The Role of Feelings of Guilt in Confessions to Unethical Behaviour Anouk Bercht
University of Amsterdam
Student number: 10196838
Brain and Cognition, Psychology, University of Amsterdam Final version
Date: 27-01-2017
First Supervisor: Linda M. Geven Doctoral Candidate
Department of Clinical Psychology
Abstract
After the performance of unethical behaviour people often feel guilty. Although research suggests feelings of guilt are one of the main reasons people confess, the direct consequences of mood on confessing behaviour have rarely been studied. Since confessions are mainly studied in retrospect (after the confession was made), the question if people are actually more likely to confess when they feel guilty, remains unanswered. Merely a few studies are
examining the role of mood on confessing behaviour and therefore this is the first study to explore the predictive value of mood on confessing behaviour. Using a difficult trivia quiz regarding a bonus participants had the opportunity to cheat and later to confess to their unethical behaviour. The Positive and Negative Affect Scale (PANAS) was used to measure participants’ mood. Of the 86 participants that cheated, 35 decided not to confess, and 39 participants confessed to the full extent. The remaining participants engaged in a partial confession. The current study was not able to find direct mood influences on confessing behaviour, since feelings of guilt or negative emotions in general were almost absent after the performance of unethical behaviour. Accordingly, in the current study mood was not able to predict the extent of confessing to cheating behaviour. Since this study was one of the first to yield results about the direct influences of feelings of guilt on confessing behaviour,
improvements are discussed to continue doing research on this topic.
Introduction
When people behave unethically, they sometimes get the opportunity to ‘clear their conscience’. Confessions are made for various reasons, including relieving guilt (Brooks, 2000). Feelings of guilt are likewise a regular consequence of unethical behaviour (de Hooge, Nelissen, Breugelmans & Zeelenberg, 2011). True confessions could be the result of a certain universal unconscious need for self –punishment concerning unethical behaviour, which together with feelings of guilt can result in a compulsion to confess (Reik, 1959). Police officers often use various tactics to make the suspect feel guilty during interrogation to motivate a confession (Gudjonsson, 2006). Whereas confessions have been extensively studied, very little is known about the direct influence of feelings of guilt on the decision to confess. In fact, reasons to confess are always studied after a confession is made, missing science on direct behavioural consequences of negative emotions. Are people actually more likely to confess when they feel guilty?
A confession can be seen as the most powerful evidence in court (Kassin et al., 2010). For example, in crimes where evidence of guilt is hard to find, a confession can solely solve the whole investigation. More horrific crimes, like murder, are mostly solved by a confession instead of any other type of evidence (Leo, 2008). Confessions have significant consequences for the individual because they often lead to convictions (Gudjonsson, 2003; Gudjonsson, Sigurdsson, Bragason, Einarsson & Valdimarsdottir, 2004; Kassin et al., 2010).
Reasons to confess
Reasons to confess to a committed crime have been extensively studied, mostly after confessions were made. These studies suggest that true confessions are caused by three factors: (1) external pressure; pressure performed by policemen and fear of custody, (2) perception of proof; that is the suspects’ belief of enough strong evidence against him or her
(even though this is not necessarily the case), and (3) internal pressure; feelings of guilt and the need to relieve those feelings by making a confession, or in other words ‘clearing the conscience’ (Gudjonsson et al., 2004; Kassin & Gudjonsson, 2004). Most of the time, it is not just one single factor that is presented prior to confessing, but a combination of different types of reasons that makes the suspect confess. This paper focusses on internal pressure and its role in confessing behaviour.
Examining reasons to confess after a confession is made, is mostly done using the Gudjonsson Confession Questionnaire – Revised (GCQ-R; Sigurdsson & Gudjonsson, 1994). While this questionnaire provides valuable insight in offender’s reasons to confess and their attitude towards their confession, an important aspect is that it is always implemented after the confession has been made (Gudjonsson & Sigurdsson, 1999).
Mood
Mood has also been examined after the opportunity to confess was given (Peer, Acquisti & Shalvi, 2014). Using a coin-tossing prediction task participants were given the opportunity to cheat and later to confess to their unethical behaviour, to the full, some, or none extent. Mood was measured after participants had gotten the opportunity to confess. Participants who cheated, felt worse compared to those who did not cheat, which is in line with findings from De Hooge et al. (2011) showing that unethical behaviour often leads to feelings of guilt. After the opportunity to confess was given, no differences were found between the non –cheaters and participants that fully confessed to their cheating behaviour (Peer et al., 2014). Accordingly, it can be suggested that a confession resulted in a reduction of negative emotions compared to emotional states right after the performance of unethical behaviour. It was also noticed that participants who only confessed partially ended up feeling the worst. Indicating that negative feelings did not disappear when participants lied about
their unethical behaviour. However Peer et al. (2014) only explained the consequences of confessing behaviour on mood, and not the consequences of mood on confessing behaviour.
Furthermore, it is meaningful to examine the consequences of mood on confessing behaviour, since it was found that suspects who are in the police station for questioning, indicate experiencing low mood due to anxiety, hopelessness and depression, compared to the witnesses (Sigurdsson, Gudjonsson, Einarsson & Gudjonsson, 2006).
Therefore, the current study will investigate whether mood has a predictive value on confessing behaviour. To improve upon the Peer et al. (2014) study a mood measurement before participants had the opportunity to confess will be inserted. It is hypothesized that reported levels of negative and positive affect will influence the extent of confessing to unethical behaviour. Participants reporting a significantly high amount of negative affect or feelings of guilt, or a low amount of positive affect might be more likely to fully confess, because they feel guilty about their wrongdoing. Participants going to conceal their cheating behaviour might not feel much internal pressure and therefore report lower amounts of negative affect and feelings of guilt, and higher amounts of positive affect than the full confessors. Medium internal pressure could lead to partial confessors since they are more likely to seek a balance between reducing negative emotions and appearing credible (Peer et al., 2014). Therefore, partial confessors might report less negative affect and feelings of guilt, and more positive affect than the full confessors, but more negative affect and feelings of guilt, and less positive affect than those who decide not to make a confession.
In addition, it is hypothesized that mood will be affected by the confession, depending on the extent of confessing. Cheaters who do not confess might not significantly feel any different as they will not have enough feelings of guilt in the first place to make a confession. Notwithstanding, the partial confessors might feel worse than they did right before their confession because they did not totally clear their conscience (Peer et al., 2014). In other
words, partial confessors do not take full responsibility for their mistakes, and end from one lie into another. This group is expected to report significantly less positive and more negative affect and feelings of guilt after the confession. Full confessors might have cleared their conscience and have ‘got it of their chest’. Resulting in positive feelings of relieve and therefore report more positive and less negative affect and feelings of guilt than right before the confession.
Personality
A better investigated topic in confessing behaviour is personality. Individual differences in personality have a predictive value when thinking of reasons to confess. Neuroticism (anxiety proneness) and Compliance are related to internal pressure; feelings of guilt resulting in the need to confess.Offenders with an anxious and compliant temperament are more likely to have feelings of remorse and feel the need to ‘get it of their chest’. Accordingly, Neuroticism is associated with confessing behaviour in general (Gudjonsson & Sigurdsson, 1999). Despite the fact that a lot of studies investigated the relation between personality and confessions, direct behavioural observations before the confession are missing since the GCQ –R is mostly used. Therefore, the actual link between different degrees of personality and different behaviours remains an understudied topic (Baumeister, Vohs, & Funder 2007).
Not much research has been conducted on the link between personality traits and ethical or dishonest behaviour as well. Only recently, it was found that unethical behaviour can be explained by a relatively new model of personality (Hilbig & Zetler, 2015): the HEXACO Model of Personality (an acronym for the six factors Honesty–Humility, Emotionality, Extraversion, Agreeableness, Conscientiousness, and Openness to Experience) (Lee & Ashton, 2012). According to the HEXACO, individual differences in honesty and
morality are an actual core factor of personality which has to be added to the Big Five (for review, see John & Srivastava, 1999). The Honesty –Humility factor is described as ‘the tendency to be fair and genuine in dealing with others’ (Ashton & Lee, 2007, p.156). The factor is supposed to describe differences in the willingness to manipulate others for personal gain, the desire to break rules, the likelihood to obtain motivation out of materialistic gain and the interest for social statuses. The other five factors of the HEXACO consists of three factors (Extraversion, Openness to Experience and Conscientiousness) that do exactly match the factors of the Big Five, and two factors (Agreeableness and Emotionality) that resemble the factors Agreeableness and Neuroticism of the Big Five, but are not identical (Lee & Ashton, 2004). Recently, a consistent medium effect sized link between the outcome of the Honesty –Humility factor and cheating behaviour was found (Hilbig & Zettler, 2015). Indicating that participants scoring lower on the Honesty –Humility factor are more likely to cheat. This was one of the first studies to examine a link between this factor and actual behaviour. No other factor showed any consistent association with dishonest behaviour.
To gain more support for the validity of the HEXACO, the current research will attempt to strengthen the link between the Honesty –Humility factor and cheating behaviour that was found in the Hilbig and Zettler (2015) experiment. It is hypothesized that participants scoring low on the Honesty –Humility factor are more likely to cheat during the trivia quiz, than participants scoring high on the Honesty –Humility factor, since this factor describes the desire to break rules, and the likelihood to obtain motivation out of materialistic gain. These results are likely to match with Hilbig and Zettler (2015). Since it was found that the personality factor Neuroticism is correlated with feelings of guilt (Gudjonsson & Sigurdsson, 1999) it is hypothesized that participants scoring high on the comparable Emotionality factor are more likely to report negative affect after cheating. This factor describes individual differences in experiencing fear and danger, the need for emotional
support, and differences in empathy and attachment towards others (Lee & Aston, 2004). Therefore, personalities with high scores on Emotionality experience more fear resulting in more negative feelings of remorse and guilt.
In short, since there are merely a few studies examining the role of mood on confessing and no studies that actually explored the predictive value of mood on confessing, it is still unclear how confessions might be affected by mood. The question how confessing changes mood and if feelings of guilt are actually reduced, remains unanswered. In addition, despite the encouraging results considering the link between the HEXACO and actual cheating behaviour (Hilbig & Zettler, 2015), more power is needed to strongly support for the theory of the HEXACO. Also, personality and the reasons behind confessing behaviour have been extensively studied, but only in retrospect (after the confession was made). Science is missing on the link between personality and feelings of guilt right before the decision to (not) make a truthful confession and therefore the consequences of unethical behaviour.
Current study
In an online study individual differences were assessed using the HEXACO and mood indications using the Positive and Negative Affect Scale (PANAS; Watson, Clark & Tegellen 1988). A trivia quiz containing two difficult questions and eight easy questions, regarding a bonus for those who answered all of the questions correctly, was used to give participants the opportunity to cheat and then to confess. Participants had to indicate their mood once as a baseline, one time after their unethical behaviour (before the opportunity to confess) and one time after they had the opportunity to confess.
Since the Peer et al. (2014) study was one of the first to yield results about the extent of confessing, the current study also aims to achieve partial confessions to be able to
investigate mood influences on partial confessions. Participants could partially confess by only being honest about cheating on one of the two difficult trivia questions.
It is hypothesized that the Honesty –Humility factor of the HEXACO predicts the likelihood of cheating. After cheating those who score high on the Emotionality factor of the HEXACO will be more likely to report negative affect. Reported mood (negative affect, positive affect and feelings of guilt) influences confessing behaviour: more negative affect, more feelings of guilt and less positive affect will increase the likelihood of confessing. After the opportunity to confess will be given, reported mood will change depending on confessing behaviour. A confession will reduce negative affect, whereas lying about cheating will induce negative affect.
Method Participants
Hundred and twenty-three participants from the University of Amsterdam, aged between 18 and 39 years (mean age = 20.4 years, SD = 2.6, 85 females) participated in this experiment, after signing an online version of the informed consent. Participants had good knowledge of the Dutch language. As a reward participants received course credits for participating.
Measures and Materials
Responses were measured using Qualtrics Software (Qualtrics, Provo, UT) and Inquisit 5 Web (Millisecond Software, Seattle, WA).
To indicate participants’ mood the 20-item Dutch version of the Positive and Negative Affect Schedule (PANAS; Engelen, De Peuter, Victoir, Van Diest & Van den Bergh, 2006) was used. The PANAS measures positive and negative affect using two 10 –item mood scales that are brief and easy to administer. Mood was assessed using the PANAS by briefly stating emotional states such as ‘excited’ or ‘guilty’ that had to be evaluated on a 5-point-Likert scale by the participants (from 1: not at all, to 5: extremely). Because participants had to evaluate their emotional state three times during the experiment, they were instructed to be precise about their state of mind at that particular moment.
The 60-item Dutch version of the HEXACO Model of Personality was used to measure individual differences in personality (Ashton & Lee, 2009). Each factor contained ten items which are presented at hexaco.org. Participants had to indicate to what extent they agreed or disagreed with different statements (from 1: strongly disagree, to 5: strongly agree).
Design and procedure
This study was an online experiment. Participants were recruited trough a participant pool website of the University of Amsterdam. The experiment was not named after the actual aim of the study but introduced as an experiment concerning relations between mood and performances. This cover story was created to make sure cheating was completely voluntarily and that none of the participants were aware of the actual aim of the study. After reading the information brochure and signing the informed consent participants started out with the HEXACO. 60 items were equally divided over six pages, to help participants keep their attention. After the HEXACO, participants had to indicate their current mood using the PANAS. The PANAS is inserted here as well to count for a baseline measure for mood and to support for the cover story. To control for serious participation, three control questions were randomly inserted in the HEXACO. Also, the ‘next button’ only appeared after 30 seconds on every page containing the HEXACO or PANAS to prevent participants from continuing without proper completion.
Then, participants had 5 minutes to solve two unrelated problem solving tasks adapted from Russano, Meissner, Narchet and Kassin (2005) inserted to support for the cover story. It was made clear participants had to complete the tasks alone. Hereafter a trivia quiz started where participants could cheat by using Google to answer the questions. Again, the instructions included that the quiz had to be made alone. This way it was indirectly communicated that it was not allowed to use Google, so that looking up the answers would really feel like cheating while at the same time the exact opposite reaction was not being provoked.
Various tactics were inserted to increase the changes of cheating. First of all, participants got a bonus (in research credits) when answering all questions correctly. Research has shown that a clear and fast reward increases dishonest behaviour (Gino &
Pierce, 2009). The trivia quiz consisted out of eight easy and two difficult questions, conducted through a pilot study, see Appendix A. Easy questions were inserted to win confidence, whereas difficult questions were supposed to evoke cheating. Previous research has shown that when facing a last choice cheating is increased (Effron, Bryan & Murnighan, 2015), therefore the difficult questions were presented as the last two questions of the quiz. Additionally, participants had a total of ten minutes to finish the quiz. Pilot studies showed that this is enough time to finish the first eight questions, remaining sufficient time to cheat.
After the trivia quiz, participants were again asked how they felt using the PANAS. By asking participants how they knew the correct answers, they were given an indirect opportunity to confess. Directly asking was considered to be too obvious, reducing the amount of confessions. To increase credibility this was also done for one easy questions and one problem –solving tasks. Participants were informed that their answers would not influence their participation or bonus in any form. Asking about the two difficult questions created the opportunity to confess partly by only admitting to have cheated on one of the two difficult question. After the confessions participants were, one last time, asked how they felt using the PANAS. A lie detection test followed for the use of another study. Lie detection was performed using the Concealed Information Test (Lykken, 1959) measuring reaction times. Participants were checked on their memory of the quiz and on their motivation to avoid detection. Memory and motivation data were obtained for the use of the other study. Lastly, participants had to provide some demographical information and a multiple choice question was used to ask whether participants used Google during the quiz, to double check for cheating and confessing behaviour.
Results
Of the 123 participants starting the experiment 91 % completed the whole experiment (N = 112). Information about gender, age and final opportunity to confess was missing of participants not finishing the experiment. Since this information was not needed for all of the data analyses, participants were not excluded. However in some analyses only participants that fully completed the experiment were included. To check whether participants completed the online experiment seriously, control questions were inserted. One participant did not answer control questions correctly and had to be excluded. Out of the remaining 122 participants, 86 participants (71 %) answered both of the difficult questions correctly, implying that cheating occurred. Eleven participants answered only one of the two difficult questions correctly, and were classified as partial cheaters. Yet, since hypotheses and analyses were not anticipated on partial cheating, those participants were excluded from the experiment. Resulting in a total (N = 111) of 86 cheaters (77%) and 25 non –cheaters(23 %).
Honesty – Humility
To predict the likelihood of cheating behaviour using the Honesty – Humility factor a binary logistic regression was performed. The small difference in the Honesty – Humility factor between cheaters (M = 3.191, SD = 0.547) and non –cheaters (M= 3.353 5, SD = 0.612) was not able to generate a statistically significant model, indicating that Honesty – Humility could not distinguish between cheaters and non –cheaters (odds-ratio = 0.022, χ2 (1) = 1.614, p = .204). Honesty –Humility could not be demonstrated as a predictor for the probability to cheat in this study (B = 0.516, p = 2.07).
To control for possible other variables that account for the variance between cheaters and non –cheaters, the above logistic regression was repeated and age and gender were added as covariates. Note that only participants that completely finished the experiment were
included in this analysis (N = 112). Adding age and gender did not significantly change the coefficient for Honesty –Humility (odds-ratio = 0.146, χ2 (1) = 0.669, p = .413), or the
predictive value of the Honesty -Humility factor for cheating behaviour (B = 0.270, p =.621). The logistic regression was also repeated to control for the remaining five factors of the HEXACO. Again this model was not able to distinguish cheaters from non -cheaters (odds-ratio = 0.053, χ2 (1) = 2.224, p = .136) and Honesty – Humility could not be demonstrated as a predictor for cheating behaviour (B =0.665, p =.138).
Affect after cheating
After the possibility to cheat was given, those who cheated reported low levels of negative affect (M = 1.330, SD = 0.489) and low levels of feelings of guilt specifically (M = 1.210, SD = 0.576) (for an overview of all mood measurement during baseline and after cheating behaviour see Appendix B). The reported negative affect –measures were also low for the non – cheaters (negative affect M = 1.420, SD = 0.449; feelings of guilt M = 1.200, SD = 0.500). Negative affect after the opportunity to cheat did not significantly differ between the two conditions (negative affect t (109) = -0.822, p = .200; feelings of guilt t (109) = 0.073, p =.942), indicating cheaters did not report more negative affect than non – cheaters. However, cheaters reported significantly higher levels of positive affect (M = 3.019, SD = 0.774) than those who did not cheat (M = 2.536, SD = 0.917) (t (109) = 2.629, p < .001). In addition, paired sampled t-test showed that cheaters increased their reported levels of positive affect after cheating (t (85) = -1.725, p < 0.05) compared to baseline, and even decreased their levels of negative affect (t (85) = 4.598, p < .001). No differences between baseline guilt, and guilt after cheating were found (t (85) = -0.185, p = .854) for those who cheated. However, non –cheaters also showed a significant decrease in reported levels of negative affect (t (24) = 2.648, p < .05) and a significant decrease in reported levels of
positive affect (t (24) = 1.824, p < .05). No differences in feelings of guilt were found before and after the opportunity to cheat was given for those who did not cheat (t (24) = -0.296, p =.770).
Emotionality
A Pearson correlation was run to determine whether variances in negative affect, positive affect and feelings of guilt after cheating were related to the Emotionality factor of the HEXACO. Only participants who cheated where included in this analyses (n =86),
because it was not expected that levels of affect were altered by the experiment for those who did not cheat. There was no significant positive relation between negative affect and
emotionality (r = .168, n = 86, p = .062).The same analysis was conducted for the relation between Emotionality and reported amounts of feelings of guilt specifically. No relation between feelings of guilt and emotionality, with statistical significance (r = .057, n = 86, p = .300) was found. Therefore, scores on Emotionality as a factor of personality were not correlated with negative affect after the performance of unethical behaviour, indicating that the comparable factor Neuroticism was not related to feelings of guilt in study.
Influences of affect on confessing behaviour
Out of the 86 participants that cheated 39 participants (45 %) fully confessed, 35 (41 %) participants did not confess and 12 participants (4%) partially confessed. An important question is whether differences in negative affect, positive affect and feelings of guilt have behavioural consequences during the opportunity to confess. Therefore, levels of negative affect, positive affect and feelings of guilt were studied as possible predictors for confessing behaviour. Table 1 shows the reported levels of negative affect, positive affect and feelings of guilt after cheating, prior to confessing behaviour.
Table 1
Reported levels of negative affect, positive affect and feelings of guilt after cheating, prior to confessing behaviour
Mean (SD)
Confession
Negative affect Positive affect Feelings of guilt
No confession 1.318(0.560) 3.100 (0.781) 1.090 (0.373) Partial confession 1.439 (0.439) 3.133(0.679) 1.420 (0.900) Full confession 1.422 (0.580) 2.910 (0.800) 1.260 (0.595)
To analyse whether mood was able to predict confessing behaviour a multinomial logistic regression was performed for negative and positive affect, see Appendix C. None of the reported affect was able to predict confessing behaviour, indicating that levels of reported affect did not significantly differ before participants decided to confess or not. Additionally a multinomial logistic regression was performed to analyse if feelings of guilt specifically could predict confessing behaviour, see Appendix D. Since feelings of guilt belong to
negative affect, this was analysed using an individual test. Feelings of guilt were also not able to predict to what extent participants were more likely to confess. Also no significant
differences were found between two groups specifically for positive affect, negative affect or feelings of guilt using post hoc tests.
Affect after confessing
Mood was also measured after the opportunity to confess was given (for an overview see Appendix E). To analyse whether mood was affected by the confession depending on the extent of confessing a repeated measures ANOVA was performed. Negative affect did not
significantly change over time (F (1, 83) = 0.094, p =.759). Also no significant interaction between confessing behaviour and time was found (F (4, 166) = .580, p =.634), indicating that reported levels of negative affect did not significantly change after the opportunity to confess was given (Figure 1).
However, additional paired sampled t –tests were conducted for each condition separately, and revealed that those who did not make a confession, significantly decreased their reported levels of negative affect after the confession was made compared to after the performance of unethical behaviour (t (37) = 2.317, p < .05). A partial confession did not lead to a change in reported levels of negative affect (t (12) = -0.417, p = .684), nor did a full confession (t (44) = -0.346, p = .731).
Figure 1. Negative affect after cheating and after the opportunity to confess. *p < .05. Error
bars represent standard error.
Figure 2 shows the reported levels of positive affect before and after the opportunity to confess was given. A repeated measures ANOVA showed a significant effect of time on reported levels of positive affect (F (1, 82) = 32.301, p < .001), but no interaction effect
0 1 2 3 4 5
No confession Partial confession Full confession
L ev el s o f n eg at iv e af fect
After cheating After confessing *
between the extent of confessing and time (F (2, 82) = 0.237, p = .790). Indicating that positive affect decreased in general after the opportunity to confess was given, not depending on confessing behaviour.
Additional paired sampled t –tests showed that those who did not confess and those who fully confessed decreased in their levels of positive affect (no confession t (37) = 4.803,
p < .001; full confession t (43) = 3.652, p < .001). Those who partially confessed did not
significantly decrease in their reported levels of positive affect after confessing compared to before the opportunity to confess was given (t (12) = 2.168, p = .053).
Figure 2. Positive affect after cheating and after the opportunity to confess. **p < 0.01. Error
bars represent standard error.
More specifically, feelings of guilt were also measured before and after the
opportunity to confess was given (Figure 3). Overall, feelings of guilt did not significantly change after the opportunity to confess was given (F (1, 83) = 3.620, p =.061), and no
significant interaction was found between type of confession and time (F (2, 83) = 3.111, p = .050), indicating that feelings of guilt did not significantly change as a result of the
0 1 2 3 4 5
No confession Partial confession Full confession
L ev el s o f p o si ti v e af fect
After cheating After confessing
** **
opportunity to confess. Additional paired sampled t –tests however showed that feelings of guilt increased after a full confession was made (t (44) = -2.847, p < .001), but did not change for those who did not confess (t (37) = -1.434, p = .160) or for those who partially confessed (t (12) = .000, p = .988).
Figure 3. Feelings of guilt after cheating and after the opportunity to confess. ** p < .001.
Error bars represent standard error. 0 1 2 3 4 5
No confession Partial confession Full confession
F ee lin g s o f g u ilt
After cheating After confessing **
Discussion
The present study aimed to gain more insight in the direct consequences of feelings of guilt on confessing behaviour by measuring mood before and after the opportunity to confess to unethical behaviour was given. It was hypothesized that higher amounts of reported
negative affect after cheating behaviour increased the likelihood of making a confession. Additionally, it was examined whether the Honesty – Humility factor of the HEXACO was able to predict cheating behaviour and if the Emotionality factor was correlated to feelings of guilt after cheating behaviour.
Firstly, findings revealed that the majority of the participants were willing to cheat for personal gain. Findings are in line with the idea that unethical behaviour is present in a lot of social and economic interactions (Mazar & Ariely, 2006). Surprisingly, there were also 11 participants that cheated partially, by only answering one of the two difficult questions correctly. Those participants did not benefit from their unethical performance, since the bonus was only earned when all of the questions were answered correctly. It might be that these participants preferred internal rewards over external rewards and were achieving to maintain their self –concept (Mazar, Amir & Ariely, 2008). Furthermore, the current study showed that people sometimes engage in partial confessions, although it was more common to confess to the full extent. It is thought that partial confessions are made to achieve a balance between appearing credible to others while reducing negative emotions after the performance of unethical behaviour (Peer et al., 2014).
Mood influences on confessing behaviour
How the cheaters felt right after their performance of unethical behaviour, did not have a significant effect on their decision to confess. Results did not support the hypothesis as positive and negative emotions, and feelings of guilt specifically were not able to predict to
what extent participants would confess. Since relieving guilt is thought to be one of the reasons to confess to unethical behaviour (Brooks, 2000 ; Gudjonsson et al., 2004; Kassin & Gudjonsson, 2004) and it was suggested that a confession to unethical behaviour resulted in a reduction of negative emotions (Peer et al., 2014) it was hypothesized that mood would influence the decision to confess.
Notably, after the performance of unethical behaviour, participants did not report feeling guilty or experience much negative affect to a high extent. Their mean scores respectively were 1.21 for feelings of guilt and 1.33 for negative emotions, indicating participants were ‘not at al’ to ‘a little’ feeling guilty or experienced other negative emotions (1 = not at all; 2 = a little; 3 = moderately; 4 = quite a bit; 5 = extremely). Also, the maximum reported level of negative affect after cheating was 3.40, indicating that negative emotions were not even ‘quite a bit’ reported. Negative emotions between cheaters and non –cheaters did not differ after the possibility to behave unethically. Cheaters even reported to feel significantly less negative after their cheating behaviour, compared to baseline. The fact that cheaters did not feel more guilty or worse after their unethical behaviour or than those who did not cheat is not in line with previous studies suggesting that feelings of guilt are a regular consequence of unethical behaviour (de Hooge et al., 2011). Findings can be explained by the idea that participants did not completely realised that using Google was not desired, until the opportunity to confess was given. Instructions implemented before the trivia questions implicitly included not to cheat by mentioning that the tasks had to be made alone, providing space for participants to think of reasons to justify their unethical behaviour (I can use Google without the help of others, all by myself) (Gino & Ariely, 2011).
Moreover, to perform a logistic regression it is advised to have at least 50 participants per dependent category, whereas in this study only between 12 and 39 participants per category could be included. To be able to predict confessing behaviour and to draw
conclusions about the influence of mood on confessing behaviour, more participants are needed.
It must be noted that there were small non –significant differences in reported levels of negative affect and feelings of guilt between the three confessing behaviours before the opportunity to confess was given. Negative feelings were slightly more present for those who were going to confess to the full extent (1.42) and to a partial extent (1.44) than for the participants who later decided not to confess (1.32). More promising were the differences in feelings of guilt specifically between the three confessing behaviours. Partial confessors reported levels of 1.42 and full confessors of 1.26 compared to 1.09 for those who were going to lie about their cheating behaviour. Although reported levels of negative emotions were more absent than present and the differences in negative emotions between confessing behaviours before the opportunity to confess was given are minor, further research is being encouraged by these results.
Positive affect was also measured after the performance of unethical behaviour. Cheaters reported a mean score of 3.02 (3 = moderately) suggesting cheaters did not feel positive emotions to a high extent, but good feelings were definitely not completely missing. Remarkably, those who cheated felt significantly better than those who did not cheat and also reported more positive emotions than before the opportunity to cheat was given. Since reported levels of positive feelings and emotions were still not above ‘moderately’ on average, minor differences could be explained by the bonus participants earned as a result of their cheating behaviour.
Mood after confessing
After the opportunity to confess was given, participants felt significantly different than right after the opportunity to perform unethically, although it must be noted that changes
were between .05 and 0.31 on a 5- point-Likert scale. Cheaters who decided not to confess had a minor decrease in reported levels of negative affect and in positive affect. It was expected that non –cheaters would not report feeling any different after the opportunity to confess since they did not have enough feelings of guilt to make a confession. Results showed that mood in general (positive and negative affect) was altered by the given opportunity to confess which suggests that cheaters that did not confess experienced all sorts of feelings and emotions to a lesser extent. Clearly, the emotional state of participants is measured which does not necessarily mean that reported levels of feelings and emotions stay constant over time. Findings can also be explained by the idea that participants were a bit bored by the experiment which might have flattened their emotions. As was expected, no changes in feelings of guilt specifically were indicated for those who did not confess. Partial confessors did not experience different feelings and emotions after their confession. It was expected that partial cheaters would feel worse after their confession because they did not totally ‘clear their conscience’ and ended from one lie into another (Peer et al., 2014). It must be noted that only 12 participants partially confessed, what might explain the fact that differences did not reach significance. Lastly, a full confession resulted in less positive feelings and more feelings of guilt, whereas it was expected that taking full responsibility would result in more positive feelings. Findings can again be explained by the idea that until the opportunity to confess was given, participants did not completely realise that using Google was not permitted.
Personality
Differences in negative affect and feelings of guilt within those who performed unethical behaviour did not correlate with the personality factor Emotionality of the HEXACO, as was however hypothesized. Emotionality describes individual differences in
experiencing empathy, fear and danger (Lee & Ashton, 2004), and the comparable personality trait Neuroticism was associated with feelings of guilt before (Gudjonsson & Sigurdsson, 1999). Therefore it was expected that participants with high scores on Emotionality were more likely to report negative feelings considering their own unethical behaviour. Results can be explained by the fact that participants that cheated did not report high levels of negative affect at all. In addition it should be mentioned that the correlation between reported negative emotions after cheating and the personality trait Emotionality, almost reached significance (p =.062).
Results also did not support for the predictive value of the Honesty –Humility factor for differences in unethical behaviour. Although, participants who cheated did report lower amounts of Honesty -Humility than participants that did not cheat, significance was not reached in this study. A possible explanation is the fact that not enough participants participated in this study. Using G*power and concerning the Hilbig and Zettler (2015) experiment an actual total of 165 participants was needed to have a high statistical power of .95, whereas in this study only 111 subjects participated so far. In addition, only 25 non – cheaters were included, whereas a logistic regression desirably requires at least 50 participants per dependent category. Only when enough statistical power is reached by using enough participants, conclusions about the validity of the HEXACO can be made. Considering the small amount of available literature about relations between the HEXACO and unethical behaviour, it is important to continue doing research on this topic.
Limitations and further research
A limitation worth mentioning is the fact that feelings of guilt after the performance of unethical behaviour were almost absent, indicating that cheating was not able to manipulate mood. Firstly it can be argued that mentioning the reward directly after the performance of
unethical behaviour (before measuring mood) might have contributed to these findings. Since in this study mood was measured directly after the bonus was mentioned, reported levels of affect might be the experimental manipulated positive feelings, which might have masked negative emotions due to cheating behaviour. To increase the changes of obtaining more negative emotions and specifically guilt feelings after the performance of unethical behaviour, it is suggested that further research implements this reward after the mood measurement.
Secondly, the inserted instructions before the trivia questions, to communicate what was expected of the participants, can be discussed. The instructions attempt to make sure participants were aware of the fact that using Google to answer the questions was not permitted, by clarifying trivia questions had to be answered alone, without the help of others. It is thought that participants probably knew using Google was not desired, but could easily think of ways to justify their unethical behaviour. In addition, results revealed that after the opportunity to confess was given participants felt worse than after cheating behaviour, which might indicate that during the opportunity to confess participants might have completely understood that using Google was not allowed. Further research might rethink how to implicitly instruct participants not to cheat while making the message clear that using Google is not allowed.
Conclusions
Although the current study fails to support the hypothesis that feelings of guilt
increase the changes of confessing to unethical behaviour, it suggests new ways in research to directly study reasons to confess. Whereas confessing behaviour is mostly studied after a confessions was made, it should be encouraged to examine the direct consequences of negative emotions on confessing behaviour. Furthermore, the current research contributes to
the science of personality in attempting to link actual behaviour to different degrees of personality (Baumeister et al., 2007). Findings also expand the available literate about partial confessions.
The absence of support in this study does not serve as the proof of absence of any relation between feelings of guilt and confessing behaviour. As limitations are discussed above, a few good reasons present why the relation between feelings of guilt and confessing behaviour was not found in the current study. It is still very likely that a guilty conscience needs to confess.
References
Ashton, M. C., & Lee, K. (2007). Empirical, theoretical, and practical advantages of the HEXACO model of personality structure. Personality and Social Psychology Review,
11(2), 150-166. doi: 10.1177/1088868306294907
Ashton, M. C., & Lee, K. (2009). The HEXACO–60: A short measure of the major
dimensions of personality. Journal of Personality Assessment, 91(4), 340-345. doi: 10.1080/00223890902935878
Baumeister, R. F., Vohs, K. D., & Funder, D. C. (2007). Psychology as the science of self-reports and finger movements: Whatever happened to actual behavior? Perspectives
on Psychological Science, 2(4), 396-403. doi: 10.1111/j.1745-6916.2007.00051.x
Brooks, P. (2000). Troubling confessions: Speaking guilt in law and literature. London: University of Chicago Press. doi: 10.1017/s0829320100007079
De Hooge, I. E., Nelissen, R., Breugelmans, S. M., & Zeelenberg, M. (2011). What is moral about guilt? Acting “prosocially” at the disadvantage of others. Journal of Personality
and Social Psychology, 100(3), 462 -473. doi: 10.1037/a0021459
Effron, D. A., Bryan, C. J., & Murnighan, J. K. (2015). Cheating at the end to avoid regret.
Journal of Personality and Social Psychology, 109(3), 395. doi:
10.1037/pspa0000026
Engelen, U., De Peuter, S., Victoir, A., Van Diest, I., & Van den Bergh, O. (2006). Verdere validering van de Positive and Negative Affect Schedule (PANAS) en vergelijking van twee Nederlandstalige versies. Gedrag en Gezondheid, 34(2), 61-70.
Gino, F., & Ariely, D. (2012). The dark side of creativity: original thinkers can be more dishonest. Journal of Personality and Social Psychology, 102(3), 445 -459. doi: 10.1037/a0026406
Gino, F., & Pierce, L. (2009). Dishonesty in the name of equity. Psychological Science,
20(9), 1153-1160. doi: 10.1111/j.1467-9280.2009.02421.x
Gudjonsson, G. H. (2003). The psychology of interrogations and confessions: A handbook. West Sussex: John Wiley and Sons. doi: 10.1016/j.paid.2003.09.020
Gudjonsson, G. H. (2006). Sex offenders and confessions: How to overcome their resistance during questioning. Journal of Clinical Forensic Medicine, 13(4), 203-207. doi: 10.1016/j.jcfm.2006.02.003
Gudjonsson, G. H., & Sigurdsson, J. F. (1999). The Gudjonsson Confession Questionnaire-Revised (GCQ-R) factor structure and its relationship with personality. Personality
and Individual Differences, 27(5), 953-968. doi :10.1016/S0191-8869(98)00278-5
Gudjonsson, G. H., Sigurdsson, J. F., Bragason, O. O., Einarsson, E., & Valdimarsdottir, E. B. (2004). Confessions and denials and the relationship with personality. Legal and
Criminological Psychology, 9(1), 121-133. doi: 10.1348/135532504322776898.
Hilbig, B. E., & Zettler, I. (2015). When the cat’s away, some mice will play: A basic trait account of dishonest behavior. Journal of Research in Personality, 57, 72-88. doi: 10.1016/j.jrp.2015.04.003
Inquisit 5 Web (2016). [computer software]. Seattle, WA: Millisecond Software. Available from http://www.millisecond.com/
John, O. P., & Srivastava, S. (1999). The Big Five trait taxonomy: History, measurement, and theoretical perspectives. Handbook of personality: Theory and research, 2(1999), 102-138.
Kassin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., & Redlich, A. D. (2010). Police-induced confessions: Risk factors and recommendations. Law and
Kassin, S. M., & Gudjonsson, G. H. (2004). The psychology of confessions a review of the literature and issues. Psychological Science in the Public Interest, 5(2), 33-67. doi: 10.1111/j.1529-1006.2004.00016.x
Lee, K., & Ashton, M. C. (2004). Psychometric properties of the HEXACO personality inventory. Multivariate Behavioral Research, 39(2), 329-358-6. doi:
10.1207/s15327906mbr3902_8
Lee, K., & Ashton, M. C. (2012). Getting mad and getting even: Agreeableness and Honesty-Humility as predictors of revenge intentions. Personality and Individual Differences,
52(5), 596-600. doi:10.1016/j.paid.2011.12.004
Leo, R. A. (2008). Police interrogation and American justice. Cambridge, MA: Harvard University Press. doi: 10.1177/13624806090130030602
Lykken, D. T. (1959). The GSR in the detection of guilt. Journal of Applied Psychology,
43(6), 385 -388. doi: 10.1037/h0046060
Mazar, N., Amir, O., & Ariely, D. (2008). The dishonesty of honest people: a theory of self-concept maintenance. Journal of Marketing Research, 45(6), 633-644. doi:
10.1509/jmkr.45.6.633
Mazar, N., & Ariely, D. (2006). Dishonesty in everyday life and its policy implications.
Journal of Public Policy and Marketing, 25 (1), 117 -126. doi:10.1509/jppm.25.1.117
Peer, E., Acquisti, A., & Shalvi, S. (2014). “I cheated, but only a little”: Partial confessions to unethical behavior. Journal of Personality and Social Psychology, 106(2), 202 -217. doi: 10.1037/a0035392.
Qualtrics, Qualtrics (2005). Provo, Utah, USA. Available from http://www.qualtrics.com Reik, T. (1959). The compulsion to confess. New York: Farrar, Straus and Cudahy. doi:
Russano, M. B., Meissner, C. A., Narchet, F. M., & Kassin, S. M. (2005). Investigating true and false confessions within a novel experimental paradigm. Psychological Science,
16(6), 481-486. doi: 10.1111/j.0956-7976.2005.01560.x
Sigurdsson, J. F., & Gudjonsson, G.H. (1994). Alcohol and drug intoxication during police interrogation and the reasons why suspects confess to the police. Addiction, 89 (8), 985 – 997. doi: 10.1111/j.1360-0443.1994.tb03358.x
Sigurdsson, J.F., Gudjonsson, G. H., Einarsson, E., & Gudjonsson, G. (2006). Differences in personality and mental state between suspects and witnesses immediately after being interviewed by the police. Psychology, Crime and Law, 12(6), 619-628. doi:
10.1080/10683160500337808
Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive and negative affect: the PANAS scales. Journal of Personality
Appendix A
The open -ended trivia questions and its respective answers. Questions were asked in Dutch. 1. Which German-born physicist developed the mass-energy equivalence formula E =
mc2?
Albert Einstein
2. In which large continent is Egypt located, famous for its Sphinx and the pyramids of Giza?
Africa
3. What instrument has six strings and can be played acoustically and electrically?
Guitar
4. What kind of animal is Garfield, created by Jim Davis?
Cat
5. According to the old proverb, all roads lead to which capital city?
Rome
6. What do the Roman numerals XIV represent?
14
7. What is the first name of English poet Shakespeare (1564-1616)?
William
8. Which of the four scientific laws ensures that a ball that was thrown up, always comes down again?
Gravity
9. Who is the author of the autobiographical book “Wishful Drinking”? *
Carrie Fisher
10. Who invented the word dinosaur? *
Richard Owen
Appendix B
Overview of all mood measurements before and after the opportunity to cheat was given for the two different cheating conditions
Mean (SD)
Cheaters Non –cheaters
Baseline After cheating Baseline After cheating Negative affect 1.442 (.518) 1.330 (.489) 1.640 (.611) 1.420 (.449) Positive affect 2.931 (.674) 3.019 (.774) 2.764 (.693) 2.536 (.916) Feelings of guilt 1.200 (.591) 1.210 (.567) 1.160 (.374) 1.200 (.500)
Appendix C
Multinomial logistic regression for positive affect and negative affect as predictors for confessing behaviour
95 % CI for Odds Ratio
B (SE) Lower Odds Ratio Upper
Full confession vs. No confession Intercept -0.004 (1.266)
Negative affect -0.732 (0.582) 0.154 0.481 1.506
Positive affect 0.280 (0.313) 0.716 1.323 2.446
Full confession vs. Partial confession Intercept -3.021 (1.796)
Negative affect 0.382 (0.592) 0.459 1.465 4.674
Positive affect 0.430 (0.450) 0.637 1.537 3.711
Partial confession vs. No confession Intercept 3.017 (1.883)
Negative affect -1.114 (0.709) 0.082 0.328 1.317
Positive affect -0.150 (0.462) 0.348 0.861 2.131
Appendix D
Multinomial logistic regression for feelings of guilt as a predictor for confessing behaviour
Note: R2= . 042 (Cox & Snell), .048 (Nagelkerke). Model χ2 (2) = 3.667, p = .160. ** p < .001. * p <.05. 95 % CI for Odds Ratio
B (SE) Lower Odds Ratio Upper
Full confession vs. No confession Intercept 0.757 (0.677)
Feelings of Guilt -0.750 (0.565) 1.560 0.472 1.428
Full confession vs. Partial confession Intercept -1.599 (0.691) * Feelings of guilt
0.316 (0.445) 0.573 1.371 3.280
Partial confession vs. No confession Intercept 2.356 (.832)**
Appendix E
Reported mood measurements after the opportunity to cheat was given (before the confession) and after the opportunity to confess was given, for the three different confessing
behaviours
Mean (SD)
No confession Partial Confession Full Confession After cheating After confessing After cheating After confessing After Cheating After Confessing Negative Affect 1.229 (.428) 1.180 (.366) 1.467 (.446) 1.508 (.480) 1.380 (.544) 1.410 (.558) Positive Affect 3.100 (.781) 2.843 (.877) 3.133 (.679) 2.855 (.826) 2.910 (.800) 2.685 (.849) Feelings of Guilt 1.090 (.373) 1.140 (430) 1.420 (.900) 1.420 (.793) 1.260 (.595) 1.620 (.877)
