Logical Thinking Under Perceived Threat
A Study about the Effects of an Expanded Logic on Public Support for Torture
Master Thesis Political Science
International Relations
Microfoundations of Conflicts
Author: Luc Driessen
Supervisor: Dr. Seiki Tanaka
Second Reader: Dr. Ursula Daxecker
Date: June 22, 2018
3
Inhoud
1. Introduction ... 5 2. Literature review ... 8 2.1. Emotions ... 9 2.2. Logic ... 10 3. Theoretical Framework ... 113.1. The Interrogation Logics ... 12
3.2. The Deterrence Logics ... 15
4. Methodology ... 16
4.1. Population and Sample ... 17
4.2. Experimental design ... 17 4.3. Dependent variable ... 19 4.4. Independent variables ... 19 4.5. Control variables ... 20 5. Results ... 20 5.1. Randomization Test ... 21 5.2. Descriptive Statistics... 22
5.2.1. Support for Torture ... 22
Figure 1 ... 22
5.2.2. Reasons for Support and Opposition ... 23
5.3. Regression Analysis / manipulation test ... 31
5.4. Power calculation ... 33
6. Discussion ... 33
6.1. Limitations & recommendations ... 37
7. Conclusion ... 38
7.1. Academic implication ... 38
7.2. practical implications ... 39
References ... 41
5
1. Introduction
Apart from Slavery, no other practice is so universally and unanimously condemned in law and human convention as torture (Shue, 1978, 124). Despite this, torture still occurs worldwide (Amnesty International, n.d.). Torture, which is often also addressed with euphemisms such as “harsh” interrogation or “enhanced interrogation techniques” has gained international attention after the CIA ’s enhanced interrogation programme came to light in 20141 (Kelman, 2018, 123). In 2016, a poll showed that “46 percent of Americans said that enemy combatants could be tortured “to obtain important military information” (Sieff, 2016). In 2017, newly elected President of the United States Donald Trump gave this group a voice by using his first tv-interview as president to state that torture “absolutely” works. He added that the US should “fight fire with fire” (Weaver & Ackerman, 2017). By nominating Gina Haspel, the former head of a CIA interrogation facility, to run the CIA, Trump has yet again made a statement with regard to his attitude towards these techniques (Smith, 2018). The results of the poll and the statements made by Trump are remarkable, considering the overall academic consensus that torture is not an effective means for intelligence collection (Janoff-Bulman, 2007, 430). The large amount of support for torture amongst Americans, combined with Trump’s remarks, raises questions with regard to what leads people to support or oppose torture and how this can be influenced.
Previous research has shown that a perception of threat leads to more public support for authoritarian policies (Moghaddam, 2007; Huddy et al., 2005; Davis & Silver, 2004). Part of this and other literature focusses on how negative emotions - mainly fear and anger - that arise under threat affect cognitive processes (Marcus, Neuman, MacKuen, 2000; Huddy et al., 2005; Valentino, 2008). Following this, academic effort has also gone into determining how threat leads to emotional incentives for supporting torture specifically (Piazza, 2015; Moghaddam, 2007).
Only little attention has been paid to the role of logical thinking in people’s decision to support or oppose torture. This is peculiar because even though public support for torture might thrive on negative emotions such as fear and anger, governments’ main defences for the application of torture are still based on its strategic uses and necessity. For example, the CIA tried to justify its harsh interrogation tactics on several occasions by arguing that they were successful in thwarting terrorist plots and that they led to the capture of terrorists. A commission later concluded this was not true (Apuzzo, Park & Buchanan, 2014). In addition,
1 Throughout this study the terms torture and harsh interrogation are used interchangeably. It is important to note that harsh interrogation in is in this case used as a euphemism for what I still consider to be torture. In the survey the terminology harsh interrogation is used in order to prevent respondent from being biased by the word torture, which could potentially lead to socially desirable answers.
6 the Israeli Supreme Court holds that necessity is an admissible defence for when torture is used by law enforcement agents against suspected terrorists who are believed to have planted a bomb that is likely to kill innocent civilians (Gaeta, 2004). Trump’s statement that the US should “fight fire with fire” suggests that he believes it expresses determination and
deterrence. It is the goal of this study to determine how, in addition to emotions that arise under threat, these and other different understandings of torture affect public support for it. Based on the existing literature on perception of threat, this study assumes that a perception of threat leads to negative emotional responses, which eventually lead people to be more likely to support the application of torture on terrorist suspects. This study extends the literature on public support for torture by providing an explanation for the remaining variation in support for torture amongst people who are under perceived threat. It does this by studying the effects of logical thinking on public support for torture. The research question for this study is therefor: To what extent does logical thinking explain for variation in support for torture on terror suspects? In response to this question, I argue that even though people under perceived threat are exposed to emotions such as fear and anger, they are also still susceptible to logic. In addition to emotions, variation in support for torture on terror suspects is also explained by logical thinking.
This study contributes to the existing literature by providing more detailed insights into the effects of logical thinking on public support for torture. As argued, much of the literature on public support for counterterrorism policies, and torture specifically, is aimed at how threat leads to irrational behaviour. Not much attention has been paid to the effects of logical thinking on people’s behaviour under threat.
More specifically, this study contributes to the existing literature on public support for torture by expanding the logic of torture. It expands the logic of torture by acknowledging two different debates: One is about its effectiveness in interrogations and one is about its effectiveness as a means of deterrence (Kearns, 2015). In turn, these debates each provide positive and negative logics about its effectiveness. These respectively lead to support or opposition to the use of torture on terror suspects.
In the debate about the effectiveness of torture in interrogations, one side suggests that torture is an effective means to acquire information from an uncooperative interrogee. This is labelled the positive interrogation logic. The other side of the debate suggests that this is not the case. Instead, it is not effective or even counterproductive when trying to acquire information (Janoff-Bulman, 2007, 430). This logic is labelled the negative interrogation logic. In the debate about the effectiveness of torture as a means of deterrence, one side suggests that torture is an effective means to deter potential terrorists. This is the positive deterrence logic. This debate also has another side that disputes its effectiveness. Instead it argues that torture is not effective or even counterproductive when trying to deter potential
7 terrorists. Instead it can cause a backlash-effect that would lead to more terrorist activity (Daxecker, 2017).
By measuring the effects of these four logics on people’s willingness to support torture, this study provides a more complete picture of the effects of logical thinking on public support for torture. This explains for remaining variation in support for torture under threat. Finally, this study contributes to the literature on public support for torture by using the Netherlands as a case. No research has previously been done on this topic in the
Netherlands.
The effects of logical thinking on support for torture is measured by conducting an experimental survey amongst 326 Dutch citizens. Contrary to the United States and Israel, the Netherlands have never been subjected to a large scale terrorist attack. Therefore, the results of the survey are less likely to be biased by experiences from a previous attack. On the other hand, respondents are logically expected to respond less emotionally and more logically to the scenario. The entire group of respondents was confronted with the same case that controlled for threat and the resulting emotions fear and anger. Then the group was divided in five subgroups: One control group and four experimental groups that were each
manipulated by exposing the participants to one of four different logics of torture. By controlling the groups for other factors, the full effect of exposure to different logics can be measured.
In addition to a scientific contribution, this study also makes a societal contribution. It does this by emphasizing the role of logic in people’s decision to support immoral acts. In this case torture. Research on emotions under threat has lead authors to discuss the effects and dangers of hate speech and fear mongering (Valentino, 2008, Huddy et al., ). However, it is also important to know to what extend people under threat are susceptible to logical
thinking. It implies that support for and opposition to torture are also created by public discourse. This means that in times of crisis it still benefits policy makers and activists to resort to reason, when they want to prevent immoral acts from happening. It also implies that public agents have the moral obligation to present the truth when they discuss the application of torture, which is that torture is not considered to be an effective means of interrogation or deterrence.
The findings of this study show that people under perceived threat are indeed also susceptible to logic. While this study did not manage to fully illustrate the effects of the deterrence logics on people’s decision to support torture, it did for the negative interrogation logic. This study indicates that amongst participants, the dominant reason for supporting torture is its perceived effectiveness as a tool to acquire information. This study also shows that support for torture can significantly be reduced by presenting a logic that disputes this effectiveness. This study therefore shows that in addition to negative emotions that arise
8 under threat, variation in support for torture can also be explained by logical thinking.
This thesis will be structured as follows. First, a literature review will provide an overview of relevant literature for this topic. This will be used to illustrate the gap in the literature that I attempt to fill with this study. After that, a theoretical framework will elaborate on the key concepts and theories. It also presents the hypotheses. After the theoretical framework follows a methodology section, in which I elaborate on the research design. After that follow a results section and an analysis section in which I elaborate on the main findings of this research. When I have discussed the limitations and recommendations of this study in, I shall conclude by providing its academic and practical implications.
2. Literature review
Central in the literature on the explanation for the occurrence of torture is the perception of threat. Overall, there is a consensus in the literature that a higher perception of threat leads to more support of antiterrorism policy. For example, Huddy et al. (2005, 604) show that a higher perceived threat of terrorism leads to heightened support for domestic and
international government actions to combat the threat of terrorism. In addition, a
heightened perceived threat of terrorism leads to lower support for civil liberties (Davis & Silver, 2004, 28). An overall view of the existing literature on threat perception leads to the conclusion that a higher perception of threat leads to more approval of authoritarian policies (Moghaddam, 2007, 438).
With regard to torture specifically, perception of threat has been linked to torture in three ways. First, when torture is analysed at the level of government policy, authors find that the perception of an active internal or external threat to the security of the state causes states to be more likely to engage in torture (Kelman, 2015; Einolf, 2007; Daxecker, 2017).
Davenport calls this the “law of responsiveness: “When challenges to the status quo take place, authorities generally employ some form of repressive action to counter or eliminate the behavioral threat” (2007, 7). Second, Kelman states that in addition to being a justification for torture on a policy level, this justification also forms an authorization of acts of torture on the level of implementation (Kelman, 2005, 128). Third, in addition to being linked to torture on an individual level and on a policy level, perception of threat is also linked to public
support for torture (Carlsmith & Sood, 2008; Piazza, 2015; Miron, Branscombe, & Biernat, 2010; Norris et al., 2010; O’Brien & Ellsworth, 2012; Tarrant, et al., 2012).
A large part of the literature that addresses public support for torture, does this by focussing on the role of negative emotions that arise under threat. In the following section, I
9 will elaborate on literature that addresses how emotional factors can lead to support for government policy and torture specifically. Following this, I will address literature on the logic of torture and argue that it has attracted insufficient attention in the debate on public support for torture.
2.1. Emotions
A large amount of literature addresses the role of emotions as a mediating factor between perception of threat and support for government policy, support for torture, information seeking and other cognitive processes (Gordon & Arian, 2001; Huddy, 2005; Moghaddam, 2007; Valentino, 2008; Marcus, Neuman, MacKuen, 2000).
Gordon and Arian argue that when people are under low threat, “both emotions and logic have a role in the process of deciding policy” (Gordon & Arian, 2001, 196). However, when a low threat turns into a high threat, their decision making process becomes dominated by emotions and not by logic or rational considerations (2001, 196). Their study therefore indicates that under threat, behaviour can be better explained through emotions. This is done extensively in the literature on behaviour under threat.
Some literature suggests that public support for torture and other coercive
counterterrorism measures is caused by a desire for retribution. Huddy et al. (2005) argue that a higher perception of threat causes people to be more supportive of coercive
government actions to combat the threat of terrorism. This is motivated by a desire for
retaliation. Carlsmith and Sood (2009) show that people are more supportive of torture when the suspects are perceived to be guilty. In this case, public support is also based on a desire for retribution on the perpetrators.
The effects of negative emotions on public support for torture are also illustrated by the difference in support for torture on out-group members and ingroup members (Miron, Branscombe, & Biernat, 2010; Norris et al., 2010; O’Brien & Ellsworth, 2012; Tarrant et al., 2012). Piazza (2015) tries to provide an explanation for this difference by demonstrating that a perception of terrorist threat leads to ingroup love and outgroup hate. This hate towards outgroups leads to a desire for punishment towards these outgroups. Piazza finds significant results that show an increase in support for harsh detention techniques when these are applied on outgroup members. However, he does not find significant results that indicate this same difference when it comes to harsh interrogation techniques. Piazza interprets these results as a partial confirmation of his theory. However, if indeed support for harsh detention is caused by emotion in the form of outgroup hate, then why is this not the case for harsh
10 interrogation? Support for both aspects was measured in the same questionnaire, so they were answered under the same perception of threat. Still, the results in both cases are different. The remaining variation suggests that Piazza’s explanation is a partial explanation at best. In the case of harsh interrogation, other factors apparently need to be taken into account.
In addition to the remaining variation in Piazza’s study, the discourse that support for government policy is dominantly caused by negative emotions (Valentino et al., 2008, 249; Giner-Sorolla & Maitner, 2013), does not correspond with the way in which governments defend it. They do this primarily by emphasizing its strategic use and necessity (Apuzzo, Park & Buchanan, 2014; Gaeta, 2004). This discourse tries to legitimise harsh interrogation through a perceived logic. It adds to the idea that the variation in support under threat cannot be fully explained through emotions.
Piazza provides two potential explanations for the remaining variation in support. The first is that the harsh interrogation techniques portrayed in the survey have been widely debated in the media. The second is that the activities have been made illegal (Piazza, 2015, 680). These explanations suggest that the debates about torture that followed after the scandal came out, have presented participants different logics about torture. It is possible that these different logics have changed the public perception about harsh interrogation techniques and have influenced public support for it.
It leads back to the division between emotion and logic, as it was made by Gordon and Arian (2001: 196). Above observations do not imply that they are wrong. Emotions can still be the dominant factors for public support under threat. However, these observations do suggest that logic at least could have played a role in people’s decision to support torture. Following but extending this, I argue that an additional explanation for support for torture can possibly be found in people’s logical understanding of torture.
2.2. Logic
While a vast amount of literature reflects on the role of emotions in public support for torture, not much literature has focussed on the role of logical thinking. Literature that does address the logics of torture does this mainly by addressing what constitutes torture, and debating its justification and legality (Kreimer, 2005; Levinson, 2005; Ignatieff, 2004; Bellamy, 2006; Shue, 1978).
A rare micro-level analysis that does connect logic to people’s support for torture is provided by Carlsmith and Sood (2008). They provide two arguments for supporting the use of harsh interrogation methods: one is about its use in interrogations and the other is about
11 retribution. The interrogation argument holds that harsh interrogation is an effective way to acquire valuable information from uncooperative suspects. The retribution argument holds that support for harsh interrogation originates from a desire to punish perpetrators (2008: 191). The results lead them to the conclusion that although people argue that they support torture because of its effectiveness in interrogations, their support is also motivated by a desire to gain retribution.
Even though their theory does address the role of logic in support for torture, their measurement is limited. Their approach is too narrow to explain the total logic behind support for harsh interrogation. Carlsmith and Sood’s measurement of logical thinking is completely based on the interrogation argument, while support for torture can also be based on the deterrence argument (Kearns, 2015, 4).
In addition, Carlsmith and Sood only include the interrogation argument that leads to support for torture (2008: 191). This also makes their measurement limited, because the different debates about torture also provide motivations to oppose it. Carlsmith and Sood do not measure the full effect of the total logic behind torture, because they only address it in a unidimensional way. By expanding the scope of the logic of torture, this study shows how logical thinking affects support for torture both positively and negatively: By creating support, but also by creating opposition. By expanding the scope of the logic of torture, this study aims to explain the remaining variation in support for torture that cannot be explained through the negative emotions fear and anger, that arise when people are under a perceived threat.
3. Theoretical Framework
The effect of logical thinking on people’s decision to support torture is determined by measuring the effects of four different logics on people’s willingness to support torture on a terrorist suspect. These logics are based on dominant discourses in the debates on the effectiveness of torture. By looking at the existing scientific literature on torture, the following section elaborates more on what these different logics entail.
The different logics that are utilised in this study are derived from the debates that are distinguished by Kearns (2015). Kearns distinguishes two arguments that are used to support torture: The interrogation argument and the deterrence argument. The interrogation
argument holds that harsh interrogation is an effective means to acquire information from a terrorist suspect. The deterrence argument holds that torture is an effective means to deter potential terrorists. Because these lines of reasoning logically lead to support for torture, they are labelled as the positive logics of torture.
12 By addressing torture through the interrogation logic and the deterrence logic, this study is linked to two different debates on torture: the interrogation logic addresses the debate about torture on a tactical level, while the deterrence logic address the debate about torture on a strategic level.
This study further expands the logic of torture by also including logics in the debates that dispute its effectiveness. These lines of reasoning logically lead to opposition and are therefore labelled the negative logics of torture. These logics are the negative counterparts of the interrogation logic and the deterrence logic.
Existing literature shows that the positive interrogation logic is primarily based on wrong assumptions. The facts on the other hand show that torture is ineffective and even counterproductive (Janoff-Bulman, 2007). The negative interrogation logic therefore holds that torture is not effective or even counterproductive when trying to acquire information. With regard to the deterrence logic, existing literature holds that torture is not effective or even counterproductive when trying to deter potential terrorists. Instead it can cause a backlash-effect that leads to more terrorist activity (Daxecker, 2017). Therefore, the negative interrogation logic holds that torture is not effective or even counterproductive when trying to deter potential terrorists.
In the following section, I will elaborate more on what these different logics hold and what they are based on. First I will discuss the different views with regard to the interrogation argument. After that, I will discuss the different views with regard to the deterrence
argument.
3.1. The Interrogation Logics
When Donal Trump stated that he supports the application of torture because he absolutely believes it works, he did that because “intelligence people at the highest level of intelligence” told him that (Persaud & Bruggen, 2017). This statement is peculiar, because there seems to be a general consensus amongst scholars and interrogation experts that torture is not an effective means for intelligence gathering: “The issue isn’t just that you want people to ‘talk’, […] but you want them to talk reliably. False information which sends you off on a wild goose chase ties up valuable resources, often leading to even more disastrous consequences”
(Persaud & Bruggen, 2017). The resulting question is why people would still believe that torture is an effective means of intelligence collection when academics and interrogation experts concur that it is not. Using Janoff-Bulman’s article (2007) as a base, the next section addresses the debate on the tactical use of torture. It illustrates that the positive interrogation logic is primarily based on assumptions and not on facts. The negative interrogation logic on
13 the other hand, is embedded in scientific literature.
In this study, the positive interrogation logic is based on the idea that torture is an effective way to acquire information from an uncooperative interrogee. As stated, this belief in the effectiveness of torture is peculiar, because the dominant discourse in the scientific literature and amongst interrogations experts is that torture is not an effective means for intelligence gathering (Janoff-Bulman, 2007; Persaud & Bruggen, 2017; Redlich, Kelly & Miller, 2014). Support for torture that is based on its perceived effectiveness must therefore be explained through misperceptions. Janoff-Bulman (2007) provides four assumptions about torture that explain why torture is (wrongfully) perceived by some to be an effective means to acquire information from an interrogee.
First, Janoff-Bulman argues that support for torture is caused by the misperception that compliance leads to reliable and accurate information. He argues that people “have an implicit understanding that extreme coercion is likely to produce its desired behavioural effect” (2007, 430). However, when one is trying to acquire intelligence, one is looking for accurate, reliable information, and not for precise submission: “Effectiveness in intelligence collection is not measured by readily available indices of behavioural compliance, but by the accuracy and reliability of information provided” (2007, 430). Although torture might make someone submissive, it does not guarantee that the subject will provide reliable information. The second factor mentioned by Janoff-Bulman has to do with the dehumanization of terrorists. While rapport-building is considered to be the most effective interrogation
technique, these techniques are not considered appropriate for less worthy terrorists (2007, 431): “[P]eople may erroneously assume that information from cruel, bad, harsh enemies can only be produced by similarly cruel, bad, harsh techniques” (2007, 432). This is because people need motives and needs they can identify with in terms of common humanity in order to see the value of relationship-based persuasion strategies (2007, 432). By derogating terrorists, people deny them their humanity. Their perceived inhumanity creates the assumption that they are more likely to be responsive to inhuman treatment (2007, 432). This is also partly attributable to a scenario that involves a ticking time bomb (2007, 431). This is a scenario in which a terrorist has planted a bomb and the interrogator believes that the interrogated terrorist knows its location, which he is not willing to share (Bellamy, 2006: 130). This scenario forms the basis for the utilitarian defence of torture. This holds that “the risks of terrorist attacks outweigh the moral and practical costs associated with using methods that many people consider torture” (Carlsmith & Sood, 2008: 191). It is a dilemma in which both choices are morally problematic. In this case, the application of torture is considered to be the ‘lesser evil’ (Ignatieff, 2004, 136). “In this situation, time pressure precludes the establishment of rapport, relationship, or any real understanding of the detainee” (Janoff-Bulman, 2007, 431). The terrorist’s wellbeing is derogated to be less
14 worthy than those of his potential victims
The third misperception that can lead to support for torture is the underestimation of people’s ability to resist it. Considering the pain, people expect that they would break and give in to the torturers demands. However, people often do not take into account other reactions that may lead to resistance instead of submission (2007, 432).
Amongst these less obvious reactions are dissociative processes. These processes provide “protection via psychological escape” (2007, 433). They involve “detachment, constricted consciousness, and the minimization of pain perception” (Herman, 1992; Janoff-Bulman, 2007, 433). Through these processes, “a torture victim may minimize the experience of pain and maximize the possibility of resistance” (2007, 433).
Another reaction is the attribution of meaning by the interrogee. This means that the pain becomes more bearable as meaning is attributed to the suffering (433). Strong devotion to a cause is likely to be associated with psychological strength in the face of torture (433). The last factor that Janoff-Bulman mentions that may cause people to support torture is the need for vengeance. In this case, people’s support for torture is caused by a desire for vengeance instead of the need for information. Carlsmith and Sood (2008) refer to this motivation as retribution and argue that it is caused by a desire to punish terrorists. What Janoff-Bulman shows is that people who perceive to be under threat, do not always have sufficient knowledge about the effectiveness of torture. In these situations, they tend to fill this void of knowledge by making their own connections and assumptions that often turn out to be wrong. As illustrated by Janoff-Bulman, the fact that some people still support the use of torture is caused by incorrect assumptions about the relation between compliance and accurate information, about the ineffectiveness of rapport-building techniques on inhumane opponents, about people’s ability to resist torture and by misperceptions about their own motivations. The idea that a void in people’s knowledge about torture causes them to make wrong assumptions about its effectiveness adds to the idea that people are also susceptible to logic to fill this void.
What this literature shows is that the debate on the effectiveness of torture can be divided into two camps: One camp that (wrongfully) understands torture to be an effective means of intelligence gathering and one camp that believes it is not effective or even
counterproductive. Regardless of how well they are supported, to test the complete effect of the interrogation argument on people’s decision to support torture, both logics have to be tested. Therefore, the interrogation logics lead to two hypotheses. For the positive
interrogation logic, the hypothesis is as follows:
H1: People who are exposed to the positive interrogation logic are more likely to support harsh interrogation methods on the terrorist suspect.
15 For the negative interrogation logic, the hypothesis is as follows:
H2: People who are exposed to the negative interrogation logic are less likely to support harsh interrogation methods on the terrorist suspect.
3.2. The Deterrence Logics
The deterrence logics are derived from the deterrent argument that was presented by Kearns (2015). The deterrence logics originate from a debate about torture that is not aimed at its tactical use in interrogations, but at its wider, strategic use. This debate is very well
illustrated by Daxecker (2017). I will therefore use Kearns’ and Daxecker’s works as guidelines for describing this debate.
The deterrence logic is based on the idea that torture can be used to deter potential terrorists by punishing the perpetrators (Kearns, 2015, 4). The deterrence theory assumes that humans are rational actors who act out of self-interest. Punishment will therefore deter them from violating the law (2015, 4). However, as is stated by Kearns, terrorists often have collective goals, due to wich deterrence involves more than the impact on an individual (2015, 4). Since research has largely focused on deterrence on an individual level, not much can be said about the effectiveness of deterrence on a group level (2015, 4)
So far, no research has specifically addressed the direct effect of torture on terrorist groups. However, literature on aggregate repression shows little support for deterrent effects (Daxecker, 2017, 1264). Lyall (2009) showed that Chechnyan villages that were shelled by Russian artillery actually showed less insurgent attacks than control groups. However, this is a single-case study and therefore does not allow to make wider claims (Daxecker, 2017, 1264). Other studies are “conditional on the timing of repressive action, the type of
government or nonstate actor involved, or the particular type of repression applied” (2017, 1264).
A substantial amount of literature suggests that repressive government policy does not form a deterrence, but actually has an opposite effect. This is referred to as a backlash effect. Instead of deterring potential terrorists, coercive policies will only further alienate members of marginalized groups (2017, 1264). In addition, it will drive innocent victims towards terrorist organisations and radicalize those who were already sympathetic to them (2017, 1264). Taken together, scientific research argues that “coercive government responses to terrorism can radicalize sympathizers, increase recruitment, and undermine community support for counterterrorism policies, leading to backlash and increased terrorist activity” (Daxecker, 2017, 261).
16 What this overview of the debate on the deterrent effect of torture shows, is that it can be divided into two camps: One camp that understands torture to be an effective means of deterrence and the other camp that believes it is not effective or even counterproductive. As was the case for the interrogation logic, the positive deterrence logic is less well supported by facts than the negative logic. This logic also seems to be based on assumptions instead of facts. Regardless of how well they are supported, to test the complete effect of the deterrence argument on people’s decision to support torture, both logics have to be tested. Therefore, the deterrence logics leads to two hypotheses. For the positive interrogation logic, the hypothesis is as follows:
H3: People who are exposed to the positive deterrence logic are more likely to support harsh interrogation methods on the terrorist suspect.
For the negative deterrence logic, the hypothesis is as follows:
H4: People who are exposed to the negative deterrence logic are less likely to support harsh interrogation methods on the terrorist suspect.
It is important to make notice of the fact that the interrogation logics and the deterrence logics do not occur separately from each other. In this case it can be argued that a possible deterrent effect is a by-product of the interrogation logic. A suspect is in the first place
tortured to acquire the information that he or she possesses. As a consequence of this action a deterrent effect comes into play. Torture can also be supported because of the positive
interrogation logic. However, the participant can also be aware of the fact that this could cause a backlash. In this case the participant would have to consider which logic he or she finds most convincing.
4. Methodology
The aim of this study is to provide a better picture of the effects of logical thinking on people’s decision to support or oppose torture. Not much has been written about the effects of logic on public support for torture. Carlsmith and Sood (2008) conducted research on different reasons for supporting torture by doing an experimental survey. This study adds to their research by analysing the effects of an expanded logic of torture. This has lead to four different logics for supporting or opposing torture.
17 experimental survey. The scenario controls for the main emotions that arise under threat: fear and anger. This provides a unique chance to isolate the effect of different logics on people’s decision to support or oppose torture.
4.1. Population and Sample
For this study it was decided to research a sample of Dutch citizens. This might seem odd, considering the fact that this study is mostly inspired by and based on American literature. A sample of Dutch citizens is used because, contrary to many other countries, the Netherlands have not yet experienced a large size terrorist attack on its soil. The Dutch intelligence service, the AIVD, states that it has prevented four terrorist attacks between 2014 and 2018 (NOS, 2018). However, the only Jihadist terrorist attack that actually occurred in the
Netherlands was the murder of Theo van Gogh in 2004 (Nu.nl, 2004). The Dutch population has not yet been exposed to a large scale terrorist attack on its own soil. Therefore it is
expected that the participants to the study possibly have a lower perception of threat than people from countries that have more experience with large scale terrorist attacks. This could potentially cause them to react less emotionally and with more logical thinking.
Participants of the survey form a convenience sample from the population of the Netherlands. All Dutch citizens were allowed to participate in the survey. The only criteria were that they had to be Dutch citizens and 18 years old or older. Due to a lack of time and resources, it was decided to stick to a convenience sample and gather respondents using the snowball method. Participants were approached through Facebook and e-mail. Furthermore, the link to the survey was shared by family, friends and other social connections. In addition, participants had the option to participate in a lottery in which they could win one of four concert gift cards. These measures resulted in 326 usable participants.
4.2. Experimental design
As stated, this study is based on an experimental survey that measures the effect of logical thinking on public support for torture. This is done by measuring the effect of exposure to four different logics on people’s willingness to support torture. The survey for this study was constructed and distributed online via Qualtrics. This made it possible to let respondents take the survey without having to physically distribute it. Qualtrics also made it possible to export the data straight to SPSS. SPSS was used to analyse the survey data.
18 treatment of terrorist suspects. All respondents were shown the same scenario. The purpose of this scenario is to control for external factors that might influence support for torture. When measuring logic, the most important interfering variables are emotions. As stated, the most prominent emotions that arise under threat are fear and anger. In order to simulate the fear and anger that arise under terrorist threat, it is essential to know what causes these distinct emotions. Giner-Sorolla and Maltner, show that “the more a terrorist group is unjust, the more anger arises, whereas the more it is seen as powerful, the more fear arises” (2013, 1069). By mentioning that the terrorist organisation primarily aims its attacks on large groups of civilians, the scenario creates an image of the organization as being unjust. This should lead to anger. In addition, by emphasizing that experts consider the organisation as a powerful terrorist organisation with many active supporters in Western countries, the organization’s power is emphasized, which should logically lead to fear. The scenario that was presented to participants is as follows:
News media report that a large terrorist attack has occurred in Amsterdam. It has killed more than fifty people and has injured a hundred more. The authorities have issued the highest terrorist alert. There are indications that another major terrorist attack is
imminent. The security services are holding an organisation responsible that is known to primarily aim its attacks on large groups of civilians. Experts consider the organisation as a powerful terrorist organisation with many active supporters in Western countries. It has proven on several occasions that it is able to conduct large scale terrorist attacks in these Western countries.
In the aftermath of the attack, the authorities have apprehended a potential terrorist who is suspected of being involved in the terrorist organisation and of having valuable
information. Experts in the security services state that there is a large certainty that the suspect is guilty of terrorism and possesses vital information about the organisation. Considering the circumstances it is decided to subject the suspect to a government
programme that uses aversive, degrading, painful interrogations that, in some cases, cause permanent physical and psychological scars.
After being shown the scenario, respondents were randomly confronted with none or one of four logics. After they were presented the scenario, and in most cases a logic, respondents were asked whether they supported “harsh interrogation techniques” on the terrorist suspect or not. The torture techniques are referred to as harsh interrogation techniques in order prevent a social desirability bias. By using the euphemism “harsh interrogation techniques”, respondents might feel less forced to provide a socially desirable answer.
To examine the mechanisms through which participants came to their decision to support or oppose torture, respondents were asked afterwards which logics played a role in
19 their decision. From these logics, respondents had to choose one logic that was the dominant reason for supporting or opposing the harsh interrogation techniques. Participants were then asked questions about control variables and about additional demographics. After collecting the data from the respondents, the data were exported to an SPSS-dataset. Using this data-set, I examine the relationship between exposure to the different logics and people’s decision to support torture.
4.3. Dependent variable
Support for torture was measured by asking whether respondents supported the application of harsh interrogation techniques. Participants could only answer this question by answering yes or no. This is done because the choice to support torture or not poses a difficult dilemma. In this case it was necessary to limit the answers to yes and no because it shows whether logical thinking actually causes a shift in opinions.
4.4. Independent variables
As stated, together with the case, participants were randomly assigned with none or one of four different logics. The logics are derived from the two debates that revolve around the effectiveness of torture: The tactical debate on torture, which focusses on its effectiveness in interrogations, and the strategic debate on torture, which focusses on its effectiveness in deterrence. As stated in the theoretical framework, these debates provide four logics about the effectiveness of torture. The exposure to each of these arguments serves to confirm or debunk one of the four hypotheses about their perceived effects.
The sentences corresponding to the logics are as follows: The positive interrogation logic:
Experts support the application of these harsh interrogation methods. They argue that it is an effective means to acquire information from terrorist suspects, which he will otherwise not disclose. They argue that the value of the information legitimizes this approach.
The negative interrogation logic:
Experts oppose the application of these harsh interrogation methods. They argue that these treatments are ineffective and even counterproductive. They argue that due to these
20 The positive deterrence argument:
Experts support the application of these harsh interrogation methods. They argue that it is an effective means against terrorism. they argue that the application of these methods will deter other terrorists and will keep them from committing more terrorist attacks.
The negative deterrence argument:
Experts are against the application of these harsh interrogation techniques. They argue that it is not an effective means against terrorism. They argue that the application of these methods will actually “radicalize sympathisers, increase recruitment and undermine community support for counterterrorism policies, leading to backlash and increased terrorist activity” (Daxecker, 2017, 1261).
4.5. Control variables
As already stated, a sample of Dutch citizens forms a good case for studying the role of logical thinking in support for torture, because they, as a people, have not yet been subjected to large scale terrorist attacks. The participants can be presented a scenario that simulates the effects that are assumed to be caused by a terrorist threat. The effects that are experienced following the scenario are unlikely to be polluted by the effects caused by previous terrorist attacks. One variable that might interfere in the correlation between the different logics and support for torture might be overall support for the government. When people do not support their government they might not support coercive policies not because they think it does not work, but because they do not see the government as suited or capable of doing it. Participants are asked about their support for the government in the survey and the effect can be controlled for statistically.
5. Results
As stated, participants in the survey were first presented a scenario, followed by one or none of the previous mentioned sentences that were each based on one of the logics. After
participants were confronted with the scenario and possibly with one of the four logics, they were first asked a polar question about their willingness to support the use of torture on the terrorist suspect. Respondents could choose to either support torture or not. The effect of the exposure to different logics is tested in two ways. First, the validity of the study was enhanced
21 by asking participants to state their motivations for supporting or opposing torture. These answers can be used to confirm whether the presented logic was the actual reason supporting or opposing torture. Second, the validity was enhanced by conducting a binary regression analysis with exposure to the different logics as the independent variable and support for torture on the terrorist suspect as the dependent variable. The regression analyses indicate which differences in means of support are significant.
5.1. Randomization Test
Before analysing the effects of the logics on public support for torture, a randomization test had to be conducted to enhance the internal validity of the study. Randomization was tested by comparing the mean levels for several demographic variables between the control group and the treatment groups using a one-way analysis if variance (ANOVA). The variables are gender, age, household income, the highest diploma received, political orientation, life conviction, parents’ origins and overall support for the government. Table 1 shows the results of the one-way ANOVA test.
Table 1. Variable F Sig. Gender 0,365 0,834 Age 1,170 0,324 Household income 0,504 0,733 Highest diploma 1,235 0,296 Political orientation 1,149 0,334 Life conviction 0,185 0,946 Parents’ origin 0,399 0,810
Overall support government 0,359 0,837
As can be seen in table 1, there is no significant difference between the groups based on any of the demographic variables. Table 1 shows that the lowest level of significance is 0,296, which is for highest diploma received. This value is sufficient for the randomisation test because it exceeds the α of 0,05 for a 95% significance level. It shows that the differences between the means in different groups is not statistically significant. The lack of significant differences between the groups suggests that the randomization of the assignment of
respondent to each of the groups has succeeded. The results from the manipulation tests are unlikely to be skewed by significant differences between the groups based on these variables. The regression analysis therefore does not have to be controlled for any of these variables.
22
5.2. Descriptive Statistics
5.2.1. Support for Torture
After participants were confronted with the scenario and possibly with one of four logics, they were first asked a polar question about their willingness to support the use of torture on the terrorist suspect. The survey result for this question are shown in table 2. Figure 1 provides a graphic overview of public support for the use of torture over all groups.
Figure 1
The first thing that stands out is that the positive logics do not have the expected effects on support for torture. Looking at the hypotheses, it was expected that the groups that were presented with the positive logics would have a higher level of support for torture. This was not the case. Instead, the group that was presented the positive interrogation logic was less supportive of torture than the control group: 43,1% against 56,1%. Contrary to what was expected, the positive interrogation logic therefore seems to have a negative effect on public support for torture. In addition, the other positive logic, the positive deterrence logic, does not seem to have a strong effect on support for torture. Support for torture in this group was slightly higher than in the control group: 56,9%. It therefore does show the expected
directional effect, however, it is small to none.
Contrary to the positive logics of torture, the negative logics do seem to have the anticipated effects on support for torture. Looking at the hypotheses, it was expected that
56,1 43,1 39,7 56,9 41,9 43,9 56,9 60,3 43,1 58,1 0 10 20 30 40 50 60 70 Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic
Support for Torture on Terrorist Suspects
23 both groups presented with negative logics would have a lower level of support for torture. The group that was presented the negative interrogation logic indeed shows a lower level of support for torture: 39,7% against 56,1%. A difference of 29,8%. The negative deterrence logic also seems to have the expected effect on support for torture. The group that was presented with the negative deterrence logic shows a lower level of support for torture of 41.9%. A difference of 25,3%. These results suggest that both logics had the anticipated effect on support for torture.
Overall, figure 1 illustrates that only the negative logics of torture seem to have the expected negative effect on support for torture. The positive logics either had a negative effect or hardly any effect. In the next section, the causal mechanism for support is tested by
analysing participants’ choices of logics for supporting or opposing torture.
5.2.2. Reasons for Support and Opposition
As stated, the effects of the different logics are first tested by asking respondents to affirm their motivations for supporting or opposing torture. After participant were asked whether they supported torture or not, they were asked to indicate why they chose to do so. The group of supporters and the group of opposers were each given three options in the survey. The questions, the possible responses, and the results for these questions can be seen in table 2. For supporters, the first two options reflected the two positive logics that logically lead to support for torture. Option 1 argued that harsh interrogation can provide valuable
information. Option 2 argued that it can deter potential terrorists and that it can keep them from committing more attacks. A third option was included that argued that the suspect deserves to be punished. This option was provided because it was argued by Carlsmith and Sood (2008) to be an additional reason for support for torture. Lastly, people were provided a fourth option to indicate that their support for torture was based on another reason.
For opposers of torture, the first two options reflected the negative logics that logically lead to opposition to torture. Option 1 argued that harsh interrogation does not provide valuable information. Option 2 argued that it will lead to a backlash-effect: It will radicalize sympathisers, increase recruitment and undermine community support for counterterrorism policies. A third option was included that argued that harsh interrogation is morally wrong. Lastly, opposers were also provided a fourth option to indicate that their support for torture was based on another reason.
Participants were first asked to point out which of the abovementioned logics were influential in their decision to support or oppose torture. Multiple answers were allowed. Subsequently, participants were asked to choose their dominant motivation for supporting or
24 opposing torture. The results for the questions amongst supporter are graphically illustrated in percentages in graphs 2 and 3. The results for the questions amongst opposers are
illustrated in percentages in graphs 4 and 5. An exact overview of the numbers can be found in table 2.
Table 2 Control Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic N
Do you support the use of harsh interrogation techniques ont terrorist suspects?
Yes 56,1% 43,1% 39,7% 56,9% 41,9%
No 43,9% 56,9% 60,3% 43,1% 58,1%
326
For what reasons do you support the use of harsh interrogation techniques on the terrorist suspect?
It can provide valuable information 91,9% 78,6% 92,6% 100% 84,6%
It can deter potential terrorists 56,8% 25% 44,4% 41,7% 30,8%
The suspect deserves to be punished 21,6% 10,7% 48,1% 36,1% 30,8%
Other reason 5,4% 35,7% 7,4% 13,9% 11,5%
154
What is your main reason for supporting the use of harsh interrogation techniques on the terrorist suspect?
It can provide valuable information 73% 78,6% 59,3% 66,7% 69,2%
It can deter potential terrorists 24,3% 3,6% 22,2% 27,8% 15,4%
The suspect deserves to be punished 2,7% 7,1% 18,5% 2,8% 7,7%
Other reason 0% 10,7% 0% 2,8% 7,7%
154
For what reasons do you oppose the use of harsh interrogation techniques on the terrorist suspect?
It does not provide valuable information 21,4% 21,6% 48,8% 7,1% 22,2%
It will lead to a backlash-effect 42,9% 45,9% 58,5% 60,7% 66,7%
It is morally wrong 78,6% 86,5% 68,3% 67,9% 77,8%
Other reason 35,7% 29,7% 14,6% 25% 22,2%
170
What is your main reason for opposing the use of harsh interrogation techniques on the terrorist suspect?
It does not provide valuable information 3,6% 8,1% 14,6% 7,1% 13,9%
It will lead to a backlash-effect 10,7% 24,3% 31,7% 32,1% 25%
It is morally wrong 67,9% 62,2% 48,8% 46,4% 50%
Other reason 17,9% 5,4% 4,9% 14,3% 11,1%
5.2.2.1. Reasons for support
When the answers that were given after exposure to different logics are compared to the answers that were given in the control group, a couple of things stand out. First, regardless of which logic participants were presented, the positive interrogation was by far the dominant reason for supporting harsh interrogation techniques on the terrorist suspect amongst all groups. However, despite this observation, there is still observable variation in
argumentation between the groups.
Looking at figures 2 and 3, the positive interrogation logic seems to have a small anticipated effect on people’s argumentation for supporting torture. Considering that the positive interrogation logic confirms the effectiveness of torture in interrogations, as is expected that more people in this group will provide the positive interrogation logic as an argument for supporting torture. When people are allowed to provide multiple answers, they actually provide the interrogation logic less often as a motivation for supporting torture: 78,6% against 91,9% for the control group. However, when asked about their dominant motivation for supporting harsh interrogation on the terrorist suspect, a larger portion of the participants indeed chose the interrogation logic as their main motivation for support: 78,6% against 73% for the control group.
Figure 2 91,9 78,6 92,6 100 84,6 56,8 25 44,4 41,7 30,8 21,6 10,7 48,1 36,1 30,8 5,4 35,7 7,4 13,9 11,5 0 10 20 30 40 50 60 70 80 90 100
Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic
Reasons for Supporting Torture
It provides valuable information It will deter potential terrorists Suspect deserves to be punished Another reason
27 Figure 3
The results also show a potential effect of the negative interrogation logic on people’s argumentation for supporting torture. Considering that the negative interrogation logic disputes the effectiveness of torture in interrogations, it was expected that less people in this group would provide the positive interrogation logic as an argument for supporting torture. Despite the fact that the logic suggested otherwise, when asked about multiple relevant motivations, a large amount of people still uses the effectiveness of torture in interrogations as an argument for supporting torture. 92% Of the participants in this group stated that it had a role in their decision. However, only 59,3% of the participants in the group still argued that it was their primary reason for supporting torture, against 73% in the control group. In addition to this argument, a larger portion of the participants provided the punishment argument as one of the motivating logics: 18,5%. This is a much larger amount than in the control group: 2,7%. The portion that based its support for torture on the deterrence logic did not differ much from the control group: 22,2% against 24,3%.
The results show that participants presented with the negative interrogation logic indeed less often provide the positive interrogation logic as their main motivation for
supporting torture. Instead, more participants provided the punishment argument as one of the motivating logics. These results suggest that the negative interrogation logic did have effect, but that the participants who still supported torture, provided different reasons for doing so. 73 78,6 59,3 66,7 69,2 24,3 3,6 22,2 27,8 15,4 2,7 7,1 18,5 2,8 7,7 0 10,7 0 2,8 7,7 0 10 20 30 40 50 60 70 80 90 100
Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic
Most Important Reason for Supporting Torture
It provides valuable information It will deter potential terrorists Suspect deserves to be puinshed Another reason
28 Participants who were presented the positive deterrence logic still primarily based their support for torture on the positive interrogation logic. All participants in this group stated that it played a role in their decision to support torture and 66,7% of them stated that it was the primary reason for them to support torture. Considering that the positive
deterrence logic confirms the effectiveness of torture as a means of deterrence, it was expected that more people in this group will provide the positive interrogation logic as an argument for supporting torture. When looking at figure 2 for multiple answers, it can be seen that the deterrence logic did not play a larger role in people’s decision to support torture. Relatively fewer participants stated it to be relevant in their choice to support torture: 41,7% of the respondents against 56,8% of the respondents in the control group. However, figure 3 shows that when asked about the dominant reason for supporting torture, indeed more participants provided its deterrent effect as the dominant reason for supporting torture. However, this difference is only small: 27,8% against 24,3% for the control group. These results suggest that the positive deterrence logic did not have a large effect on participants’ argumentation for supporting torture.
The results show that a majority of the participants who were exposed to the negative deterrence logic also provided the interrogation argument as the dominant logic for
supporting torture: 69,2%. Considering that the negative deterrence logic disputes the effectiveness of torture as a means of deterrence, it was expected that less people in this group will provide the positive interrogation logic as an argument for supporting torture. The results confirm this. They show that people who were exposed to the negative deterrence logic provided the positive deterrence logic less often as a motivation for support than the control group: 15,4% against 24,3% for the control group.
5.2.2.2. Reasons Opposition
Result of the questions about the reasons for opposing torture are presented in table 2 and graphically illustrated in figures 4 and 5. When looking at these results, a couple of things stand out. The first thing that stands out is that the distribution of the choice of arguments seems to be similar for all groups. The absolute numbers differentiate, but the variation in the argumentation is generally the same over all groups: By far, most participants oppose torture because they consider it morally wrong. After that people oppose torture because they consider it to be a method that leads to a backlash effect: it “can radicalize sympathizers, increase recruitment, and undermine community support for counterterrorism policies” (Daxecker, 2017, 1261,). Only lastly do people oppose torture because they consider it an ineffective means for valuable intelligence gathering.
29 figure 4 show hardly any difference when compared to the results of the control group.
Considering that the positive interrogation logic confirms the effectiveness of torture as a means of deterrence, it was expected that less participants in this group would provide the negative interrogation logic as an argument for opposing torture. This was not the case. For both groups, the arguments all seem equally relevant for participants’ choices to oppose torture. However, when looking at the primary reasons for opposing torture, participants provided the negative interrogation logic and the negative deterrence logic more often as reasons for opposing torture: respectively 8,1% and 24,3% against 3,6% and 10,7% for the control group. It is important to point out in this case that the group that was confronted with the positive interrogation logic more often pointed out the negative deterrence logic as a dominant reason for opposing torture. This is the opposite to what was expected.
Figure 4 21,4 21,6 48,8 7,1 22,2 42,9 45,9 58,5 60,7 66,7 78,6 86,5 68,3 67,9 77,8 35,7 29,7 14,6 25 22,2 0 10 20 30 40 50 60 70 80 90 100
Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic
Reasons for Opposing Torture
30 Figure 5
When looking at the group that was presented with the negative interrogation logic, the first thing that stands out is that overall the group took more arguments into
consideration than the control group. Considering that the negative interrogation logic disputes the effectiveness of torture in interrogations, it was expected that more participants in this group would provide the negative interrogation logic as an argument for opposing torture. This was indeed the case. A much larger portion of the group indicated that indeed the negative interrogation logic had an effect on their choice to oppose torture: 48,8% against 21,4% for the control group. When asked about their primary reason for opposing torture, the group also provided the negative interrogation logic much more often than the other groups as the main argument for opposing torture: 14,6% against 3,6% for the control group. Overall, the results for this experimental group suggest that exposure to the negative interrogation logic indeed provided people with an additional reason to oppose torture. The first thing that stands out when looking at the group that was presented with the positive deterrence logic is the strong increase of the negative deterrence logic as a reason for opposing torture. This is contrary to what was expected. Considering that the positive
deterrence logic confirms the effectiveness of torture as a means of deterrence, it was
expected that less participants in this group would provide the negative interrogation logic as an argument for opposing torture. Both when asked for multiple reasons and when asked about the dominant reason for opposing torture, this group provided the negative deterrence
3,6 8,1 14,6 7,1 13,9 10,7 24,3 31,7 32,1 25 67,9 62,2 48,8 46,4 50 17,9 5,4 4,9 14,3 11,1 0 10 20 30 40 50 60 70 80 90 100
Control Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic
Most Important Reason for Opposing Torture
31 argument much more often than the control group: 60,7% against 42,9% when asked for multiple reasons and 32,1% against 10,7% when asked for the primary reason for opposing torture. The fact that such a large portion of the group exposed to the positive deterrence argument actually provides the negative deterrence argument as a primary reason for opposing torture, suggests that the presentation of the positive deterrence argument has probably triggered opposers to argue against it.
The group that was presented with the negative deterrence logic again shows a similar distribution of the arguments that were provided. Again, most people primarily opposed based on the moral argument: 50%. Because the negative deterrence logic disputes the effectiveness of torture as a means of deterrence, it was expected that more people in this group would provide the negative interrogation logic as an argument for opposing torture. Indeed, the negative deterrence logic was used more often than in the control group. The group shows the largest portion of participants that took it into consideration deciding to oppose: 66,7% against 42,9% in the control group. Also, more participants stated it to be the dominant reason: 25% against 10,7. However, it is notable that this is less than in the groups presented with the positive deterrence argument and the negative interrogation argument. The portion of participants that took the negative interrogation logic into account in its choice to oppose torture was roughly equal to that in the control group: 22,1% against 21,4%. A larger portion stated it to be the primary motivation for support: 13,9% against 3,6% for the control group.
5.3. Regression Analysis / manipulation test
In addition to the descriptive statistics, the effect of the different logics on support for torture is further tested by conducting binary regression analyses with exposure to the different logics as the independent variables and support for torture on the terrorist suspect as the dependent variable. In order to measure the effects of the different logics on support for torture, the means of support for all four groups were separately compared to the mean of the control group in four different regression analyses. The results for the analyses are shown in table 3. A separate complete table for each of the regression analyses can be found in the Appendix.
Table 3 shows that after conducting the regression analyses, only the negative interrogation logic turned out to have a significant effect on support for torture for a
statistical significance level of 90%. The results from the other three logics correlate with the results in the descriptive analysis, but none of these results turned out to be significant. In addition to the descriptive statistics, the regression analyses suggests that the
32 group that was presented the positive deterrence logic was less supportive of torture than the control group. This is shown by the B-value, which indicates that exposure to the positive interrogation logic had a negative effect on support for torture: -0,522. The odds ratio suggests that participants were only 0,593 times as likely to support torture as people who have not been exposed to the logic.
Table 3 Positive interrogation logic Negative interrogation logic Positive deterrence logic Negative deterrence logic B -0,522 -0,661* 0,035 -0,569 S.E. 0,352 0,351 0,352 0,357 Exp(B) 0,593 0,516* 1,036 0,566 Nagelkerke’s R² 0,022 0,035 0,000 0,026 N 131 134 131 128 Statistical Power 0,566 0,778 0,052 0,633 *significant for α=0,1 ** significant for α=0,05
The negative interrogation logic turned out to be the only logic with a significant effect on support for torture. The B-value confirms that exposure to the logic had a negative effect. The odds ratio shows that participants in this group were only 0,516 times as likely to support torture as people who have not been exposed to the logic.
The results in table 3 also suggest that the positive deterrence logic only had a small positive effect on support for torture. This is indicated by the B-value of 0,035. The odds ratio of 1,036 suggests that participants were approximately just as likely to support torture when they are confronted with the logic as when they are not.
Lastly, the results in table 3 suggest that the negative deterrence logic had a negative effect on support for torture, for the B-value equals -0,569. The odds ratio indicates that participants that were exposed to the negative deterrence logic were 0,566 times as likely to support torture than those who were not. However, as stated, this effect and those for the positive interrogation logic and the positive deterrence logic did not turn out to be significant
33 for a confidence level of 90%.
5.4. Power calculation
The binary regression analyses made it possible to determine the chance of making a type I error (α) while conducting this research. This is the chance of wrongfully rejecting the null hypothesis. In addition to determining the chance of making a type 1 error, it is also important to determine the chance of making a type 2 error (β). This is the chance of wrongfully failing to reject the null-hypothesis. This makes it possible to determine the statistical power for each of the tests (1-β). Because this study has four hypotheses and four mean comparisons, four power calculation needed to be conducted. The results for the power calculations are displayed in table 3.
The results show that the statistical power differentiates strongly for the different groups. The comparison of means between the control group and the group presented with the positive interrogation logic provides a statistical power of 0,566. This means that even though the chance of wrongfully rejecting the null hypothesis is too high in order to accept the hypothesis, there is also a chance of roughly 44% that the null hypothesis is wrongfully not rejected and that unbeknownst there is still a significant effect.
The statistical power for the test between the control group and the group presented with the negative interrogation logic is 0,778. This means that there is a chance of 22,2% percent that the null hypothesis is wrongfully not rejected. However, as the regression analysis already indicated, a significant effect has been found for a confidence level of 90%. The test regarding the positive deterrence logic showed a very low statistical power: 0,052. This means that there is still a 94,8% chance that the model wrongfully did not reject the null hypothesis.
The last model, comparing the means of the control group and the group that was presented the negative deterrence logic, shows a statistical power of 0,633. This means that there is still 36,7% chance that there is an effect that has not been detected by this model. The power calculations indicate that although only one model provided a significant effect, there is still a relatively high chance for each of the three other models, that there is a significant effect present that has not yet been found.
6. Discussion
It is the aim of this study to provide an explanation for the remaining variation in support for torture amongst people who are under perceived threat. I argued that the remaining variation
