The effect that I am looking for in this thesis is essentially a policy level version of Background Contrast Effect and how it affects perception.
Background Contrast Effect is a concept most often seen in marketing research. It is an extension of consumer choice theory and was first coined by Itamar Simonsen and Itavar Tversky.47 They were studying the behavior of consumers with relation to
marketed products. In their research, they used coupons, computers, cameras and tires as hypothetical goods which research subjects could choose between. Their method
consisted of giving the subjects a choice between products (the background set) and then having them make a second choice among a new set of similar products. They varied the background sets between the experimental groups, and they found that the background set had a significant predictive power on the subsequent choice. Furthermore, they showed that the power of this background effect is still there with only a single choice in the background set.
Taiwanese citizens are consumers of healthcare and they form perceptions of it.
This research isn’t saying that consumer choice theory is necessarily applicable or that a background contrast effect is present. Instead, I theorize that something similar is happening, and that “something similar” to a background contrast effect in relation to policy perceptions is important.
47 (Simonson & Tversky, 1992)
21
For this research, it is very important to talk about the meaning of perception.
Fortunately, there has been an immense amount of work over the last decade in this area, and scholars have worked to evaluate different studies and methodologies.
In the 1980's, healthcare perception as something separate from satisfaction was not widely considered. Measurements of healthcare perception were generally done as part of measurements of healthcare satisfaction. Instruments such as the Service Evaluation Questionnaire have desirable characteristics like uniformity and ease of use but they tend to hide negative experiences.48 As survey instruments and methodology improved (for example, the development of the Evaluation Ranking Scale), reported satisfaction generally diminished.49 Governments and scholars use many of these surveys to formulate official policy and assess public opinion. While these have continued to get more accurate, they don't always reflect a person's direct experiences.
The Meaning of Healthcare Satisfaction: an Explanation of High Reported Levels, Brian Williams et al.50 is one of many studies that tried to move beyond patient satisfaction to examine actual patient experience. Williams was assessing the experience of patients in the British mental health system. According to his research, the way satisfaction has traditionally been conceptualized is misleading. Patients might report a high level of satisfaction while at the same time reporting a wide variety of negative experiences. This suggests that satisfaction surveys, such as those conducted by the NHI every year, are a misleading indicator of patient experience. Thus, relying on them as a basis for preserving or altering policy outcomes is risky since they are not necessarily reflections of the actual experiences of patients.
One of the things that Williams suggests is that the possible determinant of satisfaction is expectations. A patient who had a bad experience, but had the health system work as they expected it would, might still be satisfied. Dissatisfaction arises largely when they see some part of the system not fulfilling its expected responsibilities.
This was interesting because the same effect might be at play for why Taiwanese people and Americans are both largely satisfied with their respective healthcare despite the vast difference in cost and accessibility. Extrapolating from Williams's research to my
48 (Nguyen, Attkisson, and Stegner 1983 p 312.)
49 (Pascoe and Attkisson 1983 p. 335)
50 (Williams, Coyle, and Healy 1998)
22
research, it seems possible that it is the change in expectations, not the change in actual services, which determines how people perceive their healthcare. Americans are not dissatisfied with expensive care or lack of access in the American system because that's what experience has taught them to expect. Similarly, Taiwanese people might see any attempts to raise their premium as a broken contract, since gradually rising premiums are not what people have been conditioned to expect.
In The Measurement of Satisfaction with Healthcare: Implications for Practice from a Systematic Review of the Literature51, the authors evaluated over 3,000 articles about ways of measuring satisfaction and eventually narrowed them down to a few hundred articles. Based on those studies, they were able to make some generalizations about the problems with healthcare satisfaction as a metric. Several of the problems were methodological, for example, measures of satisfaction tend to omit high risk groups like the old or the sick. Low response rates have a tendency to introduce systemic bias. They found further problems with the use of satisfaction because only 20% of the surveyed articles assessed expectations as well as satisfaction.
Broadly, this review had two main conclusions: satisfaction does not mean superior service, only adequate or acceptable service; and satisfaction is a relative
concept, what satisfies one person may dissatisfy another. For my research, both of those conclusions are important. The fact that satisfaction is relative and based on expectations of what's acceptable is important for how people perceive their healthcare systems, especially if a move to another country has altered their expectations. This is part of why this study didn't use satisfaction as a metric.
With relationship to Taiwan, the NHI every year gives out the “National Insurance Satisfaction Survey.” (全民健康保險滿意度調查). In 1998,52 for example, this survey was given to four different groups: the Joint service centers in Southern Taiwan, the general public, insurance units, and medical providers. Strangely enough, all of the surveys, with the exception of the one for the general public, specifically ask for
expectations as well as satisfaction. While it is important that the NHI has a method for assessing the role of expectations in determining satisfaction, the fact that the actual
51 (Crow et al. 2002)
52 (張 1988)
23
consumers of healthcare are not polled about their expectations is likely to leave a blind spot in public policy.
Additional important research was done by the team of Shou-Hsia Cheng, Yu-Jung Wei, and Hong Jen Chang. In Quality Competition and Perceived Expensiveness on Healthcare Consumers,53 they wanted to find out how perceived quality and
expensiveness of hospitals affected consumers in Taiwan. They designed a survey
focusing on patients who were in the hospital for stroke, diabetes, asthma and pneumonia.
The survey measured perceived expensiveness, actual out-of-pocket costs, perceived quality of care, and whether they'd recommend the hospital. Interestingly enough, they discovered that perceived expensiveness was determined by the quality of care and the actual out-of-pocket cost. They posited that “perceived expensiveness” was the
intervening variable between the actual out-of-pocket costs and the propensity to recommend the hospital. Furthermore, they noticed that perceived expensiveness was negatively associated with whether a patient would recommend the hospital, but that the actual out-of-pocket cost was not.
Cheng, Wei, & Chang’s findings, that it is the cost and the quality together which determine the perceived expensiveness, provide a possible mechanism for the results of my study. In the US, Taiwanese respondents were dealing with a similar quality of care to the NHI but at a far higher out-of-pocket cost. My study suggests that living abroad in the US affects the intervening variable of perceived expensiveness. Put simply, a Taiwanese person who has lived in the United States is more likely to consider the given cost of both premiums and direct medical care available in Taiwan as “cheap.”
For the purposes of this study, it is very important to talk about perception over satisfaction. This thesis is laying the groundwork for larger research in the future, but such a study would be contingent on having some idea of what drives the perceptions of value and cost. With regards to this question, we are within Kaplan's54 “context of discovery,” so this research needs to be open- ended.
The separation between satisfaction, perception, and experience was enumerated
53 (Cheng, Wei, & Chang, 2006)
54 (Kaplan 1998 p. 308)
24
by Sofaer and Firminger55 In their review of several (mostly American) studies, they found that healthcare perceptions were more complex than just satisfaction. From the studies they evaluated, they pointed out several problems with the existing methodology.
Expectations and experiences are often confounded and consumers are often aware of it.
A person finds knowing how long someone actually waited to see the doctor a more relevant metric than whether that same patient was satisfied with the wait time. Their model for perception goes as follows,
“In this model, patient perceptions of quality are in response to their experiences, whether a single episode of care or a number of episodes over time. These
perceptions result from the interaction of the patient’s expectations and their experiences. Then, as patients apply their own definitions and/or criteria regarding quality, their perceptions of the quality of the care they received crystallize. It is important, however, to recognize that these definitions or criteria are rarely if ever consciously articulated or named by the individual. They are typically implicit rather than explicit.”
My research is aimed at bringing these implicit definitions and criteria to light so they can be examined more robustly with a later study.
Many of the studies they evaluated are part and parcel of what's called
“Benchmarking.” There are several definitions of benchmarking but it can be defined as
“a process of comparison between the performance characteristics of separate, often competing organisations intended to enable each participant to improve its own performance in the marketplace.”56 This is an approach that is often mandated by government agencies and used by policy researchers to improve the effectiveness of healthcare systems.
Internal benchmarking is the comparison and two-way communication within an organization of performance indicators. This is done within the NHI itself as well as within providers and organizations. External benchmarking requires a comparison of external organizations in order to discover new ideas, methods, products and services.57 Benchmarking is performed because comparison within an organization or with
55 (Sofaer and Firminger 2005 p. 520)
56 (Kay 2007 p. 22)
57 (Cox and Thompson 1998 p. 3)
25
organizations outside allows the organization to objectively assess its own performance as well as giving an idea of what more could be achieved. Measures of healthcare
satisfaction, assessments of clinical outcomes, international cost comparisons are all part of this process. Healthcare policy experts, as part of their professional careers, would be expected to encounter such benchmarking studies.
Benchmarking is traditionally done on an organizational level, but there is a concept from marketing called “reference price,” that outlines how this could happen unconsciously on an individual level. The literature on this is extensive and is generally concerned with selling specific retail goods; however, it may play a role in how people perceive the value of their healthcare systems. It is defined by Mayhew and Winer as
“The final price construct is the internal reference price, the price the consumer expects to encounter for the brand or thinks is normal or fair for that brand.” 58 This idea is used in marketing and it has several aspects. One of the most important is that for a consumer to buy a product, that product can't be more expensive than the consumer’s internal reference price. While the NHI itself is insurance provided by a public organization (the Bureau of National Health Insurance), it is likely that people in Taiwan view it like a product. The fact that it has existed for twenty years with very cheap premiums has arguably caused the internal reference price of Taiwanese consumers to be very, very low.
Rajendran and Tellis59 suggested that “contextual reference price” i.e. the price that consumers form based on comparison with similar products at the same point of sale, is very influential in changing the internal reference price. With healthcare, people don't just spend money, they also spend time. While reference price in the literature is monetary, a similar process could be at work with regards to the frequency of visiting the doctor, or the wait time.
Price is not the same thing as value, but this idea is interesting when considering the willingness of people to pay more for healthcare. National health insurance systems fall into a trap because consumers rarely live abroad and reset their reference price.
Policy experts who by default take the international perspective often encounter
contextual prices in their research and thus are likely to have an internal reference price
58 (Mayhew and Winer 1992 p. 64)
59 (Rajendran and Tellis 1994 p. 22)
26
set relative to international standards. While this research won't hinge on the precise behavioral mechanisms that people use to form their perceptions of healthcare, it is important to note that this could be a possible mechanism.
Kahneman et. al. in Economic Preferences or Attitude Expressions?: An Analysis of Dollar Responses to Public Issues, explore the interaction between preferences and attitudes and the way subjects respond to questions designed to put dollar values on the public's willingness to pay. They examine the problems that arise from the contingent valuation model and make the point that the “stated willingness to pay” that comes from these questionnaires is best interpreted as expressions of attitudes, not actual economic expressions. They also go into depth about the limits and biases that arise from the use of monetary scales and the discrepancies between bounded categorical scales and monetary ones. One of the most important points is that monetary scales and bounded categorical scales are very strongly correlated but that the scope of monetary answers is not
consistent from person to person.
Another important point is labeled 4-1, where they say,
“An object that is considered in isolation evokes a comparison set of similar objects. The valuation of the object is relative to the set that it evoked. Features that are common to the evoked set play no role in relative judgments and valuations.” 60
Essentially, when asked whether a cake is big, a person compares it to other cakes. If asked whether a car is big, a person compares it to other cars. However, if asked whether a car or a cake is big at the same time, the internal valuation of what “big” means is altered by putting the two unlike things in the same comparison. They mentioned how this applies to stated willingness to pay when dealing with public policy issues in isolation or together.
Methodologically, they explained in detail about the use of monetary scales.
Essentially, a monetary scale is an unlimited magnitude scale with a lower bound at zero and no modulus. This makes the scales vary a great deal between individuals. However, they did notice that monetary answers did follow the answers given on categorical scales.
This is important for my research because it was crucial that I explored what is expensive
60 (Kahneman et al. 2000 p. 217)
27
as a perception and then elicited what that means to the respondent in dollar terms.
Altering any of these mechanisms could potentially make it easier for
policymakerss in Taiwan improve the NHI. David Webber’s paper details the importance of analyzing political feasibility at each of six stages in the Policy cycle.61 The
government in Taiwan has historically been very good at agenda setting, policy
formulation, policy legitimation, and policy implementation. The problems for the NHI come at the evaluation and revision phases where the public's strong preference for the status quo impedes the ability of policymakerss to raise premiums. That said, this paper is not concerned with policy revision legitimacy at this stage. Whether or not it is possible will depend on whether the difference in perceptions even exists in the first place.
For the methodology, it is important to peruse past research. I am using a
Grounded Theory approach as a foundation. Grounded Theory, according to Corbin and Strauss, is useful for measuring small sample sizes and exploring thoughts and
perceptions. This thesis relies on the subjects to illuminate their own perception on healthcare in depth. Because of this, a qualitative research method, based on grounded theories and semi-structured interviews is the best approach.62 This variety of interview generally elicits a wealth of detail as it gives an open forum for subjects to tell their own stories and generate different perspectives. It has been used very successfully in the past by other researchers looking at the quality of care in Britain,63 at perceptions about the cost and quality of primary care in America,64,65 the quality of nursing care for oncology patients in America66, nursing care in Australia,67 and several others. This method then lays the groundwork for further quantitative study. For instance, Radwin's qualitative study based on grounded theory and in depth interviews was later used to develop the Oncology Patients' Perceptions of the Quality of Nursing Care Scale (OPPQNCS) in 2003.68
Grounded theory is also good for this type of research because it is iterative and
28
ongoing. As I did the interviews, new ideas were revealed and new patterns and concepts emerged.
“A key feature of grounded theory is not that hypotheses remain unverified, but that hypotheses (whether involving qualitative or quantitative data) are constantly revised during the research until they hold true for all of the evidence concerning the phenomena under study, as gathered in repeated interviews, observations or documents.”69
During this study, questions were revised and removed or added to address newly
revealed information. The conceptual framework became smaller and more streamlined.
Another issue that is important for a study that relies on past experience is to minimize recall bias. An example of recall bias is given in “Recall Bias in
Epidemiological Studies” describing a study of women:
“The percentage of women who correctly recalled within a year their age at menarche, age of natural or surgical menopause, and age at first use of oral contraceptives ranged from 75 to 90%. The percentage who correctly recalled their menstrual cycle length within one day, however, ranged from 0 to 60%.”70 In my survey, I found that when dealing with past experiences, although the cutoff for living in the United States was within the last ten years, many people couldn't remember certain information, and the survey was altered to remove the questions which caused the most difficulty. This is detailed in the Variables section.
4 Definitions
Several terms will be used in this research.
Healthcare system – This research will use the more general definition favored by the World Health Organization. “A health system consists of all organizations, people and actions whose primary intent is to promote, restore or maintain health. This includes efforts to influence determinants of health as well as more direct health-improving
69 ((Corbin & Strauss, 1990 p. 11))
70 (Coughlin 1990 p. 88)
29
activities.71” In Taiwan, the primary component of the healthcare system is the NHI, which would allow for a more specific definition, but the more complex US system necessitates that the definition be broad enough to include the myriad of private insurers, private providers, and public agencies that a patient interacts with in the US.
Layperson – The broad definition of layperson is a person who is not qualified in a given profession and/or does not have specific knowledge of a certain subject. For this paper, a layperson is a member of the general public whose only interaction with the
Layperson – The broad definition of layperson is a person who is not qualified in a given profession and/or does not have specific knowledge of a certain subject. For this paper, a layperson is a member of the general public whose only interaction with the