four different kinds of attributes are examined: the familiarity of the general practitioner (GP), the type of online communication channel, waiting time, and health insurance coverage. As this particular subject has not yet been studied extensively, this study offers interesting and valuable insights for healthcare managers. One of the general insights gained from this study is that the answer to what Dutch consumers find most important regarding online consultations is not an obvious one; relatively equal high importances are given to the communication channel used, the coverage of health insurance, and the familiarity of the general practitioner. Waiting time, on the other hand, seems to be of relatively least importance to the Dutch consumer.

All that may come to my knowledge in the exercise of my profession

or in daily commerce with men, which ought not to be spread abroad,

I will keep secret and will never reveal.

ABSTRACT  ...  1  

1. INTRODUCTION  ...  4  

2. THEORETICAL FOUNDATION  ...  5  

2.1 A CHANGING HEALTHCARE LANDSCAPE  ...  5  

2.2 E-HEALTH CONSUMERS: MORE THAN JUST A PATIENT  ...  6  

2.3 DETERMINING CHANNEL PREFERENCES  ...  7  

2.4 DETERMINING THE ATTRIBUTES FOR ONLINE DOCTOR CONSULTATIONS  ...  8  

2.4.1 Doctors and their patients: The importance of a good relationship  ...  8  

2.4.2 A Thing Called Trust  ...  9  

2.4.3 The Unavoidable Confrontation in Services: Waiting time  ...  10  

2.4.4 Governmental Regulations: Obligatory Health Insurance  ...  11  

2.4.5 A Treasure Chest of Online Channels  ...  12  

2.4.6 Today’s Buzzword: Privacy  ...  13  

2.5 CONCEPTUAL RATIONALE AND HYPOTHESES  ...  15  

3. RESEARCH DESIGN  ...  17  

3.1 RESEARCH METHODS  ...  18  

3.1.1 Conjoint Analysis  ...  18  

3.1.2 Sawtooth Software SSI Web  ...  19  

3.1.3 Latent Class Analysis  ...  20  

3.2 VARIABLES AND DATA COLLECTION  ...  20  

4. RESULTS  ...  21  

4.1 RESPONDENT DESCRIPTIVES  ...  21  

4.2 AGGREGATE MODEL VALIDATION AND ESTIMATION  ...  23  

4.3 SAWTOOTH LATENT CLASS  ...  24  

4.4 DESCRIPTIVES AND PART-WORTHS FOR THE 7-CLASS SOLUTION  ...  25  

4.4.1 Class 1: “It’s all about communication channels, but we hate Social Media”  ...  26  

4.4.2 Class 2: “Not interested unless it’s fully covered”  ...  26  

4.4.3 Class 3: “We would only like to see our own GP, whom we already see very often”  ...  27  

4.4.4 Class 4: “Time is of the essence”  ...  27  

4.4.5 Class 5: “Even if it was fully covered, the answer would still be no”  ...  27  

4.4.6 Class 6: “Only e-mail/Instant Messaging”  ...  27  

4.4.7 Class 7: “A random GP? No thank you!”  ...  28  

5. DISCUSSION  ...  30  

5.1 THE MOST IMPORTANT ATTRIBUTE  ...  30  

5.2 GENERAL (DEMOGRAPHIC RELATED) FINDINGS  ...  30  

6. CONCLUSIONS  ...  32  

7. RECOMMENDATIONS  ...  34  

8. LIMITATIONS AND FUTURE RESEARCH  ...  35  

APPENDIX A: THE ONLINE SURVEY  ...  37  

1. Introduction

‘This morning I woke up with a little red lump on my lower eyelid that hurt whenever I touched it. Having absolutely no idea what it was or how it even got there, I started to ‘google’ to hopefully find some useful answers. Within seconds, dozens of links appeared. But none of them gave me a definitive answer as to what this odd little red lump could possibly be, leaving me with even more questions I wanted answered. So, how about just calling the doctor’s office to make an appointment to have my eye checked out? Obviously, I’m feeling slightly worried about this tiny, but rather distracting annoyance. But it’s just a little red lump. How life threatening could it possible be? Why can’t I just send my doctor a ‘tweet’ with a description of my symptoms and attach a picture of my eye? I seem to be able to use the Internet and Social Media for everything else, so why not for such a little thing as this?’

In the U.S., ‘high-tech’ healthcare has already made it to practice. “Hello Health”, a Brooklyn-based primary care practice, uses web-Brooklyn-based social media to interact between doctors and patients. Video chatting or sending Instant Messages to Dr. Sean Khozin – an internist, who, according to his Facebook-inspired profile page, enjoys “downtempo” jazz and mountain biking – seems to be common practice at Hello Health. In addition, receiving a quick e-mail from him is free of cost, but “cyber-visits” that take longer have to be paid for. And for those patients who have not yet embraced the offered online services, it is still possible to make an appointment to drop by the office for a consultation or for the doctor to come over to their own home.

In the Netherlands, however, patients are currently offered less of an online healthcare experience. Although patients do have the option to find trustworthy and professional information about health related issues on the Internet (e.g. www.dokter.nl and www.dokterdokter.nl), and over half of the Dutch population does so, approximately 80 per cent of Dutch Internet users claim they would also like to make use of online doctor’s consultations if possible (nhg.artsennet.nl). Currently, patients are (mostly) still required to drop by the office to actually have a doctor’s consultation. Even though there are Dutch websites to be found that offer ‘e-consults’ to patients, these are few and offer limited online services (e.g. www.emaildokter.nl and www.dokterdevalk.nl). The limitations mostly concern the fact that only online contact is possible via e-mail and most importantly, this service is available for mainly existing patients of the doctor owning the website (e.g. www.huisdokteronline.nl). Either way, there is growing interest in the use of online applications for doctor-patient interaction, or so-called ‘e-Health’, which has also led to the ‘Dutch Association of eHealth’ (i.e. www.nveh.nl). But despite this growing interest and the use of some of these websites, online healthcare still is not used on a wide scale in the Netherlands, which can be considered a missed opportunity.

To help gain further insights into this relatively new field of research on online GP healthcare consultations in the Netherlands, this study will use a Choice-Based Conjoint study to determine the most important attributes with regard to the use of online channels for doctor consultations strictly from the consumers’ point of view. In other words, if consumers could make use of such online applications, what would they consider to be of main importance? Therefore, the primary research objective addressed in this study is stated as follows:

What attributes are most important in online doctor consultations from a consumer’s point of view? In addition, several consumer factors (e.g. socio-demographical variables and psychological variables) that influence the potential usage of such online applications will be examined, leading to the following sub-research question:

Which consumer factors are key to making the decision of making use of such online doctor consultations?

Considering the little amount of academically published work with regard to the adoption of the Internet and Social Media in the healthcare delivery system, this research will offer some interesting insights that can be taken into consideration with regard to the imminent online development of today’s healthcare structure in the Netherlands. The results from this research can assist healthcare managers in making decisions with regard to the direction online communication tools and Social Media in the healthcare delivery system could, and perhaps should, follow in the future.

Subsequent sections of this thesis will discuss literature relevant to this topic, such as the rise of the e-health consumer, the determinants of channel preference, doctor-patient relationships, and the issue of privacy and trust. Following will be the conceptual rationale and discussion about the attributes that will be of interest in this thesis. After conducting an online survey in order to gather data about the preferences of attributes and other important personal data, the results will be analysed and discussed, and conclusions will be drawn. This paper will conclude with the limitations encountered during this research and several recommendations for the development of online healthcare delivery.

2. Theoretical Foundation

2.1 A Changing Healthcare Landscape

Instant Messaging (35 per cent) and that the use of such social networking sites (SNS) has increased to an overall 57 per cent in 2012 (CBS, 2013). In addition, with Social Media no longer being at an infantile stage, many organisations have also put these versatile tools to use. The internet and social networking sites are no longer just a means of staying in touch with family and friends, instead they have evolved and have redefined how we communicate, offering consumers new and meaningful ways to engage with people, brands and events.

This technology-mediated and online world has not only transformed the methods of communication significantly, but it has also changed the way for consumers to search, gather, and exchange all kinds of information about products and services (Hennig-Thurau et al., 2010). Consumers have become empowered and more computer-literate, which as a result for the healthcare industry has changed the delivery method of healthcare (Ball & Lillis, 2000). With public access to medical and health information having increased, the use of the Internet as a health source has gained popularity, leaving the doctor/patient relationship never to be the same again (Ball & Lillis, 2000; Lustria, Smith & Hinnant, 2011; Wald, Dube & Anthony, 2007). These online health consumers, or otherwise called ‘e-Health’ consumers, are becoming progressively more pro-active in their own healthcare, showing more “interest in and demand for onlineadministrative processes, information-rich health portals, and access to physician web pages and e-mail” (Ball & Lillis, 2000).

2.2 e-Health Consumers: More than just a patient

In contrast to the consumer who wishes to rely entirely on their doctors to make decisions on their behalf, the e-Health consumer wishes for themselves to play a more significant role with regard to their own healthcare, especially now that they are responsible for a greater portion of their healthcare costs (Swenson et al. 2004; Thielst, 2011). There are three main things that interest them (Ball & Lillis, 2000):

1. Convenience: Being well-educated, working full-time, and badly wanting to save some free time, the modern consumer has grown to wish and expect the highest level of convenience with regard to every service imaginable. The healthcare industry, however, has been taking its time to fulfil the increasing demand for convenience, as consumers are still subject to undergoing lengthy waiting time for appointments and inconvenient scheduling that interfere with office hours.

2. Control: A growing number of patients are using the Internet to find useful information about their health conditions, symptoms, or diseases, which they take with them to their doctor’s appointments. This pro-active behaviour shows their need of playing a key role in their own health, which at first may let doctors feel somewhat threatened but actually could benefit both parties in due course.

data that allows them to make comparisons between these services and/or products to help them making the best decisions possible. This also translates to the consumer’s demand of wanting to know more about the performance levels of the different doctors, hospitals and even medication that they have at their disposal before appointments are made or prescriptions are used.

These e-Health consumers will grow in numbers as more people integrate the Internet into their lives to manage parts of their daily activities (Wald, Dube & Anthony, 2007). Concurrently, modern and new technologies will continue to confront consumers with the decision to obtain services via online technologies or traditional non-technological alternatives (Looney, Akbulut & Poston, 2008). Sitting behind a computer to get a simple doctor’s consultation for minor health issues instead of visiting the doctor’s office could be the next big thing in modern healthcare delivery. But in order for this to take place, consumer acceptance of such online channels must be relatively high. Moreover, consumers will have to determine which of these competing offline and online channels they prefer to use for these minor health related issues.

2.3 Determining Channel Preferences

Many researchers have taken a closer look at different factors that affect the liking or disliking of a specific type of channel, specifically online channels (e.g. Black et al. 2002; Broekhuizen, 2006; Gehrt and Yan, 2004; Girard et al. 2003; Li et al. 1999). Regardless of the specific terms that have been used in all of these studies, the factors that influence channel choice can be roughly categorized in the following five groups: consumer-, retailer-, product-, channel-, and situational factors as depicted in Table 1 (Broekhuizen, 2006).

Factor Examples

Consumer Socio-demographics, lifestyle, past behaviour

Retailer Trust/reputation, goods, service

Product Complexity, product risk

Channel Ease of use, accessibility, channel risk

Situational Time availability, mood, weather

2.4 Determining the attributes for Online Doctor Consultations

There are several obvious elements that are needed in order for a doctor’s consultation to take place: the doctor himself, the patient, and a time and place. These same elements are necessary with regard to online doctor’s consultations, except place plays less of a role as technology and the online world have made it possible to communicate with one another across even the vastest of distances. In fact, place in this context is easily replaced with online channel. Needless to say, these four elements will be taken into account with regard to the construction of the conceptual model of this study. In addition, the wishes of the modern e-Health consumer, and the different factors pointed out in the previous subchapter will also be taken into account. The following subchapters will discuss the different elements of the conceptual model.

2.4.1 Doctors and their patients: The importance of a good relationship

As the healthcare industry continues to evolve, so must the relationship between doctors and their patients. It is this particular relationship that is arguably the most important interaction in healthcare (e.g. Shipman, 2010; Street Jr. & Haidet, 2010). Studies in social psychology show that forming meaningful relationships is what lies at the heart of human nature (Baumeister & Leary, 1995; Bowlby 1969, 1973). It is hardwired as one of the core needs of people to build and maintain enduring, positive, and relevant interpersonal relationships (Ibid). For this ‘need to belong’ to be satisfied, the person in question needs to feel a certain degree of caring or even love from the other party involved. But the interpersonal relationship should also be characterized by stability, affective concern, and must prove to be worth continuing into the near future (Baumeister & Leary, 1995). Not being able to establish such feeling of belongingness could lead to considerable distress and cause a number of poor effects on the self (Ibid).

Once the development of interpersonal relationships has taken place, it is of importance to maintain a continuous bond of compassion in order to maintain a satisfactory relationship, as a lack of it will only make for partially satisfactory relations (Baumeister & Leary, 1995). Research even shows that the more often particular individuals are exposed to one another, the likelier a friendship or other form of attachment will develop (Ibid). In addition, upholding lasting relationships with the same individuals instead of engaging in a continuous varying series of companions will be more satisfactory; with every substitution of a relationship, time and effort must be devoted yet again to slowly build up intimacy and shared experiences, which in itself can form an obstacle in building a relationship. Therefore, relationships that endure the test of time bring greater satisfaction and a sense of belonging as opposed to interactions with strangers or new relationships (Ibid). Ending valuable relationships even causes distress and does not go without protest – a response that has proven to be virtually universal (Hazan & Shaver, 1994a).

strong relationship and good communication can result in better decision-making by both parties and improves the quality of care and health outcomes (Dibbelt et al., 2009). But in order to build a prosperous relationship between patients and doctors, several key elements must be present: respect, trust, integrity, honesty, and open lines of communication (Shipman, 2010). These elements foster the patient’s trust in his or her doctor, which will result in better acceptance to adhere to the doctor’s treatment plan (e.g. Fishbein & Cappella, 2006; Godin & Kok, 1996; Janz & Becker, 1984; Shipman, 2010; Street Jr. & Haidet, 2010). Also, they lead to consumers getting a better understanding of what their options are and the state of their current medical condition (Bridgeman & Malinoski, 2009). And with the information that doctors gain simply by letting patients disclose more information about their condition and lifestyle, a more effective treatment plan can be set up to the patient’s own advantage (e.g. Elwyn et al., 2000; Shipman, 2010; Street Jr. & Haidet, 2010).

2.4.2 A Thing Called Trust

Consumers buy things from people they can trust, which is no different in the healthcare industry (Shipman, 2010). Trust is often a vital and prominent element of good, valuable, and fulfilling relationships (Holmes & Rempel, 1989). It is also widely accredited as a fundamental ingredient in doctor-patient relationships (e.g. Anderson & Dedrick, 1990; Farin, Gramm & Schmidt, 2012; Katz, 1984; Krupat et al. 2001; Thom et al. 2004). When discussing trust in this context, it mostly refers to the belief that doctors are open and always tell the truth, are reliable and competent, and put your medical needs above all other considerations, including costs (e.g. Anderson & Dedrick, 1990; Henman et al., 2002; Krupat et al. 2001; Thom et al. 2004). Even ‘being treated like a person, and not a number’ has been mentioned to be associated with trust (Henman et al. 2002). Especially in the healthcare industry, the presence or absence of trust can play a life-changing role: consumers who have more trust in their healthcare provider are more likely to seek care, follow doctor’s orders, and are more likely to revisit for follow-up care than those who have little trust (Thom, Hall & Pawlson, 2004). Also, trusting the doctor and general patient satisfaction seem to correlate significantly and highly (Farin et al. 2012; Thom et al. 2004). It has been shown that satisfaction is a predictor of vital health-related behaviour (i.e. following doctor’s orders and upholding continuity of care), but that trust is even strongly associated with it (Thom, Hall & Pawlson, 2004). Moreover, when doctors and patients hold similar beliefs, it would appear that patients are more likely to endorse and trust their doctors (Krupat et al. 2001; Thom et al. 2004).

Trust is a common theme in patients’ explanations of how they make decisions with their doctor (Henman et al. 2002). This trust – and confidence – also seems to be conveyed through the doctor having a good reputation, and providing the necessary information not only about the medical issue the patient is dealing with, but also what might happen during and after the treatment (Ibid). In addition, reports show a direct link between the degree of trust expressed in the doctor and the role the patient wants to play in his or her treatment; decreased trust showed the patient’s desire to be more involved in his or her treatment decisions (Kraetschmer, Sharpe, Urowitz, & Deber, 2004).

about the trustworthiness of other by interacting with them over time (e.g. Lewicki & Bunker, 1996; Mayer et al., 1995; Ring & Van de Ven, 1994; Shapiro, Sheppard & Cheraskin, 1992; Williams, 2001). The development of trust is proposed to be influenced by perceived trustworthiness and interpersonal affect (e.g. Lewicki & Bunker, 1996; Lewis & Weigert, 1985; Mayer et al. 1995; McAllister, 1995; Ring & Van de Ven, 1994). By experiencing repeated social interactions, people are enabled to update their information about others’ trustworthiness (e.g. Axelrod, 1984; Gabarro, 1978; Lewick & Bunker, 1996; Mayer et al. 1995; McAllister, 1995; Rempel, Holmes & Zanna, 1985; Sheppard & Sherman, 1998; Williams, 2001; Zand, 1972).

Feelings or affect also influence trust as research from sociology, psychology and organizational theory show (e.g. Johnson-George & Swap, 1982; Jones & George, 1998; Lewicki & Bunker, 1996; Lewis & Weigert, 1985; McAllister, 1995; Rempel et al., 1985). It is stated that affective attachments are fundamental for caring and compassionate actions that build trust, and that affective responses (e.g. happiness, irritation, disappointment) influence how people evaluate their feelings for, attachment to, and trust in others (e.g. Jones & George, 1998; Lewicki & Bunker, 1996; McAllister, 1995; Mayer et al., 1995).

2.4.3 The Unavoidable Confrontation in Services: Waiting time

Waiting in line to pay for groceries, waiting on the phone for customer services, or waiting at the doctor’s office for an appointment: waiting has become a universal phenomenon that consumers experience on a daily basis (Lovelock & Wirtz, 2011). The fact is that time is a powerful force to which many parts of life are subject to (Haan, Millsap, and Hartka, 1986; Lytle et al., 2000; Varendi, Porter, and Winberg, 1997). It is a powerful force in general that even “transforms people’s preferences” (Hernandez, 2001; Quoidbach, Gilbert and Wilson, 2013). Due to the fact that time plays a central role in most services, researchers have recommended paying more attention to improving the understanding of how consumers perceive, budget, consume and value time (Lovelock & Gummesson, 2004). After all, services face the challenge that they cannot normally be packed away to be used at a later point in time: they are to be enjoyed in real time (Lovelock & Wirtz, 2011).

implications for consumer decision making (Ibid).

But there is more to time than just temporal length. With the concept of time being the main interest in a variety of studies, several reoccurring aspects of this particular phenomenon have come to light which has lead to the following classification of time:

(1) Objective – the elapsed time as measures by a stopwatch by consumers before being served (Davis & Vollman, 1990; Katz et al., 1991; Taylor, 1994); (2) Subjective – the consumer’s estimation of time waited or ‘perceived’ waiting time (Hui & Tse, 1996; Pruyn & Smidts, 1998). The perceived waiting time depends on objectively measured elapsed time (Antonides et al., 2002; Hornick, 1984; Pruyn & Smidts, 1998); (3) Cognitive – the consumer’s evaluation of the wait as being (or not being) acceptable, reasonable, tolerable, as well as considered to be short versus long (Durrande – Moreau, 1999; Pruyn & Smidts, 1998); and (4) Affective – the emotional responses to waiting, such as annoyance, boredom, stress, joy, etc. (Taylor, 1994; Hui & Tse, 1996; Pruyn & Smidts, 1998).

It would appear from past research that the objective and subjective waiting time negatively influence cognitive and affective responses to having to wait. Consumers often tend to think that the objective waiting time is longer than it actually is (Lovelock & Wirtz, 2011). Delay, for example, - measured by a combination of objective and subjective aspects – has a significant effect on the feelings of anger (Taylor, 1994; Lovelock & Wirtz, 2011). Additional factors playing a role in waiting time satisfaction are the information provided in case of delay and the characteristics of the waiting environment (e.g. Antonides et al., 2002; Hui & Tse, 1996; Pruyn & Smidts, 1998). A psychological effect of having to wait is that consumers who are having to face uncertainty about the length of the waiting time experience significant stress. However, if consumers are informed about the possible waiting duration, this uncertainty can be reduced and the overall stress level experienced can be lowered (Maister, 1985; Lovelock & Wirtz, 2011). Thus, what determines customer satisfaction with regard to waiting depends on the gap between the perception and expectation for the waiting experience (Maister, 1985).

2.4.4 Governmental Regulations: Obligatory Health Insurance

insurance company (Rijksoverheid.nl). It means that the first part of the received healthcare service is to be paid by the insured party. For 2013, this amount was increased to €350 euro (Ibid). The government is prohibited to regulate prices and contractual conditions of private health insurance policies. This means that for the use of online doctor’s consultations, for instance, some health insurance companies can determine themselves whether or not they wish to (fully) cover the use of this service or not. Consumers who are dissatisfied with either their healthcare or with aspects of their healthcare plan itself (e.g. costs or paperwork) show an increased rate of plan switching (e.g. Allen et al., 1994; Grazier et al., 1986; Juba, Lave & Shaddy, 1980; Patrick et al., 1997). This could imply that – concerning costs – if the costs get too high or disagreeable, because/or certain services are not (fully) covered, consumers can get unenthusiastic; having a relatively inexpensive and fully covering health insurance could therefore play a significant role in seeing the doctor. Research also shows that relatively healthy consumers evaluate their health insurance performance mainly in terms of concerns about paperwork, costs, or other unrelated issues that concern the quality of the service being delivered (Allen et al., 1994; Long, Settle & Wrightson, 1988; Mechanic, 1989). On the other hand, consumers with considerable healthcare experience (i.e. the relatively less healthy consumers) are most likely to evaluate the service from a performance perspective; their focus on cost will be less, compared to healthier consumers (Schlesinger, Druss & Thomas, 1999). It would even appear that consumers with chronic illnesses are as likely to seek out information about alternative health plans as are healthier consumers (Knutson, Fowles, Finch, et al., 1996). There is growing evidence that price is an important determinant of health insurance choice decisions (Stromblom, Buchmueller & Feldstein, 2002) and price-sensitivity declines with age (Ibid).

2.4.5 A Treasure Chest of Online Channels

In today’s changing environment where digital technologies have taken over a significant part of people’s lives, the doctor-patient relationship should be paralleled by relying on methods of communication that reach beyond one-on-one communication, including interactions through e-mail and the Internet (Hopkins & American Healthways, 2004). There are various different online channels to choose from to replace the traditional face-to-face doctor’s consultations to examine mild healthcare issues. Regarding this study, a closer look will be taken at the currently popular Social Networking Sites (SNS), Video chatting, Instant Messaging, and E-mail.

Social Networking Sites

or representations of users, which contain personal details about the person in question. Some of the most popular social networking sites in the Netherlands include Facebook, Twitter and LinkedIn, which have the potential to play a useful role in the Dutch health care sector (e.g. Thielst, 2011). These three SNS’s by themselves are good for almost 1.2 billion users all together worldwide as of May 2012. Facebook and Twitter in particular are two of many other social media channels (i.e. blogs and other SNS) that are about speeding up and enriching communication. Instead of one-to-one communication – for instance, talking on the phone – it enables communication from one to many (e.g. via a blog post), or from many to many (e.g. via one’s Facebook wall). Also, Twitter’s most useful characteristic is that it enables real-time activity and tweets can be delivered in a constant stream without crowding an e-mail inbox. In addition, it can be used to build great relationships (Newlands, 2011). Online doctor’s consultations that would make use of these types of online channels would offer consumers the possibility to send a written direct message to their familiar or even unfamiliar doctor in order to receive a consultation. However, this type of communication is open to noteworthy issues regarding privacy as will be discussed in the following chapter.

Video Chatting

Besides the use of social networking sites, doctors and their patients could also turn to web-based applications such as online video telephone services (e.g. Skype). It is a tool that offers the possibility of real-time communication and comes close to the traditional face-to-face consultations at the doctor’s office but adding the convenience of place and potentially time. Although the visual quality of Video chatting might not be most suitable for doctors to examine a patient’s rash, for example, it does allow for decent enough communication for the consumer to explain his or her symptoms and for the doctor to give his or her professional advice.

Instant Messaging and E-mail

Other options with regard to written communication between doctor and patient is the use of Instant Messaging (IM) or E-mail. Similar to Video Chatting, Instant Messaging allows for real-time communication, and unlike social networking sites it can be considered a more private way of communication with one’s doctor. E-mail, on the other hand, used appropriately, offers possibilities that neither traditional consultations, nor consultations via SNS, Video chatting or Instant Messaging directly offer; during the course of a consultation, it very well may happen that consumers forget to ask a certain question or to write down the response of their doctor. The use of e-mail offers the possibility to quickly relate follow-up questions and go over the doctor’s responses. Moreover, considering the flexibility of e-mail, consumers are able to access and read their e-mail whenever convenient (Ball & Lillis, 2000).

2.4.6 Today’s Buzzword: Privacy

transmitted over the Internet. It is the fear of the increase in databases, volume of collected personal data, the possibility of privacy violations, loss of control during the process of collecting, accessing, and how the information could be utilized that worries consumers considerably (Culnan, 1993; Hiller & Cohen, 2001). In order to reassure consumers to some extent and to build trust, most online organisations place their privacy policies on their websites, which generally state that personal information will not be disclosed. But even long before the invention of computers, privacy has been a very delicate subject. The term ‘privacy’ itself has had many definitions, one of them being ‘the desire of people to choose freely under what circumstances and to what extent they will expose themselves, their attitude, and their behaviour to others’ (Westin, 1967). A more recent definition includes ‘the right of an individual to be left alone and being able to control the release of his or her personal information (Chang, Marchewka, Lu & Chun-Sheng, 2005).

With the coming of social networking sites (SNS) also came the ethical dilemma of how people should act on Social Media in general. Social networking sites mainly revolve around openness, networking and informality, putting many private and perhaps sensitive information about people ‘out there’. So where to draw the line to minimise disclosure of such information? What many people might underestimate is that not only can data about consumers be collected during purchase or other transactions, but also by simply monitoring online activities (Mascarenhas, Kesavan, & Bernacchi, 2003). This includes online activities on social networking sites. Such information gathered, stored and used through this cannot be fully controlled by consumers (Sackmann, Struker & Accorsi, 2006). In fact, chances are that many are unaware of such gathering and analysing of information done by websites (Milne, 2000).

Privacy concerns in such context are also defined as ‘SNS users’ feeling of apprehension about their loss of privacy due to the collection of information by SNS providers and/or SNS advertisers’ (Taylor, Lewin & Strutton, 2011). The content that people share on their personal profiles can be used by many websites for targeting purposes, an issue that is considered rather intrusive by many. In fact, one privacy advocate warned of “an incredibly sophisticated, ever advancing system for profiling online users” of social networking sites as Facebook that captures detailed personal information (Tessler, 2009).

exchange of low equivocal information (Rice, 1993; Steinfield, 1986). In other words, when the test results of a patient shows bad news, using web-based tools to inform patients might not be the best option to put to use.

2.5 Conceptual Rationale and Hypotheses

The conceptual model in Figure 1 combines the different factors that have just been discussed. It integrates most of the key elements necessary for general practitioner’s (GP) healthcare delivery (i.e. healthcare provider, consumer, channel/place, and waiting time), which are compatible with most of the factors mentioned in Table 1 (i.e. consumer-, healthcare provider-, channel, and situational factors), and other factors mentioned in chapter 2.4 (i.e. governmental regulations/rising costs).

The attributes that will be tested in this study have been labelled Service Factors in order to distinguish them clearly from the Consumer Factors, which in this study will be used as descriptive variables. Especially with regard to demographic variables, studies have shown that they have proven to be less appropriate for explaining why consumers (do not) use channels (Dabholkar & Bagozzi, 2002; Gehrt & Yan, 2004).

Having chosen the attributes, levels must be assigned to them. This has been done based on studies by Ryan, McIntosh & Shackley (1998), and in consultation with a small focus group of consumers. This has resulted in the following subdivisions and hypotheses:

Familiarity of GP – a regular doctor, a doctor of choice, or a random doctor

Based on the discussed literature, it is expected that consumers are most likely to choose the option of having an online doctor’s consultation given by their own familiar GP, as there will most likely already be a relationship of trust and shared experiences in addition to the fact consumers prefer to maintain good relationships and would rather not have to replace these relationships (Baumeister & Leary, 1995). Additionally, due to the importance of trust in relationships, it is expected that trust will influence the utility of using online doctor’s consultations. Trust itself will be measured by using the Health Care Relationship Trust Scale developed by Bova, Route, Fennie, Ettinger, Manchester & Weinstein (2012). Therefore:

H1a: The ability to communicate with a familiar general practitioner (GP) will have greater preference over unfamiliar general practitioners (GP) with regard to using online doctor’s consultations.

H1b: The preference for having a familiar general practitioner (GP) is more pronounced with a higher level of trust.

Waiting time – 12 hours, 24 hours, or 36 hours

responses to having to wait (Bielen & Demoulin, 2007; Lovelock & Wirtz, 2011). Moreover, delay has proven to significantly affect feelings of anger (Taylor, 1994; Lovelock & Wirtz, 2011). It is therefore expected that consumers will be more willing to use online doctor consultations if they are able to receive them in a short period of time. Especially considering that the online channel should be of competing quality with the traditional way of receiving doctor’s consultations.

H2: Longer waiting time negatively influences the preference for online doctor’s consultations usage.

Health Insurance Coverage – full coverage, partial coverage, or no coverage

Health insurance coverage is obligatory in the Netherlands and insured parties are subject to the goodwill of insurance companies who have the privilege of determining themselves what they will (not) cover with regard to healthcare. Considering how consumers always want to spend the least and get the most, it will be of no surprise that the level of Health Insurance Coverage will play an important role for consumers in determining whether or not to make use of online doctor’s consultations. It is therefore expected that the higher the coverage, the more willing consumers will be to make use of online doctor’s consultations.

Channel – written private (e.g. Instant Messaging or e-mail), written public (e.g. SNS), or video chatting

Technology has provided consumers with the possibility to communicate with one another in an online world through many different channels. There are different ways of communicating through writing, which nowadays besides the somewhat more private options of e-mail and Instant Messaging also include the use of the popular social networking sites (SNS) such as Facebook and Twitter. However, there is also the possibility of making use of Video chatting, which would come closest to the traditional manner of receiving doctor’s consultations in the online world. Depending on their own personal preferences, consumers will make their own choices as to which channel they would consider best for receiving consultations. An important variable that will influence this choice is privacy concerns, as most consumers concern themselves with the reliability of the Internet, especially with regard to private and sensitive information that is cast into this digital realm. Video chatting usually does not involve the storage of data (i.e. recording of video chat conversations) and can therefore arguably be considered to be the channel that is least sensitive to privacy infringement at face value. However, even with a foolproof privacy policy, consumers still cannot be guaranteed that video consults are not being recorded in some way or for some purpose. The uncertainty that can build up from this thought alone can be enough for consumers to not want to use Video chatting for those private doctor’s consultations. Besides, some consumers simply might find the idea of having a consultation via a webcam rather awkward. Because of this, and considering the different advantages and disadvantages of the discussed communication channels, it is expected that consumers will prefer privately written communication. The main reason being that not only will consumers have the ability to write their health related questions at any time and place they please, but they can also go over the doctor’s advice as often as they wish and at any place and time they seem fit. With regard to the measurement of privacy concerns, the scale used by Mary J. Culnan (1993) will be practiced. This leads to the last two hypotheses:

H4a: The option of communicating through written private channels will have greater preference compared to having to communicate through written public or video chatting options.

H4b: The preference of the communication channel is more pronounced with a higher level of privacy concerns.

3. Research Design

3.1 Research methods

The research methods relevant in order to answer the research questions stated in the previous chapter include multivariate techniques, such as conjoint analysis and latent class analysis. Sawtooth Software SSI Web and Sawtooth Software Latent Class Segmentation Module are used to conduct both analyses.

3.1.1 Conjoint Analysis

Conjoint analysis is considered to be a research method that realistically depicts consumer’s decisions as trade-offs among multi-attribute products or services (Huber, 1987). It is best suited to gain perspective of consumers’ reactions to and evaluations of fixed attribute combinations that correspond to potential products or services (Hair et al., 2010). Conjoint analysis grants researchers significant insight into the composition of consumer preferences while maintaining a high degree of realism, and allows for a definition of the most favorable combination of attribute levels for an object, or the seclusion of certain groups of consumers who place conflicting importance on the attributes (Ibid). Depending on the basic characteristics of the proposed research, there are three conjoint methodologies to choose from (Ibid). Table 3.1 below provides a summary of these methods:

Characteristic Traditional Conjoint Adaptive/Hybrid Conjoint

Choice-Based Conjoint Upper Limit on Number

of Attributes 9 30 6

Level of Analysis Individual Individual Aggregate or Individual

Model Form Additive Additive Additive + Interaction

Choice Task Evaluating Full-Profiles One at a Time

Rating Profile Containing Subsets of Attributes

Choice Between Sets of Profiles

Data Collection Format Any Format Generally Computer-Based Any Format Table 3.1: A Comparison of Alternative Conjoint Methodologies (Hair et al., 2010)

3.1.2 Sawtooth Software SSI Web

As stated before, Sawtooth Software SSI Web is used to conduct the choice-based conjoint analysis. With regard to the design and implementation of this specific methodology, researchers differ in their preferences. With CBC analyses, a choice can be made between a so-called ‘fixed orthogonal’ design or a random design (Sawtooth Software, 2009). The first provides all respondents with the same version of the survey. The latter provides all respondents with a unique version of the survey. Both designs have their advantages and disadvantages with regard to efficiency. However, random designs, as compared to fixed orthogonal designs, do allow for all interaction effects to be measured. This can be of particular interest as some interaction effects may not be deemed of interest at the time the study is designed, but actually do become of interest in a later stage of the study. Furthermore, with random designs, biases due to order and learning effects can also be reduced, relative to fixed designs (Ibid). Therefore, for this particular research, a randomized design is used.

Presenting the survey to the respondents

When presenting the survey to the respondents, there is a choice to be made with regard to the presentation of the profiles. The most commonly used profile presentations are full-profile and pairwise comparisons (Hair et al., 2010). The first entails a presentation where each profile is described separately and contains all product or service attributes. The latter involves comparing two profiles at a time with a limited number of the attributes used in the study (Ibid). In this study, the full-profile presentation is used considering it is the recommended method when the number of factors is 6 or fewer, and it offers better realism (Ibid).

Including the None-option

As mentioned previously, a CBC analysis allows for the inclusion of a so-called ‘none-option’ with which respondents can express their lack of interest when the alternatives in the presented choice sets seem unappealing to them. For this particular study, a none-option is also included for the reason of added realism, considering that consumers are not required to choose products or services that do not satisfy their needs. In addition, the none-option can also be used as an indicator for respondents to point out that they would prefer to continue using their current product or service (Sawtooth Software, 2009).

3.1.3 Latent Class Analysis

The second multivariate analysis method used for this research is a Latent Class Analysis. This type of cluster analysis is used for segmentation purposes. It classifies the respondents based on a set of selected characteristics (Hair et al., 2010). In this particular study, the selected characteristics involves demographics and other variables that deal with differences among respondents, such as use and frequency of social media, and their overall health status. The final amount of clusters is determined by the information criteria provided by the Sawtooth Latent Class analysis, these being the CAIC (Consistent Akaike Information Criterion) and the Relative Chi-Square.

3.2 Variables and Data collection

The data has been collected by means of an online survey. The survey is divided into three parts: the actual CBC choice sets, questions concerning the usage of Social Media and other online communication applications, and other questions which are used for segmentation purposes. In order to stay within the acceptable margins of error, a sample size of 195 respondents is used. This allows for sufficient sample sizes per segment after segmentation to reflect on the differences between preferences of the different segments (Hair et al., 2010).

The variables that are used in this CBC analysis are summarized in Table 3.2. There are four different attributes, each with 3 levels. By limiting the number of attributes as well as the number of levels, the respondents should experience little difficulty with completing the task, which in turn adds to the reliability of the conducted research. In total, respondents had to complete 12 randomized choice sets and 3 fixed choice sets, which are included to measure the validity of the CBC analysis design.

In addition to the choice sets, respondents also had to answer several simple questions with regard to their use of Social Media and other online communication applications (e.g. which applications are used and how often). Furthermore, standard questions with regard to demographics (e.g. gender, age, location) and socio-economic variables (e.g. profession and income) have been included for segmentation purposes. For a full overview of the survey, see Appendix A.

Attribute Levels Attribute Levels

Familiarity General Practitioner (GP)

Regular Of choice Random

Waiting Time 12 hours 24 hours 36 hours

Channel E-mail/Instant Messaging Video chatting

Social Media

Health Insurance No coverage Partial coverage Full coverage

4. Results

This chapter discusses the results of the analyses performed with the data collected from the online survey. First, some general descriptives with regard to the respondents will be presented. Second, the model estimation and validation will be addressed, followed by an overview of the latent class analysis that indicates the optimal number of groups for segmentation. Finally, each of these segments will be described in terms of their descriptives and part-worth utilities.

4.1 Respondent Descriptives

A total of N=195 respondents completed the online survey with almost an equal amount of men as women (see Table 4.1) More than half of them are highly educated and work on a full-time basis. Due to the fact that this study focuses on a subject that concerns all those who can visit a general practitioner without necessary supervision (such as a parent), the range with regard to age is high (see Figure 4.1). The average age is 36 years.

Figure 4.1: Population age range

With regard to the self-proclaimed health status of the respondents, more than 70% says to be hardly ever or even never ill (see Table 4.2). Most of the respondents also see their general practitioner about once a year or less.

Variable Level Frequency Percentage Gender Male 94 48,2% Female 101 51,8% Nationality Dutch 192 98,5% Western immigrant 3 1,5% Non-Western immigrant - - Location City 75 38,5% Village 100 51,3% Countryside 20 10,3%

Education Low (max. vmbo, mavo) 12 6,2%

Middle (i.e. havo, vwo, mbo) 63 32,3%

High (i.e. hbo, wo) 120 61,5%

Profession Student 29 14,9%

Full-time employee 106 54,4%

Part-time employee 31 15,9%

Employer 1 0,5%

Self-employed 2 1,0%

Stay home mum/dad 16 8,2%

Unemployed 10 5,1%

Income Less than €600 euros 3 1,5%

€600-€1199 12 6,2%

€1200-€2399 17 8,7%

€2400-€2999 27 13,8%

€3000-€4799 38 19,5%

Not willing to say 98 50,3%

Table 4.1: Demographics of complete sample

Variables Level Frequency Percentage

Health status Basically always ill 10 5,1%

Regularly ill 46 23,6%

Hardly ill 102 52,3%

Never ill 37 19,0%

Amount of appointments About once a month 10 5,1%

About once per 6 months 74 37,9%

About once per year 74 37,9%

Less than once a year 55 28,2%

Table 4.2: General health statistics of complete sample

Variable Level Frequency Percentage

Number of active hours spent on SNS (daily)

About an hour or less 72 36,9%

About 1 ½ hours 41 21,0%

About 2 hours 26 13,3%

More than 2 hours 40 20,5%

Not applicable 16 8,2%

Number of active hours spent on Skype (daily)

About an hour or less 34 17,4%

About 1 ½ hours 4 2,1%

About 2 hours 1 0,5%

More than 2 hours - -

Not applicable 156 80,0%

Number of active hours spent on e-mail/IM (daily)

About an hour or less 94 48,2%

About 1 ½ hours 29 14,9%

About 2 hours 3 1,5%

More than 2 hours 7 3,6%

4.2 Aggregate Model Validation and Estimation

For the purposes of testing the validity of the aggregate model and in order to calculate the part-worths, the Sawtooth Hierarchical Bayes Estimation Module (CBC/HB) was used. Standard settings for estimation were used, which includes 10.000 preliminary iterations, followed by 10.000 further iterations in order to estimate the parameters. Table 4.4 summarizes the parameters and part-worths of the estimation of the aggregate model.

Variable Part- Worth Variable Part -Worth

General Practitioner (GP) Waiting Time

Regular GP 2,90 12 hours 0,92

Random GP -3,00 24 hours 0,33

GP of choice 0,09 36 hours -1,25

Comm. Channel Insurance Coverage

Videochatting -0,05 Full coverage 5,58

Social Media -3,87 Partial coverage -1,34

E-mail/IM 3,92 No coverage 4,24

None 7,86

Table 4.4: Part-worths of total sample population With this data, the relative importance of the attributes can be calculated simply by dividing the range between the minimum and maximum part-worths of an attribute by the sum of all the ranges of all attributes. The results of these calculations are presented in Table 4.5, which indicates that the communication channel with which an online doctor’s consultation is to be given is relatively of greatest importance to the respondents of this particular study. However, the results also show that insurance coverage is of relatively great importance considering the importance percentage of over 30%. The relative importance of the familiarity of the General Practitioner (GP) also shows to be noteworthy with almost 26%, while the waiting time before being able to get an online doctor’s consultation is considered to be of least importance relatively speaking (9,53%).

Attribute Part-worth Min Part-worth Max. Range Importance

General Practitioner (GP) -3,00 2,90 5,90 25,90%

Communication Channel -3,87 3,92 7,79 34,20%

Waiting Time -1,25 0,92 2,17 9,53%

Insurance Coverage -1,34 5,58 6,92 30,38%

SUM= 22,78

Table 4.5: Relative importance of attributes with regard to total sample population

would be expected by chance and the value for a perfect fit (Sawtooth, 2014). The RLH value is also derived from the likelihood of the data. With regard to this study, the RLH for the chance model would be 0,333 as the RLH would be 1/k, in which k is the number of alternatives per choice task (three in this study). A perfect fit would mean a RLH of 1. In this case, the RLH for the aggregate model proved to be 0,779, which can be interpreted as the aggregate model being just better than over two times the chance level.

As for the Average Variance and Parameter RMS, considering that both are a less direct indicator of goodness of fit, they will not be used in this study (Sawtooth, 2014).

4.3 Sawtooth Latent Class

A latent class analysis was performed using Sawtooth Latent Class in order to detect groups of respondents with similar preferences and to estimate average part-worths within these groups. To determine the optimal amount of segments, estimations were done for 1 to 10 classes, all with a random starting seed. Table 4.6 shows the different measures of fit for all classes, which are used to determine the optimal amount of classes.

Log-likelihood Per. Certainty CAIC Chi Square Relative Chi-Square 2 Classes -3275,31341 44,97320 6730,08281 5353,80204 281,77905 3 Classes -2980,47901 49,92655 6234,86455 5943,47083 204,94727 4 Classes -2798,50325 52,98383 5965,36355 6307,42235 161,72878 5 Classes -2633,38181 55,75795 5729,57120 6637,66523 135,46256 6 Classes -2476,14779 58,39955 5509,55368 6952,13328 117,83277 7 Classes -2416,90486 59,39486 5485,51834 7070,61914 102,47274 8 Classes -2348,66499 60,52452 5445,48913 7205,09888 91,20378 9 Classes -2323,31604 60,96720 5487,24176 7257,79677 81,54828 10 Classes -2294,70384 61,44790 5524,46787 7315,02118 73,88910 Table 4.6: Measures of fit for different amount of classes

4.4 Descriptives and Part-worths for the 7-Class Solution

Each of the 7 classes that has been identified has its own set of part-worths with regard to the levels of the four different attributes. An overview of these worths can be found in Table 4.7. These part-worths have already been rescaled for easier comparison purposes. In addition to these part-part-worths, Table 4.8 and Figure 4.2 summarize the relative attribute importance expressed in percentages per class. At first glance, class 3 and 7 appear to be rather similar. However, they differ greatly with regard to the None-option (see Table 4.7). Whereas class 3 shows to have the highest None value, indicating that it is very unlikely it will willingly make use of online doctor consultations as proposed in this study, class 7 seems to have less of an issue to make use of such consultations. Furthermore, only class 2 – a class that values insurance coverage highly and is less interested in the communication channel used – seems to be eager to make use of online doctor’s consultations, as it is the only class that shows a negative None value.

Attribute/Level Part-Worths per Class

Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Class 7

Familiarity General Practitioner (GP) Regular GP 16,041 53,519 113,286 26,959 21,245 30,268 69,212 Random GP -18,601 -40,987 -82,236 -25908 -22,424 -20,930 -104,315 GP of choice 2,560 -12,532 -31,050 -1,050 1,179 -9,338 35,103 Communication Channel Video-chatting 49,410 -8,325 -3,304 -3,077 -8,762 -19,245 26,452 Social Media -113,026 2,744 -44,171 -2,499,7 -49,794 -8,239,9 -58,845 E-mail/IM 63,615 5,581 47,475 28,075 58,556 101,645 32,392 Waiting Time 12 hours 41,717 32,492 2,411 80,307 10,603 11,890 1,672 24 hours 10,018 -1,568 7,857 7,269 1,367 6,407 7,518 36 hours -51,735 -30,924 -10,268 -87,576 -11,971 -18,298 -9,191 Insurance Coverage Full coverage 46,311 118,161 52,191 66,903 127,381 77,981 71,968 Partial coverage 2,638 -8,155 -9,678 -7,631 -29,359 -21,395 -25,412 No coverage -48,950 -110,006 -42,512 -59,271 -98,021 -56,586 -46,556 NONE 4,883 -4,586 174,065 173,705 132,334 121,331 115,463 Table 4.7: Part-worths per class

Attribute Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Class 7

Familiarity GP 8,66% 23,62% 48,88% 13,21% 10,91% 12,79% 43,38%

Communication Channel 44,16% 3,47% 22,91% 13,26% 27,08% 46,01% 22,80%

Waiting Time 23,36% 15,85% 4,53% 41,97% 5,64% 7,54% 4,17%

Insurance Coverage _{23,81% 57,04% 23,67% 31,54% 56,35% 33,64% 29,63%}

Figure 4.2: Relative attribute importance per class

Covering only 4,1% of the total population sample, class 1’s most relatively important attribute is the type of communication channel used for online doctor’s consultations. This class shows the greatest distaste for the use of Social Media, whilst showing the most interest in the use of Video chatting compared to all other classes. The attribute they are least concerned with is the familiarity of the general practitioner. Other characteristics of this class that stand out is that compared to the other classes, class 1 is relatively best represented online expressed in percentages and also one of the three classes that shows to have the least privacy concerns (see Table 4.10). Also, it is the second most highly educated and full-time working segment, and scores highly with regard to their financial status. With regard to their average health status, class 1 seems to be one of the two most relatively healthiest segments and also does not mind waiting longer than a day to get an online doctor’s consultation.

4.4.2 Class 2: “Not interested unless it’s fully covered”

4.4.3 Class 3: “We would only like to see our own GP, whom we already see very often”

Class 3 has the highest average age (43 years) and is also the only one that shows clear preference for their own GP. Relatively speaking, they care very little for the possible waiting time involved with online doctor’s consultations. A very striking result, however, is that even if class 3 could make use of online doctor’s consultations from their own GP, they show the most resistance to using online consultations as a whole as the value of the None-option reveals (see Table 4.7). They also appear to be one of the least online active segments and the ones relatively most worried about their (online) privacy. Furthermore, they are the very top segment that visits their GP most often within six months compared to all other segments and they also seem to be most patient when looking at the preferred average response time (31 hours). Other interesting results with regard to this particular segment is that it consists mostly of women (71,4%, see Table 4.9) and it is the only segment with a very high percentage with regard to wanting to keep their income secret (Table 4.9).

4.4.4 Class 4: “Time is of the essence”

The only class to place waiting time at the top of their preference list is class 4; their average preferred waiting time is 15 hours before getting an online doctor’s consultation. Nevertheless, this class also seems to show a lot of resistance to the idea of using online channels for their doctor’s consultations (Table 4.7). Furthermore, it is relatively speaking the second oldest segment and also the highest scoring segment with regard to not making use of any social networking sites at all. Moreover, it is a segment that has relatively many members who claim to be ill most of the time.

4.4.5 Class 5: “Even if it was fully covered, the answer would still be no”

Although more than 56% of relative attribute importance goes to (full) insurance coverage, it immediately becomes apparent that class 5 is less than interested in making use of online doctor’s consultations according to their score on the None-option (see Table 4.7). An interesting finding is that this segment is dominated by males (72,5%). It is also a relatively high-educated segment with the highest percentage of full-time working members, but of which more than half does not wish to share his or her income information. Although this class places least importance on waiting time, it does wish to receive a consultation rather quickly compared to other segments (average waiting time of 15 hours).

4.4.6 Class 6: “Only e-mail/Instant Messaging”

4.4.7 Class 7: “A random GP? No thank you!”

The relatively youngest (31 years) and most highly educated segment is class 7. What appears to be of least interest for this class is waiting time. Instead, this segment also places the familiarity of the GP first and shows an extremely clear distaste for random GP’s (Table 4.7). They also show the highest feeling of trust toward their own GP’s and are one of the ones most concerned with their (online) privacy. Fascinatingly, this segment also appears to spend more than twice the second highest percentage of time spent on social networking sites: over 56% of class 7’s members spend more than 2 hours on these sites.

Variable Level Class

Class 1 (4,1%) Class 2 (10,0%) Class 3 (14,8%) Class 4 (11,9%) Class 5 (20,5%) Class 6 (23,4%) Class 7 (15,3%) Gender Female 50% 40,0% 71,4% 43,5% 27,5% 63,0% 63,3% Male 50% 60,0% 28,6% 56,5% 72,5% 37,0% 36,7% Average Age 35 36 43 39 36 35 31 Nationality Dutch 100% 100,0% 100,0% 95,7% 100,0% 95,7% 100,0% Non-Dutch - - - 4,3% - 4,3% - Location In a city 75% 60,0% 17,9% 39,1% 27,5% 37,0% 50,0% In a village 25% 40,0% 64,3% 56,5% 57,5% 47,8% 46,7% Countryside - - 17,9% 4,3% 15,0% 15,2% 3,3% Education Low - - 10,7% 26,1% 5,0% 2,2% - Middle 25% 35,0% 64,3% 30,4% 25,0% 30,4% 16,7% High 75% 65,0% 25,0% 43,5% 70,0% 67,4% 83,3% Function Student 12,5% 10,0% 7,1% 21,7% 12,5% 13,0% 26,7% Full-time employee 62,5% 55,0% 46,4% 47,8% 72,5% 45,7% 53,3% Part-time employee 25,0% 20,0% 28,6% 13,0% 10,0% 15,2% 10,0% Employer - - - 2,2% - Self-employed - - - - 2,5% 2,2% -

Stay home mum/dad - 10,0% 14,3% 13,0% - 10,9% 6,7%

Unemployed - 5,0% 3,6% 4,3% 2,5% 10,9% 3,3%

Income Less than €600 - 5,0% - - - 4,3% -

€600-€1199 50,0% 5,0% - 17,4% - 4,3% 3,3%

€1200-€2399 - 30,0% 7,1% 8,7% 5,0% 6,5% 6,7%

€2400-€2999 12,5% 20,0% 7,1% 8,7% 15,0% 13,0% 20,0%

€3000-€4799 25,0% 20,0% 7,1% 21,7% 25,0% 21,7% 16,7%

More than €4800 - - - -

Not willing to say 12,5% 20,0% 78,6% 43,5% 55,0% 50,0% 53,3%

Variable Level Class

Class 1 Class 2 Class 3 Class 4 Class 5 Class 6 Class 7

Type of SNS used None - - - 13,0% 12,5% 2,2% - Facebook 100% 100,0% 96,4% 87,0% 87,5% 97,8% 93,3% LinkedIn 87,5% 55,0% 14,3% 43,5% 60,0% 54,3% 43,3% Google+ 37,5% 20,0% 7,1% 34,8% 22,5% 17,4% 6,7% Twitter 50,0% 40,0% 7,1% 26,1% 35,0% 50,0% 43,3% MySpace - - - - Tumblr - - - - 2,5% - 10,0% Instagram 25,0% 10,0% 3,6% 4,3% 17,5% 39,1% 53,3% Pinterest 12,5% 5,0% 7,1% 4,3% 10,0% 17,4% 26,7% Other - - - - Amount of active hours spent on SNS

About an hour or less 37,5% 50,0% 53,6% 39,1% 27,5% 32,6% 30,0%

About 1 ½ hours 25,0% 20,0% 17,9% 17,4% 22,5% 30,4% 10,0%

About 2 hours 25,0% 15,0% 10,7% 8,7% 12,5% 23,9% -

More than 2 hours 12,5% 10,0% 7,1% 21,7% 22,5% 8,7% 56,7%

Not applicable - 5,0% 10,7% 13,0% 15,0% 4,3% 3,3%

Amount of active hours spent on video-chatting

About an hour or less 37,5% 40,0% 7,1% 13,0% 5,0% 26,1% 13,3%

About 1 ½ hours - 10,0% 3,6% - 2,5% - -

About 2 hours - - - 3,3%

More than 2 hours - - - -

Not applicable 62,5% 50,0% 89,3% 87,0% 92,5% 73,9% 83,3%

Amount of active hours spent on e-mail/IM

About an hour or less 37,5% 65,0% 25,0% 43,5% 47,5% 60,9% 46,7%

About 1 ½ hours 37,5% 25,0% 10,7% 21,7% 15,0% 6,5% 13,3%

About 2 hours 12,5% - - - - 2,2% 3,3%

More than 2 hours - 5,0% - 4,3% 5,0% 2,2% 6,7%

Not applicable 12,5% 5,0% 64,3% 30,4% 32,5% 28,3% 30,0%

Average response time in hours

28 19 31 15 18 24 21

Health status Basically always ill - 5,0% - 8,7% - 10,9% 6,7%

Regularly ill - - 39,3% 4,3% 27,5% 32,6% 26,7% Rarely ill 100,0% 70,0% 46,4% 52,2% 45,0% 50,0% 46,7% Never ill - 25,0% 14,3% 34,8% 27,5% 6,5% 20,0% Amount of doctor’s appointments

A few times a month - - - -

Approx. once a month

- - 7,1% 8,7% - 10,9% 3,3%

Approx. once every 6 months

25,0% 20,0% 50,0% 21,7% 27,5% 19,6% 36,7%

Approx. once a year 25,0% 45,0% 21,4% 34,8% 40,0% 50,0% 33,3%

Less than once a year 50,0% 35,0% 21,4% 34,8% 32,5% 19,6% 26,7%

Average Privacy concerns

3,3 3,2 3,9 3,4 3,3 3,7 3,7

Average Trust 3,8 3,9 4,0 4,0 3,7 4,0 4,4

Table 4.10: Active hours online, health status, and privacy and trust per class