Holism Vs. Individualism: Risk Perception and Analysis

33
Paul Chakalian 1 Philosophy of Social Science Term Paper April 4 th 2015 Holism Vs. Individualism: Risk Perception and Analysis 1. Introduction There exists a debate on the merit of methodological holism vs. individualism in the social sciences. This distinction is especially clear in the study of risk perception. The concept of risk has been discussed by many scholars in many fields and as such has acquired various definitions (Aven, Renn, 2009). The most common definition is that risk is the probability of an unwanted outcome, or stated in utility theory as simply the expected loss, or disutility. Using this definition of risk, uncertainty is commonly understood to be ignorance in regards to that probability. However, other scholars have described risk more broadly, as uncertainty with regards to something of human value that is at stake (Aven et al 2009). Using these definitions an individual’s perception of risk is either, a. their belief of the probability of a given unwanted outcome or event, or b. their perception of a state-of-affairs where something of human value is at stake, and its fate is uncertain. What neither of these necessarily describe is an individuals risk preference, which is ones opinion on how much risk (how likely an event, or what degree of stakes) they are willing to accept for a given pay off (a degree of a positive stake, or a positive event itself), in utility theory: how risk

Transcript of Holism Vs. Individualism: Risk Perception and Analysis

Paul Chakalian 1

Philosophy of Social Science

Term Paper

April 4th 2015

Holism Vs. Individualism: Risk Perception and Analysis

1. Introduction

There exists a debate on the merit of methodological holism vs. individualism in

the social sciences. This distinction is especially clear in the study of risk perception.

The concept of risk has been discussed by many scholars in many fields and as such

has acquired various definitions (Aven, Renn, 2009). The most common definition is

that risk is the probability of an unwanted outcome, or stated in utility theory as simply

the expected loss, or disutility. Using this definition of risk, uncertainty is commonly

understood to be ignorance in regards to that probability. However, other scholars

have described risk more broadly, as uncertainty with regards to something of human

value that is at stake (Aven et al 2009). Using these definitions an individual’s

perception of risk is either, a. their belief of the probability of a given unwanted

outcome or event, or b. their perception of a state-of-affairs where something of

human value is at stake, and its fate is uncertain. What neither of these necessarily

describe is an individuals risk preference, which is ones opinion on how much risk (how

likely an event, or what degree of stakes) they are willing to accept for a given pay off

(a degree of a positive stake, or a positive event itself), in utility theory: how risk

Paul Chakalian 2

seeking or risk averting they are. To illustrate these distinctions consider this example,

if you believe that in a car accident on the highway without wearing your seatbelt you

are 80% more likely to die than if you were wearing your seatbelt, than how risky

driving on the highway without a seatbelt is, is 80% more than wearing it. In this

situation an individual’s risk perception of driving without a seatbelt is that it is 80%

more likely to result in death than wearing it. Their risk preference is at what

percentage they decide to wear or not wear the seat belt. Different people can have

different risk perceptions and different risk preferences, that is, one person can believe

that not wearing a seatbelt is 40% more likely to cause death in a highway accident

(their risk perception), and that person can decide therefore not to wear a seatbelt

because they don’t want to it if its is only 40% more likely to cause death (risk

preference), while another person can believe that not wearing your seatbelt is 20%

more likely to result in death in a highway accident (risk perception) and decide

therefore to wear their seatbelt to be more safe (their risk preference). It can be seen in

this example how risk perception and risk preference are related but independent.

Even if we define risk by consequences and uncertainty rather than probability, the

difference between risk perception and risk preference remains. Using this latter

definition risk perception simply has to do with how meaningful one considers the

uncertainty associated with seatbelt use on the highway, and whether or not they

believe it makes them relatively safer or less safe to wear or not wear it. Their risk

Paul Chakalian 3

preference would still be their opinion about at what level of meaningfulness of steaks

under uncertainty they act to either increase or decrease their exposure to that

uncertainty – using the highway example, when they decide to wear their seatbelt or

not, or drive or not.

Risk and uncertainty are integral to the development of our species and our

civilization. Our ability to assess a degree of uncertainty, a stake at risk and the

likelihood of an outcome allows us to make those choices. Decision making, for its part,

is what allows us to realize our own agency. Academics have naturally been curious as

to how we do this– how we are able to assess and manage endless risks and uncertain

decisions daily. Quick decision making under uncertainty has been aptly explained by

several psychological studies, most famously by Daniel Kahneman’s and Amos

Tversky’s studies of heuristics. Kahneman and Tversky laid down the foundation for

how we make decisions under uncertain circumstances, and judge the likelihood of

various outcomes with relative ease. There was still a question about how we perceive

the risks that we are judging themselves, and more specifically, how we perceive the

size of the stakes involved. Many psychological, sociological, anthropological and

political theories have attempted to explain how different individuals and societies

perceive risk. Often, understanding risk perception becomes most important at the

intersection of an environmental or technological innovation, scientific analysis of the

potential hazards, and public fear of those perceived hazards. In this way, by

Paul Chakalian 4

understanding how people perceive risk, one can begin to better analyze and manage

risk in the real world.

There have been two distinct methodologies with which risk perception has

been studied: holistic and individualistic. Using the holistic approach, groups,

organizations and institutions are treated with agency to perceive, analyze and manage

risk. In the individualistic approach, only individual persons are treated as having

agency to perceive, analyze and manage risk. I am defining holism and individualism

using List and Spiekermann’s 2013 article, Methodological Individualism and Holism in

Political Science: A Reconciliation. Placing the risk perception scholarship under this

analytic framework will help inform future interdisciplinary discourse on risk perception

and analysis by highlighting what works and what doesn’t of each methodological

approach and where opportunities for inconsistent discourse may exist between

disciplines. Providing an overview of the current state of risk perception scholarship

across disciplines will highlight where skewed discourse1 between experts and the

public may exist, and thus demonstrate opportunities to adapt more effective risk

assessment and management polices.

2. Background

1 See Finan, T., 2003: Climate science and the policy of drought mitigation in Ceará , 400 Northeast Brazil. Weather, Climate, Culture, S. Strauss and B. Orlove, Eds., Berg, 401 203–216.

Paul Chakalian 5

The study of risk perception started in the 1960’s with the advent of nuclear

technology and the general technological and environmental transformation that

followed the Second World War. Some of the earliest studies were done by Chauncey

Starr. He focused on how risky common activities were in modern society and looked

for patterns in the differences between activities. Around the same time Kahneman and

Tversky conducted psychological studies on heuristics and decision making under

uncertainty. These studies were followed by Paul Slovic and his colleagues with

Decision Research at the University of Oregon. They designed the psychometric study

of risk perception. Not long after, Mary Douglas first wrote about Cultural Theory,

which sought anthropological explanations for differences in risk perception across

populations. By the 1990’s and 2000’s new scholars were enhancing and building on

the work done in the 70’s and 80’s. Karl Dake, Aaron Wildavsky and eventually Susanne

Rippl applied quantitative analysis to the cultural theory framework. Roger Kasperson

with colleagues at Decision Research designed the Social Amplification of Risk

Framework (SARF), which attempted to combine all the various frameworks being used

at the time. By the late 2000’s risk perception had found a new focus– climate change.

Currently, a wealth of interdisciplinary research is being done on the topic of climate

change, including the application of the SARF, as well as the creation of new risk

perception frameworks (Helgeson, Linden, Chabay 2012).

Paul Chakalian 6

Chauncey Starr had an engineer’s perspective on the public perception of

nuclear energy hazards. Using aggregated historical data Starr’s theory focused on risks

that were perceived as being controllable or uncontrollable. He believed that people

were willing to accept risks with much higher likelihoods of negative consequence if

they were in control of the exposure to the risk (e.g. driving a car). Risks where

exposure was out of the exposed’s control produced high levels of discomfort and

concern, even if the likelihood of the hazard being realized was much less (e.g. nuclear-

energy plant meltdowns). Starr concluded two rules: a. the public was willing to accept

a hazard when its perceived payoff was ~3x greater than its perceived risk, and b. that

the natural probability of risk of death from disease in a society was proportional to the

average risk a society was willing to accept from an uncontrollable hazard (e.g.

commercial flight) (Starr 1969).

Concurrently, Kahneman and Tversky conducted psychological studies

demonstrating individuals heuristic problem solving techniques. They showed how

individuals used these heuristic techniques, or cognitive short cuts, to answer questions

under uncertainty, determine the likelihood of an outcome and the magnitude of

stakes. These studies were groundbreaking in the field and provided not only evidence

that individuals do not make decisions based on statistical probabilities, but

demonstrated precisely how they used heuristics to do so instead. By answering

unknown questions with known answers to different questions, individuals could make

Paul Chakalian 7

quick decisions without intensive effort. For example the availability heuristic: when

asked how common x is, individuals answer how many times or how easily they can

remember x happening, instead of actually attempting to innumerate every instance of

x happening (Kahneman 2013). Understanding this phenomenon began to establish

the inadequacy of traditional rational-choice theory and technical risk analysis (i.e. risk

= probability x stake / time), in particular it demonstrated the public’s irrationality with

regards to risk perception.

At the same time that Tversky and Kahneman were developing their theory on

heuristics, the Decision Research Group in Oregon, lead by Paul Slovic, Baruch

Fischhoff and Sarah Lichtenstein was developing what would become a foundation in

risk perception scholarship. The psychometric paradigm they designed focused on a

subject’s expressed preference to risk, as opposed to simply the risk they were willing

or not willing to accept based on historic data. This study accepted, but did not

analyze, differences in culture and socialization that effect risk perceptions, outside of

any cognitive heuristics, and regardless of whether or not exposure was voluntary. In

this way the psychometric paradigm was designed to create a taxonomy of hazards

and predictable public perceptions. Building off Starr, and Tversky and Kahneman’s

work psychometric studies sought to outline what hazards US society perceived as the

most or least risky, the payoff or value of those hazards, the degree to which efforts

should be taken to mitigate the risks of those hazards, and where the general public

Paul Chakalian 8

and experts disagreed on these points. The biggest revelation from the Oregon

research was the idea that the more a risk was associated with feelings of dread, lack of

control, or catastrophic potential, and perceived as unknown, new, or having hazards

with delayed onsets, the more a society would pay to have it mitigated– regardless of

the actual probabilities of disaster, or the objective degree of the stake. When experts

were surveyed they were equally likely to place the same hazards in the same

descriptive categories as the public (e.g. both groups agreed that food additives have

delayed, unknown and uncontrollable hazards), however experts did not use these

measures to assign the amount of energy that should be placed into regulating and

mitigating these risks. Instead, experts focused on the statistical rates of harm from

given hazards, regardless of the level of dread or uncertainty they produced.

Paul Slovic’s research through Decision Research empirically demonstrated that

risk perception was not only mediated by cultural background, cognitive biases, and

emotional biases that surround dread and uncertainty. It was also mediated quite

strongly, and sometimes irrationally by reputation. He called this a ripple effect (Slovic

1987). Using the example of the Three Mile Island disaster Slovic pointed out that

when hazards are realized there are higher order consequences on public risk

perception that go far beyond any quantitative first order damages or losses (e.g.

persons killed, or property destroyed). The realization of a disaster from a perceived

risk can cause a society to perceive many other technically unrelated risks as “riskier,”

Paul Chakalian 9

forcing additional costs on the society by demanding heightened regulation and risk

mitigation. After Three Mile Island there was enhanced regulation and scrutiny of many

other unknown technologies (e.g. genetic engineering) (Slovic 1987).

While psychologists were experimenting with how individuals and groups

perceive and manage risk, sociologists and anthropologists were asking how structures

of society influenced the answers. Mary Douglas and Aaron Wildavsky’s work on

cultural theory asserts that in order to understand why some risks are perceived as

more or less severe, and why some risks become politicized while others remain latent,

one needs to understand how risks are both constructed and selected (Tansey &

O’Riordan 1999). By viewing risk perception as ingrained in social norms and values

one can begin to understand the reasons behind the biases discovered from the

Oregon work. They argued that only by understanding and addressing structural

reasons can science engage a productive discourse about risks and hazards with the

general public. Cultural theory suggests that only by understanding the effects of

concepts like reputation, prestige, credibility and status can one understand how the

public will respond to information from experts. Most of the work on this topic has

been theoretical, while applications have been subjected to criticisms of functionalist

explanation, and unproven assumptions.

Mary Douglas is known for her grid-group typology that places individuals’

worldview of society into two dimensions. The grid dimension relates to the level of

Paul Chakalian 10

externally forced structures imposed on an individual’s agency. The group dimension

relates to the degree to which an individual is subject to a larger group of individuals.

In this way one can build a grid box which if placing increasing grid aspects on the top

end of the y axis and increasing group aspects on the right end of the x axis would

have individualism occupy the lower left quadrant, fatalism occupy the upper left

quadrant, hierarchy occupy the upper right quadrant, and egalitarianism occupy the

lower right (Thompson, Ellis, Wildavsky 1990). Douglas applies cultural theory

framework to analyze risk by first placing a population in the typology. Then, in

understanding those groups’ internal relationships to each other and external

relationships to nature (by the extent to which they impose external structure or

internal organization) can predict how they will perceive hazards, and manage risks

(Douglas & Wildavsky 1983; Tansey & O’Riordan 1999).

Karl Dake made notable advancements in cultural theory and risk in the 1990’s.

Using Riley Dunlap’s theory on human expansionist vs. ecological worldviews, and

Stephen Cotgrove’s distinctions of cornucopian vs. catastrophic positions, as a starting

point. Dake set out to distinguish, “who fears what and why?”(Dake 1991), and

quantitatively test the grid-group typology hypothesis of cultural theory. Dake tested

this theory using aggregated psychometric studies. He surveyed large stratified sample

populations from California measuring their risk perceptions and risk preferences (what

they perceived as more or less risky, and how risk seeking they were) their

Paul Chakalian 11

demographic backgrounds, and their positions on egalitarian / hierarchical / individual

worldviews. He measured worldviews from answers of rated agreement / disagreement

to statements such as, “the welfare state tends to destroy individual initiative.” Dake

did not analyze the “fatalist” grid-group type. Dake found strong correlations between

personality types, risk perceptions, risk preferences, and worldviews that are

hypothesized from cultural theory, and from his and Wildesky’s earlier analysis (Dake

and Wildesky 1990).

In 2002, Susanne Rippl reexamined Dake’s approach to determine whether or

not his conclusions were warranted given his findings, and if his approach was

reasonable given the theory he was testing (Ripple 2002). Rippl ultimately criticized

Dake’s approach. She provided new enhanced methodology to assess cultural theory

empirically by using structural equation modeling to measure multiple latent analytical

constructs (the four original grid-group typologies) and multidimensional loadings of

each (grid & group). Rippl’s approach lends further analytical applicability to cultural

theory based risk perception analysis. Her work was subject to some criticism by the

cultural theory school however. By using aggregated individual level results in order to

interpret a cultural frame, her work has been criticized as failing to accurately assess the

cultural frame itself by those who believe the frame has its own distinct ontology

separate from the aggregate reality of intervals. That is, if a cultural frame exists

Paul Chakalian 12

independently from the individuals within it, than it would need to be studied on its

own level to understand its effects.

Although cultural theory was making steps toward establishing empirical

analytical tools, by the end of the 20th century there were still two academic schools

studying risk perception: cultural theorists and psychometric analysts. The

psychometric paradigm school put forth new theories in attempts to bridge the gap

between the technical and social approach of implementing risk assessment and

management, as well as between cultural theory and the psychometric paradigm

(Pidgeon, Kasperson, Slovic 2003). The result is what has become known as the Social

Amplification of Risk Framework (SARF) first proposed in 1988 by R. Kasperson, Renn,

Slovic, Brown, Emel, Goble, J. Kasperson, and Ratick. Like cultural theory, the SARF

places importance on social framing in risk perception analysis; like the psychometric

paradigm it provides tools with testable applications. It is similar to Dake and Ripple’s

quantitative applications of cultural theory, with a focus more on the individual

psychology that is produced by certain cultural frames as opposed to the aggregated

social / cultural frame itself. Importantly, the SARF sought to incorporate into one

framework all of the various approaches that were in use to analyze risk perception and

assessment. Adding to the theoretical foundations of the framework Ortwin Renn

isolated 7 different classification approaches that were in use at the time: actuarial

(statistical methods), toxicological and epidemiological, engineering (probabilistic),

Paul Chakalian 13

economic (utility theory, risk-benefit), psychological (psychometric), social theory

(qualitative), cultural theory (grid-group). He then cross analyzed these classifications

with their respective applications and attributes in seven categories: base unit,

predominate method, scope of risk concept, basic problem area, major application,

instrumental function, and social function. Renn provided this technique as way for

SARF to systemically use all available methods to understand risk perception (Renn,

Krimsky and Golding 1992). The SARF thus provides the following two maps for

analyzing risk perception:

Figure 1. Pidgeon et al. 2003

Paul Chakalian 14

Figure 2. Renn 1992

In this way SARF provides tools to look at a given risk and analyze its impacts in

multiple dimensions. It provides the framework to question social, individual and

heuristic factors that will intermediate the risk perception and ultimately its potential

Paul Chakalian 15

impacts. Renn’s map provides the tools necessary to implement risk management and

assessment using the SARF. Importantly, SARF considers not only cultural framing, but

that culture (including any of its norms, values, traditions and institutions) may itself be

perceived as at-risk. Additionally, it considers the idea that cultural framing and cultural

vulnerability may either attenuate or amplify perceived risk. An example of this can be

seen in Greenland, currently there is large public debate on the impacts of foreign

workers on Greenlandic culture. Greenland has an incredibly small population, ~5600

people, because global warming is creating new minable land on the island the

country may soon have a situation where the local domestic population is

outnumbered by foreign mineral and petroleum extractors. Public officials are worried

about the risk this poses to Greenlandic culture (the integrity and homogony of their

traditions and values), at the same time, Greenlandic culture itself has been historically

very isolated from the outside world and thus lends itself to a potential amplification of

the risk perception from new in-migration. This is in line with the ripple effect theory of

reputation posited by the psychometric paradigm, (which indicates secondary effects

from hazard events to things like trust in the competency of government to regulate

and manage risks, and to the acceptance of new technologies at large). The SARF goes

further by systematically analyzing the information system surrounding a particular risk–

the ways and degrees to which it is publicized, politicized, or polarized. As R.

Kasperson and J. Kasperson describe at length in their 1996 article, The Social

Paul Chakalian 16

Amplification and Attenuation of Risk, SARF seeks to acknowledge the underlying

system of values, norms and information dispersion that cause either amplification or

attenuation of risk. It seeks for example, to acknowledge that one society may not feel

dreadful of something like handgun violence, or may feel that handgun regulation is

not a productive way to mitigate the risk of handgun violence (possibly because of

fears of other risks, like having your home invaded and not owning a handgun to

protect your family with), while another society will regulate handgun distribution in

order to mitigate the risk of handgun violence because they either feel dreadful of

handgun violence, and/or because they don’t feel dreadful of not owning a handgun.

Importantly, these two societies will therefore either proceed or not proceed to actively

and effectively manage handgun violence by controlling handgun distribution. The

SARF can acknowledge that marginal populations are often subjected to attenuation of

risk and underprepared management, while risks that affect a majority class, or

otherwise appeal to the institutional mores of a society, will often have swift and

prepared responses. At the same time, risks that resonate strongly with normative

feelings of dread, fear, or injustice become amplified, pulling disproportionate

resources and creating a state of over-preparedness. This gives SARF a wide and deep

perspective on risk perception in society, though implemented frequently through

traditional psychometric survey methods cultural theory propionates criticize SARF for

Paul Chakalian 17

still failing to provide tools for understanding why culture affects risk perception, even

if it attempts to classify how it does.

Most recently there has been a surge in academic discussion of risk perception

in regards to climate change. It seems the majority of the literature on this topic has

used the SARF (Mase, Cho, Prokopy 2015; Renn 2011). Some scholars have put forth

new perspectives as well. Jennifer Helgeson, Sander van der Linden and Ilan Chabay

published an article titled The role of knowledge, learning and mental models in public

perceptions of climate change related risks 2012, which describes five processes that

frame perceptions of climate-related risks: cognitive, subconscious, affective, socio-

cultural and individual, though this did not provide a new scientific framework as much

as a new analytical perspective. Others have simply added to existing bodies of

knowledge on the cognitive research of risk perception, but with new content (L. Zaval,

E. A. Keenan, E. J. Johnson and E. U. Weber 2014). It is too early to tell exactly what

these new studies will contribute to the literature on risk perception analysis. So far

they seem to add more new content for risk analysis rather than new theories on risk

perception.

3. Analysis and Discussion

The study of risk perception has evolved over more than 40 years to include a

dizzying array of theories, methodologies and tools to better prepare the potential risk

Paul Chakalian 18

analyst. In fact, many subsequent theories attempt to reconcile differences from the

past, and only add to the mix. At the same time, the social sciences in general have

been separated on the issue of methodological holism vs. individualism, and whether

there is room for both at various levels of analysis, or if one or the other is the “correct”

choice. The argument for methodological individualism is that without being able to

logically defend and test a theory’s mechanisms at the level at which they are directly

interacting, it is irresponsible to attempt to use that theory to explain phenomena. The

argument for methodological holism is that some phenomena are manifestly different

than their respective parts and therefore it is not only fair to analyze their interactions at

higher-order levels, but doing so provides the most useful explanation. I will examine

how these different risk perception theories use methodological holism or

individualism, and in doing so determine what works and what doesn’t from each

approach. Let us look at each case.

Kahneman and Tversky’s work is methodologically individualistic. They

experimented with individual risk perceivers and drew conclusions about those

individuals’ cognitive biases. They used empirically sound statistical methods to posit

that these biases hold true for other individuals, but did not explain this phenomenon

by any outside structure or system. This benefited their work greatly by providing it

with a solid epistemological foundation and strong explanatory power. However this

reductionism limited the breadth of that explanatory power significantly. Many other

Paul Chakalian 19

studies and approaches were needed to understand how the cognitive heuristics they

discovered play out in the real world.

Chauncey Starr’s work is more layered; it operated at the individual level in

gathering data, but was analytically holistic. By using aggregated historical data of

individual actions in order to assess how groups of people related to their environment

and risks within those environments he treated groups as agents. He did not consider

the individual perceptions of each of the actors in the historical data, nor their

intentions and motivations in accepting the risks recorded. This is important because

he asserted that the US society as a group accepted higher risks for voluntary actions

and lower risks for non-voluntary actions. This was regardless however, of the opinions

of the individual members of the group at the time they were taking on those risks–

their past or present perception / thoughts / feeling about “accepting” those risks. In

doing so he did not establish, for example, that it was not in fact a few elite who

decided how risky air travel would be, which had nothing to do with how people

buying the tickets perceived the risk. The risk takers may have been very unhappy, in

fact, about the size of the risk they were taking on, and saw the pay-off of the risk as

mere necessity. For this reason Starr’s holism was a serious weakness to his work.

Though I will show later that I believe holistic approaches have merit, in this case Starr

did not provide robust enough explanatory power to justify the lack of reduction in his

Paul Chakalian 20

analysis. Using any of the other risk perception framework discussed here would yield

more thorough and responsible results.

I argue that Paul Slovic and colleagues’ work incorporated methodological

holism and individualism. Sampling discrete individuals in survey formats and placing

them in multivariate factor analyses provided evidence of individual perceptions, and

conclusions about society’s perceptions and management strategies as a whole. This

was carried out using methodological individualism. However, Slovic’s theory on ripple

effects of risk perceptions gave agency to groups (the public) to cause ripple effects

(the public demanding higher risk mitigation) without appealing to lower order causes

(such as individual feelings of dread or fear). This methodological holism could be

warranted because through his mathematical analysis he satisfactorily demonstrated

that the higher order explanans was robust to changes in, and the configuration of, the

lower order explanans (i.e. individual perception). This demonstrated that these higher

order phenomenon were ontologically distinct from their inner components at the level

of analysis. To provide an example, he did not establish for instance, that every

individual risk perceiver who recently experienced a nuclear fiasco subsequently felt

dread and perceived the risk as much greater than it actually was for a certain period

after the event, and that therefore they demanded to their politicians that they increase

nuclear regulations. He did however, demonstrate that regardless of the individuals’

feeling or actions, when the public perceived risks as great and when that perception

Paul Chakalian 21

was coupled with availability heuristics the result was increased regulation of that risk.

In this way he established the higher-order association regardless of lower order

mechanisms. Therefore, his documentation demonstrated that the cause and effect

(cause= actualization of a hazard event, effect= increased regulation of that hazard,

regardless of changes, or lack of there of, to the statistical probabilities associated with

that risk) was robust to changes to the individual risk perceivers, or their individual

personalities / perceptions / opinions. Thus, I believe Slovic’s work benefited from its

higher-level holistic analysis. The major weakness of the psychometric paradigm is its

inability to adequately explain or assess social framing involved in risk perception,

perhaps due to a fear of departing from reducible metrics.

Mary Douglas’s work is methodologically holistic. It directly suggests that

contexts of society themselves predict the society’s perception of and response to risk.

It operates solely with higher order units of analysis, both as explanandum and

explanans. This provides a powerful ability to analyze social frames as consequences

and drivers of risk perception. However, stark anti-reductionism weakened its ability to

be taken seriously as scientific approach to understanding risk perception, and

handicapped its applicability, as it was difficult to test without appealing to a lower

order phenomenon.

Karl Dake’s work was important by bridging the gap between individual and

collective risk perception analysis, demonstrating how individuals both internalize a

Paul Chakalian 22

cultural frame, and express that frame, and thus achieving an empirically demonstrable

dialectic ontology of cultural theory’s unit of analysis (the grid/group type). Dake’s work

used individual level analysis to explain social structures that he claimed embody their

own risk perception. In this way Dake’s work gave agency to group types (holism) but

he explained those types by reducing them to individual beliefs (individualism). In this

way Dake’s work is very much inline with what List and Spiekermann (2013) call causal-

explanatory holism. Dake uses a lower-level phenomenon (individual tendencies

toward grid/group configuration) to explain a higher-level phenomenon (social risk

perception), which then gives agency to another higher-order explanans (the group is

egalitarian, and perceive risk like egalitarians). In order to defend this he would have to

establish that this is sound because the higher-level causes of risk perception x (i.e. that

they’re egalitarian) are: a. robust to changes in the lower-level cause (the individual risk

preferences) and b. can be realized by multiple different configurations of them. I

believe this to be true in Dake’s theory. Different people can feel slightly differently

and correlate slightly differently, but still end up in the same groups, making the

groups inherently robust to changes at the lower level. Therefore making it fair to say,

“egalitarians do x,” supposing you’ve adequately defined what egalitarians are.

Establishing egalitarians’ ontology is done by individual level personality and political

preference surveys. In order for this to be sound explanans at the higher-level you

would need to suppose this definition was also robust to changes at the individual

Paul Chakalian 23

level, and could arise from multiple configurations of that level. Dake and others

scholars have demonstrated this to be true. I believe this makes Dake’s use of

methodological holism strong. Though, as a risk analysis strategy it is unable to

account for psychologically intermediating factors on risk perception directly, and

therefore only explains part of the variation in risk perception across populations. This

matters because the most useful risk perception theory should be able to explain more

than one component of variation.

Susanne Ripple’s work strengthens Dake’s. It enhances the lower order analysis

by including measurable units of degrees of grid and group components respectively

and establishes with greater empirical soundness that: a. individual level phenomenon

create aggregate truths, and b. those aggregate truths can be justifiably analyzed at

their higher level. This is important because without doing so Rippl’s use of

methodological holism would be vulnerable to lack of justifiability. Rippl’s work is

subject to the same critique however of not being fully capable of incorporating

individual level abnormalities into the analysis. In theory this wouldn’t be necessary if

the theory created higher-order groups that have strong explanatory power on their

own, however even if the group-level explanans are statistically significant, they still

only explain a fraction of the overall variation in the risk perception of a society.

Therefore a fully mature theory of risk perception should be able to not only explain

that people group in x ways that create y aggregates that perceive z risk in w ways and

Paul Chakalian 24

that those aggregates are robust to changes at the lower level. But also explain how

individuals have x traits that cause them to directly perceive risks in q ways, which

could intermediate higher level risk perception, or change it altogether over time. For

example, take a population of people, within it there is a group of individuals called

practicalists that all feel that nature is inherently resilient and their actions will not have

great effects on it, and they all group together in statistically significant ways. Within

the population there are also people who feel quite the opposite, that nature is fragile

and highly susceptible to harm caused by humans, called cautionists, and they also

group together in statistically significant ways. However in the population the

practicalists and cautionists are intermixed with other individuals that lie anywhere on

the spectrum and do not group into statistically significant sub populations. In this

situation cultural theory and Riple’s expansion of it would capture what was happening

with practicalists and cautionists, but would not capture the interactions between those

individuals and all the other people in the populations. It would not capture for

instance, if there were actually different specific individuals in different groups

depending on the particular risk being perceived, and if there were quantitative trends

in the amount of people in a group over time or by risk. For instance, if Jane was a

practicalists toward nuclear power and an cautionists toward gun control, but Bob was

a practicalists toward gun control and a cautionists toward nuclear power, making the

number of people in each group the same but the individuals different, or if slowly

Paul Chakalian 25

more people were becoming part of one or another group. It would also not capture if

the people outside of either group, in the statically unexplained portion of the

population, actually held disproportionate influence or power on regulatory or political

processes that would have much larger influences on how the population as a whole

perceived or acted on risks. For instance if Joe was not a cautionists or a practicalists

but was the head of FEMA, whatever his perception was would be very important, and

totally unexplained by cultural theory.

The SARF does what it set outs to do. It combines both the methods and units

of analysis of earlier work on risk perception. Generally methodologically individualistic,

the framework acknowledges that individuals do create aggregate phenomenon which

feedback to those same individuals and intermediate risk perception. The SARF has

wide and deep explanatory power, although its lengthily methodology and complex

application may be its weakness. If the best theories are the simplest then SARF is

clearly no winner, requiring multiple different methods and lengthy analysis. However,

the SARF represents the combination of the total current extent of our understanding

of risk perception and our ability to analyze and mange those risks. It is the best option

we have now. Ultimately, I agree with List and Spiekermann in their 2013 article. Any

feasibly useful theory of risk perception will need to be able to do three things. A.

acknowledge “Supervenience individualism: the individual-level facts fully determine

the social facts; i.e., any possible worlds that are identical with respect to all individual-

Paul Chakalian 26

level facts will necessarily be identical with respect to all social facts” (List and

Spiekermann, 2013, p.633). But b. acknowledge, “token holism: some particular

objects in our social ontology are distinct from (and not redescribable as) any objects in

the individual-level ontology” (List and Spiekermann, 2013, p.634). And c.

acknowledge “type holism: some social properties are distinct from (and not

redescribable as) any individual-level properties” (List and Spiekermann, 2013, p.634).

This is apparent if you consider the risk of nuclear energy. Though feelings of lack of

control over the risk play into emotional fears, and the availability heuristic plays into

ripple effects from recent nuclear meltdowns, social groups, values and identities also

play a role intervening in risk perception. Those individuals who possess the qualities

sufficient to be considered egalitarians will be more wary of expert opinions on safety

levels and individualists will be less concerned with lack of control. On an even higher

order, institutions and organizations that reify their social ontology through things like

reputation and posses their own agendas through acts of drafting mission statements

and falling into traps of habituated status quo behavior, may have structurally

determined risk perceptions that operate outside of the individuals who embody the

organization. Often only time corrects the inconsistency between how the organization

will perceive and respond to risk, and how its members perceive and respond to risk.

This can be seen in the following example. Take an anti-nuclear NGO that was created

after a disaster that was staffed by egalitarians and they all individually felt the traits

Paul Chakalian 27

necessary to be considered egalitarians and have the perceptions of nuclear risk that

cultural theory would predict based on that. Over time some of those individuals leave

the organization and are replaced by individualists who care about the environment

and are worried about particular nuclear issues, but who otherwise perceive the risk as

cultural theory would predict an individualist would, not an egalitarian, then, it would

take time before the agenda of the NGO as an organization changed. Certainly the

NGO as its own entity would still perceive the risk of nuclear energy the way an

egalitarian would for sometime after the change in individual members. Cultural theory

alone would be blind to the discrepancy between the mid-level aggregates (the

egalitarians and individualists in the organization), and the higher-level aggregate (the

egalitarian organization), and would be blind to the ways in which the lower-level

individualists in the organization perceive the risk of nuclear energy. Psychometric

analysis would be blind to the way the organization would perceive the risk regardless

of the views of the individualists at first. This leaves SARF as the sole framework that

could explain both phenomena, which is a necessary quality of a robust risk analysis

framework.

4. Conclusions

Starting with List and Spiekermann’s 2013 article on methodological holism vs.

individualism I conducted a thorough review of the risk perception literature since the

Paul Chakalian 28

1960’s. I conclude that their arguments for a combination of individualist and holistic

approaches to social science theory hold true in risk perception theory. Any one theory

on risk perception failed to adequately analyze the entire issue, except for Drake and

Ripple’s quantitative application of cultural theory and the SARF, which both use

methodological holism and individualism at different levels. However, because of the

limitations to Drake’s and Ripple’s approach I have concluded that the SARF represents

currently the most useful and mature theory and framework on risk perception we have

available. I agree with Lennart Sjoberg 2002 that every scientific study needs to be

simplifiable, and the SARF is not perfect. It suffers quite noticeably from over-

complexity and a laborious application. I believe there is still a need for a risk

perception theory that adheres to supervenience individualism, allows for holistic

analysis, and can be used to explain easily and straightforwardly why and how both

individuals and groups will perceive a given risk. Until that time I believe the careful

and attentive application the SARF benefits from its inclusion of individualism and

holism and is a sound and substantial method of risk perception analysis.

Paul Chakalian 29

References:

Aven, T., & Renn, O. (2009). On risk defined as an event where the outcome is

uncertain. Journal of Risk Research, 12(1), 1-11.

Dake, K. (1991). Orienting Dispositions in the Perception of Risk: An Analysis of

Contemporary Worldviews and Cultural Biases. Journal Of Cross-Cultural

Psychology, 22(1), 61-82. doi:10.1177/0022022191221006

Finan,, T., Strauss, S., & Orlove, B. (2003). Climate science and the policy of

drought mitigation in Ceará, Northeast Brazil. Weather, Climate, Culture,

203-216.

Fischhoff, B., Slovic, P., Lichtenstein, S., Read, S., & Combs, B. (1978). How safe is safe

enough? A psychometric study of attitudes towards technological risks and

benefits. Policy Sci, 9(2), 127-152. doi:10.1007/bf00143739

Heise, D., Gross, J., & Rayner, S. (1987). Measuring Culture: A Paradigm for the

Analysis of Social Organization. Contemporary Sociology, 16(4), 587.

doi:10.2307/2069980

Helgeson, J., van der Linden, S., & Chabay, I. (2012). The Role of Knowledge, Learning

and Mental Models in Perceptions of Climate Change Related Risks. In A. Wals &

P. Corcoran, Learning for sustainability in times of accelerating change (1st ed.).

Wageningen, NL: Wageningen Academic Publishers.

Paul Chakalian 30

Hillson, D., & Murray-Webster, R. (2007). Understanding and managing risk attitude.

Aldershot, England: Gower.

Kahneman, D. (2013). Thinking, fast and slow. New York: Farrar, Straus and Giroux.

KASPERSON, R., & KASPERSON, J. (1996). The Social Amplification and Attenuation of

Risk. The ANNALS Of The American Academy Of Political And Social Science,

545(1), 95-105. doi:10.1177/0002716296545001010

Kasperson, R., Renn, O., Slovic, P., Brown, H., Emel, J., & Goble, R. et al. (1988). The

Social Amplification of Risk: A Conceptual Framework. Risk Analysis, 8(2), 177-187.

doi:10.1111/j.1539-6924.1988.tb01168.x

Kingston, P., Douglas, M., & Wildavsky, A. (1983). Risk and Culture: An Essay on the

Selection of Technological and Environmental Dangers. Contemporary Sociology,

12(4), 414. doi:10.2307/2067477

LIST, C., & SPIEKERMANN, K. (2013). Methodological Individualism and Holism in

Political Science: A Reconciliation. Am Polit Sci Rev, 107(04), 629-643.

doi:10.1017/s0003055413000373

Marris, C., Langford, I., & O'Riordan, T. (1998). A Quantitative Test of the Cultural

Theory of Risk Perceptions: Comparison with the Psychometric Paradigm. Risk

Anal, 18(5), 635-647. doi:10.1023/b:rian.0000005937.60969.32

Paul Chakalian 31

Mase, A., Cho, H., & Prokopy, L. (2015). Enhancing the Social Amplification of Risk

Framework (SARF) by exploring trust, the availability heuristic, and agricultural

advisors' belief in climate change. Journal Of Environmental Psychology, 41, 166-

176. doi:10.1016/j.jenvp.2014.12.004

Pidgeon, N., Kasperson, R., & Slovic, P. (2003). The social amplification of risk.

Cambridge: Cambridge University Press.

Rayner, S., & Cantor, R. (1987). How Fair Is Safe Enough? The Cultural Approach to

Societal Technology Choice1. Risk Analysis, 7(1), 3-9. doi:10.1111/j.1539-

6924.1987.tb00963.x

Renn, O. (1992). Concepts of Risk: A Classification. In S. Krimsky & D. Golding, Social

Theories of Risk (1st ed., pp. 53-79). Westport, CT: Praeger.

Renn, O. (2011). The social amplification/attenuation of risk framework: application to

climate change. Wiley Interdisciplinary Reviews: Climate Change, 2(2), 154-169.

doi:10.1002/wcc.99

Renn, O., Jaeger, C., Rosa, E., & Webler, E. (1999). The Rational Actor Paradigm in Risk

Theories: Analysis and Critique. In M. Cohen, Risk in the Modern Age. Social

Theory, Science and Environmental Decision-Making (1st ed., pp. 35-61). London:

MacMillan.

Paul Chakalian 32

Rippl, S. (2002). Cultural theory and risk perception: a proposal for a better

measurement. Journal Of Risk Research, 5(2), 147-165.

doi:10.1080/13669870110042598

Rosa, E. (2003). The logical structure of the social amplification of risk framework

(SARF): Metatheoretical foundations and policy implications (1st ed., pp. 47-79).

Short, J. (1984). The Social Fabric at Risk: Toward the Social Transformation of Risk

Analysis. American Sociological Review, 49(6), 711. doi:10.2307/2095526

Short, J. (1990). Hazards, Risks, and Enterprise: Approaches to Science, Law, and Social

Policy. Law & Society Review, 24(1), 179. doi:10.2307/3053792

Sjöberg, L. (2000). The Methodology of Risk Perception Research. Quality And

Quantity, 34(4), 407-418. doi:10.1023/a:1004838806793

Slovic, P. (1987). Perception of risk. Science, 236(4799), 280-285.

doi:10.1126/science.3563507

Slovic, P., Fischhoff, B., Lichtenstein, S., & Roe, F. (1981). Perceived Risk: Psychological

Factors and Social Implications [and Discussion]. Proceedings Of The Royal Society

A: Mathematical, Physical And Engineering Sciences, 376(1764), 17-34.

doi:10.1098/rspa.1981.0073

Slovic, P., Fischhoff, B. and Lichtenstein, S., (1985). Characterizing Perceived

Paul Chakalian 33

Risk. R. W. Kates, C. Hohenemser, & J. X. Kasperson (eds.), Perilous

progress: Managing the Hazards of Technology, pp. 91-125

Starr, C. (1969). Social Benefit versus Technological Risk. Science, 165(3899), 1232-

1238. doi:10.1126/science.165.3899.1232

Tansey, J., & O'riordan, T. (1999). Cultural theory and risk: A review. Health, Risk &

Society, 1(1), 71-90. doi:10.1080/13698579908407008

Thompson, M., Ellis, R., & Wildavsky, A. (1990). Cultural theory. Boulder, Co: Westview

Press.

Weber, E. (2006). Experience-Based and Description-Based Perceptions of Long-Term

Risk: Why Global Warming does not Scare us (Yet). Climatic Change, 77(1-2), 103-

120. doi:10.1007/s10584-006-9060-3

Wildavsky A. and Dake K., (1990) Theories of Risk Perception: Who Fears What

and Why? Daedalus pp. 41-60

Zaval, L., Keenan, E., Johnson, E., & Weber, E. (2014). How warm days increase belief

in global warming. Nature Climate Change, 4(2), 143-147.

doi:10.1038/nclimate2093