SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

Research on counterfactuals (‘If only…’) has seldom considered the effects of counterfactual communication, especially in a defensive context. In three studies, we investigated the effects of counterfactual defences employed by politicians. We assumed that self-focused upward counterfactuals (‘If only I…, the outcome would have been better’) are a form of concession, other-focused upward counterfactuals (‘If only they…, the outcome would have been better’) are a form of excuse, and self-focused downward counterfactuals (‘If only I…, the outcome would have been worse’) are a form of justification. In Study 1, a counterfactual defence led to a more positive evaluation of the politician than a corresponding factual defence. Of the two types of defence, the counterfactual defence reduced the extent to which the politician was held responsible for the past event and was perceived as more convincing. In Study 2, counterfactual excuse and counterfactual justification were equally effective and led to a more positive evaluation of the politician than counterfactual concession. In Study 3, the higher effectiveness of counterfactual justification was independent from perceived ideological similarity with the politician, supporting the strength of this defence. These results show that counterfactual defences provide subtle communication strategies that effectively influence social judgements. Copyright © 2013 John Wiley & Sons, Ltd.

When blamed for a negative outcome, individuals can react in a number of different ways. For instance, they can deny responsibility and blame someone else. Alternatively, they can acknowledge their role in the negative event or try to justify their actions, for example, by asserting that the outcome was not as negative as suggested (Austin, 1961; McGraw, 2001). Each of these defences can be formulated in various ways, which may have different effects on the audience.

In the present research, we tested the effects of counterfactual defences, that is, of defences focused not on the negative event itself, but on possible alternatives. In a counterfactual statement, an antecedent of a past event is postulated to have changed in order to hypothetically alter the outcome of the event (e.g. ‘If I had studied harder, I would have passed the exam’; Kahneman & Tversky, 1982; Roese, 1997). Generating and being exposed to counterfactuals influence the way people explain past events and attribute responsibility for those events (Nario-Redmond & Branscombe, 1996; Wells & Gavanski, 1989). Counterfactuals are therefore likely to be effective when used as defensive statements. However, the use and effects of counterfactual defences and more generally of counterfactual communication have not been investigated extensively (but see Bertolotti, Catellani, Douglas, & Sutton, 2013; Catellani, 2011; Tal-Or, Boninger, Poran, & Gleicher, 2004; Wong, 2010).

The three studies presented herein aim to examine whether counterfactuals can be used as an effective way to defend oneself against criticism and whether counterfactual defence can provide a less explicit but more persuasive way to account for past events than ‘factual’ defence. We asked participants to read fictional texts in which people use various types of factual and counterfactual defences to defend themselves against criticism. We expected counterfactual defences to induce a more positive evaluation of the person giving the account than explicit, factual defences. Furthermore, we expected some types of counterfactual defences to be more effective than others, in particular those that refer to a better possible outcome and focus on the past actions of someone else or, conversely, those that refer to a worse possible outcome and focus on the past actions of the person giving the account.

DEFENSIVE ACCOUNTS

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

When an action has a negative outcome, or a decision has an unexpected result, people often attempt to provide an account, that is, a ‘more acceptable or satisfactory explanation of the event than that contained in a worst-case reading’ (Schlenker, 1980, p. 136). The function of such accounts is twofold: They can be used to reduce the degree of negativity attributed to the event, and they can be used to reduce the level of culpability for the event. Making reference to these two functions, McGraw and colleagues (McGraw, 1990, 2001; McGraw, Timpone, & Bruck, 1993) have proposed a typology of four main types of defensive account. The first is denial, where the individual neither recognises the negativity of the outcome nor admits involvement in it. The second is concession, which can be considered the opposite of denial because the individual acknowledges both the negativity of an outcome and their responsibility for it. In the third, excuse, the individual acknowledges the existence of a negative outcome but denies responsibility for it, either completely or in part. Conversely, in justification, the individual acknowledges their role in the outcome but denies or diminishes the negativity of their role. Justifications attempt to reframe the negative event by highlighting previously unconsidered positive aspects of it or by setting it against events that are even more negative.

Most empirical studies agree that denial is the least effective type of account, but there is no general consensus on which of the other three types of account is the most effective. For example, some studies (Kim, Ferrin, Cooper, & Dirks, 2004; Riordan, Marlin, & Kellogg, 1983; Schlenker, 1980) have found excuses and justifications to be more effective defences than concessions, but others (Hodgins & Liebeskind, 2003) have pointed out that concessions are more effective than other types of defences when the party receiving the account shares a close relationship with the party providing the account. Also, some studies have found excuses to be more effective defences than justifications (Gonzales, 1992; Shaw, Wild, & Colquitt, 2003; Tata, 2000), whereas other studies have found the opposite (Conlon & Ross, 1997; McGraw, 1991; McGraw et al., 1993). Such differences may be due to the fact that these studies have covered a wide range of scenarios, have operationalised defensive accounts in many different ways, and have included different types of relationships between those who give the account and those who receive it.

In the aforementioned studies, individuals defend themselves by referring to actual negative past events and the facts or behaviours that resulted in the present undesirable situation. In our research, individuals defend themselves by focusing on how the circumstances could have been different rather than on what occurred. We assume that a less confrontational, indirect approach could be used effectively to improve the persuasiveness of defensive communication. As suggested by linguistic research, mitigation of statements by reduction of the illocutionary force (Caffi, 1999) can be used to relieve both the speaker and the audience from fully committing to the content of the account (Thaler, 2012). Thus, a hypothetical statement such as ‘what would have happened if…’ is more likely to be perceived as genuine speculation on the part of the speaker than as a self-serving reconstruction of the events and could be considered more convincing by the audience.

COUNTERFACTUAL AS DEFENCE

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

The target of the antecedent and the direction of the hypothesised outcome are of particular importance in the study of counterfactuals in defensive communication because these features are most closely related to the explanation of past events and to suggestions of culpability.

The counterfactual target is the actor whose actions are mutated in the hypothetical scenario. A common distinction is made by research on counterfactual thinking between self-focused and other-focused counterfactuals (Epstude & Roese, 2008). For example, after being involved in a car accident, the driver may generate a self-focused counterfactual such as ‘If I had slowed down at the intersection, the accident could have been avoided’. However, the driver may also generate an other-focused counterfactual such as ‘If the other driver had slowed down at the intersection, the accident could have been avoided’.

The counterfactual direction refers to whether the outcome envisaged in the hypothetical scenario is better or worse than the actual outcome (Markman, Gavanski, Sherman, & McMullen, 1993). For instance, after finishing a race in second place, one might generate an upward counterfactual such as ‘If I had run just a little faster, I would have won the race’. Alternatively, one might generate a downward counterfactual such as ‘If I had run just a little slower, I would have finished third’. After a negative event, generating upward counterfactuals increases the perceived negativity of the event (Branscombe, Owen, Garstka, & Coleman, 1996; Branscombe, Wohl, Owen, Allison, & N'gbala, 2003; Johnson, 1986) and the negative emotions associated with it (Zeelenberg et al., 1998). In addition, the target of an upward counterfactual is more likely to be perceived as responsible for the negative event as compared with other actors involved in the event (Markman & Tetlock, 2000; Nario-Redmond & Branscombe, 1996; Wells & Gavanski, 1989). In a study on counterfactual thoughts after negative events, Markman and Tetlock (2000) found that individuals consistently attributed less responsibility for a negative outcome to themselves when they generated other-focused upward counterfactuals. Given this link between the target of upward counterfactuals and the attribution of responsibility, it could be assumed that in a defensive context, individuals may use upward counterfactuals focused on themselves as a concession to recognise both the negativity of the event and their responsibility for it, whereas they may use upward counterfactuals focused on other people as an excuse to attribute responsibility to others for the negative outcome. Whereas upward counterfactuals are associated with negative emotions, downward counterfactuals are associated with positive emotions (Markman et al., 1993; Medvec, Madey, & Gilovich, 1995; Sanna, Turley-Ames, & Meier, 1999). The hypothetical simulation of a worse outcome triggers an ‘affective contrast effect’, which leads to a comparatively positive evaluation of the actual outcome (McMullen & Markman, 2000, 2002). These findings suggest that, in a defensive context, people may use downward counterfactuals as a justification to reduce the perceived negativity of the event in which they were involved.

In addition to the target and the direction of counterfactuals, in our research, we took into account the counterfactual structure, namely whether the scenario is hypothetically altered by adding or removing an antecedent (Roese & Olson, 1995). Additive counterfactuals introduce a new element that would have led to a different outcome (e.g. ‘If I had taken an umbrella, I wouldn't have got wet’). Conversely, subtractive counterfactuals remove an actual event, action, or decision, thereby simulating how the scenario would have unfolded in the absence of such element (e.g. ‘If it hadn't rained, I wouldn't have caught a cold’). Previous research has not found a clear link between counterfactual structure and attribution of responsibility (Mandel, Hilton, & Catellani, 2005); hence, we had no specific expectation regarding the role of structure in the effectiveness of counterfactual defences. Nevertheless, we included counterfactual structure in our design to control whether different combinations of direction and structure would have different evaluative consequences (Epstude & Roese, 2008; Zeelenberg, van der Pligt, & de Vries, 2000).

THE PRESENT RESEARCH

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

In three studies, we investigated the effects of counterfactual defences under the following assumptions: (1) self-focused upward counterfactuals are a form of concession (e.g. ‘If I had argued more convincingly, the outcome would have been better’); (2) other-focused upward counterfactuals are a form of excuse (e.g. ‘If the others had accepted our proposal, the outcome would have been better’); (3) and self-focused downward counterfactuals are a form of justification (e.g. ‘If I had abandoned my proposal, the outcome would have been worse’). We did not investigate counterfactual denial because it is not possible to hypothetically alter an event if one wishes to deny that this event occurred.

We decided to situate the scenarios of our studies in the political context. This context is particularly apt to study defensive accounts because elected officials often defend their past actions and decisions, and accountability is a crucial issue in politics. Politicians devote a great deal of effort to facework (Bull & Fetzer, 2010) to defend their positive face and aggravate the face of their adversaries.

We manipulated the text of an interview between a fictional journalist and a fictional politician. In the first section of the interview, which was the same across all experimental conditions, the journalist criticised the politician for the negative outcome of their economic policy. The interview ended with a defence by the politician, which varied across experimental conditions.

In Study 1, we compared self-focused and other-focused factual and counterfactual defences. Because of their indirect form, we expected counterfactual defences to induce participants to assign less responsibility to the politician for the negative outcome and to evaluate the counterfactual defence as a more convincing account than the factual defence and that this would result in a more positive general evaluation of the politician. We expected that this would be the case both for self-focused and for other-focused counterfactuals. However, we also expected that the advantage of the counterfactual version would be especially high for the other-focused defence: The indirect form of counterfactual defence would reduce the backlash effect often observed against individuals criticising other individuals (Roese & Sande, 1993; Skowronski, Carlston, Mae, & Crawford, 1998).

In Study 2, we focused on counterfactual defence only and compared a broader range of defences, manipulating the target, direction, and structure of the counterfactual statements employed by the politicians when defending their actions. In line with the aforementioned research on account giving (Kim et al., 2004; Riordan et al., 1983; Schlenker, 1980), we expected counterfactual excuse (i.e. other-focused upward counterfactuals) and justification (i.e. self-focused downward counterfactuals) to be more effective than counterfactual concession (i.e. self-focused upward counterfactuals). We did not expect counterfactual structure to influence these effects given that, as also mentioned earlier, past research has not shown evident links between the structure of counterfactuals and the attribution of responsibility.

Finally, in Study 3, we investigated the direction of counterfactual defences in more depth. We tested whether the positive effect of counterfactual justification (i.e. self-focused downward counterfactuals) on the evaluation of the politician was due to a perception of reduced negativity about the past performance of this individual. We expected counterfactual justification to induce a more positive evaluation of past performance than counterfactual concession thanks to the fact that downward counterfactuals trigger a positive contrast with a negative hypothetical outcome (McMullen & Markman, 2000, 2002). This would in turn lead to a positive evaluation of the defending individual. We also tested whether the effects of counterfactual concession versus justification (i.e. self-focused upward counterfactuals vs self-focused downward counterfactuals) were moderated by perceived ideological similarity between the politician and the person receiving the account. As people are generally biased in favour of those they perceive as similar (Mackie, Worth, & Asuncion, 1990; Petty, Briñol, & Tormala, 2002), we expected participants to evaluate a politician who shared their political beliefs more positively than one with opposing political views. However, we did not expect perceived similarity with the politician to undo the higher effect of counterfactual justification as compared with counterfactual concession. As discussed earlier, previous research on account giving has shown that concession is more effective than justification only if there is a close relationship between the party providing the account and the party receiving it. This is usually not the case when a citizen evaluates a politician.

STUDY 1

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

The main aim of Study 1 was to assess whether counterfactual defences are more effective than the corresponding factual defences. Participants read a fictional interview ending with a defence by the politician that was either self-focused or other-focused and was formulated in factual or counterfactual terms, according to the experimental condition.

Method

Participants and Procedure

A total of 112 students (34.9% men; mean age 21.7 years) at the Catholic University of Milan participated in the study as volunteers. The study was presented as a research project about how people form opinions of political leaders. Participants were invited to imagine being a citizen of a fictitious country where political elections were soon to be held. They were given the text of an interview between a journalist and a leader of the incumbent government running for re-election.

In the interview, the journalist asked the politician to comment on the unfavourable economic conditions that the country was experiencing. The first part of the interview was the same in all experimental conditions:

Journalist: From an economic point of view, you must admit that we can't really say we're in a good condition. In the last few days there's been heated debate on the way your cabinet managed the national budget. What would you say in that regard?

Politician: Actually, we approved several measures to reduce government spending. Among the most important of those was a drastic cut in the high costs of the political machinery. When we first came into office, the national budget was in a terrible state, while now, thanks to the job we did in government, it's back under control.

Journalist: But as a matter of fact, the statistical data doesn't show a significant reduction of the national debt and voters are very sceptical about the effect of your policies. Indeed, many people think that you could have done more.

After this dialogue, the politician answered with a defensive statement, which varied across experimental conditions. In the self-focused factual condition, the politician answered directly, focusing on their own unsatisfactory past actions (‘Certainly, I didn't state my position firmly enough within the coalition and I didn't fully implement my own ideas. I wasn't forceful enough about my ideas and proposals. In general, I didn't act decisively enough’). In the other-focused factual condition, the politician pointed out the wrongdoings of the opposition (‘Certainly, it should be said that the opposition didn't revise its ideological stance and it didn't keep their extreme wing under control. They didn't support our proposals. In general, they did not act in a collaborative manner’). In the self-focused counterfactual condition, the politician used counterfactual sentences focused on themselves (‘Certainly, things would have been better if I had stated my position firmly enough and if I had fully implemented my own ideas. If I had been more forceful about my ideas and proposals, it would have been easier to reach consensus. In general, if I had acted decisively enough, I would have reduced the national debt’). Finally, in the other-focused counterfactual condition, the politician used counterfactual statements that were focused on the opposition (‘Certainly, things would have been better, if the opposition had revised its ideological stance and had kept their extreme wing under control. If the opposition had supported our proposals, it would have been easier to reach consensus. In general, if the opposition had acted in a more collaborative manner, I would have reduced the national debt’).

Participants were randomly assigned to one of the four aforementioned experimental conditions. After reading the interview, they were asked to answer a short questionnaire, and later, they were thanked and debriefed.

Measures

We asked participants to give a general evaluation of the interviewed politician (‘Generally speaking, how would you rate the described politician?’) using a 10-point scale ranging from 1 (very negatively) to 10 (very positively) and to report to what extent they considered the politician responsible (‘How responsible do you think the politician is for the state of the national debt?’) using a 7-point scale ranging from 1 (not responsible at all) to 7 (very responsible). We also asked participants to indicate to what extent they found the politician's defence convincing on a 7-point scale ranging from 1 (not at all) to 7 (very much). Finally, as a control measure, we asked participants to rate the believability of the politician's defence, indicating how clear, sincere, relevant, and exhaustive the defence was, using the same 7-point scale. The four items, based on the four Gricean maxims of conversation (Grice, 1975), were used to compute a single believability index (Cronbach's α = .829).

Results and Discussion

Responsibility Attributed to the Politician

Using a 2 (defence target: self vs other) × 2 (defence style: factual vs counterfactual) ANOVA design, we analysed the degree of responsibility attributed to the politician for the state of the national debt. A main effect of defence style emerged, F(1, 108) = 17.54, p < .001, η2 = 0.09. As expected, participants attributed less responsibility to the politician using a counterfactual defence than to the politician using a factual defence (M = 3.82, SD = 1.34 vs M = 4.75, SD = 1.07). Defence target also had a main effect, with the politician being judged less responsible when using an other-focused defence than when using a self-focused defence (M = 3.96, SD = 1.26 vs M = 4.59, SD = 1.27), F(1, 108) = 7.17, p < .005, η2 = 0.04. The main effects were qualified by an interaction effect that approached significance F(1, 108) = 3.47, p = .07, η2 = 0.02. The responsibility attributed to the politician using a counter factual defence rather than a factual defence was much lower under the other-focused conditions (M = 3.32, SD = 1.19 vs M = 4.65, SD = 0.94), t(52) = 4.56, p < .001, whereas the difference between the two self-focused conditions was not statistically significant (M = 4.32, SD = 1.31 vs M = 4.83, SD = 1.18), t(56) = 1.57, p = .12.1 This result showed that the advantage of counterfactual defence over factual defence in reducing the responsibility attributed to the politician was especially evident in the case of other-focused counterfactuals.

Convincingness of the Defence

Using the same 2 × 2 ANOVA design employed earlier, we then analysed the effects of defence target and style on the evaluation of the convincingness of the defence. Again, a main effect of defence style emerged, F(1, 108) = 6.44, p < .05, η2 = 0.02. Counterfactual defences were evaluated as more convincing (M = 3.32, SD = 1.78) than factual defences (M = 2.55, SD = 1.30). The defence target was also significant, with other-focused defences being judged as more convincing (M = 3.42, SD = 1.65) than self-focused defences (M = 2.50, SD = 1.44), F(1, 108) = 9.48, p < .01, η2 = 0.04. The interaction effect between defence style and target was instead not significant, F(1, 108) = 0.35, p = .55, η2 < 0.01.2

Evaluation of the Politician

The same 2 × 2 ANOVA design was also employed to analyse how the manipulated variables affected the participants' judgement of the politician. Once again, a main effect of defence style emerged, F(1, 108) = 6.40, p < .05, η2 = 0.02. The politician received a more positive general evaluation when presenting a counterfactual, rather than factual, defence (M = 4.48, SD = 1.90 vs M = 3.71, SD = 1.22). As in the earlier analyses, the effect of defence target was also significant. The politician was evaluated more positively when using an other-focused defence than when using a self-focused defence (M = 4.43, SD = 1.63 vs M = 3.79, SD = 1.59), F(1, 108) = 4.13, p < .05, η2 = 0.02. No significant interaction between defence target and defence style emerged, F(1, 108) = 0.53, p = .47, η2 < 0.01.3

Mediation Analysis

A further aim of our study was to test whether the effect of counterfactual defences on the general evaluation of the politician depended on there being less responsibility attributed to the politician and on the counterfactual defences being more convincing than the factual defences. The evaluation of the politician was positively correlated with the convincingness of the defence, r(110) = .637, p < .001, and negatively correlated with responsibility attribution, r(110) = −.505, p < .001. Responsibility attribution and convincingness were also negatively correlated, r(110) = −.381, p < .001. We used AMOS 20 (IBM Corp., Armonk, NY, USA) software to test three alternative mediation models of the effect of defence style (coded 0 for factual defence and 1 for counterfactual defence) on the evaluation of the politician. The first model consisted of a dual-pathway mediation where responsibility attribution and convincingness of defence independently mediated the effect of defence style on the evaluation of the politician. The second model consisted of a three-pathway mediation where the effect of defence style was mediated by the convincingness of the defence, whose effect on the evaluation of the politician was in turn mediated by responsibility attribution. Finally, the third model reversed the order of mediators in a three-pathway mediation so that the effect of defence style was mediated by responsibility attribution, whose effect on the evaluation of the politician was in turn mediated by the convincingness of the defence. All three models were estimated with 5000 bootstrapping resamples to assess indirect effects according to the procedure proposed by Preacher, Rucker, and Hayes (2007).

The first model showed a poor fit to the data, χ2(2, N = 112) = 12.53, p < .001, with less than adequate fit indexes, adjusted goodness-of-fit index (AGFI) = 0.747, normed fit index (NFI) = 0.884, comparative fit index (CFI) = 0.897, and root mean square error of approximation (RMSEA) = 0.218. The second model had a small and not significant difference in model fit as compared with the first model, ∆χ2(1) = 1.891, p = .169, AGFI = 0.563, NFI = 0.902, CFI = 0.906, RMSEA = 0.295. The third model, instead, had a significantly better fit, ∆χ2(1) = 11.09, p < .001, AGFI = 0.936, NFI = 0.987, CFI = 0.996, RMSEA = 0.063, and parameter estimation confirmed our expectations (Figure 1). Counterfactual defences led to a lower responsibility attributed to the politician, β = −.361, bootstrapping 95% confidence interval (CI) for 5000 samples CI [−0.506, −0.200], p < .001, which in turn increased the convincingness of the defence, β = −.381, CI [−0.542, −0.198], p < .001. Finally, the convincingness of the defence strongly predicted the evaluation of the politician, β = .520, CI [0.378, 0.646], p < .001. The effect of defence style on the convincingness of the defence was fully mediated by responsibility attribution, which retained a direct effect on the evaluation of the politician, β = −.306, CI [−0.435, −0.161], p < .001, in addition to an indirect effect through the convincingness of the defence, β = −.198, CI [−0.308, −0.103], p < .001.

image

Figure 1. Path model showing the indirect effect of defence style on the evaluation of the politician through responsibility attributed to the politician and convincingness of the defence (Study 1)

Download figure to PowerPoint

In sum, the results of Study 1 fully confirmed our expectations that counterfactual defences would be more effective than factual defences. By reducing the extent of responsibility attributed to the politician for the negative outcome, counterfactual defences were perceived as more convincing and led to a better evaluation of the politician. In addition, our results showed that other-focused defences (i.e. excuses) led to a better evaluation of the politician than self-focused defences (i.e. concessions), which suggests that shifting responsibility to others may be more advantageous than recognising one's own responsibility. However, the advantage of the other-focused defence over the self-focused defence in terms of reducing the level of responsibility was especially evident in the case of the counterfactual defence, which suggests that an excuse that shifts responsibility to others may work especially if it is formulated in indirect terms.

STUDY 2

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

After finding that counterfactual defences were more effective than factual defences, in Study 2, we focused on counterfactual defences only, manipulating the target, direction, and structure of the counterfactual statements employed as defences. Here, we aimed to compare the effects of counterfactual concession (i.e. self-focused upward counterfactuals), excuse (i.e. other-focused upward counterfactuals), and justification (i.e. self-focused downward counterfactuals) in a single experimental design. We also manipulated the structure of the counterfactuals in order to control for possible differences between additive and subtractive counterfactual defences. Finally, we added one new dependent measure, the evaluation of the defending politician's leadership (Bertolotti et al., 2013). Given that the attack and consequently the defence concerned the performance of the leader, we expected the defence to influence not only the general evaluation of the politician but also the evaluation of the politician's leadership.

Method

Participants and Procedure

A total of 202 students (43.3% men, mean age 22.4 years) from the Catholic University of Milan participated in the study as volunteers. Participants were asked to read the same interview excerpt employed in Study 1, and again, only the final defence by the politician was manipulated. Eight different versions of the defence were created, which varied according to target (self-focused or other-focused), direction (upward or downward), and structure (additive or subtractive). The self-focused and other-focused upward additive counterfactual defences were the same as those used in Study 1. In addition, one experimental condition employed self-focused downward additive counterfactuals (‘Certainly, but things would have been worse, if I had hesitated to state my position within the coalition and had restrained from fully implementing my own ideas. If I had been weak in promoting my ideas and proposals, some decisions wouldn't have been taken at all. In general, if I had acted indecisively, I wouldn't have reduced the national debt’), and one condition employed other-focused downward additive counterfactuals (‘Certainly, but things would have been worse, if the opposition had maintained its ideological stance and if it had loss control of its extreme wing. If the opposition had blocked our policy proposals, some decisions wouldn't have been taken at all. In general, if the opposition had acted in a less collaborative manner, I wouldn't have reduced the national debt’). To create the four subtractive versions of the defence, we proceeded as follows. The upward subtractive versions were created by replacing the downward additive versions with their negations, both for self-focused counterfactuals (‘Certainly, things would have been better, if I hadn't hesitated to state my position within the coalition…’) and for other-focused counterfactuals (‘Certainly, things would have been better, if the opposition hadn't maintained its ideological stance…’). Symmetrically, the downward subtractive versions were created by replacing the upward additive versions with their negations, both for self-focused counterfactuals (‘Certainly, but things would have been worse if I hadn't stated my position firmly enough…’) and for other-focused counterfactuals (‘Certainly, but things would have been worse, if the opposition hadn't revised its ideological stance…’).

Measures

The evaluation of the politician and the responsibility attribution were assessed using the same measures employed in Study 1. In addition, participants were asked to evaluate the politician's leadership. The politician's leadership was measured by asking participants to complete the statement ‘Based on your impression after reading the interview, could you tell us to what extent the politician is…’ followed by four trait adjectives: determined, energetic, tenacious, and dynamic. Participants rated each item using a 7-point scale, ranging from 1 (not at all) to 7 (very much), and ratings from the four items were used to compute a politician's leadership index (Cronbach's α = .847).

Results and Discussion

Responsibility Attribution

We performed a 2 (counterfactual target: self vs other) × 2 (counterfactual structure: additive vs subtractive) × 2 (counterfactual direction: upward vs downward) ANOVA on the responsibility attributed to the politician. No significant main effect of counterfactual target, structure, or direction emerged, F's(1, 194) < 1.17, p's > .28, η2's < 0.01, whereas a counterfactual direction-by-target interaction was found, F(1, 194) = 9.30, p < .005, η2 = 0.05 (Figure 2). Planned contrasts confirmed our expectations and showed that responsibility attribution based on other-focused upward defence (M = 3.41, SD = 1.35) and self-focused downward defence (M = 3.62, SD = 1.19) was similar, p = .40, and that it was lower than the responsibility attribution based on self-focused upward defence (M = 4.13, SD = 1.30), both p's < .05, or on other-focused downward defence (M = 3.96, SD = 1.04), both p's < .05. No further significant two-way or three-way interaction effects were found, F's(1, 194) < 1, p's > .50, η2's < 0.01.

image

Figure 2. Evaluation of the defending politician as a function of counterfactual (CF) target and counterfactual direction (Study 2)

Download figure to PowerPoint

Evaluation of the Politician

The same 2 × 2 × 2 ANOVA was then performed on the evaluation of the politician. Results mirrored those found in the case of responsibility attribution. No main effect emerged as significant, F's(1, 194) < 1, p's > .30, η2's < 0.01, whereas a significant interaction effect between counterfactual target and direction was found, F(1, 194) = 4.23, p < .05, η2 = 0.02. A self-focused downward defence and an other-focused upward defence (M = 4.87, SD = 1.43 and M = 4.55, SD = 1.66) led to a similarly high evaluation of the politician, p = .40, higher than the evaluation of the politician after a self-focused upward defence (M = 4.17, SD = 1.63), both p's < .05, and after an other-focused downward defence (M = 4.31, SD = 1.67), both p's < .05. Again, no significant two-way or three-way interaction effects were found with counterfactual structure, F's(1, 194) < 1.20, p's > .28, η2's < 0.01.

Evaluation of the Politician's Leadership

The evaluation of the politician's leadership was positively correlated with the general evaluation of the politician, r(200) = .397, p < .001, and negatively correlated with responsibility attribution, r(200) = −.224, p = .001. Finally, as in Study 1, the evaluation of the politician was significantly and negatively correlated with responsibility attribution, r(200) = −.424, p < .001.

The same 2 × 2 × 2 ANOVA performed in the previous analyses was also performed on participants' evaluations of the politician's leadership. Results showed a main effect of counterfactual direction on leadership ratings, F(1, 194) = 5.66, p < .05, η2 = 0.03, with downward defence leading to a more positive evaluation of the politician's leadership (M = 4.02, SD = 1.09) than upward defence (M = 3.64, SD = 1.11). A counterfactual target-by-direction interaction effect also emerged, F(1, 194) = 10.56, p < .01, η2 = 0.05, and again confirmed our expectation. Post hoc comparisons revealed that the evaluation of the politician's leadership was lower after the use of a self-focused upward defence (M = 3.32, SD = 1.15) than after the use of the other three types of defence (self-focused downward, M = 4.17, SD = 1.07; other-focused downward, M = 3.85, SD = 1.09; and other-focused upward, M = 3.97, SD = 0.96), all p's < .05. Differences between other-focused upward and self-focused and other-focused downward defences were not significant, all p's > .13. No significant main or interaction effects of counterfactual structure were found, F's(1, 194) < 2.3, p's > .13, η2's < 0.01.

In sum, the results of Study 2 fully confirmed our expectation that counterfactual excuse (i.e. other-focused upward defence) and justification (i.e. self-focused downward defence) would be more effective than counterfactual concession (i.e. self-focused upward defence). Counterfactual excuse and justification were more effective than counterfactual concession in reducing the responsibility attributed to the politician, as well as in improving the general evaluation of the politician and the evaluation of the politician's leadership. These data suggest that when defensive accounts are formulated counterfactually, excuse and justification are equally effective and both are more effective than counterfactual concession. Finally, counterfactual structure did not have any effect on any of the dependent measures, indicating that participants had no specific preference for additive or subtractive counterfactual defences.

STUDY 3

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

After finding through Study 2 that counterfactual justification (i.e. self-focused downward defence) can be employed as an effective defence, in Study 3, we further investigated the effects of this defence. Here, we wanted to assess whether the effectiveness of counterfactual justification can be explained by a reduced negative perception of the defending individual's past performance and whether it is moderated by perceived similarity with the defending individual.

Method

Participants and Procedure

A total of 108 students (51.4% men, mean age 21.5) from the Catholic University of Milan participated in the study as volunteers. Participants were asked to read the same interview excerpt employed in the earlier two studies, but this time, only the counterfactual direction of the defence was varied. The final defensive statements therefore consisted of self-focused upward additive counterfactuals (i.e. ‘If I had…, things would have been better’) or self-focused downward additive counterfactuals (i.e. ‘If I had…, things would have been worse’) and were the same ones employed in Study 2. In addition, in Study 3, the political affiliation of the politician was manipulated by presenting the politician as the leader of either a centre-left or centre-right coalition. Political affiliation was indicated at the beginning of the interview and repeated a further three times in the text. Participants were randomly assigned to one of the four experimental conditions derived from the combination of the direction of counterfactuals (upward vs downward) and the political orientation of the politician (centre left vs centre right).

Measures

We asked participants to indicate their evaluation of the politician using the same scale employed in Studies 1 and 2. As a control measure, we also assessed the believability of upward and downward counterfactuals, using the same four-item measure employed in Study 1. In addition, we asked participants to evaluate the politician's past performance on the issue of national debt using a 7-point scale ranging from 1 (very negative) to 7 (very positive). Finally, participants were asked to indicate their political orientation by choosing among five possible options: left, centre left, centre right, right, or ‘none of the above’. Participants were later regrouped into three groups: those with a left or centre-left orientation (‘centre left’; n = 32), those with a right or centre-right orientation (‘centre right’; n = 43), and those who did not express their orientation (‘not placed’; n = 33).

Results and Discussion

Evaluation of the Politician

We performed a 2 (counterfactual direction: upward vs downward) × 2 (politician orientation: centre left vs centre right) × 3 (participant orientation: centre left vs centre right vs not placed) ANOVA on the evaluation of the politician. A main effect of counterfactual direction emerged as significant, F(1, 96) = 4.98, p < .05, η2 = 0.05, with the downward defence leading to a better evaluation of the politician (M = 4.85, SD = 1.69) than the upward defence (M = 4.23, SD = 1.59). There was no main effect of the politician's orientation (M = 4.50, SD = 1.64 for centre left vs M = 4.56, SD = 1.66 for centre right), F(1, 96) = 0.11, p = .91, η2 < 0.01, nor of the participants' political orientation (M = 4.41, SD = 1.68; M = 4.81, SD = 1.51; and M = 4.27, SD = 1.55 for centre left, centre right, and not placed, respectively), F(2, 96) = 1.62, p = .20, η2 = 0.02. However, a significant interaction did emerge between the two, F(2, 96) = 13.46, p < .001, η2 = 0.25, confirming the predicted partisan bias of participants in favour of the politician sharing their own ideology. The centre-left politician was evaluated more positively by participants with a centre-left orientation (M = 5.41, SD = 1.51) than by those with a centre-right orientation (M = 4.23, SD = 1.31) and those not placed (M = 3.94, SD = 1.85), F(2, 53) = 4.40, p < .05, η2 = 0.14. Likewise, the centre-right politician was evaluated more positively by centre-right participants (M = 5.43, SD = 1.50) than by centre-left participants (M = 3.27, SD = 1.03) and those not placed (M = 4.63, SD = 1.63), F(2, 49) = 10.06, p < .001, η2 = 0.29. No interaction however emerged between counterfactual direction and political orientation of the politician or the participant, F's < 1.7, p's > .20, η2's < 0.01, showing that the effect of counterfactual direction was independent from political orientation.4

Evaluation of the Politician's Past Performance

The same 2 × 2 × 3 ANOVA model was employed to analyse the effect of counterfactual defence and political orientation on the evaluation of the politician's past performance. A main effect of counterfactual direction was found, F(1, 96) = 5.81, p < .05, η2 = 0.05, with downward defence leading to a more positive evaluation (M = 3.69, SD = 1.16) of the politician's past performance than upward defence (M = 3.13, SD = 1.24). An interaction effect between the politician's political orientation and the participant's political orientation was also found, F(1, 96) = 5.43, p < .01, η2 = 0.09, with differences closely mirroring those found in the earlier analysis. The centre-left politician's past performance was evaluated more positively by participants with a centre-left orientation (M = 3.88, SD = 1.45) than by those with a centre-right orientation (M = 2.88, SD = 1.05) and those not placed (M = 3.45, SD = 0.96), F(2, 53) = 4.88, p < .05, η2 = 0.16. Likewise, the centre-right politician's past performance was evaluated more positively by centre-right participants (M = 4.90, SD = 1.09) than by centre-left participants (M = 2.67, SD = 1.04) and those not placed (M = 3.37, SD = 1.46), F(2, 49) = 4.77, p < .05, η2 = 0.16. Again, no interaction between counterfactual direction and political orientation of the politician or the participant was found.5

Mediation Analysis

The evaluation of the politician's past performance was strongly correlated with the evaluation of the politician, r(106) = .60, p < .001. To test whether the positive effect of downward counterfactual defence was mediated by the evaluation of the politician's past performance, we performed a mediation analysis, following the bootstrapping procedure proposed by Preacher et al. (2007). We regressed the evaluation of the politician's past performance on counterfactual direction (coded 0 for upward counterfactual defence and 1 for downward counterfactual defence) and found the previously observed main effect of downward defence, β = .39, t = 2.06, p < .05. We then regressed the politician's evaluation on the direction of counterfactual defence, β = .37, t = 1.97, p = .05. The evaluation of the politician's past performance was then added as a predictor, β = .58, t = 7.29, p < .001. As we expected, counterfactual direction was no longer significant after controlling for the evaluation of past performance, β = .15, t = 0.93, p = .36. The significance of the indirect effect was tested using 1000 bootstrapping samples with the bias-corrected 95% CI method. As the CI did not include zero, CI [0.02, 0.22], the Preacher et al. (2007) criteria for mediation were satisfied, indicating that the effect of downward counterfactual defence was mediated by the evaluation of the politician's past performance.

In conclusion, the results of Study 3 further confirmed the results of Study 2 as regards the higher effectiveness of counterfactual justification (i.e. self-focused downward counterfactuals) as compared with counterfactual concession (i.e. self-focused upward counterfactuals). This result was independent from the political orientation of both the politician and the participant (and from the interplay between the two), thus confirming that the positive effect elicited by counterfactual justification was not moderated by the audience's prior attitudes towards the politician. Furthermore, the findings from Study 3 confirmed our hypothesis that the effect of counterfactual justification would be mediated by a reduction in the perceived negativity of the individual's past performance. By focusing attention on the possibility of a worse outcome, the politician reduced the perceived negativity of their past performance and was consequently evaluated more positively.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES

The results of our research show that counterfactuals are effective defence strategies. A counterfactual defence is more convincing and leads to a more positive evaluation of a defending individual than the corresponding factual defence. This advantage of counterfactual over factual defence can be ascribed to counterfactual defence creating the impression that the defending individual is less responsible for the negative outcome. Furthermore, our findings show that counterfactual excuse (i.e. ‘If they… things would have been better’) and counterfactual justification (i.e. ‘If I… things would have been worse’) are similarly effective and are more effective than counterfactual concession (i.e. ‘If I… things would have been better’). Finally, the effectiveness of different types of counterfactual defence is not affected by ideological similarity between the audience of the counterfactual statement and the individual being evaluated, which further supports the strength and generality of our results.

These findings add to our knowledge regarding the effectiveness of defensive accounts, by showing how and why an indirect defence expressed through counterfactual statements may be more effective than a direct factual defence. Counterfactual defence may appear to be less robust, and therefore less effective, than factual defence because it is formulated using conditional terms (‘If only…’). However, this is not the case. Evidently, counterfactual defence provides a hypothetical point of comparison with actual behaviour, which can influence how people judge this behaviour. This mechanism is reminiscent of the findings of Tormala, Jia, and Norton (2012), who showed that providing information about the potential achievements of a person induces more positive evaluations of that person than information about their actual achievements in a range of professional, athletic, and academic tasks.

As discussed in the introduction, past research on account giving has yielded conflicting results on the effectiveness of different kinds of defence. Some research, for example, has indicated that excuse is more effective, whereas other research has suggested that justification is (Shaw et al., 2003). In our research, we found that counterfactual excuse and counterfactual justification were equally more effective than counterfactual concession. The effectiveness of the counterfactual excuse (i.e. ‘If they… things would have been better’) is especially notable, because it involves a suggestion that others are to blame for the negative outcome. Prior research has found that criticising other people for negative outcomes is likely to induce a backlash effect against the person making the criticism (Roese & Sande, 1993; Skowronski et al., 1998). This was not the case in the present research, probably because counterfactual excuse blames others indirectly (as part of a hypothetical scenario) rather than explicitly. Indirectness thus seems to make counterfactual excuse more ‘palatable’ than factual excuse, confirming that oblique forms of communication can have greater effectiveness than straightforward ones (Catellani & Bertolotti, 2013; Fiedler, 2008; Fiedler & Mata, 2013).

The effectiveness of counterfactual justification (i.e. ‘If I… things would have been worse’) is another interesting finding of our research. A counterfactual justification reduces the perceived negativity of the defending person's actions by comparing them with hypothetical actions that would have led to a worse outcome. This in turn leads to a better evaluation of the individual. This finding is consistent with previous research that shows that downward counterfactuals trigger an ‘affective contrast effect’ that reduces the perceived negativity of the actual event (McMullen & Markman, 2000, 2002). Previous research has also shown that downward counterfactual thinking may lead to a shift in the standards of comparison that are employed to evaluate both ourselves and others. When people consider how they have refrained from immoral behaviour in the past, they may feel more justified in adopting a lower moral standard in the future (Effron, Monin, & Miller, 2012). A downward comparison with a more negative reference point can also improve evaluations of the past behaviour of others, by lowering the standard of comparison (Markman, Mizoguchi, & McMullen, 2008). Our research showed that defensive accounts that focus on how a situation could have been worse can also trigger such shifts in the standards of comparison.

Future research might explore the factors that moderate the effects of the various defensive accounts. In the case of counterfactual justification, one of these factors may be the mental availability of the downward hypothetical alternative. A downward comparison with a worse hypothetical scenario is more likely to be convincing if the worse scenario is readily accessible, as research on close-call counterfactuals suggests (Markman, Elizaga, Ratcliff, & McMullen, 2007; Markman & Tetlock, 2000; Tetlock, 1998). Another moderating factor may be the severity of the actual event. When one needs to defend oneself against a particularly serious allegation, the suggestion of a worse possible alternative might be less plausible and therefore less convincing when evoked in a counterfactual justification. A third moderating factor may be the content of the allegation and whether it is, for example, leadership or morality that is called into question (Kim, Dirks, Cooper, & Ferrin, 2006). In our research, the individual defended themselves against a charge related to their leadership. Further research could investigate the use of counterfactual defences when allegations are instead related to morality.

An in-depth exploration of the use of counterfactuals in defensive communication should also pay particular attention to the issue of ecologic validity, which is inevitably limited by the experimental manipulation used in scenario studies. An analysis of actual political speeches would be essential to investigate how counterfactuals are indeed employed by politicians. A recent research investigating actual political speeches (Catellani & Covelli, 2013) seems to confirm that politicians make wide use of (both implicit and explicit) counterfactuals to defend themselves, but further studies in this field would be desirable.

In our research, we used written transcripts of fictitious politicians' defences, thus asking participants to base their judgments only on a written text. In other studies, participants might instead be exposed to (real or fictitious) politicians who defend themselves through a television broadcast or YouTube video. In this case, recipients' attention would be drawn not only to the verbal content of the defence but also to the nonverbal cues accompanying it. As defensive contexts can arouse strong emotional reactions in speakers, nonverbal behaviour is likely to play a relevant role in defensive communication.

A limitation of our research derives from the specific context that we investigated, that is, political communication. From a certain point of view, politicians have the same defensive options as everyone else. Therefore, the effects that clearly emerged in our research are also likely to emerge in other contexts. The strength of these effects may, however, vary from one context to another. For example, in our research, the excuse employed by an incumbent politician focused on the opposition, namely, the ‘expected’ rival of the politician. In contexts other than the political one, identifying a natural and accepted rival may be more difficult. Future research could therefore include an in-depth investigation of what ‘others’ may successfully be chosen as targets of defensive counterfactuals.

Another finding of our research that should be tested in a different context is that effective defences led to a strong reduction in the responsibility attributed to the politician using the defence but improved to a lesser degree the convincingness of the defence and the evaluation of the politician. The fact that the defending person was a politician may have affected the lower strength of these effects. Past research has shown that people tend to be suspicious of persuasive messages coming from politicians (McGraw, Lodge, & Jones, 2002). Replicating a similar scenario in a nonpolitical context might therefore yield even stronger effects than those observed in the present research as regards the convincingness of the defence and the resulting evaluation of the defending individual.

In conclusion, our research showed that when someone needs to explain unsatisfactory results without losing credibility, one way of doing it is to shift the focus from what actually happened to what could have happened instead. By employing counterfactuals, people subtly influence perceptions of their past performance, thereby inducing their audience to modify their attribution of responsibility for negative outcomes, and thus the evaluation of the actors involved. Counterfactual defences therefore are indirect and subtle communication strategies that effectively influence social judgements.

  1. 1

    To control for possible effects of defence believability, a 2 × 2 analysis of covariance (ANCOVA) on the evaluation of the politician was also conducted, with defence believability as a covariate. The covariate had a significant effect on responsibility attribution, F(1, 107) = 6.80, p < .05, η2 = 0.04, indicating that in general, more believable defences resulted in lower responsibility attributed to the politician. Defence style, F(1, 107) = 15.73, p < .001, η2 = 0.11, defence target, F(1, 107) = 7.39, p < .01, η2 = 0.05, and the interaction between the two, F(1, 107) = 3.84, p = .07, η2 = 0.03, maintained effects similar to those found in the 2 × 2 ANOVA. This confirmed that the effect of counterfactual direction was robust also after controlling for the believability of the defence.

  2. 2

    The same 2 × 2 ANCOVA carried out on responsibility attribution was also carried out on the convincingness of the defence. Results showed that defence believability had a strong effect on convincingness, F(1, 107) = 66.51, p < .001, η2 = 0.33. The effects of defence style, F(1, 107) = 4.84, p < .05, η2 = 0.04, target, F(1, 107) = 15.33, p < .001, η2 = 0.08, and their two-way interaction F(1, 107) = 0.46, p = .50, η2 < 0.01, remained however similar to those found in the 2 × 2 ANOVA.

  3. 3

    As in previous analyses, a 2 × 2 ANCOVA with defence believability as a covariate was conducted also on the evaluation of the politician. A strong effect of believability was again found, F(1, 107) = 94.26, p < .001, η2 = 0.42. Once again, the effects of defence style, F(1, 107) = 5.31, p < .05, η2 = 0.02, target, F(1, 107) = 7.07, p < .05, η2 = .03, and their two-way interaction F(1, 107) = 1.36, p = .25, η2 < 0.01, remained similar to those found in the 2 × 2 ANOVA.

  4. 4

    To control for possible effects of defence believability, a 2 × 2 × 3 ANCOVA on the evaluation of the politician was conducted, with defence believability as a covariate (see Study 1). An effect of defence believability was found, F(1, 95) = 24.35, p < .001, η2 = 0.20, indicating that, in general, the politician was evaluated more positively when the defence was considered more believable. The other results closely mirrored those obtained with the 2 × 2 × 3 ANOVA, with similar effects of counterfactual direction, F(1, 95) = 9.86, p < .01, η2 = 0.06, and of the interaction between the politician's and participants' political orientations, F(2, 95) = 8.89, p < .001, η2 = 0.11. The other main and interaction effects remained nonsignificant, F's < 2.07, p's > .13, η2's < 0.04.

  5. 5

    A 2 × 2 × 3 ANCOVA on the evaluation of the politician's past performance was also conducted, with defence believability as a covariate. Again, an effect of believability was found, F(1, 95) = 3.98, p < .05, η2 = 0.04, but the other results closely mirrored those found in the 2 × 2 × 3 ANOVA. Counterfactual direction, F(1, 95) = 4.71, p < .05, η2 = 0.04, and the interaction between the politician's and participants' political orientations, F(2, 95) = 4.34, p < .05, η2 = 0.07, were still significant, whereas the other main and interaction effects remained nonsignificant, F's < 2.48, p's > .09, η2's < 0.04.

REFERENCES

  1. Top of page
  2. Abstract
  3. DEFENSIVE ACCOUNTS
  4. COUNTERFACTUAL AS DEFENCE
  5. THE PRESENT RESEARCH
  6. STUDY 1
  7. STUDY 2
  8. STUDY 3
  9. GENERAL DISCUSSION
  10. REFERENCES