Special Thematic Section " Collective Memories and Present-Day Intergroup Relations : A Social-Psychological Perspective " It ’ s About Valence : Historical Continuity or Historical Discontinuity as a Threat to Social Identity

National identity is underpinned by historical representations. Recent research shows that narratives presenting an in-group’s history as discontinuous rather than continuous raise collective angst, suggesting that historical discontinuity threatens social identity. This previous research has focused on positive aspects of an in-group’s past. The present research aims to extend the findings to go beyond positive histories. We suggest that when the in-group’s actions in the past are presented as negative, historical continuity instead of discontinuity will increase perceived identity threat because a negative, continuous history threatens group members’ need for a positive social identity in the present. In an experiment with a sample size of N = 316, we manipulated the narrated valence of in-group actions during the historical event of the approval of the German constitutional law by framing the group’s actions in either positive or negative terms. In addition, we presented the in-group’s history as connected or disconnected to the in-group’s present. Results demonstrate that historical continuity only decreased identity threat compared to historical discontinuity when the in-group’s past behavior was presented as positive. When the in-group’s past was presented as negative, continuity even increased identity threat compared to historical discontinuity. These results were particularly pronounced for people who strongly identified with their national in-group. We discuss implications of the findings for political communication and managing a nation’s perception of social identity threat.


Non-Technical Summary
Background Media frequently presents narratives of national history.These narratives affect how people feel and think about their national identity.Research suggests that narratives that present people's national history as broken from its past threaten people's identity whereas narratives about a continuous national history benefit people.

Why was this study done?
We suggested that historical narratives affect peoples' perceptions of identity threat differently depending on whether the narratives focus on a positive historical episode or on a negative one.More precisely, we suggested that for negative episodes of people's national past -episodes during which the national in-group actively violated present-day established norms and values -narratives of historical continuity amplify perceptions of threat because they imply that the in-group continues to misbehave in the present.By contrast, we expected historical discontinuity to relieve people from the burden of a negative past by explicitly separating the past from the present.serves as a psychological resource, whereas discontinuity between the in-group's past and present constitutes a social identity threat.
While targeting an important topic, this previous research on historical continuity of the national in-group mostly focused on positive aspects of the in-group history, neglecting historical episodes where the in-group behaved badly (see Peetz, Gunn, & Wilson, 2010).Negative in-group actions can undermine collective self-esteem and thus, threaten a positive social identity (Tajfel, 1982).In the case that historical narratives focus on negative aspects of the in-group's past behavior, continuity of the in-group's history might not be beneficial.Instead, negative aspects of the in-group's past might threaten in-group members' identity.The present research aims to extend knowledge about the effects of narratives of the in-group's past on group members' perception of identity threat.We suggest that the valence of the historical episode determines whether a continuous or a discontinuous representation of the in-group's history threatens their identity more or less (for a review of social identity threat, see Branscombe, Ellemers, Spears, & Doosje, 1999).

Representations of In-Group History
Members of social groups usually share a system of knowledge including norms, values, and beliefs (Liu & Sibley, 2009).These representations of the in-group include information about the in-group's past and present.We refer to this shared system of knowledge about the in-group's past as social representations of history (Liu & Hilton, 2005).According to Liu and Sibley (2009), social groups, especially "peoples", seek to establish traditions and institutions that maintain temporal continuity between the past, present, and future.Historical continuity has been theorized as a prerequisite for the social identity of peoples because it enables group members to understand, through the durability of the group's heritage, a legacy supporting how their group is distinct from other groups (Jetten & Hutchinson, 2011;Jetten & Wohl, 2012;Sani et al., 2007).In other words, the in-group's distinct norms, values, and beliefs can be explained by the continuity of the in-group's historical phases and events.Thus, perceived continuity of the in-group's past can provide the grounding for the in-group's present.
A collective history underscores the group's unique identity that endures through the vicissitudes of time (Liu & Hilton, 2005).However, the impact of particular narratives and their contents often remain controversial.Some of these narratives present the in-group's history as continuous, and their current norms, values, and behaviors as directly linked to its past (Sani et al., 2007).In contrast, other narratives stress discontinuity and dissimilarities of the in-group in the present as opposed to the in-group in the past.For example, Japan and Germany, as villains and losers of the "greatest war in human history" (see Liu et al., 2005), have a burden of history placed upon them that has caused Japan to alternately apologize for or deny its war crimes (see Liu & Atsumi, 2008); and for young people in Germany to alternately feel collective shame for the past (Dresler-Hawke & Liu, 2006) or seek to "draw a line in the past" to avoid such a negative identity position (Hanke et al., 2013).By contrast, in the victorious English speaking nations of the United Kingdom (Jetten & Wohl, 2012) or New Zealand (Liu & Robinson, 2016), it has been shown that historical continuity can be mobilized as a resource to argue for policy decisions affecting the future.

Benefits of Historical Continuity
With respect to the personal self, continuity between the past self and the present self has structuring and stabilizing functions and is associated with positive consequences (e.g., Sadeh & Karniol, 2012).Self-discontinuity, in contrast, has been associated with anxiety and other negative emotional and behavioral outcomes (e.g., Milligan, 2003).
Consistent with theory and research about continuity of the personal self, research suggests that perceived continuity at the collective level has beneficial effects for group members, too (Chandler, Lalonde, Sokol, & Hallett, 2003;Fivush, Bohanek, & Duke, 2008;Sani et al., 2007).For instance, research on mergers has shown that a continuation of the initial in-group has positive consequences for dealing with challenges brought by mergers (Jetten & Hutchinson, 2011;van Knippenberg, van Knippenberg, Monden, & de Lima, 2002).These results indicate that the perception of a continuous in-group -even through times of change -can protect social identity and increase the capability to cope with challenges.
Perceived historical discontinuity refers to the perception that the group's history is interrupted and that the ingroup's norms, values, and behaviors in the present are barely connected to the in-group in the past (Sani, Bowe, & Herrera, 2008).Jetten and Wohl (2012) considered historical discontinuity as a threat to social identity, because it arouses angst that the group's traditions, norms, and values might be lost (see also Smeekes & Verkuyten, 2014b).Empirical evidence supports this claim.Studies have demonstrated that perceived historical continuity has beneficial consequences and that perceived historical discontinuity has detrimental consequences for social identities.For instance, Sani and colleagues (2007) have shown that perceived continuity of the in-group's history is associated with variables that are advantageous for social identity such as collective self-esteem, positive feelings towards the in-group, and the perception that the in-group fulfills basic needs (see also Chandler & Lalonde, 1998).Jetten and Wohl (2012) investigated causal effects of perceived historical continuity versus discontinuity on group members' collective angst.They presented English participants with narratives about their nation's history as being either connected to the present or disconnected.Results showed that people who read about England's past as being disconnected from the present experienced more collective angst than those who read about the in-group's past as being connected to the present.
The reviewed research underscores the notion that perceived historical continuity is beneficial whereas perceived historical discontinuity has harmful effects on social identity.However, the aforementioned studies investigated the impact of perceived historical continuity with respect to positive or neutral aspects of in-group history.For example, Jetten and Wohl (2012) presented a text passage focusing on positive aspects of England's past (a gentle, fabled land) being either connected or disconnected to England's present that was also described in positive terms (vibrant cities, great nightlife, pleasant countryside).We are not aware of any research that has addressed valence of the in-group's historical episode as a boundary condition for beneficial effects of historical continuity on group members' identity.

Historical Continuity as a Value Threat to Social Identity
In general, people strive for a positive view of themselves (Tajfel & Turner, 1979;Wilson & Ross, 2003).Negative episodes in a person's past can threaten self-esteem (Roth, Steffens, Morina, & Stangier, 2012).As personal self and social self are different but related aspects of the self-concept (McConnell, 2011;Tajfel, 1974), negative group-images (similarly as negative self-images) constitute a threat to positive identity (Tajfel & Turner, 1979).
Typically, research has focused on social identity threat by taking an intergroup perspective (for overviews see Branscombe et al., 1999;Riek, Mania, & Gaertner, 2006).From this perspective, social identity threat results from negative intergroup comparison (Riek et al., 2006).Some authors have pointed out that social identity threat can stem from the in-group itself (Branscombe et al., 1999;Doosje, Branscombe, Spears, & Manstead, 1998).Narratives about the in-group's past -revealing unfavorable values and in-group's actions in the past -may threaten the ingroup's integrity in the present.Thus, a historical episode may become a value threat when group members in the past acted in ways that violated in-group values and norms of the present.As a consequence, when group members become aware of this negative part of their group's history, they feel threatened, guilty, or ashamed (e.g., Branscombe et al., 1999;Dresler-Hawke & Liu, 2006;Gausel, Leach, Vignoles, & Brown, 2012;Imhoff, Wohl, & Erb, 2013;van Leeuwen, van Dijk, & Kaynak, 2013).
We suggest that a negative episode of the in-group's past is particularly threatening when connected to the ingroup's present values and norms in a continuous manner.Thus, in the case of a negative in-group past, in-group historical continuity threatens social identity compared to historical discontinuity because past misdeeds continue casting a damning light on the in-group in the present.This is the opposite effect to when the in-group's past is positive.Concisely, we suggest that for negative episodes of the in-group's past -episodes during which an ingroup actively violated present-day established norms and values -perceptions of historical continuity amplify identity threat because this implies that the in-group continues to misbehave in the present.By contrast, historical discontinuity can relieve in-group members from the burden of a negative past by explicitly separating the past from the present.

The Moderating Role of Social Identification
People differ in how strongly group-based threats affect them (Branscombe et al., 1999).In fact, previous research has shown that people who strongly identify with their national in-group show more collective angst when presented with a positive history that was discontinuous (i.e., a threat to the group) compared with a condition where history was presented as continuous (Jetten & Wohl, 2012).Thus, social representations of history should have stronger effects on group members' reactions the more people identify with the in-group.
Research suggests that social identification is complex, consisting of five components that hierarchically relate to two dimensions: a group based self-investment dimension consisting of satisfaction, solidarity, and centrality; and a group based self-definition dimension with in-group homogeneity and individual self-stereotyping (Leach et al., 2008).Centrality, one of the identification components of the self-investment dimension, represents the extent to which group members perceive the group as a central aspect of their self-concept.Leach et al. (2008) demonstrated that centrality -out of the five identification components -best predicted perceived threat when people were exposed to an in-group threat.Therefore, we expect that the identification component of centrality will moderate the effects of historical narratives on group members' perceptions of threat.
Postmes, Haslam, and Jans (2013a) however, have suggested that the self-investment dimension constitutes the core of social identification and that differentiating between its individual components is not necessary in most cases.Given that the multicomponent structure of social identification has received considerable evidence (e.g., La Barbera & Capone, 2016;Lovakov, Agadullina, & Osin, 2015;Roth & Mazziotta, 2015) and that we derived specific predictions about which component is relevant for threat related reactions (Postmes, Haslam, & Jans, 2013b), we implemented the multicomponent questionnaire and analyzed the five components separately as moderators of identity threat.

The Present Experiment
The purpose of the present experiment is to examine the effect of perceived historical (dis)continuity of the national in-group on group members' perceptions of identity threat as a function of the valence of the historical episode.
We go beyond previous research to suggest that beneficial effects of historical continuity compared to discontinuity can be observed only when positive historical episodes are highlighted.Instead, when narratives highlight the ingroup's historical misdeeds, we expect perceived historical continuity to increase identity threat.Historical discontinuity in the face of such negative historical episodes will be perceived as less threatening because it implies an improved contemporary trajectory for the in-group, having broken from its negative past.Thus, we hypothesize that the valence of a past historical episode moderates whether historical continuity compared to historical discontinuity results in more or less perceived identity threat.Furthermore, we hypothesize that identification with the national in-group moderates these effects.Specifically, we expect that the hypothesized effects of historical (dis)continuity by past valence of the narratives will be exacerbated when the national in-group is more central to a group member's self-concept.

Method Design and Study Materials
The experimental design was a 2 (historical continuity: continuous vs. discontinuous) × 2 (past valence: positive vs. negative) between subject design.Since the experiment was conducted in Germany, we developed the description of the historical context based on a historical event in Germany.In order to achieve an unconfounded manipulation of the valence of the past, we aimed to hold the historical event constant across experimental conditions and presented the in-group's actions as either positive or negative in valence.We conducted a pilot test to select a historical event that was perceived to be very important, but neutral in valence.Selecting an important but neutral historical event would allow us to manipulate the valence (positive versus negative) of the same historical episode by merely changing its framing.
In selecting the historical event, we had sixteen participants rate the importance (1 = not important, 2 = important, 3 = very important) and valence (-2 = negative, -1 = mostly negative, 0 = neutral, +1 = mostly positive, +2 = positive) of 13 historical events (e.g., Great Depression, Holocaust, German Reunification, winning the soccer world cup championship, approval of the German constitutional law).The four historical events that were rated as very important were: Holocaust (M = 2.94, SD = .25),German Reunification (M = 2.75, SD = .45),foundation of the two German states after World War II (M = 2.69, SD = .48),and approval of the German constitutional law (M = 2.69, SD = .48).All other historical events were rated as important or not important (average ratings ≤ 2.50).Out of these four very important historical events, two events were rated positively (approval of the German constitutional law, M = 1.00,SD = .73,and German Reunification, M = 1.47,SD = .64),and two events were rated negatively (foundation of the two German states after World War II, M = -1.13,SD = .62,and Holocaust, M = -1.94,SD = .25).
Overall, based on descriptive statistics, approval of the German constitutional law was perceived to be the most neutral among the historical events while it was also perceived to be very important.The approval of the German constitutional law was a crucial historical event in May 1949.It was at the same time the foundation of the Federal German Republic after World War II.Since its approval, the German constitutional law incorporates the base of a direct democracy in Germany.It regulates the rights and duties of institutions including the government.One core aspect are the basic rights that are primarily a defense of the citizens against the state.In addition, the basic rights embody an objective order of values.This historical milestone is part of the German school curriculum and even though people may not remember the exact date, we believe that Germans in general are familiar with the approval of the German constitutional law in response to the Hitler regime as the beginning of a new, ongoing era.Based on the high importance ratings and the most neutral valence rating among the historical events presented to our pre-study participants, we chose this event as the historical context for the main experiment.

Participants
We conducted the experiment online.The survey was open for participation in February and March 2014.People were invited to participate through social networks.They participated without any incentive.Before they started with the survey, we informed them that the study is about their perception and opinion about recent German history.We also informed them that their data would remain anonymous.At the end of the study, we thanked participants and debriefed them carefully.A total of 321 participants completed the entire survey.
i Data from five participants were excluded from the analyses because they were not members of the national group.The final sample consisted of 316 Germans (49% female) with an average age of 43.91 years (SD = 17.32).The majority was well educated (58% had a university degree).Participants were politically diverse; on a scale ranging from 1 = left to 7 = right, the average was 3.37 (SD = 1.22).More information about the sample can be found in Appendix B, Table B.1.

Measures and Procedure
First, participants read an introduction where they learned that the present survey is about their perception and opinion regarding modern German history.

Manipulations of In-Group's Past
Subsequently, we assigned participants randomly to one of the four experimental conditions.In all four conditions, participants read an alleged newspaper article.To increase the credibility of the article, participants read that the article recently appeared in an established German newspaper.Furthermore, we formatted print, font, and column width of the article to fit common German newspaper articles.
The newspaper article consisted of two parts.In each condition, participants read one of the combinations of both parts resulting in the four experimental conditions (positive past-continuous, negative past-continuous, positive past-discontinuous, and negative past-discontinuous).

Positive Versus Negative Past
The first part described Germans' past political involvement -thus, representing norms, values, and behavior of the in-group in the past -directly after the approval of the German constitutional law as positive versus negative.
The negative valence manipulation is presented in brackets […] ii .
It's been 65 years since the deputies of the Parliamentary Council developed the German Basic Law.(…) In the recently adopted constitutional law, it was scheduled that Germans elect their political representatives [only] every four years.Possibly, this is why amongst Germans the interest in political affairs kept growing [diminishing] in the following decades and new generations showed more and more [less and less] enthusiasm in personal political involvement.(…) Thus, the entire German population contributed to the development [failure] of an actively practiced democracy.(…)

Historical Continuity Versus Historical Discontinuity
The second part described historical continuity or discontinuity in Germans' political involvement until the present.
The discontinuity version is presented in brackets […].We kept the text versions of the four experimental conditions similar regarding length, wordings, and perspectives as well as other critical narrative features (e.g., perspective, agency).

Perceived Identity Threat
We adapted two short questionnaires that have been successfully implemented in previous research measuring perceived identity threat (nine items, e.g., Given the article I just read I feel threatened in my identity, The German history described in the article contributes to the preservation of my values and customs [reverse coded], Cronbach's α = .75;de Hoog, 2013; Falomir-Pichastor, Gabarrot, & Mugny, 2009; for all items see Appendix A).We assessed each of the answers on a scale ranging from 1 = totally disagree to 7 = totally agree and randomized item order.
Positively worded items were reverse coded so that higher scores indicate more perceived identity threat.

Manipulation Check: Historical Continuity Versus Historical Discontinuity
After the dependent variable, four items that we presented in random order measured perceived historical continuity (adapted from Sani et al., 2007): The German history described in the article is a sequence of interconnected events, There is no continuity between the different episodes of the German history described in the article (reverse coded), The German history described in the article is a sequence of not interconnected events (reverse coded), and The German history described in the article takes a continuous course (1 = totally disagree to 7 = totally agree).Higher scores indicate perceptions of more historical continuity (Cronbach's α = .76).

Manipulation Check: Positive Versus Negative Past
Then, two items that we presented in random order assessed perceived valence of the past: In the article, the Germans are presented as rather negative in the past (reverse coded) and The article portrays the Germans in the past as rather positive (1 = totally disagree to 7 = totally agree).Both items correlated substantially, r = .73,and we averaged them.Higher scores denote more perceived positive valence of the past.
iii Finally, participants completed demographic questions including their national identity, their gender, level of education, and political orientation.

Manipulation Checks Historical Continuity Versus Historical Discontinuity
A 2 (historical continuity: continuous, discontinuous) × 2 (past valence: positive, negative) ANOVA with perceived historical continuity as the dependent variable showed the expected main effect of historical continuity.The continuous representation resulted in higher perceived historical continuity (M = 4.80, SD = 1.18) than the discontinuous representation (M = 4.38, SD = 1.24),F(1,312) = 9.29, p = .003,η 2 p = .029.Neither the main effect of past valence nor the interaction was significant, Fs < 1.

Positive Versus Negative Past
A 2 (historical continuity: continuous, discontinuous) × 2 (past valence: positive, negative) ANOVA with valence of the past as the dependent variable showed the expected main effect of past valence.The positive past representation resulted in perceiving the past as more positive (M = 5.45, SD = 1.28) than the negative past representation (M = 3.20, SD = 1.23),F(1,312) = 246.33,p < .001,η 2 p = .441.Neither the main effect of historical continuity nor the interaction was significant, Fs < 2.

Moderation of Historical Continuity by Past Valence
Our first hypothesis stated that the effect of historical continuity on perceived identity threat depends on the valence of the historical episode.Consequently, we hypothesized a two-way interaction between historical continuity and past valence on perceived identity threat.

Moderation by Social Identification
Our second hypothesis stated that the interaction between historical continuity and past valence on perceived identity threat (see Figure 1) would be more pronounced the more people identify with their group.Specifically, we expected that people to whom the German identity is more central to their self-concept would show a stronger reaction to historical continuity depending on the valence of the historical episode than people to whom the German identity is less central.We conducted a 2 (historical continuity: continuous, discontinuous) × 2 (past valence: positive, negative) ANCOVA with the continuous measure of centrality (the expected moderator, mean centered) as the covariate and perceived identity threat as the dependent variable.In this analysis, the expected moderation of centrality would show up in a three-way interaction of centrality, historical continuity, and past valence on perceived identify threat.This is exactly what the results showed; F(1,308) = 4.96, p = .027,η 2 p = .016(see Figure 2); identity centrality moderated the effect of historical continuity and past valence on identity threat.None of the main effects were significant: historical continuity, F(1,308) = 1.43, p = .232,η 2 p = .005;past valence, F(1,308) = 3.63, p = .058,η 2 p = .012;centrality, F(1,308) = 3.43, p = .065,η 2 p = .011.The only significant two-way-interaction remained the one between historical continuity and past valence, F(1,308) = 35.22,p < .001,η 2 p = .103;[historical continuity × centrality, F < 1; past valence × centrality, F(1,308) = 1.407, p = .237,η 2 p = .012].
To decompose the pattern of the three-way interaction, we report findings separately for high and low centrality based on a median split.For participants who showed high identity centrality, results replicated the two-way interaction between past valence and historical continuity, F(1,142) = 28.96,p < .001,η 2 p = .169(see Figure 2 and compare with Figure 1).Following the data analysis above, simple comparison showed that when the past was  For participants who showed low identity centrality, the two-way interaction between past valence and historical continuity was also significant, but the effect size was much smaller, F(1,166) = 8.78, p = .003,η 2 p = .050(see Figure 2).When the past was represented as positive, continuity decreased identity threat (M = 2.98, SD = 0.92) compared to discontinuity (M = 3.39, SD = 0.81), F(1,166) = 5.41, p = .021,η 2 p = .032.When the past was represented as negative there was only a tendency that historical continuity increased identity threat (M = 3.37, SD = 0.67) compared to historical discontinuity (M = 3.06, SD = 0.71), F(1,166) = 3.43, p = .066,η 2 p = .020.
In sum, as predicted, for participants to whom the German identity was more central to their self-concept, the effects of historical continuity for the positive and negative past on perceived identity threat were significantly stronger than for participants to whom the German identity was less central.
Notably, we additionally conducted the same analysis separately with the other four identification components as moderators (see Leach et al., 2008).None of the two components of the group-level self-definition dimension moderated the effect of historical continuity and past valence on identity threat: individual self-stereotyping, F(1,308) = 1.14, p = .287,η 2 p = .004;in-group homogeneity, F(1,308) = 0.035, p = .851,η 2 p < .001.However, solidarity and satisfaction -the other two components of the group-level self-investment dimension -showed a similar moderating pattern as centrality with the strongest effect for satisfaction, F(1,308) = 6.55, p = .011,η 2 p = .021,and a trending effect for solidarity F(1,308) = 2.61, p = .108,η 2 p = .008(descriptive statistics and correlations for identification components can be found in Appendix B, Table B.2).

Discussion
We investigated two hypotheses: first, whether valence of the in-group's past is a boundary condition for the identity benefiting effect of an in-group's historical continuity; and second, whether social identification exacerbates this effect.We examined these hypotheses using a large and diverse sample of Germans -varying in age, political orientation, and education.This work conceptually replicates and extends previous research in important ways.Most importantly, we manipulated the valence of the same historical episode.This procedure allows unconfounded causal conclusions about the effects of historical (dis)continuity of a positively or negatively framed past episode.Results replicate previous findings that historical continuity decreases perceived identity threat compared to historical discontinuity (Jetten & Wohl, 2012).Notably, the present results showed that historical continuity only decreased perceived threat compared to historical discontinuity when the in-group's past behavior was represented positively.When the in-group's past was represented negatively, historical discontinuity rather than historical continuity decreased perceived identity threat.These results support our claim that historical discontinuity of the in-group only threatens group members' identity when the in-group's past is of positive valence.When the ingroup's past is of negative valence, historical continuity instead of discontinuity threatens identity.This pattern of results was stronger for highly identified group members than for those who identified less.
Whereas most of the previous research has investigated the effects of historical (dis)continuity on group members' social identity, in the present experiment we asked about how the in-group's history was perceived as personally threatening.Given that personal identity and group identity are interwoven (Smith & Henry, 1996), it is not surprising that group-based threats are also perceived as personally threatening.We assessed social identification at the beginning of the study and this could have increased the salience of national identity augmenting the relevance of group based threats for further processing of information.Consequently, in the present experiment, we believe that threats to the personal self were perceived through the lens of the national in-group.
Notably, in the present experiment we framed the historical episode in a way that the in-group was an active agent of their past.Thus, the in-group either lived in line with current norms and values or violated them.In an intergroup scenario, a negative in-group's past can either be a perpetrator history (the in-group perpetrated a crime with respect to an out-group) or a victim history (the in-group suffered from an out-group perpetrating crimes).Whereas a perpetrator history mirrors the in-group as an active agent of wrongdoings, a victim history is not caused by ingroup behavior.Instead, the in-group suffered wrongdoings from an out-group.Whereas histories of both, perpetrators and victims, are negative, reactions to the negative in-group's past of perpetrators and victims diverge (e.g., Sahdra & Ross, 2007).In the present experiment, we manipulated the valence of the in-group's past in a way that the in-group was the agent responsible for its negative past.This is more comparable to a perpetrator history than it is to a victim history.In fact, the present results are in line with previous intergroup research showing that members of a perpetrator group feel less collective guilt when their negative history was framed as having no impact on contemporary group members (Imhoff et al., 2013) or when these negative events are subjectively perceived to be farther away in the past than close to the present (Peetz et al., 2010).
We have brought forward a motivation-based explanation about why historical (dis)continuity threatens identity depending on the valence of the in-group's historical episode.Within this approach, we suggested that a continuing negative past leaves group members with a negative image of their in-group in the present, making continuity a double-edged sword (Sani, Bowe, & Herrera, 2008).On the one hand, a continuous in-group nourishes people's need to know where they come from and provides guidance and reason.On the other hand, membership in a group that violated norms and values in the past threatens collective self-esteem (Tajfel, 1974), driving a need for historical closure by the former perpetrators of negative deeds (Hanke et al., 2013).Only an in-group that distances itself from its past wrongdoings, and is thus changed for the better satisfies people's need for a positive identity.

The Moderating Effect of Social Identification
In line with research on reactions to in-group related threats, highly identified group members showed stronger reactions to historical (dis)continuity and past valence than less identified group members (Branscombe et al., 1999;Jetten & Wohl, 2012).As we predicted, effects of historical (dis)continuity more strongly depended on the valence of the past when group membership was central to the person's self-concept (see Leach et al., 2008).
These findings corroborate that the centrality of the in-group to the individual's self-concept amplifies how narratives of the in-group's past affect identity threat.However, results showed a similar pattern for the other two components of the group-level self-investment dimension (i.e., satisfaction and solidarity).None of the components of the selfdefinition dimension (i.e., individual self-stereotyping, in-group homogeneity) showed any effect.Thus, the present results support Leach and colleagues' (2008) claim that centrality is related to group-based threats.However, the satisfaction component showed the same effect with a descriptively stronger effect size.In line with Postmes and colleagues' (2013a) suggestion that the self-investment dimension represents social identification better than the self-definition dimension, only the components of the self-investment dimension showed the predicted effect of social identification.Finally, the finding that all three components of the self-investment dimension show a similar pattern may suggest that aspects of self-investment represent a homogenous construct when people are perceiving group-based threat.Future research could test this prediction.

Implications for Theory
It has been argued that collective continuity constitutes a basic social identification motive that plays a crucial role in group dynamics (Smeekes & Verkuyten, 2013;Vignoles, Regalia, Manzi, Golledge, & Scabini, 2006).In line with this argument, recent research has indicated that the need for collective continuity increases social identification and that violating collective continuity results in defensive reactions such as opposition to out-groups (Jetten & Wohl, 2012;Smeekes & Verkuyten, 2013, 2014a).The present results demonstrate that collective continuity compared to discontinuity can increase perceptions of identity threat when the in-group is perceived negatively in the past.These results suggest that collective continuity per se might not be the crucial motive, but that group members specifically strive for a positive social identity in the present (Tajfel, 1974;Vignoles, Chryssochoou, & Breakwell, 2002;Vignoles et al., 2006).Thus, when collective continuity does not satisfy the self-esteem motive because the in-group was an agent of historical wrongdoings, continuity increases identity threat.Historical closure -drawing a line between the past and present (see Hanke et al., 2013) rather than continuity becomes a psychological benefit in the case of former perpetrators.Thus, even though collective continuity is among the identity motives (Smeekes & Verkuyten, 2014a), it appears to interact with the identity motive of positive collective esteem.
As demonstrated in the present study, collective continuity benefits identity only when the in-group's past was perceived as positive but it threatens identity when it was negative.Further research could more directly investigate the interplay between the collective continuity motive and the collective self-esteem motive.

Further Implications
Narratives about the national in-group history are a powerful tool for politicians and institutions (see Liu & Hilton, 2005).The present research indicates that political agendas communicating positive or negative aspects of the nation's past as continuous or discontinuous can threaten or benefit peoples' identities.In an intergroup context, a focus on historical discontinuity may be useful for a group to deal with its shameful past.However, the (victimized) out-group can perceive it as a denial of its history of victimization.Thus, only where properly handled, can the emphasis on historical discontinuity have benefits not only in terms of reduced identity threat of members of a previous perpetrator group, but also in terms of intergroup relations.For example, Japan's attempts at historical closure over its war past have been met with resistance from China and South Korea, who do not accept that Japan has shown genuine contrition over past misdeeds, and therefore refuse to forgive or accept closure (see Hanke et al., 2013;Liu & Atsumi, 2008).One of the conditions for intergroup apologies to work is that the representatives of the perpetrator group dissociate the present political system from the system responsible for an historical transgression thereby, disconnecting the in-group from its past (Blatz, Schumann, & Ross, 2009;Licata, Klein, Saade, Azzi, & Branscombe, 2012).Liu and Sibley (2009) have pointed out that social representations of history become particularly important for politics when the in-group faces challenges.Immigration of foreign nationals -as it is happening these days in Europe -might constitute a challenge for the respective societies.Particularly in these situations, identity-entrepreneuring (Reicher & Hopkins, 2001) narratives from politicians about a nation's history may help or hurt in integrating and accepting immigrants.In these situations, narrating a positive history that is discontinuous instead of continuous, or a negative history that is continuing instead of discontinuing could increase perceived identity threat (see Jetten & Hutchinson, 2011).

Limitations
In the present study, we manipulated the valence of the past while holding the historical event constant across experimental conditions.An alternative design would have been to select a naturally occurring negative historical episode in the German history and compare it with another positive historical episode.Despite the advantage of high ecological validity of this alternative operationalization, selecting different historical episodes would have come along with several confounds (e.g., time passed since the episode, importance of the episodes, cultural and historical context of the episodes) that would have limited the strength of the causal conclusions.Therefore, we decided for the unconfounded option and presented the same historical episode as either positive or negative in valence.
Notably, the pretest indicated that the selected historical event -even though it was perceived to be the most neutral among the ones we had presented -was evaluated as slightly positive.The manipulation check clearly shows that we were able to present it either in a more positive/less negative or in a less positive/more negative way.However, if we had selected a more negative event, the effects between continuity and discontinuity in the negative condition might have come out even stronger.
One could criticize that we included a reference to "recent" in our manipulation of the past historical event.Even though this might have directed participants' attention to the present instead of the past, the results of the manipulation checks show that the manipulation lead to the intended perceptions of connection versus disconnection to the past.However, future research should carefully select words that refer consistently and clearly to the past.
We replicated and extended previous research on historical (dis)continuity by investigating the identity of Germans -which is an interesting case because Germans' larger representation of history is overshadowed by the extremely negative events of the Holocaust and World War II (Hanke et al., 2013).Future research could test whether the present findings hold true for different national groups and other social groups such as ethnic groups and organizations.We predict that the findings are general and that group members perceive identity threat when their positive in-group's past is discontinuous (rather than continuous) while continuity (rather than discontinuity) of a negative in-group's history will provoke identity threat -at least when the in-group was the agent of past wrongdoings.
It remains an avenue for future research to test whether and under what conditions an in-group with a victimized past profits from being connected or disconnected to its present.In an intergroup conflict, even though the victimized group suffered a negative event, the negative in-group past was caused by an agentic out-group and not by ingroup members.Thus, the present results can only be generalized to an in-group who was an active agent of a positive or negative event in the past.For victim groups, both positive and negative continuous historical events may serve to preserve social identity because the notion that these representations of the in-group's past are shared brings a sense of cohesion among group members.Perceiving victimhood as continuous may not increase perceptions of identity threat but may fuel negative responses towards the perpetrating out-group (e.g., Schori-Eyal, Klar, & Ben-Ami, 2017;Schori-Eyal, Klar, Roccas, & McNeill, 2017).Consider for example, the plight of the Palestinians: their recent suffering may strengthen a social identity that in turn fortifies their collective motivation to fight Israel and achieve an independent state (see Pratto et al., 2014).

Conclusion
Historical continuity instead of discontinuity of the in-group is beneficial for group member's identity when positive aspects of its past are salient, but reaction to negative histories may be more complex.A negative episode of an agentic in-group's past that is represented as continuous increased perceived identity threat compared to a perceived discontinuous past.Thus, as a powerful strategy to create identity threat, political leaders, institutions, and organizations can portray a national group's past in a specific manner (with positive or negative valence; Liu, Sibley, & Huang, 2014;Reicher & Hopkins, 2001).These narratives particularly strongly affect group members who value their in-group and to whom the group membership constitutes an important aspect of their self-concept.
Since perceived identity threat is related to politically relevant outcomes such as attitudes toward immigrants and discrimination, framing a national history in a specific manner can help or hurt nations to deal with contemporary pressing societal issues.

Notes
i) In total, 484 persons followed the link to the study, 436 of them started the questionnaire and 321 completed it.From those who started the questionnaire 22.3% dropped out before the manipulation started.The dropout after the manipulations was distributed equally across conditions (positive past-continuous: 4.5%; positive past-discontinuous: 4,3%; negative past-continuous: 4.0%; and negative past-discontinuous: 2.7%), χ 2 (3) = 3.04, p = .386.Finally, we examined whether experimental groups differed in gender, age, education, political orientation, voting behavior, whether they grew up in East or West Germany, or in national identification.Experimental groups did not differ significantly in any of these variables.Statistics can be found in Appendix B, Table B.1.
ii) The implemented original text passages in German are available upon request from the first author.
(…) According to a recently published study (…) today's Germans still [hardly] show great [any] similarities to their past [anymore].Concerning the generations currently living in Germany, no considerable changes [considerable changes] in the ways of thinking or behavior can be observed.The perspectives of Germans have basically remained the same [changed fundamentally].(…)

Figure 1 .
Figure 1.Perceived identity threat depending on whether the in-group's history is presented as continuous or discontinuous and the valence of the past episode.Note.Error bars represent standard errors.

Figure 2 .
Figure2.Perceived identity threat depending on whether the in-group's history is perceived as continuous or discontinuous and the valence of the past episode separated by high versus low identity centrality based on a median split (Mdn = 4.00 on a 7-point scale) for ease of presentation.Note.Error bars represent standard errors.
Note.ISS = individual self-stereotyping; IGH = in-group homogeneity.Table B.2 Descriptive Statistics and Correlations for Five Components of Social Identification self-stereotyping; IGH = in-group homogeneity.*p < .001.