To Apologize or to Compensate for Colonial Injustices? The Role of Representations of the Colonial Past, Group-Based Guilt, and In-Group Identification

In the aftermath of an intergroup conflict, along with instrumental reparations, victims may request an apology on behalf of the perpetrators, yet such political apologies are often not given. Whereas we know a lot about the motivations of the victims, less is known about when and why the perpetrators are willing to apologize. In this study, from the perspective of the perpetrator group, we simultaneously examined and compared support for a political apology and for instrumental reparations (e.g., financial support and other forms of assistance) offered to both former colonies and colonial-origin minorities living in the country responsible for the past colonization. We considered the indirect role of positive and negative representations of the colonial past via feelings of group-based guilt. Using a community sample of the native Dutch population (N = 763), we showed that the Dutch were more supportive of instrumental reparations than of political apology. They also agreed with both the positive and negative aspects of their colonial past, but they did not experience much collective guilt. Agreement with positive representations of the Dutch colonial past was, via weaker feelings of group-based guilt, related to less support for both political apology and instrumental reparations. In contrast, negative representations of the past were, via higher guilt, related to more support for these reconciliatory outcomes. These processes were similar for higher and lower in-group identifiers. Importantly, the association between guilt and support for political apology was twice as large as the one between guilt and support for instrumental reparations, suggesting that political apology is more effective in restoring the in-group’s moral self-image.

In the aftermath of an intergroup conflict, victims often request an apology from the perpetrators, to be issued by their representatives, such as the government or a leader (Brooks, 1999). However, in many cases, such political apologies for past transgressions have not yet been given, and this also holds in the particular context of colonial history. For instance, even though the Dutch government and the Crown have recently apologized for some of the atrocities committed in their former colony Indonesia, as shown in the quote above, no apology has been issued to their other former colony Suriname and its people for the slave trade (Stack, 2020), and Spain has rejected Mexico's demand to apologize for the harm done during the colonization (López-Obrador, 2019). The absence of such symbolic gestures is problematic because political apologies, when paired with instrumental reparations (e.g., economic compensation) and post-apology engagement, are most effective in facilitating reconciliation and victims' empowerment (Hornsey, Wohl, & Philpot, 2015;Wohl, Hornsey, & Philpot, 2011).
Whereas research on support for instrumental reparations for transgressions has mostly focused on perpetrators (e.g., Doosje et al., 1998;Iyer, Leach, & Crosby, 2003;Klandermans, Werner, & van Doorn, 2008), research on symbolic reparations has predominantly adopted the victim perspective and examined the victims' need for or willingness to accept an apology (see Blatz & Philpot, 2010). This is unfortunate because it is important to know not only when and why victims need an apology, but also when and why members of the perpetrator group are ready to respond to these needs. Therefore, in this research, we assess support for political apology among members of the perpetrator group.
We zoom in on the role of the representations of the past and feelings of group-based guilt in explaining perpetrator groups' support for an apology alongside their support for instrumental reparations. People can interpret past events in different ways (Liu & Hilton, 2005), and research has shown that both positive and negative representations of the past play a crucial and contrasting role in people's willingness to offer instrumental reparations for the transgressions committed by their in-group members (e.g., Cabecinhas, & Feijó, 2010;Mari et al., 2010). There is also convincing evidence that group-based guilt as a moral emotion is associated with more support for this form of reparation (e.g., Bonnot et al., 2016;Doosje et al., 1998;Figueiredo et al., 2010;Licata & Klein, 2010). Whereas we know that guilt also motivates an apology (Allpress et al., 2010;McGarty et al., 2005), we do not know whether positive and negative representations of the past, via guilt, relate to the willingness to issue a political apology, and whether this mechanism differs in strength for symbolic and instrumental reparations. Thus, our contribution to the literature is that we study political apology and instrumental amends in tandem and compare the underlying processes.
Furthermore, we consider the moderating role of ingroup identification. There is experimental evidence that the associations between representations of the past and group-based guilt can differ for high and low in-group identifiers (Doosje et al., 1998); however, whether the relationships between guilt and the perpetrator group's willingness to offer political apology as well as instrumental reparations vary for those strongly and weakly identified remains an open question.
We address these questions in the context of the Dutch colonial past. In addition to support for political apology for the colonial crimes, we take into account support for instrumental reparations, not only limited to financial compensation but also considering other forms of assistance. Furthremore, we distinguish between reparations offered to former colonies and those offered to colonial-origin minorities in the Netherlands. Collective memories of colonialism can power the dynamics not only between the formerly colonized and colonizing nations but also between immigrants from the former colonies and the descendants of colonizers who today live together in European host societies (Bobowik, Valentim, & Licata, 2018). Yet, research on support for instrumental reparations for the colonial past has neglected this differentiation.
In contrast to the existing experimental (e.g., Doosje et al., 1998;Lastrego & Licata, 2010) and survey research (e.g., Mari et al., 2010) on colonialism that relied on convenience samples, we use cross-sectional data from a large community sample of native Dutch adults. This allows us to study actual opinions and psychological processes in relation to the colonial past, which is particularly relevant given the current emphasis in Western societies on the need for a more nuanced remembering of the colonial times (Bobowik et al., 2018;van Nieuwenhuyse & Valentim, 2018). With this sample, next to answering our explanatory questions, we can provide a descriptive overview of the societal level of endorsement of positive and negative representations of the colonial past to find out whether the silencing of troublesome national past is still common. Furthermore, we can establish whether there is societal support for political apology and how this compares to support for other forms of reparation.

Political Apology and Instrumental Reparations
When remembering or being reminded of in-group's wrongdoings, people may want to repair damaged intergroup relations. Amends can be offered symbolically, in words (e.g., acknowledgement of responsibility, an apology, truth-seeking, or guarantees of non-repetition) and instrumentally, in deeds (e.g., financial compensation, but also restitution of rights and rehabilitation through care and services for victims beyond monetary payments, Barkan, 2000;OHCHR, 2005). Research on symbolic reparations has revealed that victims request an apology (Hornsey, Okimoto, & Wenzel, 2017), usually from a political representative of the perpetrator group on behalf of that group. Furthermore, such political apologies have some positive effects for both groups, namely, they tend to improve intergroup feelings and reduce concerns that the harms of the past will be repeated (e.g., Blatz, Schumann, & Ross, 2009;Wohl et al., 2013). Yet, political apologies are not always effective, particularly when it comes to promoting forgiveness among the victimized groups (Philpot & Hornsey, 2008;; for reviews see Blatz & Philpot, 2010;Blatz et al., 2009;Hornsey, 2016;Hornsey & Wohl, 2013). This is because political apologies usually fail to contain necessary elements of a successful apology, such as sincere declarations of regret, admissions of responsibility and injustice, acknowledgements of victim's suffering, credible promises to behave better in the future, and offers of repair (see Blatz et al., 2009). That is, victims are not convinced enough by apologetic words alone that the perpetrators truly regret what has happened or perceive that the perpetrators fail to offer compensation for current inequalities or injustices (Wohl et al., 2011). Empirical research dedicated to real-life apologies for serious transgressions has shown that members of the victimized groups are usually skeptical of the motives behind a political apology and may perceive such an apology as a cynical statement (e.g., Ferguson et al, 2007).
It is thus important to study political apologies in parallel with other types of reparations, including the instrumental ones. According to the needs-based model of reconciliation (Shnabel & Nadler, 2008), the victimized groups seek empowerment, and this can be achieved not only by means of an apology but also through instrumental reparations. In the context of colonial history, these can entail improving the situation of colonial-origin minorities or delivering structural financial help to former colonies. Whereas there is substantial research on the perpetrator group's willingness to grant instrumental reparations (e.g., Bonnot et al., 2016;Doosje et al., 1998;Figueiredo et al., 2010;Lastrego & Licata, 2010;Licata & Klein, 2010;Mari et al., 2010;Sibley, 2010;Sibley, Liu, Duckitt, & Khan, 2008), studies examining their willingness to offer an apology are still scarce (but see Barlow et al., 2015;Harth, Hornsey, & Barlow, 2011;Hornsey et al., 2017;.
Importantly, to our knowledge, the role of representations of colonial past in motiving the descendants of colonizers to apologize for colonial crimes has not been yet examined. Given the importance of both amends in words and deeds, we investigate support for a political apology and instrumental reparations in parallel. We argue that the motivation to grant both forms of reparation depends on the way colonial history is remembered by the descendants of the perpetrators. We also explore whether the same mechanism guides instrumental reparations directed at former colonies and those directed at colonial-origin minorities in the Netherlands.

Representations of Colonial History and Reparations
Representations of colonial history form an important part of collective memory. Whereas throughout the centuries former colonizing powers depicted colonial times as an era of voyages and discoveries, more recently representations of the colonial period have become ambivalent and polemic among the descendants of the colonizers ( Bobowik et al., 2018). These critical or negative representations of colonialism can coexist with more benevolent or positive ones, and, respectively, descendants of colonizers can remember the colonial past as a history of exploitation and violence (Doosje et al., 1998;Licata & Klein, 2010) as well as a history of paternalism, which commemorates the colonizers as national heroes and good-natured people (Cabecinhas, & Feijó, 2010;Doosje et al., 1998;Licata & Klein, 2010;Vala, Lopes, & Lima, 2008).
Emphasizing the exploitation during colonial history instead of portraying it as 'voyages of discovery' (Cabecinhas, & Feijó, 2010) or civilizing missions can go hand in hand with acknowledgement of responsibility and willingness to grant reparations to the members of the formerly colonized groups. Indeed, studies have shown that critical representations of colonial past are associated with more willingness to offer instrumental reparations for colonial crimes (Berndsen & McGarty, 2009;Lastrego & Licata, 2010;Mari et al., 2010), whereas post-colonial ideologies are linked to more reluctance towards compensations (Sibley, 2010;Sibley et al, 2008). We expect to find a similar pattern for political apology.

The Explanatory Role of Group-Based Guilt
Categorizing oneself as a member of a group that is perceived to have unjustly treated another group can give rise to feelings of group-based guilt (Branscombe, Slugoski, & Kappen, 2004). Importantly, depending on how colonialism is framed or remembered, the descendants of colonizers may or may not experience group-based guilt.
Positive representations that emphasize the colonizers' role in the development of the colonies have been argued to reduce feelings of guilt, as a result of the motivation to maintain a positive group image (Branscombe & Miron, 2004). In contrast, negative construals of colonialism have been suggested to evoke more guilt, because responsibility for the suffering caused is acknowledged. Robust evidence corroborates that reminders of past colonial crimes can trigger feelings of guilt among the descendants of the colonizers whereas positive narratives tend to have the opposite effect (Doosje et al., 1998;Figueiredo et al., 2010;Klein, Licata, & Pierucci, 2011;Licata & Klein, 2010).
Further, as group membership and associated attributes can become part of the individual's self-concept (Tajfel & Turner, 1986), people may feel responsible for making the past right. Group-based guilt is thus usually accompanied with the wish to right the previous collective wrongs (for a review see Wohl, Branscombe, & Klar, 2006;Lickel, Steele, & Schmader, 2011). There is ample evidence that group-based guilt is related to a higher willingness to provide instrumental compensation to the victims of a transgression (e.g., Bonnot et al., 2016;Doosje et al., 1998;Figueiredo et al., 2010;Iyer et al., 2003;Klandermans et al., 2008;Leach et al., 2006;Licata & Klein, 2010), and several studies reveal that guilt also motivates symbolic reparations (i.e., an apology: Allpress et al, 2010;McGarty et al., 2005;Wohl et al., 2013). Whereas there is also research on global indicators of compensations or averaged scores for instrumental and symbolic forms of amends in relation to group-based guilt (Berndsen & McGarty, 2010;Brown et al., 2008;Doosje et al., 1998;Figueiredo et al., 2010;Klein et al., 2011;Licata & Klein, 2010;Zebel, Zimmermann, Viki, & Doosje, 2008), hardly any studies have simultaneously examined support for instrumental and symbolic reparations as two separate outcomes of guilt. An exception is the work done by Allpress and colleagues (2010) but even in this study there was no theorizing about a possible differential effect of guilt on political apology compared to instrumental reparations, nor were these paths estimated in parallel and statistically compared in terms of strength.
Intergroup Emotions Theory suggests that specific group-based emotions should increase the likelihood of specific behavioral tendencies rather than just negative or positive behavioral intentions in general (Mackie, Devos, & Smith, 2000). Following this idea, we argue that the association between group-based guilt and political apology might differ in strength from that between guilt and instrumental reparations. Emotion scholars suggest that group-based guilt is more self-focused (or in-groupfocused) than initially thought (Leach, Iyer, & Pedersen, 2006). Further, ingroup-focused guilt is associated with the more abstract goal of reparation and less with more concrete action necessary to bring systemic compensation (Leach et al., 2006). It is thus reasonable to expect that apology, a more abstract (symbolic) form of restitution, would be more strongly related to guilt than instrumental forms of reparation. Although guilt predicts motivations to make amends, it seems that these intentions are mainly driven by a desire to reduce one's feelings of guilt (see Lickel et al., 2011). This is in line with the needs-based model of reconciliation, where -once a transgression is committed -the perpetrator group desires to restore its moral image (see Shnabel & Nadler, 2008). A political apology, as a symbolic or abstract form of reparation, provides a relatively easy way to regain group morality (Barlow et al., 2015). In turn, instrumental reparations can be more costly for the in-group in terms of economic resources and power (Wohl et al., 2006).
Based on this theoretical reasoning, it can be hypothesized that the association between group-based guilt and support for political apology is stronger than the one between guilt and support for instrumental reparations. Altogether, we expected positive and negative representations of the colonial past to be more strongly related to political apology than to instrumental reparations via feelings of group-based guilt.

The Role of In-Group Identification
We further considered the moderating role of in-group identification in this process. According to Intergroup Emotions Theory (Mackie et al., 2000), self-categorization as a member of an in-group is a prerequisite for experiencing group-based emotions, and higher in-group identifiers tend to experience such emotions more strongly. For instance, higher identifiers have been found to express more anger in the face of perceived injustices (e.g., Pennekamp et al., 2007), and to be more nostalgic about their in-group's past (Smeekes, 2015). Yet, when it comes to moral emotions, such as group-based guilt or shame, that are threatening to the image of the group, higher identifiers might be particularly motivated to avoid experiencing these emotions because they care particularly about maintaining a positive group image (Branscombe & Miron, 2004;Wohl et al., 2006). Experimental research has shown that, when confronted with a clearly negative or positive narrative about the colonial history of their group, high and low identifiers reacted in the same way: by expressing less guilt in the positive condition and more in the negative condition.
However, when confronted with an ambiguous representation of the past (one containing both positive and negative elements), higher national identifiers pick those elements from their in-group's history that are least damaging to their moral image and hence experience less guilt (Doosje et al., 1998). Our study is not experimental but correlational. Yet, as we are measuring agreement with both positive and negative representations of the colonial history, we are making both aspects salient in participants' minds, which comes closest to being confronted with an ambiguous narrative. Therefore, we expect higher identifiers to rely more heavily on the positive than negative representations of the colonial past when evaluating their feelings of group-based guilt. Thus, we hypothesize that the negative association between positive representations of colonization and group-based guilt will be stronger among higher than lower national identifiers. We also hypothesize that the positive association between negative representations and guilt will be weaker among higher than lower national identifiers.
Furthermore, to the extent that members of the perpetrator group experience moral emotions such as guilt, higher identifiers might be more willing than low identifiers to act in order to restore the moral image of their group. According to Social Identity Theory (Tajfel & Turner, 1979), high identifiers are particularly motivated to ensure that their group is of a positive standing in society, and identification has been linked to collective action on behalf of the in-group (Van Zomeren et al., 2008). In a similar vein, when feeling guilty about the transgressions of their ancestors during the colonial times, higher identifiers might be more willing to repair the damage by offering an apology or providing financial or other instrumental compensation.
Therefore, we expected the associations between guilt and both the apology and instrumental reparations to be more positive for higher national identifiers.

The Dutch context
We tested our hypotheses in the Netherlands, a country that has historically had a long colonial presence in Indonesia, Suriname, the Netherlands Antilles and Aruba (Essed & Hoving, 2014). While the Dutch invested in the infrastructure and institutions in these countries, they also exploited the land and the people. For example, during the colonization of Indonesia, natural resources were extracted and Indonesians were used as cheap labor. After a violent struggle, Indonesia declared independence in 1945 (Oostindie, 2014), followed by Suriname in 1975. The Netherlands Antilles and Aruba still form part of the Kingdom of the Netherlands (Sharpe, 2005). First attempts to reconcile the colonial past have been taken since the late 1990s when the Dutch government started supporting initiatives that commemorate the Atlantic slave trade (Oostindie, 2014), although no apologies have been issued to Suriname or Antilles yet. However, in 2011, the Netherlands officially apologized for the atrocities caused by Dutch soldiers in the Indonesian village of Rawagede in 1947 during the Indonesian independence war. This apology was accompanied by a compensation worth twenty thousand euros given to each of the nine remaining surviving relatives (Bijl, 2012).

Participants and Procedure
We used data from an online survey conducted in the Netherlands in 2016 through I&O Research, a Dutch research agency that maintains an online panel with approximately 22,000 panel members with a diverse background in terms of age, gender, education, and province of residence. A random sample of native Dutch participants (i.e., with both parents of Dutch ethnicity) was drawn from this panel. The survey focused on different societal topics in the Netherlands related to history, migration, and ethnic diversity. Participants were informed about this general topic. The questionnaire consisted of several separate modules. One module was reserved for questions about colonial history, and we used all items that were asked in relation to the constructs analyzed in this study. 1 Participants gave their informed consent, and their participation was voluntary and anonymous. Small rewards were offered in return for participation, following the standard policy of I&O Research. The median time for the completion of the survey was 19 minutes. 2 The research agency filtered out incomplete answers and only participants who completed all the measures were retained in the data set (N = 765). We excluded two participants who had missing values on the control variables, resulting in a valid sample of 763 ethnic Dutch adults. Females made up 42%, and the age ranged from 18 to 90 years (M = 56.4, SD = 12.9). Regarding education, 18% of the participants completed a low level (primary or lowest secondary), 52% a medium level (high school or vocational education), and 30% had tertiary education.

Measures
All items were measured on a 7-point Likert scale ranging from (1) 'strongly disagree' to (7) 'strongly agree'. Because we used latent variables, we do not report Cronbach's alpha but composite reliability scores (see the section on the measurement model). Composite reliability statistic is superior to Cronbach's alpha because it does not assume equal factor loadings across items (Raykov, 2017).

Positive representations
This measure was based on a previous study by Lastrego and Licata (2010) and consisted of three items: ' Generally, the Dutch dealt humanely with people from their colonies', 'Dutch colonization has introduced new techniques, whereby the colonies were able to modernize', and 'Despite what some people say about colonization, the colonized peoples have also benefited from the Dutch example by getting new schools, hospitals and better infrastructure'.

Negative representations
We adapted three items from the same study by Lastrego and Licata (2010): 'During the colonial period, the Dutch have caused much pain and suffering to the local population', 'Dutch colonial history was violent and barbaric', and 'The colonial system of the Dutch government was morally unacceptable'.

In-group identification
We focused on national identification, and we used two items borrowed from Verkuyten and Martinovic (2006): 'I identify strongly with the Netherlands' and 'I feel strongly attached to the Netherlands'.

Support for a political apology
Given the relative recency of the actual governmental apology to the victims in Indonesia and the increasing importance of fostering harmonious intergroup relations in immigration countries, we asked about the willingness to apologize to colonial-origin minorities living in the Netherlands. We assessed two items: 'I think we the Dutch must apologize to our minorities from Suriname, Netherlands Antilles and Indonesia, who are now living in the Netherlands', and 'The Dutch government does not need to give a public apology to the colonial minorities living in the Netherlands' (reverse coded).

Support for reparations towards colonial-origin minorities
It was measured with four items: 'The Dutch must make more effort to improve the position of people from former colonies who live in the Netherlands', 'Colonial minorities in the Netherlands do not deserve special Treatment' (reverse coded), 'We need to welcome more migrants from our former colonies in the Netherlands', and 'The Dutch government should do its best to stop further immigration from former colonies' (reverse coded).

Support for reparations in the former colonies
We measured support for reparations in the former colonies with three items: 'The Netherlands should help the former colonies solve their current financial problems', 'It is the responsibility of the Netherlands to help the former colonies in their fight against corruption', and 'The Netherlands should continue to provide economic aid to the former colonies to help them develop further'.
We controlled for age, gender (female = 1, male = 0), and education (7-point scale ranging from primary to postgraduate level).

The Measurement Model
Confirmatory factor analysis in Mplus (version 8.0) with maximum likelihood estimator was used to test a measurement model consisting of seven latent constructs: two types of representations of colonization, three reparation outcomes, group-based guilt, and in-group identification. Table 1 shows that this model had a good fit (Model D), and that it outperformed simpler models with fewer factors (Models A-C). However, the fit of the proposed seven-factor model further improved significantly after dropping one item measuring positive representations 'generally the Dutch dealt humanely with people from their colonies' (reverse coded) that had very low explained variance (R 2 = 0.157). Therefore, we proceeded with the latent factors as specified in Model E.

Descriptive Findings
The descriptive findings can be found in Tables 2 and 3. A weight provided by the data collection agency was applied to correct for over-or underrepresentation of respondents from certain age, gender, and education categories, as well as province of residence, so that the means and correlations presented here are representative of the ethnic Dutch population with respect to these four socio-demographic characteristics. 3 On average, participants did not show much support for political apology, nor did they agree with reparations toward colonial-origin minorities in the Netherlands or toward the former colonies (Table 2). Moreover, they indicated low feelings of group-based guilt and high national identification. The endorsement of both positive and negative representations was also high. Yet, a significant Wald-test revealed that participants endorsed negative representations more than the positive ones, χ 2 (1) = 78.149, p < 0.001. They were also more inclined for their country to offer reparations to immigrant minorities from former colonies (χ 2 (1) = 61.605, p < 0.001) and    Note: NL = The Netherlands. † p < 0.10, * p < 0.05, ***p < 0.001.
to help former colonies develop further (χ 2 (1) = 44.021, p < 0.001) than to offer an apology. That is, there was more support for instrumental than for symbolic reparation. The levels of support for the two types of instrumental reparation were similar, χ 2 (1) = 0.174, p = 0.672. The correlations between the latent variables were mostly significant and always in the expected direction ( Table 3). Endorsement of positive representations was negatively correlated with guilt and all three measures of reparations. The opposite pattern was found for negative representations. National identification was positively related to the positive representations, and negatively to the negative ones. Identification also correlated with less willingness to apologize and to compensate minorities in the Netherlands but not with offering financial help to former colonies.

Mediation by Guilt
We used structural equation modeling in Mplus (version 8.0) to test whether positive and negative representations of colonial history are related to support for political apology and two types of instrumental reparations, and whether these relationships can be explained by group-based guilt. We applied bootstrapping with 5000 replacement samples to obtain confidence intervals for the indirect effects (Preacher & Hayes, 2008). Figure 1 shows the unstandardized regression coefficients for the main paths in the mediation model. Control variables age, gender, education, and Dutch identification are not presented but were included in the analysis both in relation to the mediator and the three dependent variables. The decomposition of the total, indirect, and direct effects of positive and negative representations of colonization on the three types of reparation can be found in Table 4. For the unstandardized parameter estimates of the control variables see Table 5.
The total effects of positive representations of colonial past on apology, reparations in the Netherlands, and reparations in colonies were all negative and significant, whereas the total effects of negative representations were all positive and significant, as expected. These total effects were largely explained by group-based guilt, in line with our hypotheses. Participants who endorsed more strongly the positive representations of colonial history experienced lower feelings of group-based guilt (Figure 1). In contrast, the more the participants endorsed the negative representations, the more guilty they felt. Feelings of group-based guilt were, in turn, positively related to apology and the two forms of instrumental reparation. Bootstrapped confidence intervals for all six indirect paths ( Table 4) did not include zero, meaning that there is proof of indirect effects (Preacher & Hayes, 2008).
To assess whether guilt was differently associated with apology compared to the two types of instrumental reparations, we conducted Wald tests in which we constrained, one by one, each respective pair of coefficients to be the same. Group-based guilt turned out to be significantly more associated with support for an apology compared to reparations towards colonial-origin minorities in the Netherlands, χ 2 (1) = 60.269, p < 0.001, and in the former colonies. χ 2 (1) = 27.6, p < 0.001, confirming our hypothesis. 4

Moderation by In-Group Identification
In the mediation model discussed above, in-group identification was included as a control variable and it was negatively related to intergroup apology and reparations to colonial-origin minorities in the Netherlands, and also negatively but not significantly to reparations in the former colonies ( Table 5). This means that higher identifiers were less willing to offer both symbolic and instrumental reparations. However, identification and guilt were unrelated.
In the next step, we estimated three latent interaction models to find out whether the mediation process was conditional upon levels of in-group identification. First, we interacted guilt with identification in relation to the three forms of reparation and we found no significant interaction effects (ps > 0.313). This refutes our hypothesis about higher identifiers being more willing than lower identifiers to support reparations to the extent that they feel guilty.
Next, we estimated interactions between positive representations and in-group identification and found them not to be significant in relation to guilt or any of the reparation measures (ps > 0.268). The interaction between negative representations and in-group identification on guilt was also not significant (B = 0.001, s.e. = 0.070, p = 0.987). Altogether, and in contrast to our hypothesis, the relationships between the two representations and guilt turned out to be the same for high and low identifiers. However, the direct paths from negative representations to apology (B = -0.108, s.e. = 0.052, p = 0.036), reparations in the Netherlands (B = -0.134, s.e. = 0.048, p = 0.005), and reparations in former colonies (B = -0.131, s.e. = 0.054, p = 0.015) were all negatively and significantly moderated by in-group identification. For higher identifiers, the association between negative representations of the colonial past and willingness to repair the damage was always less positive than for lower identifiers. Note: Unstandardized coefficients presented. CI = confidence interval. Bootstrapping with 5000 replacement samples applied. * p < 0.05, ** p < 0.01, *** p < 0.001.

Discussion
The few studies on political apologies that have adopted the perpetrator perspective have primarily focused on moral emotions and shown that group-based guilt is an important driver of the willingness to issue an apology (Allpress et al., 2010;McGarty et al., 2005;. Our study has contributed to this literature by considering the representations of the group's troublesome past as a more distal and group-based guilt as a more proximal explanation of the willingness to issue a political apology in the context of colonial transgressions. Furthermore, as both representations of the past and feelings of group-based guilt matter for the willingness to offer instrumental reparations (Cabecinhas, & Feijó, 2010;Doosje et al., 1998;Licata & Klein, 2010;Mari et al., 2010;Sibley, 2010;Sibley et al., 2008;Zebel et al., 2008), we examined support for political apology alongside two more specific forms of reparation (assistance offered to former colonies and colonial-origin immigrants) and thereby we compared the underlying processes for both symbolic and instrumental forms of reparations. We focused on the Netherlands -a country that used to be a colonial power -and we zoomed in on the mediating role of group-based guilt and the moderating role of in-group identification.
Our study is the first to show that representations of in-group's wrongdoings relate to the willingness to apologize among the descendants of perpetrators: whereas negative representations were linked to more support for political apology, positive representations were related to less support. We found the same for instrumental reparations, which resonates with previous research showing that negative representations of in-group's colonial past translate into more willingness to offer compensation (Doosje et al., 1998;Lastrego & Licata, 2010;Mari et al., 2010), whereas compensation is less offered by those who endorse ideologies that legitimize past colonial crimes (Sibley, 2010;Sibley & Liu, 2012;Sibley et al., 2008).
Further, we have shown that group-based guilt is an important mechanism through which representations of the past explain support for political apology. In line with previous evidence, feelings of group-based guilt were positively associated with the wish to apologize (Allpress et al., 2010;McGarty et al., 2005;Wohl et al., 2013) and engage in reparations both in the former colonies and toward colonial-origin minorities in the Netherlands (Allpress et al., 2010;Doosje et al., 1998;Figueiredo et al., 2010;Licata & Klein, 2010). Our study adds to this research by showing that positive representations of colonial history are related to less guilt, and indirectly to less support for apology, whereas the opposite holds for representations that focus on exploitation and violence. In addition, we found support for a similar mechanism regarding instrumental compensation, in line with the study by Doosje and colleagues (1998). Importantly, we showed that the mechanism operates in a similar way in relation to compensating former colonies and colonial-origin minorities in the Netherlands.
Importantly, as hypothesized, group-based guilt was more strongly related to support for political apology than to support for instrumental forms of reparations, with the effect on the former being twice as large as on the latter. Moreover, our model explained 53% of the variance in support for an apology and only 32% and 31% of the variance, respectively, in support for reparations towards colonial-origin minorities in the Netherlands and reparations in the former colonies. The reason for this difference could be that an apology is by default accompanied by acknowledgement of in-group responsibility for the past transgressions: one can only apologize if one feels guilty, and issuing an apology is an easy way to restore the group's moral self-image (Barlow et al., 2015). In contrast, instrumental reparations imply higher costs (Wohl et al., 2006) and could moreover be offered by people who do not necessarily feel guilty about the past but who empathize with the disadvantaged groups and are willing to support initiatives that address the present unfavorable position of immigrant minorities in the country and the difficulties in the developing countries.
We further examined the moderating role of in-group identification. In contrast to our expectations, we found that group-based guilt was an equally important mediator in the link between representations of colonial past and support for reparations, irrespective of participants' level of Dutch identification. This resonates with experimental results from Doosje and colleagues (1998) who found that, when presented with a clearly negative or clearly positive representation of the colonial past, people report more guilt in the former case and less in the latter, regardless of how strongly they identify with their in-group. At the same time, our findings from bivariate correlations show that higher in-group identifiers do agree more with the positive and less with the negative representations, compared to lower identifiers, which is in line with the proposition that higher identifiers tend to pick and choose the narratives that present their in-group in a more favorable light (Doosje et al., 1998).
An explanation for the absence of moderated mediation could be sought in the type of measure used to capture identification. We relied on general questions about in-group attachment, but identification is a multi-dimensional concept (Ashmore, Deaux, & McLaughlin-Volpe, 2004). The fact that in-group attachment was very high and did not vary much in our sample does not preclude the possibility that our participants had different content of their national identity in mind. For instance, ethnic and civic conceptions of nationhood might be worth considering in future research. Majority members who endorse an ethnic conception have a more exclusive definition of who belongs to the nation, namely only those with ethnic majority background, whereas a civic conception of national belonging is more inclusive, and the two have been found to have opposite effects on support for minority rights (e.g., Verkuyten & Martinovic, 2015). Furthermore, research on blind and constructive patriotism (Schatz, Staub, & Lavine, 1999) shows that these two ways of thinking about one's in-group, with the former implying an uncritical and the latter a questioning stance towards values and behaviors of one's in-group, can have different consequences for inter-group relations. This resonates with the distinction that Roccas and colleagues (2006) made between in-group glorification and in-group attachment, and the finding that the former is more predictive of lower feelings of guilt. Future research could examine whether in-group glorifiers, blind patriots, and those with an ethnic conception of nationhood are particularly likely to reject the negative and embrace the positive version of colonial history. These individuals might also feel less guilty in response to critical narratives of the colonial past and therefore be less willing to offer reparations.
However, once guilt was accounted for, lower identifiers were more willing to issue an apology and compensate in instrumental ways, the more they agreed with the negative representations of the colonial times. This resonates with the argument that high national identifiers tend to detach present-day inter-group relations and existing power dynamics between majority and minority groups from the negative in-group past (see Sibley, 2010;Sibley & Liu, 2012;Sibley et al., 2008). In contrast, low identifiers probably feel less threatened by the shadow that the colonial past casts on the in-group's moral image and are thus more prone than high identifiers to acknowledge and act upon the in-group's responsibility for the crimes committed in the distant past for reasons other than feeling guilty. Perhaps other group-based moral emotions, such as image and essence shame (Allpress et al., 2010) are elicited more by the memory of the negative past among low in-group identifiers and this could be the reason why they are still more willing to offer compensation after guilt has been taken into account. Feelings of shame are distinct from feelings of guilt in that guilt focuses on wrongful behavior in a concrete situation, while shame is about more stable attribution to the self or the group. While image shame is positively related to support for an apology, essence shame tends to be related to support for concrete financial and material reparations (Allpress et al., 2010). Future research could take image and essence shame into account as additional mechanisms through which negative representations of the past motivate reparations in the present.
A strength of our research is that we used weighted survey data from a large community sample of the ethnic Dutch population. This allowed us to investigate actual opinions in relation to the colonial past and draw conclusions about the pattern of associations in the real world. We found little evidence for the denial of the negative past: both positive and negative representations of the colonial past were acknowledged by the Dutch, and there was even significantly more agreement with the latter. Our results are in line with previous research based on qualitative data and/or convenience samples from Portugal and Belgium showing that positive and negative representations of the past may coexist in a society (Cabecinhas, & Feijó, 2010;Licata & Klein, 2010), even though their endorsement may differ across generations, with younger people holding a more critical view of the in-group past (Licata & Klein, 2010).
The coexistence of positive and negative representations might also be the reason why feelings of group-based guilt remained relatively low in our sample. That a substantial agreement with negative representations did not translate into much support for either an apology or other types of compensation might be because positive representations, which attenuate guilt, were also clearly endorsed. Low levels of guilt, shame, and responsibility for colonial crimes have typically been detected in various national contexts and with different samples (Allpress et al., 2010;Doosje et al., 1998;Klein et al., 2011; see also Leach, Zeineddine, & Čehajić-Clancy, 2013 for a review on self-criticism for mass violence), and our study is no exception.
Whereas support for the three forms of reparation was also generally low in our study, Dutch participants were more willing to offer instrumental compensation than to apologize. This finding is particularly interesting in the light of the finding that the association between guilt and political apology was twice as large as that between guilt and instrumental reparations. The Dutch are thus less willing to apologize than to compensate in other ways, but guilt motivates apology more than that it motivates instrumental reparations (Leach et al., 2006). One explanation for the difference in the willingness to offer symbolic and instrumental reparations may be that contemporary Dutch are not willing to accept the responsibility for the crimes committed by their ancestors, whereas they may desire to reduce existing inequalities between their national in-group and either colonial-origin minorities or former colonies who are usually less privileged than the formerly colonizing nations. That is, many people may view colonial injustices as a remote and distant past, and therefore may feel no urgency to apologize for the colonial past. Instead, they may perceive instrumental reparations as a more specific and useful form of supporting lower-status groups in the present.

Limitations
Several limitations have to be considered with regard to the present study. First, even though the experimental studies by Doosje and colleagues (1998) showed that representations of colonial history can predict feelings of group-based guilt, and these feelings are predictive of reparations, we cannot make claims about causality in our study. Future experimental research could examine the causal effects of representations of the past on the willingness to issue a political apology.
Second, we did not differentiate between the ways in which colonial injustices in the Caribbean and Indonesia are remembered and negotiated. Yet, these contexts likely differ in the recognition and perceived severity of the colonial injustices. The colonial history of the East (Indonesia) is more acknowledged compared to its Western counterpart (Suriname and the Dutch Antilleans) (Wekker, 2016). While the Netherlands officially apologized for the colonial atrocities in Indonesia in 2011, a formal apology to Suriname and the Caribbean is still pending. Future research could focus on comparing these colonial contexts and investigating whether the representations of the colonial past, the extent to which Dutch people feel guilty, and the extent to which they are willing to offer reparations differ with respect to different former colonies.

Conclusion
We have shown on a large and diverse sample of ethnic Dutch participants that positive and negative representations of the colonial past are, indirectly via guilt, related to the willingness to issue a political apology. Whereas the process is similar for instrumental reparations, guilt tends to be a stronger explanation of support for an apology. We did not find evidence that this pattern of findings differs among high and low in-group identifiers. Future research could consider other emotions, such as shame and empathy, as well as different dimensions of in-group identification, in an attempt to further explain the willingness among the descendants of colonizers to apologize and offer instrumental help.

Notes
1 This module also included an additional set of questions about the perceived characteristics of the colonizers (moral, aggressive, fair, unjust), but we did not consider these because they were not part of our research question. 2 We did not receive a variable with start and end time or any other measure of duration, so we could not exclude speeders. 3 For each combination of demographic characteristics, the number of participants was obtained from the data set. The same was extracted from the Dutch population. Two weights were then calculated: one for the combination of province, age, and gender, and one for education. The following formula was used: (total number of participants/number of participants in the category) * (number of persons in this category on the population level/total population size). These two weights were then multiplied and thus combined into an overall weighing factor. We used this overall weight in our analyses. 4 Group-based guilt also predicted reparations in the former colonies more strongly than reparations towards colonial origin-minorities, χ 2 (1) = 6.247, p = 0.012. Positive and negative representations of the colonial past were equally strongly related to group-based guilt, χ 2 (1) = 0.445, p = 0.505.