Terrorism, lightning and falling furniture

ABSTRACT

From time to time, opinion pieces appear in the media that point out that the risk of being harmed by terrorism is very low. This much is true, at least from an actuarial perspective. These opinion pieces are often accompanied by lists of other, usually absurd, ways that a person is more likely to die, including being struck by lightning or crushed by falling furniture. When asked, people do state a likelihood of being harmed by terrorism that is much greater than the actuarial odds. But risk perception is complex and to many people the actuarial odds of being killed by terrorism versus being killed by falling furniture do not adequately reflect the differences in the nature of risks from these two things. A discussion about risk perception and terrorism cannot start and end with the conclusion that people simply overestimate the risk. To do so would be to overlook the nuances of risk perception and decision-making under conditions of risk and uncertainty. An understanding of the complex ways in which risk perceptions are shaped is essential for those who would seek to accurately characterise, compare and regulate risks in the terrorism context.

KEYWORDS:

• Risk perception
• terrorism

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Peter J. Phillips is an Associate Professor (Finance) at the University of Southern Queensland in Australia. He works in the field of decision theory, especially applications to terrorist behaviour and the practice of counter-terrorism and law enforcement. His earliest work was primarily concerned with terrorist attack method choice and risk preference. Using orthodox theoretical frameworks to draw inferences about terrorists’ profiles from the characteristics of their attack method choices was a natural extension of this work into ‘investigative economics’. Since then, his work has expanded to encompass behavioural models of choice, like prospect theory, which he has applied not only to terrorist decision-making, copycat behaviour and spree shootings but also to law enforcement and counter-terrorism. His most recent work explores the problem of terrorism watch-lists and suspect ranking within a prospect theory framework.

Gabriela Pohl is a Lecturer in Social Science at the University of Southern Queensland in Australia. She works in the field of decision theory, especially applications to terrorist choice in a context where media attention is a desired but uncertain outcome of terrorist activity. Her work in this area has been undertaken in the terrorism context prevailing in the Federal Republic of Germany during the 1970s, with special emphasis on the Red Army Faction (RAF) and 2nd of June Movement. Alongside her work in the field of ‘terrorist choice and the media’, she has used prospect theory to develop a behavioural analysis of terrorist choice that helps to shed some light on the choices that terrorists make when they seek to copy the actions of others. Her most recent applications of behavioural decision theory are an attempt to resolve some of the problems that affect the optimality of suspect ranking procedures, especially concerning terrorism watch-lists and surveillance.

Notes

1 Royal Statistical Society (2017).

2 For example, the Insurance Information Institute’s (2017), Facts + Statistics = Mortality Risk.

3 For example, Muggah’s (2017) blog entry published by the World Economic Forum or Jetter and Stadelmann’s (2017) blog entry published by The Conversation.

4 For example, Andrews (2017) writing for Forbes Magazine and Shaver (2015) writing for The Washington Post.

5 The populations of the two countries being approximately equal.

6 In this paper, we do not debate the appropriateness of using actuarial odds as a measure of risk. Consider, though, that the odds of being killed in a car accident (1-in-583) are greater than the odds of being killed in a motorcycle accident (1-in-846). This is because many more people die in car accidents (Insurance Information Institute, 2017). Despite this, most people probably perceive motorcycle riding to be riskier than driving a car. And it is difficult to argue that they are wrong. It is also the case that the actuarial odds have no ‘logically necessary implications for acceptability of risk’ (Slovic & Peters, 1987, p. 285; Fischhoff, Lichtenstein, Slovic, Derby, & Keeney, 1981).

7 Before the 2005 London bombing attacks.

8 An estimate of approximately 20% was also recorded in a study by Lerner, Gonzalez, Small, and Fischhoff (2003). Most studies of this type were completed during the 2000s. Having established that people’s perceptions of terrorism risk are high compared to the actuarial odds, research has moved on to explore the reasons for this and its implications (see Braithwaite, 2013) for a summary of these studies.

9 The problems with comparing the risk of death by lawnmower with the risk of death by terrorism are discussed by Ritchie (2018) in a blog entry published by Our World in Data.

10 The language differs across jurisdictions. In the UK, the threat levels refer to likelihood. In Australia, the threat levels refer to attacks being probable and in the US, the threat levels refer to risk.

11 For example, Peters, Slovic, Hibbard, and Tusler (2006) and Van der Pligt (1998).

12 By 2006, ‘terrorism’ ranked 1st (Slovic & Peters, 2006, p. 323).

13 For research on risk perception, social costs and ecology see Lennox et al. (2018, p. 283), Carter and Linnell (2016) and Gallagher (2016).

14 Shedler & Manis (1986, p. 26) define a heuristic as ‘ … a strategy that people use in making inferences; it is a shortcut that takes the place of an exhaustive approach to the problem at hand’.

15 In the literature there is also a slightly different definition or interpretation of availability that stresses the ‘ease’ with which examples can be called to mind, not the number.

16 The full list of factors is called the psychometric model. Although the model contains more than 20 factors, it can explain just 20% of the variation in risk perception (Sjöberg, 2000, p. 8).

17 ‘Humans perceive and act on risk in two fundamental ways. Risk as feelings refers to individuals’ instinctive and intuitive reactions to danger. Risk as analysis brings logic, reason, and scientific deliberation to bear on risk management. Reliance on risk as feelings is described as the affect heuristic’ (Slovic & Peters, 2006, p. 322).

18 Dread is different from fear and has different neurological effects. The difference may be expressed as follows. A person experiences dread while sitting in the dentist’s waiting room. He experiences fear as the drill comes towards his mouth (Kellerman, 2014). To say, then, that people dread the consequences of terrorism is different from saying that they fear terrorism or are afraid of it.

19 Among the many research papers produced by Gigerenzer and his team, see Gigerenzer (1991, 1996, 2002, 2004, 2007, 2008), Gigerenzer and Edwards (2003), Gigerenzer and Engel (2006), Gigerenzer and Goldstein (1996), Gigerenzer and Hoffrage (1995) and Gigerenzer and Todd (1999).

20 See especially, Gigerenzer (2008) and Gigerenzer and Brighton (2009).

21 That is, the subscript i might be just two or it could be 50 or any other number depending on the context and the possible outcomes that characterise it.

22 There are others besides these that are not obvious from the two equations presented but which reside elsewhere in the structure of the two theories.

23 This figure shows probability increasing along the horizontal axis and the way in which decision weights change as it does so. At first, the weights are higher than the probabilities (low probability outcomes are overweighted). Then the weights gradually fall below the probabilities (high probability outcomes are underweighted). We have drawn the inflection point where probability equals 0.50 but some research suggests that inflection might be observed between probabilities of 0.30 and 0.40 (see Prelec, 1998).

24 We are dealing here with measures that people might take themselves, such as cancelling travel, rather than measures that might be put in place by the government, such as more rigorous baggage inspections at airports.

25 Even if the risk were zero and ‘no prevention’ was optimal, probability weighting would lead to over-prevention efforts. If the risk of being harmed by an act of terrorism were so great that maximum prevention was optimal, probability weighting would lead to under-prevention (see Baillon et al., 2018, p. 4).

26 l’Haridon and Vieider (2018) studied the risk preferences of people from thirty different countries to determine the typical size of the insensitivity region of the probability weighting function. There is a fairly marked similarity across most countries and the global average is an insensitivity region between 0.0769 and 0.8443.

27 The steeper the ends of the probability weighting function, the more pronounced the effect. The degree of curvature depends on the values for the parameters of the probability weighting function. These are determined by fitting the function to data, usually gathered experimentally. For discussion on the determination of the shape of the probability weighting function, see Wu and Gonzalez (1996) and Gonzalez and Wu (1999).

28 Risk involves clear probabilities while ambiguity involves vague probabilities (Trautmann et al., 2008, p. 225).

29 Al-Najjar and Weinstein (2009) review the large amount of literature that this debate has produced.

30 The reader who believes that the structure of information is otherwise, can simply reverse the following reasoning. That is, if terrorism contains information that reduces ambiguity while the investigation that follows makes things less clear, then the marginal benefits of prevention increase with the terrorist attack and decrease with the investigation that follows.

31 Gollier and Kimball (2018a, 2018b) work through these in detail.

32 Our discussion refers to this precise definition of prudence. Thinking in terms of the everyday meaning of prudence will only lead to confusion.

33 Also see Krieger and Mayrhofer (2016), Felder and Mayrhofer (2017) and Brianti, Magnani, and Menegatti (2018).

34 This is consistent with Eeckhoudt and Gollier’s (2005) conclusion that prudence lowers prevention. A very rough explanation is that prudence encourages precautionary savings. Prevention now costs money and if a person wants to keep on preventing in an uncertain future, he or she will need money going forward. Prudence will encourage a reduction in costly prevention now in order to prepare for costly, uncertain, prevention later. As such, if a person is prudent he or she will prevent less now because prudence increases the perceived marginal costs of prevention.