‘If you are dissatisfied, talk to me’ : how general practitioners feel about online rating systems

(1)

‘If you are dissatisfied, talk to me’

How general practitioners feel about online rating systems

Saskia Mekers (10552650)

saskiamekers@outlook.com

Master Thesis

Social Problems, Social Policy

Supervisor: Olav Velthuis

Second reader: Patrick Brown

Amsterdam, 09-07-2018

Number of words

1

_{: 22.165}

(2)

1

Acknowledgements

First and foremost, I would like to thank my supervisor Olav Velthuis for his continued support in the past six months. He has always had faith in me and my thesis, even when I felt somewhat lost in the process. His suggestions and feedback helped me to create better work each and every time.

My sincere gratitude goes to all participants for their time and honesty. Without them, this thesis would not be possible. The interviews carried out gave me a glimpse into a world that I only knew from a patient’s perspective. From the insight gained from the participants, I know one thing for sure: my visits to the doctor will never be the same.

I would like to thank Mr. Van Ek for his willingness to answer various questions about ZorgkaartNederland. The information given by him was truly valuable forcontextualising the platform and learning the rationale behind it. I hope that this thesis may be of some value for the Patiëntenfederatie in the future.

I would also like to thank Patrick Brown for being my second reader and for his contributions in the early stage of this thesis. Many thanks also go out to Shaun for proofreading this thesis and correcting my mistakes in English, as he has been doing since we first met.

Finally, I am very grateful for my family and friends who have encouraged me throughout this process. Thank you all for listening to me and providing me with the confidence to succeed. Most importantly I would like to thank Lars for being my rock throughout this whole process. Thank you for enduring my stressed moments and for struggling together. I cannot wait for us both to be Masters of Sociology and to explore life after graduation. Thank you all so much!

(3)

2

Summary

This thesis is explores online rating systems that evaluate general practitioners and has the Dutch website www.zorgkaartnederland.nl as its main focus. The aim of the thesis is to identify how general practitioners react to their reputation on online rating systems. This aim is inspired by the idea of reactivity by Espeland and Sauder (2007), which can be described as the change of behaviour in response to being evaluated, and the three patterns of that reactivity these authors have distinguished. Online rating systems host online reputations, which are collective evaluations of the functioning of people or organisations which can in their turn help other people decide whether or not they want to work with that actor (Dellarocas, 2010). They thus host public images of general practitioners that are constructed by patients.

Online rating systems are located in a more comprehensive audit culture where services and institutions are increasingly evaluated (Shore & Wright, 2015), and in a time where the doctor-patient relationship is developing into one that has a patient-centred approach (Mead & Bower, 2000). The audit culture is subsequently located in a broader field of new public management policies that stimulate public providers to embrace market-like management (Simonet, 2011). Tonkens, Bröer, Van Sambeek and Van Hassel (2013) notice that this commodification is present in the Dutch healthcare system and distinguish five different reactions that healthcare professionals have to cope with this increasing commodification. Using mainly a snowball sampling method (Bryman, 2012), fifteen general practitioners in Amsterdam and Almere are contacted and interviewed about their experiences with online ratings. The fieldwork shows that online ratings do not play an active role in the everyday lives of general practitioners, and not many signs of reactivity are found (Espeland & Sauder, 2007). There are a few indications of manipulation, or ‘gaming’, reactivity (Espeland & Sauder, 2007), signified by general practitioners who recruit positive ratings to balance a received negative rating. One indication of maximization reactivity (Espeland & Sauder, 2007) is found, but this is done to improve online visibility, rather than to improve an online reputation. It can be concluded that overall, general practitioners do not change their behaviour in their everyday work in response to online ratings, contrasting to what previous research focused on cosmetic surgeons found (Menon, 2017).

This lack of reactivity can be explained by the lack of legitimacy that online ratings have in the eyes of the participants. First of all, the number of online ratings that the participants have received is considered to be too low to mean anything substantial to them and to create a representative online reputation. Promoting online rating systems is met with a lot of resistance. A small number of younger participants are more open to this possibility, however have not acted upon it.

(4)

3 Online rating systems do not seem to be compatible with the relationship between the participants and their patients, in which a patient-centred approach (Mead & Bower, 2000) is central. When patients do not talk about their dissatisfaction but instead put it on the internet, they undermine the relationship with their general practitioner and any opportunity for improvement is lost. The fieldwork also demonstrates a lack of necessity for having a good online reputation because patient loyalty, a characteristic of the doctor-patient relationship (Gérard, François, De Chefdebien, Saint-Lary & Jami, 2016), takes precedence over recruiting new patients.

The fact that online ratings are seen as emotional outbursts of extreme emotions by patients, with a tendency to steer towards negativity also delegitimises online rating systems and online reputations. These outbursts provide the participants with no opportunity for improvement and decrease the credibility of online reputations in the eyes of the participants. Quantifying a patient’s experience into an online rating, a form of commensuration (Espeland & Sauder, 2007), also makes it difficult for the participants to improve, because they do not understand which situation the rating refers to. Knowing who writes the rating and what context it refers to is crucial to the participants.

Finally, the fact that there are already a lot of alternative feedback systems in place for general practitioners added to uselessness of online ratings. Examples of these alternatives are face-to-face feedback, the accreditation process and complaints. These are all deemed more beneficial and useful than online rating systems. The fact that the participants have plenty of alternatives to monitor how they perform can be seen as the main reason why most participants are not aware of their online reputation.

These findings provoke a discussion about the future of online rating systems. It can be argued that online rating systems for general practitioners should be deactivated because they do not provide opportunities for improvement and oppose the current relationship between general practitioners and their patients. Another option would be changing aspects of ZorgkaartNederland to form it into a system that is useful for general practitioners. This includes revising the notification system of the system, creating a feature that allows patients to alter or delete their given rating within a set time frame, or broadening the current trial of verified online ratings (Redactie ZorgkaartNederland, 2018) to general practitioners.

(5)

4

1. Introduction

Just like how people strive for wealth and power, they will strive for an immaculate reputation (Haegens, 2017). This prediction by Dutch journalist Haegens is not that unorthodox, based on what is happening in modern society. Whether it is a review, star rating or a ranking, it is clear that feedback is getting more integrated in the lives of people and in the structure of society. Hotels and AirBnbs are getting rated by the minute, companies are getting reviewed on Facebook, restaurants display certificates from TripAdvisor, and universities are ranked by numerous organisations.

Healthcare is no exception to this development of increasing online evaluation. In the United States, online rating tools for patients are omnipresent in the world of healthcare, with websites such as www.ratemds.com, www.vitals.com, and www.zocdoc.com. The use of RateMDs has grown a 100-fold between its foundation in 2004 and the first month of 2010 (Gao, McCullough, Agarwal, & Jha, 2012). On ZocDoc, some healthcare professionals are even actively sponsoring their profile for it to get seen and therefore attract patients.

In the Netherlands, only a few websites are aimed at evaluating and comparing healthcare, such as www.zorgkaartnederland.nl and parts of www.independer.nl. The first website does not only allow patients to rate healthcare institutions, such as hospitals or general practices, but also individual healthcare professionals such as general practitioners, dentists and physiotherapists. This is different from the well-known and more broad Google reviews, which provide an opportunity to rate any organisation of your choice. ZorgkaartNederland is designed to specifically evaluate healthcare institutions and providers by collecting patients’ experiences. With one million visitors per month ("Wat is ZorgkaartNederland?”, n.d.) its scope is extensive. The rating opportunities for patients in the Netherlands are thus becoming more widespread and more personalised by the ZorgkaartNederland platform.

Evaluation tools have in other contexts been described as “consequential and controversial” (Espeland & Sauder, 2007, p. 2) and are not implemented without hesitance. Espeland and Sauder (2007) show that on the one hand, these tools are seen as enhancing the transparency and functioning of organisations, while on the other hand, these tools are argued to have negative externalities such as a loss of quality. Quantitative research about ZorgkaartNederland in relation to quality has been done by Stehmann, Goudriaan, In ’t Veen, Kollen, and Verheyen (2016). They conclude that the system cannot be viewed as a valid measure of the quality of care, mainly because the ‘reliability threshold’ (which will be discussed in chapter 2) of nine ratings per healthcare professional is only met in a small percentage of cases. However, this statistical conclusion is not where the discussion about these platforms ends. There is much more that can be said about online rating systems.

(7)

6 This thesis will focus on the effects of online rating systems on individual healthcare professionals and their everyday practices. These systems are a highly personalised form of healthcare evaluation. What kind of impact do these personalised rating systems have on the people that are being rated and evaluated? This thesis is focused on Dutch general practitioners, mainly because these professionals work in a highly personal branch of healthcare. General practitioners often have long-lasting relationships with patients, sometimes for multiple decades, which are signified by a continuity of care and the development of a bond between the general practitioner and the patient (Gérard, François, De Chefdebien, Saint-Lary, & Jami, 2016). Considering this fact, what will the reactions of general practitioners to online rating systems be? Do they influence their work in any way? Studying how general practitioners feel about and react to the fact that patients can write about their experiences online can be extremely valuable in understanding the implications of online rating systems on both a professional and individual level.

All the above considerations form the basis for the question guiding this thesis: How do general practitioners react to their reputation on online rating systems? The word ‘react’ must be interpreted as a derivative of the concept of ‘reactivity’ by Espeland and Sauder (2007). Reactivity is a form of reflexivity and can in short be described as the idea that “individuals alter their behavior in reaction to being evaluated, observed, or measured” (Espeland & Sauder, 2007, p. 6). This concept will be explained further in the theoretical section of this thesis, chapter 3. Attention will also be paid to theemotional reactions that general practitioners may have in response to online rating systems and that underlie the presence or absence of reactivity.

Shortly after commencement into the fieldwork, it became clear that most general practitioners do not quite ‘react’ actively to their reputation on online rating systems. As it will become clear in chapter 3, this conflicts what might be expected from the literature on this topic. The general practitioners are not actively changing the way they do their job and are not striving for an immaculate reputation, like Haegens predicts (2017). The main reason for this is that they question the legitimacy of these websites. However, the reasons for this lack of legitimacy are extremely interesting and important to showcase. For this reason, a second question was formulated during the research: What are the reasons for general practitioners to not see online rating systems as legitimate entities?

Because there are less online rating systems of healthcare professionals in the Netherlands than in the United States, it could be the case that not all general practitioners are aware of their online reputation, or that they only look at one specific system and not at others. That is why the following subquestion is formulated: To what extent are general practitioners aware of their online reputation?

(8)

7 Looking for an answer to this question can help say something about the size of the influence of the rating phenomenon on Dutch general practitioners.

Before discussing the theoretical framework, methodology, and of course the most important findings of this research, we need to take a closer look at these online rating systems and what measures they encompass. The ZorgkaartNederland website will be the focus of this, since it is a one of a kind platform in the Netherlands.

(9)

8

2. The Dutch case: ZorgkaartNederland

ZorgkaartNederland was founded in 2009 by the Patiëntenfederatie (which is the main Dutch interest group for patients)2_{, a publisher and a health insurance company, but is currently owned by the}

Patiëntenfederatie (Stehmann et al., 2016). Its character can be considered a unique one. To compare, Google and Facebook are platforms on which all types of people, institutions, services and so on can be rated. The Dutch website Independer allows for comparison of insurances, mortgages, pensions and healthcare institutions (general practices included). In personal communication with the Patiëntenfederatie, it is made clear that the website is founded as a platform for patients to share their experience with healthcare and is contributing to transparency in the healthcare system (A.-J. Van Ek, personal correspondence, April 6, 2018). By giving patients a voice, the Patiëntenfederatie also hopes to give the patient influence in the healthcare system (A.-J. Van Ek, personal correspondence, April 6, 2018). What makes ZorgkaartNederland so interesting and valuable to research is the fact that individual general practitioners can be rated. In a way, this website can be interpreted as the Dutch equivalent of the personal rating websites in the United States. By doing quantitative research, Stehmann et al. (2016) found that of all medical specialists working in the hospitals that are listed on ZorgkaartNederland, 56,6% is rated. Ultimately, all these written experiences are argued to help other patients to make a well-advised choice between different healthcare providers (A.-J. Van Ek, personal correspondence, April 6, 2018). As discussed, the platform attracts a large number of visitors each month which currently amounts to one million (“Wat is ZorgkaartNederland?”, n.d.). So, judging by the numbers, this platform has a significant role in the Dutch healthcare system.

On ZorgkaartNederland, each healthcare professional has an overview page that shows his or her average rating as a grade between 1 and 10 and the total number of ratings that that professional has received. All individual ratings can be consulted as well as the average rating per year. All these ratings do contain a date but are published anonymously. Every individual rating on ZorgkaartNederland contains a ‘quantitative’ part in which patients give a grade between 1 and 10 on six indicators: appointments, treatment, interaction with employees, information, listening and accommodation. In comparison, Independer lists 9 indicators on which people can rate a certain general practice, while Google uses a star system without any specified indicators. Every individual rating also contains a ‘qualitative’ part where patients are obliged to write something about their experience. Each individual rating can also be given a ‘thumbs up’ (but no ‘thumbs down’) if it is deemed useful by the reader. The healthcare professional is

(10)

9 able to respond to a given rating. Healthcare institutions and practices can request three different packages (two of them are paid packages) at ZorgkaartNederland which they can use to analyse reviews statistically, add information about their institution on the website and promote the website to their patients with flyers and posters (“Pakket voor praktijken”, n.d.).

Overall, the ratings and reviews seem to range from very positive to extremely negative, and the number of them differs greatly across practitioners. The average grade of all general practitioners, when writing this paragraph, is an 8.23_{. But what does this grade mean? Does this mean that Dutch general}

practitioners are doing their job well? The Patiëntenfederatie argues that ZorgkaartNederland was not founded to be a quality measurement but to be a website on which patients can read and write experiences to make an informed choice between healthcare providers (A.-J. Van Ek, personal correspondence, April 6, 2018). However, there does seem to be a correlation between the aggregated rating and official quality measures of hospitals which increases when the number of received ratings increases (Geesink, 2013). The Patiëntenfederatie argues that this correlation is stable when professionals have received nine ratings and organisation have received thirty ratings (A.-J. Van Ek, personal correspondence, April 6, 2018), what I refer to as the ‘reliability threshold’. If a professional has 9 or more received ratings, and thus reaches this threshold, the circle that shows the average rating will get dark blue instead of grey. While it is unsure how many hospitals in Geesink’s (2013) research meet the reliability threshold, Stehmann et al. (2016) find that for specialists working in hospitals, the threshold is only met in 7,6% percent of all cases. At the time of writing this paragraph, 2097 of 9049 general practitioners on ZorgkaartNederland (23%) meet the threshold of nine ratings.

What about general practitioner that do not meet the threshold? Can their online ratings simply be dismissed as ‘unreliable’ when they come in low numbers? While these are no official quality measure, only correlated with it when the number is high enough, they do seem to have significance. The online ratings on ZorgkaartNederland construct a certain online image of that healthcare professional. Something that normally stays behind closed doors, the interaction between a patient and a doctor, is now made visible through online ratings. While the professional may be capable of responding to a rating, (s)he has no say in what patients can and cannot write.

This thesis focuses on every online reputation of general practitioners, not just the ones that meet a reliability threshold. It will go beyond statistical analyses and use a qualitative approach to investigate

3_{This grade is of course prone to change since new ratings are added every day. The following weblink}

can be consulted if the reader is interested in an up to date average grade:

(11)

10 the phenomenon of online rating systems and their impact. The next chapter will cover theoretical insights from previous research that may be of use in understanding the impact of an online rating system such as ZorgkaartNederland on the feelings and practices of general practitioners.

(12)

11

3. Theoretical framework

3.1. Audit culture

Online ratings are not a standalone phenomenon. Feedback systems have grown more and more ingrained in society throughout the years. Shore and Wright (2015) even argue that in modern society, an “audit culture” (p. 421) is prevalent. Auditing is the constant monitoring and evaluation of the performance of institutions, services, programmes and many more on the basis of numbers and calculations (Shore & Wright, 2015). Online rating systems are located within this bigger audit culture. More specifically, the systems that are the focus of this research can be seen as public evaluations of the performance of healthcare professionals. While auditing has its roots in the financial world, it is omnipresent in modern society and measures the quality, efficiency and transparency of many different actors (Shore & Wright, 2015).

The increased measuring of performance can be traced back to the 1980s when many countries solidified neoliberal ideas in ‘new public management’ policies (Simonet, 2011). In general, these policies aim to apply market logic to public services to improve their quality and to cut governmental costs (Simonet, 2011). In other words, this means that public services are encouraged to behave more like private firms. This includes competing with others and treating users as customers with certain wants and needs that are to be fulfilled (Simonet, 2011). The latter aspect seems to be prevalent in the Dutch healthcare system, since the work of healthcare professionals is increasingly becoming commodified and more ‘demand steering’ to serve the patient well (Tonkens, Bröer, Van Sambeek, & Van Hassel, 2013).

Behind this new form of governing lies the idea that subjecting healthcare professionals to market-like auditing will motivate them to improve the quality of their service (Shore & Wright, 2015). After all, evaluating the performance of professionals with the use of numbers, makes it easier to compare them. This can in its turn increase competition among those professionals. Indeed, Shore and Wright (2015) found that many services are nowadays managed in a way that aims to increase their performance outcomes. Online rating systems can be viewed as both a result and an engine of this increasing commodification of healthcare. On the one hand, they are a form of auditing that measures if healthcare professionals are performing well on patient satisfaction. On the other hand, their public character can stimulate professionals to compete with each other on the basis of numbers by increasing their performance outcomes on these systems. Next to online rating systems, there are many more forms through which the performance of healthcare systems can be audited.

(13)

12 Kumar and Prasuna (2016) differentiate between government, people’s and social audit in which the former is the most governed, regulated way of checking the outcomes of a certain service by using, for example, standardised quality measures. In the Netherlands, the Health and Youth Care Inspectorate4_,

a formal inspection authority by the government, takes on this role. Every healthcare provider and institution are subjected to this inspection. Multiple instruments can be used by the inspectorate to influence and stimulate healthcare providers if they do not meet the set quality standards (Hout, Stibane, Frederiks, Legemaate, & Robben, 2010). Professional and institutions can even be sued by the inspectorate to a disciplinary board, although this does not happen on a regular basis (Hout et al., 2010). In contrast to this type of audit, Kumar and Prasuna (2016) distinguish the people’s audit, which measures customer or user satisfaction. This form of auditing can thus give a glimpse into the perceptions of receivers of healthcare. The ratings that are central in this research seem to fit into the group of people’s auditing measures, since they focus on the experience of patients with general practitioners. One characteristic of the people’s type of audit is however that the results are generally not accepted (Kumar & Prasuna, 2016). This is different with ZorgkaartNederland. In a letter to the parliament, the former Dutch Minister of Health Edith Schippers (2015) has spoken out positively about ZorgkaartNederland and has described it as a platform that contains reliable information. The website also gets some financial support from the government and is part of a plan to improve the transparency of healthcare services in the Netherlands (Schippers, 2015). Does this means that ZorgkaartNederland exceeds a people’s audit and can be considered a form of a social audit?

Kumar and Prasuna (2016) understand a social audit as an integration of multiple perspectives and as an active collaboration between affected people, other stakeholders, and the government. Hill, Fraser, and Cotton (1998) have a slightly different definition of social auditing and describe it as “an open, participatory process of dialogue which assesses and reports on the social relationships and performance of the health centre” (p. 1481). Professionals can respond to received ratings on ZorgkaartNederland, as established in chapter 2, but after looking through the website, I noticed that this only seems to happen in a small number of cases. Because of this observation, we cannot speak of a direct, reciprocal dialogue between the general practitioner and patient on the system. And although the former Minister of Health, Welfare and Sport supports ZorgkaartNederland as a way of enhancing transparency of healthcare, the platform is not conceptualised by her as an official way of measuring the performance of healthcare (Schippers, 2015). ZorgkaartNederland does not facilitate a multi-perspective evaluation of healthcare

(14)

13 and thus cannot be seen as a full-fledged social audit programme in terms of either Kumar and Prasuna (2016) or Hill et al. (1998).

So, ZorgkaartNederland can be viewed as a people’s auditing form (Kumar & Prasuna, 2016) that evaluates if healthcare professionals provide good service to patients, which fits the idea of new public management (Simonet, 2011). However, the governmental recognition and support does raise the impression that the platform is and can be more than ‘just’ patient experiences. The Patiëntenfederatie itself has the ambition of posting quality measures on their platform in the future to support patients even more in choosing a certain healthcare professional or institution to go to (A.-J. Van Ek, personal correspondence, April 6, 2018).

An example of such a quality measure is accreditation. This is a yearly-repeated process that practices can sign up for. There are different organisations that practices can contact to do the accreditation with. The NHG-accreditation is a well-known one, that involves both a pre-audit, by which practices develop a plan for improvement, and a regular audit (“NHG-Praktijkaccreditering”, n.d.). If a practice passes that regular audit, they receive a quality mark and develop a new plan of improvements that they can implement in the coming year (“NHG-Praktijkaccreditering”, n.d.). This accreditation focuses on a lot of aspects such as hygiene, transparency of the practice’s policy plans, and privacy, but also requires the practices to distribute surveys amongst patients to ask about their experience with the healthcare provider (NHG Praktijk Accreditering b.v., 2015). While every healthcare practice is subjected to the evaluation by the Health and Youth Care Inspectorate, it is not mandatory to follow the accreditation process.

Understanding the idea of auditing is an important step in trying to understand the place of online rating systems in current society. The functioning of a general practitioner is measured by the Health and Youth Care Inspectorate and via the accreditation process that already includes patients’ experiences. The question now arises as to how online rating systems stand in relation to these forms of evaluation that are already in place. Are online rating systems, from a general practitioners’ view, part of this audit culture of measuring and evaluation? Or do they mean something else to general practitioners? The next subchapter will discuss one way of making sense of online rating systems; as them forming a reputation.

3.2. Reputation and reputation systems

To properly research the reactions of general practitioners to their online reputation, we need to delve deeper into the actual idea of a reputation. Dellarocas (2010) describes this concept as follows:

(15)

14 Reputation is a summary of one’s relevant past actions within the context of a specific community, presented in a manner that can help other community members make decisions with respect to whether and how to relate to that individual (and/or the individual’s works). (“What is a reputation system?”, para. 2).

Using this definition, the online reputation of a general practitioner includes everything he or she has done during interactions with patients that is written down on a rating website. This summary of actions can then be used by patients to make certain decisions such as choosing which general practitioner they want to register at.

However, a reputation does not only exist online. Other forms through which a reputation can be formed are for example judgement by other professionals or word of mouth. What makes online reputations unique is the fact that they are located on reputation systems, which are systems that facilitate evaluation and comparison of different reputations (Dellarocas, 2010). ZorgkaartNederland is such a reputation system. On the website, every general practitioners’ reputation is portrayed in a quantified way, as well as every individual rating by standardised grading scales. The reputation of a general practitioner is simplified to a grade between 1 and 10. Google is an even more simplified system, since it does not use any indicators but just has the option to give a star rating between 1 and 5. This presentation of someone’s reputation makes it easily understandable and comparable for many people. What makes these systems even more special is the fact that they are highly public. The online reputations are available to everyone, provided they have a computer. This public character is crucially different than other reputation mechanisms such as offline word of mouth. It allows anyone to write about their experience whenever, wherever they want and allows anyone to read those experiences. Together, the patients are creating an image of the general practitioner as a nice, trustworthy, angry, unfriendly, et cetera, person.

Because of the calculative and public character, the reputation of a general practitioner can also be described as a “collective measure of trustworthiness” (Beldad, De Jong, & Steehouder, 2010, p. 866), just like that of a vendor, restaurant, hotel and so on. It is an image constructed by patients that forms a prediction of what someone can expect when they go to a general practitioner that they are not familiar with. So, when speaking about the online reputation of general practitioners, it can be operationalised as the standardised grades or the stars that exist on public online rating systems such as ZorgkaartNederland and Google.

But reputations do more than just ‘exist’; they generate effects. Power, Scheytt, Soin, and Sahlin (2009) formulated the effects of a reputation eloquently: “Like trust, reputation is a complex relational

(16)

15 concept – a quality of social actors’ perceptions, and perceptions of perceptions. Perceptions may or may not be true, but they are social facts which generate actions and reactions” (p. 307). Reputations can thus be conceptualised as social facts that have certain effects. One of those effects is reactivity (Espeland & Sauder, 2007), which chapter 3.4 will elaborate on. There are some other interesting insights from previous research about the effects of professionals’ reputations that need to be discussed first.

Research conducted in Chile suggests that having a good reputation as a healthcare professional has a positive statistical effect on the patients’ loyalty towards and trust in that professional (Torres, Vasquez-Parraga, & Barra, 2009). In other words, this means that the better a healthcare professional is evaluated by third parties, the more trusting and loyal its patients will be. Torres et al. (2009) measured the reputation of professionals by letting patients answer a set of questions on a questionnaire. Research conducted in the United States, also took online ratings into account when assessing the reputation of specifically cosmetic surgeons (Menon, 2017). For cosmetic surgeons, their reputation and online ratings are important, mainly because their line of work is highly consumer-driven in the United States (Menon, 2017). It seems that in their case, having a good reputation can lead to more customers. Menon’s (2017) research shows that many surgeons feel vulnerable because of online ratings as they may be damaging to their reputation, both in the online and offline sense of the word. Rationally and statistically, having an immaculate reputation is arguably beneficial to healthcare professionals. It seems however that emotionally, not all healthcare professionals are in favour of online reputations.

While the findings of Menon (2017) do give some insights, it is difficult to predict what the general practitioners that are the focus of this research will think about the existence of these reputation systems. As to my knowledge, no previous academic research has been done about exclusively general practitioners and these systems. There has been media coverage on this topic in which general practitioners and other specialists gave their opinion on online reputations. For example, the chairman of a regional cooperation of general practitioners stated in an interview that he does not mind getting rated, just like many of his colleagues (Trompert, 2017). However, seven years earlier, a previous chairman of that same cooperation saw nothing beneficial to online rating systems (“Huisarts: rapport opweb is niks”, 2010). This chairman also points out that an online rating is commenting on the relationship between a doctor and patient that ought to be built on trust (“Huisarts: rapport opweb is niks”, 2010), which suggests that he may see these systems as undermining this relationship.

By seeing online ratings as building blocks for an online reputation, we can understand them as simplified images that at times seem to benefit but at other times seem to impede a professional’s reputation. The

(17)

16 reputation of a general practitioner can be viewed as the public image of that practitioner, which is made visible on reputation systems i.e. online rating systems, and is constructed by patients. Torres et al. (2009) researched the connections between reputation, loyalty and trust. However, loyalty is not considered by Menon (2017), which makes sense because cosmetic surgeons are not as involved in a patients’ medical life as general practitioners. When trying to theoretically make sense of general practitioners’ feelings towards online reputations, patient loyalty and trust are important to take into account. Because like the chairman suggests, trust seems to play a part in the relationship between the general practitioners and their patients (“Huisarts: rapport opweb is niks”, 2010). This notion of trust is also emphasised by academic research, as we will see in the next chapter. Before being able to study how general practitioners react to their online reputation, we need to look in more depth into the relationship that they have with their patients and the role that online rating systems may play in this.

3.3. The doctor-patient relationship

A helpful differentiation of the types of doctor-patient relationships is made by Szasz and Hollender (1956) by their three models of “activity-passivity”, “guidance-cooperation” and “mutual participation” (p. 586-587). They argue that most of the healthcare provisioning falls under the second category. In this relationship, the patient comes to the professional with its problem, acknowledges that the professional has more power than the patient based on its medical knowledge, and agrees to what is said and recommended by the professional (Szasz & Hollender, 1956). This is an image that some people might have when they think about general practitioners in the early- and mid- 20th_century.

In the last decades however, there has been a significant change in the relationship between healthcare professionals and their patients, especially in the Western world (Kaba & Sooriakumaran, 2007). The third model of mutual participation (Szasz & Hollender, 1956) is argued to become increasingly central in modern society (Kaba & Sooriakumaran, 2007). One distinctive feature of this type of relationship is that the professional and the patient have equal power in the interaction, as opposed to a hierarchical relationship in the first two models where the healthcare professional has and typically performs his or her power over the patient (Szasz & Hollender, 1956). This emancipated idea of the patient fits the development in healthcare to be more accommodating to the patients’ wishes (Tonkens et al., 2013), like discussed in chapter 3.1. The emergence and growing use of the internet also plays a significant role in this changing power dynamic because patients have access to medical information and are more informed than ever before (Kaba & Sooriakumaran, 2007). Patients can thus proactively look for possible causes and treatments online and discuss those findings in the consultation with the general practitioner.

(18)

17 However, power is not the only factor that characterises the relationship between a general practitioner and its patients. The relationship that these people have is often longstanding and continuous, which makes patient loyalty a distinctive feature of that relationship (Gérard et al., 2016). It is difficult to pinpoint exactly how and when patient loyalty arises because the concept is strongly associated with other factors, but it is mostly interwoven with trust (Gérard et al., 2016). In a way, relying on the general practitioners’ knowledge and following it without hesitation like Szasz and Hollender (1956) describe in their second model, can be considered a form of trust. However, they do not explicitly discuss this concept in their theory. Recent scholars like Gérard et al. (2016) show that trust is indeed a very important attribute of the relationship. Bending (2015) also pleas that trust must be recognised as a central feature of the medical relationship, specifically trust in the fact that the general practitioner works in the best interest of the patient without any covert conflict of interests. By putting the insurance of this type of trust central in laws, the autonomy and power of the patient will be strengthened, according to Bending (2015). As discussed, online rating systems allow patients to assess the trustworthiness of a general practitioner without having to meet with the professional. But online rating systems can also undermine trust. If loyal, satisfied patients look up the online reputation of their general practitioner and that is a bad reputation, it is possible that their trust in the professional erodes.

Conceptualising the doctor-patient relationship as one that is characterised by more than just power fits the current trend of the emancipation of the patient through a patient-centred approach in healthcare (Mead & Bower, 2000). According to Mead and Bower (2000), not only power but also the responsibility of becoming healthy is equally divided between the healthcare provider and receiver in a patient-centred approach to healthcare. They also argue that patient-centred approach includes a “therapeutic alliance” (p. 1090), which means that it recognises the importance of a likeable, positive interaction between the professional and the patient. Another aspect that is interesting to mention is the awareness of the individual character and situation of the patient in this approach (Mead & Bower, 2000). Online rating systems showcase individual experiences and can heightened this awareness. These characteristics of the patient-centred approach (Mead & Bower, 2000) seem far away from the traditional hierarchical, one-sided relationship in which a general practitioner is a person with knowledge and power.

Evidently, these changing relational dynamics have implications for the role of the healthcare professional. The medical profession is traditionally based on a form of professionalism that it is characterised by a high education and skill-level, and by the fact that professionals control the practices of the specific profession (Noordegraaf, 2007). Online rating systems seem to be interfering with the latter aspect, because they evaluate the practices of the professional from a patient’s point of view, even though

(19)

18 patients have no official control over them. What is deemed to be good practice by the general practitioners can be contested, or of course supported, by the opinion of patients which are made visible via online rating systems.

Szasz and Hollender (1956) argue that “patients often choose physicians not solely, or even primarily, on the basis of technical skill. Considerable weight is given to the type of human relationship which they foster” (p. 592). So, when patients are choosing a general practitioner, they will try to find one that fits their expectations and preferences. Online rating systems are useful for this as they make the relationship between healthcare professionals and patients visible to a broad public. They showcase the reputation of a certain general practitioner which a person looking for a new general practitioner can use to assess whether that professional fits the patient’s needs and wants or not.

This brings forth the question that Noordegraaf (2007) poses: “What does it mean to be professional in changing times?” (p. 781). The professionalism of general practitioners in modern day society to be no longer based on their knowledge and on the trust of the patient that the professional knows best. It is rather based on their skill to come to an agreement about the best course of action and on the match that a patient and professional have (Szasz & Hollender, 1956; Mead & Bower, 2000). But this relational change should not only be regarded as a decline of professional power (Tonkens et al., 2013). Tonkens et al. (2013) found different ways in which the professional can change its ‘traditional’ form of professionalism to adapt to current changes in the field of health. This idea of adaptation will be further discussed in chapter 3.4.

To summarise, the relationship between general practitioners and their patients is nowadays characterised by loyalty, trust, equality, and attention to both persons as subjects who influence each other. Throughout the years, the patient has become more central in this relationship. Online rating systems can be construedas tools to enhance the patient-centredness in general practice. To research how general practitioners react to online ratings and reputations, it is important to take the above theorisations about the doctor-patient relationship into consideration. Do general practitioners see this patient-centredness as a positive development or not? How do they see the impact of online rating systems on their relationship to patients? As Tonkens et al. (2013) suggest, the change in the relationship between general practitioners and their patients also sparks a change in the professional identity of general practitioners. The next subchapter will look at possible reactions to and changes because of online rating systems and the increasing patient-centredness.

(20)

19

3.4. Reactivity and changing professionalism

As is clear, this research not only tries to map how general practitioners feel about their online reputation and the increasing evaluation of their work but also how they react to it. Or in other words, what they do because of their online reputation and if or how they try to influence it. The concept of reactivity, coined by Espeland and Sauder (2007) is very useful to use in this regard. They describe it as the idea that “individuals alter their behavior in reaction to being evaluated, observed, or measured” (p. 6). Espeland and Sauder came up with the concept of reactivity by studying the influence of rankings on law schools. This idea can be applied to many actors and organisations that are evaluated in modern life, such as universities and their professors, restaurants and their chefs, books and their writers, films and their directors and so on. Using this concept, it can be easier to understand and frame the reactions of general practitioners to their specific online reputation and to the mere possibility of being evaluated by patients.

According to Espeland and Sauder (2007), there are two mechanisms that generate reactivity: self-fulfilling prophecy and commensuration. Especially the latter is useful to apply to the online rating systems that are central in this thesis. Commensuration is the transformation of qualitative aspects into measurable, quantitative indicators (Espeland & Sauder, 2007). In this case, it is the translation of the patients’ perception of and experience with a general practitioner into measurable components: the six indicators. As discussed in chapter 3.2, this commensuration creates an online reputation of the general practitioners. Transforming real-world aspects into numbers gives new meaning to it because it puts information and knowledge in an easily understandable, simplified context (Espeland & Sauder, 2007). Espeland and Sauder (2007) argue that this commensuration leads to different immediate and longer term responses. This thesis will question what their online reputation means to general practitioners and how they change because of it.

Espeland and Sauder (2007) observe three patterns of reactivity in law schools: maximization, redefinition and manipulation. The first two are changing how money is distributed and changing work tasks both to boost ratings (Espeland & Sauder, 2007). In my research, it could for instance be the case that general practitioners decide to allocate more money to the design of their office or to doctors’ assistants to reduce the waiting time if they find that there are a lot of negative ratings about those aspects. It could also be that they adapt their behaviour in the doctor’s office to accommodate to specific indicators that are embedded in the online ratings, which fits the second effect of reactivity, which is redefinition (Espeland & Sauder, 2007). The manipulation pattern of reactivity is also called a ‘gaming reaction’, which means people are trying to get a high score without substantially improving the quality of that which is being rated (Espeland & Sauder, 2007).

(21)

20 Reactivity has also been researched in contexts other than that of law schools. When studying cosmetic surgeons and their online reputation, Menon (2017) observes that “some surgeons were reluctant to say no to patients outright, no matter how impossible a patient’s request, for fear of how they might be reviewed” (p. 6). She finds however that they are not changing their behavior to directly boost ratings but rather to prevent people from being upset and by extension from writing a negative rating that could possibly affect their online reputation. It is thus a less proactive reaction than the redefinition idea of Espeland and Sauder (2007). Menon’s research also shows that some cosmetic surgeons ask satisfied patients to write a positive online review. But in her research, not all participants changed their behavior to prevent bad reviews. The reasons that are given for this are that the online ratings do not say anything about the skills of the surgeons and that patients are not seen fit to comment on this medical skill (Menon, 2017).

One would expect that it may be easier for customers in restaurants to say something about food than for patients to review a medical procedure. However, Beuscart, Mellet, and Trespeuch (2016) find that customers are not seen as a qualified group to comment on the quality of restaurants. Although the restaurant owners devalue and delegitimise these customer reviews, they are much more involved with their online reputation than the surgeons. While cosmetic surgeons check their online reputation and only take measures when things get out of hand (Menon, 2017), many restaurant owners tend to respond to (negative) ratings or ask customers to give a rating on a more regular basis (Beuscart et al., 2016). And while cosmetic surgeons do not change their practice in response to individual online ratings (Menon, 2017), more than half of the restaurant owners do say they change small organizational aspects because of them (Beuscart et al., 2016). This comparison shows that different people react quite differently to the fact that they are being evaluated by online rating systems.

Menon’s (2017) research is embedded in an interesting context because cosmetic surgery is a strongly consumer-driven branch of healthcare in the United States. Tonkens et al. (2013) see that this consumer-driven character is also spreading throughout the Dutch healthcare system. They argue that the commodification of healthcare influences relationships between doctors and patients because the latter are more and more seen as people that need to be pleased, as consumers. Hirschman theorises that when consumers are dissatisfied with a certain service or exchange, they can choose between the options of loyalty, exit or voice (as discussed in Singh, 1990). In Hirschman terms, this means that they can either stay with the provided service or good, leave and go look for another, or voice their dissatisfaction with the aim of stimulating a change in the provider (as discussed in Singh, 1990). The latter option can thus function as a stimulus for professionals to improve their work.

(22)

21 Online rating systems for ‘patients-as-consumers’ add an interesting view on Hirschman’s theory (as discussed in Singh, 1990). On the one hand, they can be perceived as a new way through which patients can ‘voice’ their dissatisfaction. In healthcare, this can be done in many ways, for example by filing a complaint, discussing the problem with the professional, or filling in patient surveys (Brüggemann, 2017). On the other hand, online rating systems make it easier for patients who ‘exit’ to look for a higher rated one which may serve them better. When information about healthcare providers is transparent, the likelihood of patients exiting instead of voicing grows (Brüggemann, 2017). This seems to be fitting to online rating systems, as they provide transparent, simplified information of healthcare providers. Brüggemann (2017) does address that voice usages are based on the expectation that professionals will be stimulated by it to improve their practice. This thesis can find out if that is indeed the case for general practitioners.

Tonkens et al. (2013) argue that this increasingly consumerist and demand-steered character of the healthcare system, embodied by online rating systems, brings forth different ways in which healthcare professionals redefine their profession: through entrepreneurialism, activism, bureaucratisation, pretending and performing. In the first form, professionals are “typically more concerned with efficiency, competition, and patient-friendly behaviour” (Tonkens et al., 2013, p. 375). Connecting this to this thesis, it could be the case that professionals use online ratings to look for improvements to their ‘service’ of patients. Tonkens et al. (2013) do argue that there are professionals who adopt this type of professionalism while opposing the commodification of healthcare. Other professionals are also critical but actively resist the commodification of healthcare, and stay true to their traditional form of professionalism, which is described as activism (Tonkens et al., 2013). Some professionals are adopting a bureaucratized working method which means that professionals follow bureaucratic procedures, regardless of how tedious it may be, to account for what they do (Tonkens et al., 2013).

The last two types of professionalism seem to overlap with some forms of reactivity. The pretending form of professionalism means that professionals are “pretending to work by the rules, thereby disguising the act of following your own rules in order to protect the quality of your work” (Tonkens et al., 2013, p. 378). This fits the gaming reactivity that Espeland and Sauder (2007) observed in law schools who sometimes report different numbers to ranking organisations “than to their accrediting body” (p. 30). The performance type is described as a deliberate change in behaviour of the healthcare professional in interactions with patients, in response to the possibility that they are being watched and evaluated (Tonkens et al., 2013). This seems congruent with the general definition of reactivity (Espeland & Sauder, 2007). This last type of professionalism is already observed in Menon’s (2017) research on

(23)

22 cosmetic surgeons. I expect it to be a reaction that the participants of this thesis also express, mainly because of the similarities between the consumer-driven character of Dutch healthcare (Tonkens et al., 2013) and that of cosmetic surgeons in the United States.

3.5. Conclusion

This theoretical framework provides many insights that are of help in this research on general practitioners and their online reputation. Online rating systems such as ZorgkaartNederland or Google create and host an entirely new form of reputation (Dellarocas, 2010); one that is public and quantified. Their existence can be viewed as a result of a new public management that has started in the 1980s, in which competition between and customer-satisfaction about public service providers is central (Simonet, 2011). The growing focus on the patient seems to translate to more equal and patient-centred interactions of general practitioners and their patients (Mead & Bower, 2000). However, the effects of online rating systems on the daily practices of general practitioners are not known.

The three patterns of reactivity (Espeland & Sauder, 2007) and the five forms of changing professionalism (Tonkens et al., 2013) can help with interpreting the responses of general practitioners to their online reputation in my research, regardless of whether these responses are active or passive, positive or negative. This thesis can discover if online ratings are used by general practitioners to improve their practice, like other auditing measures (Shore & Wright, 2015), if they impact them differently, or if they do not impact them at all.

It must be noted that the idea of reactivity (Espeland & Sauder, 2007) has the underlying assumption that people are indeed aware of online rating systems or other evaluation forms. Because of the relative novelty of online ratings for general practitioners in the Netherlands, it could be the case that general practitioners are unaware of their online reputation and thus express no reactivity in response to it. It also assumes that people comply with the ratings with the aim to boost them (Espeland & Sauder, 2007). While Beuscart et al. (2016) found that this seems to be the case for restaurant owners, Menon’s (2017) research shows that it cannot blindly be assumed that this is the case for healthcare professionals, because some alter their behaviour simply out of fear for judgement. Tonkens et al. (2013) also pointed out that some healthcare professionals showcase an ‘activist’ form of professionalism, in which they actively resist the commodification of care. There is a possibility that this reaction is also observed in my research in connection to online ratings. Now that the theoretical framework of this thesis is established, the next chapter will elaborate on the practical side of the research.

(24)

23

4. Research design

As shown in chapter 3, previous research has generated important insights about online ratings and healthcare professionals. How general practitioners feel about and act upon online ratings to my knowledge has not been studied. Because of this reason, I have decided to take an exploratory approach. This chapter will provide a detailed description of the research strategy that is used to answer the research questions, with the aim of increasing the transparency and therefore the reliability of this thesis (Silverman, 2014).

4.1 Delineation of the research population

The research is carried out by conducting open-ended interviews with general practitioners in Amsterdam (approx. 850.000 residents) and Almere (approx. 200.000 residents). The selection of these two cities is determined by a number of considerations. First of all, Amsterdam was chosen because of practical concerns. Because the interviews were done face-to-face, it was more straightforward to search for participants in the city that I am based in. Second, the two cities are connected in the ‘Huisartsenkring Amsterdam/Almere’ (loosely translated as the ‘Association of General Practitioners in Amsterdam/Almere’). An email was sent to this association, asking to share a short online questionnaire about the topic of online ratings among their members. The questionnaire ended with the option to leave an email address if the respondent wanted to participate in an interview. The association unfortunately never responded, but its mere existence is important in the delineation of the participant group. Third, at the time of conducting research, five of the twenty general practitioners with the most ratings on ZorgkaartNederland work in Almere. The general practitioner with the most ratings of the whole country is also located within that city. This makes Almere an interesting city to look for participants, because it may be an indicator that online ratings play a substantial role there. Almere has a unique system of healthcare provision because next to regular general practices, every neighbourhood also has a healthcare centre. These are big centres where not only general practitioners, but also physiotherapists, dentists, child care workers and many more professionals work. These centres perhaps have a more managed, organisational character and are therefore more likely to adopt auditing measures compared to smaller general practices.

Another consideration made was the attention to the age of participants when establishing contact. Throughout the fieldwork, I strived towards getting as much diversity in age as possible. The reason for this was that the rise of the internet has been influential for the change in professional-patient

(25)

24 dynamics as discussed in chapter 3. Older general practitioners have likelyworked in a time when the internet was less prominent. Younger general practitioners, on the other hand, are likely more familiar with the internet and also with online ratings. It would be interesting to see if there are differences between the participants in relation to their role as a general practitioner and their opinions about online ratings based on their difference in age. When discussing the findings in chapter five, any remarkable differences or similarities between the older and younger participants will be addressed.

4.2 Data collection and analysis

The main method of data collection in this thesis is snowball sampling. This means that participants are found by contacting a small group of people, who can then suggest others to the researcher that are relevant to the research (Bryman, 2012). The reason for choosing this method was the expectation that general practitioners were extremely busy and that reaching out to them randomly would result in a lot of rejections. Using the snowball method could help find ‘a way in’ and may lead to more response from general practitioners.

The starting snowballs of this research have been rolled out via social media and by word of mouth. I posted on my personal LinkedIn and Facebook profiles that I was looking for general practitioners to interview. Posts were also placed in the ‘Amsterdam Durft Te Vragen’ and ‘Almere Durft te Vragen’ Facebook groups. In these groups, people can post all types of questions and other members can answer those. However, only the post on my personal Facebook profile led to contacts with general practitioners who became actual participants. Some of the other posts did lead to contact with general practitioners but resulted in either no response or in rejection. Finding participants by talking to friends and acquaintances proved to be effective in a couple of instances.

The most important and effective snowballs were the ones that originated from the participants after the interviews. Especially the participants in the beginning stages of the fieldwork were asked if they knew any other general practitioners that might be interested in doing an interview. This proved to be a very effective method for gathering participants and keeping the snowballs ‘moving’. This also proved to be an effective way of recruiting participants based on their age. Participants were specifically asked if they know young general practitioners (in the age of 45 or younger) or older practitioners (with the age of 55 or older) who would be interested in participating in the research because those were lacking in the participant sample at the beginning stages of the fieldwork.

Even though the snowball method worked, it did not lead to a sufficient number of participants in the set timeframe of this research. This method was useful for getting in contact with general

(26)

25 practitioners working in Amsterdam, but less for those working in Almere. Therefore, an additional sampling method was used to reach out to general practitioners in Almere; that of systematic sampling (Bryman, 2012) using general practitioners listed on ZorgkaartNederland. The default order of general practitioners on ZorgkaartNederland is based on ‘relevance’. It is however not exactly clear how the ‘relevance’ order is established and how often it changes. Because of this, the names of all general practitioners working in Almere that were listed on ZorgkaartNederland (a total of 109) were sorted in alphabetical order. An online random number generator was used to select a number between 0 and 9. The general practitioner that was on the place of that number was contacted. 10 points were added, and that general practitioner was contacted as well. Adding another 10 points, another general practitioner was contacted and so on until the end of the list. This method resulted in three new participants volunteering and only had to be used once.

During this data collection process, a total of fifteen general practitioners were interviewed over the course of two and a half months. Fourteen interviews were done face-to-face and took place in the doctors’ offices. One interview was done via telephone at the request of the participant. Because he seemed slightly reluctant in earlier contact about participating in the research and about the credibility of online ratings, I tried to accommodate the participant on this point. This interview was recorded, just like all other interviews, by using a mobile app. Most interviews lasted between 30 and 40 minutes but depending on the availability of the participant some lasted a bit shorter or longer.

All interviews were carried out by following a topic list, which can be found in appendix 9.1. The topic list is in Dutch and all interviews were also conducted in Dutch. The quotes that will be presented throughout this written report are thus translations. In some cases, the quotes have been slightly altered to increase for example the legibility of the text. To minimise the loss of context and meaning, the English quotes presented in this thesis have been reread multiple times while listening to the audio it stems from. Even though a topic list formed the foundation of the interviews, each interview had its own dynamic. This also means that some questions were left out, included, shifted around etcetera during the interview.

Each interview was transcribed as soon as possible after the interview took place. After that, the interviews were coded in English using the program Atlas.ti. The first step was initial coding, which means that the coding stayed close to the text. The first seven interviews were coded after they were all transcribed. This already brought up a lot of interesting findings. After that, 7 other interviews were carried out and initially coded. The initial coding phase generated more than 400 codes. After this phase, the most useful and important codes were hierarchised and grouped together. This was first done in a

(27)

26 text document, and later by making code groups in Atlas.ti. The last interview was coded when most of the analysis was done already, mainly because of scheduling reasons.

Throughout the whole process of searching for participants, interviewing, transcribing, and coding, memos were written. These contain particularities of the participants, descriptions of the atmosphere of the interviews, preliminary findings, connections between different interviews and so on. Both the coding and the writing of memos played crucial roles in analysing the data and in crystallising the findings of this thesis. Especially the memos played an important role in the outline of the next subchapter, which contains a description of the participants and a reflection on the fieldwork.

4.3 The fieldwork

A comprehensive table about characteristics of the participants such as sex, age, working location and number of ratings, is presented in appendix 9.2. As seen, five participants work in Almere, including one substitute general practitioner who works both in Almere and Amsterdam. Only a few of the participants reach the reliability threshold or nine ratings, as discussed in chapter 2. Several general practitioners with a higher number of ratings had been contacted, but either did not respond or declined to participate in the research. At the time of writing this report, 77 of 495 general practitioners in Amsterdam (16%) and 20 of 109 general practitioners in Almere (18%) on ZorgkaartNederland meet the reliability threshold of nine ratings. The fact that only three of the fifteen participants (20%) in this research meet the threshold is therefore not surprising. The average grade of the participants that are rated on ZorgkaartNederland is 7,9 compared to the national average of 8,2.

Despite the fact that a lot of the participants have less than nine, or even no, personal ratings, they were able to say a lot about online ratings during the interviews and have strong opinions about the topic. Some are extremely negative about online ratings, while others have a more neutral opinion or are positive about them. It is likely that my participant sample consists of more critical doctors than when a random sampling method is used. After all, people that feel strongly about a topic, whether positive or negative, might be more inclined to participate than people who are more moderate in their opinion. The fact that participants referred me to other general practitioners can be viewed as another selection distortion. These issues of representativeness are common when using the snowball method, which rarely leads to a completely representative sample (Bryman, 2012). However, the nature of this research is an exploratory one. It is likely that the diversity of opinions would be greater if general practitioners with a lot of online reviews are included in the sample. But in general, it is difficult to assess someone’s opinion about a topic before having interviewed them. Adding to this the fact that getting general practitioners to

‘If you are dissatisfied, talk to me’ : how general practitioners feel about online rating systems