How to blame and make a difference: perceived responsibility and policy consequences in two Swedish pro-migrant campaigns

In this paper, we explore the assumption that blame-attribution can be an effective rhetorical strategy for non-elite interest groups who want power holders to be attentive to their demands. Through a qualitative analysis of two pro-migrant campaigns led by grassroot activists in Sweden, one taking place in 2005 and the other in 2017, we offer a nuanced empirical examination of non-elite initiated blame-games. We show how perceived responsibility influences these blame-games, and explore which policy consequences might emanate from them. We demonstrate that blame-making, under certain conditions, can be a successful strategy to gain policy influence, but that this strategy is conditioned by the complexity and transparency of the institutional arrangements of accountability within the policy sector. The focus on non-elite blame-making in order to change policies enables us to contribute to the theoretical discussion on the relationship between anticipatory and reactive forms of blame-avoidance behaviours, and to discuss the democratic implications of blame-games in both shorter and longer time perspectives. One implication of this study is that successful non-elite blame-making at one point in time actually can lower the chances of successful blame-making in the future.


Introduction
Blame is an important rhetorical weapon in modern political life. There is a widely held assumption that elected politicians, office holders, and front-line bureaucrats will care more about avoiding personal blame than gaining credit for their decisions, because being blamed for causing harm can lead to severe reputational loss, which may imply a loss of legitimacy, position, benefits and future career options (Weaver 1986;Hood 2011;Leong and Howlett 2017). To publicly blame power holders for causing harm of some kind can therefore be an effective strategy for contenders of status quo. Yet, the growing literature on blame in political life has mainly focused on power holders' blame-avoidance behaviour (Brändström and Kuipers 2003;Hood 2011;Resodihardjo et al. 2016;Hansson 2018b;Hinterleitner andSager 2017, 2020), leaving the blamegenerating side of political blame-games-the strategies, considerations and objectives of the blame-makers-unexplored.
The aim of this paper is to investigate the rationale behind non-elite groups' blamemaking, how such strategies are influenced by perceptions of responsibility, and how various policy consequences can emanate from such attempts. The concept of non-elite here refers to groups with no clear capacity and position to influence policies and laws, in contrast to elite groups consisting of individuals with either formal power to influence policies and laws, or people around them with capacity to exert such influence in informal ways (Lasswell 1952;Mills 1956). Through a qualitative analysis of two pro-migrant campaigns initiated and led by grassroot activists in Sweden, one taking place in 2005 and the other in 2017, we offer a nuanced empirical examination of non-elite initiated blame-games and their consequences. The research questions guiding our analysis are: How do perceptions of responsibility determine the pressure non-elite blame-makers can put on power holders, and which implications do these perceptions have for our knowledge about the conditions for and consequences of blame-games?
The few studies that do explore blame-makers' rationales and goals have exclusively looked at elite players such as political opposition parties and elite interest groups (Weaver 2018;Hansson 2018a;Hinterleitner 2018;Resodihardjo 2020). This has led to an exaggerated scholarly focus on blame-making as a strategy to win electoral advantages over political opponents and a neglect of other, less competitive and career-oriented, rationales for using blame strategies, such as policy influence. Even if both Weaver (2018) and Hinterleitner (2018) discuss policy changes as blame-game consequences, they do so from the perspective of elite players' rationales for blame-making, thus leaving non-elite's rationales for blame-making unexplored.
In this paper, we argue that by acknowledging the rhetorical resources blame-games can offer to non-elite groups in society, a hitherto understudied democratic potential of blame-games can be seen. Often, interest groups lack political power and resources to give them leverage in policy-making processes, making them employ creative strategies to alert power holders to their political demands. One such strategy, recognized in social movement scholarship but largely neglected in the blame-game literature, is to attribute blame to power holders for harms being done to particular groups in society (Klandermans et al. 1999;Benford and Snow 2000;Amenta et al. 2010). By bringing such a perspective into the blame-game literature, we hereby attempt to bring a new angle to this literature, which can deepen the discussion about the democratic implications of blame-games.
The two Swedish campaigns are selected based on a most similar system design, as both cases centred on political demands to offer denied asylum seekers regularization programmes, and both campaigns publicly attributed blame to power holders for the negative conditions these groups suffered. However, the cases differ in terms of how responsibility within the asylum policy sector was perceived. Previous research (Hinterleitner and Sager 2017;Figenschou and Thorbjørnsrud 2018) demonstrates that perceptions of responsibility may differ depending on how the institutional arrangements of accountability within a policy sector are structured, and this, in turn, may determine the opportunities for power holders to avoid blame. With clear and hierarchical arrangements of accountability within a policy sector, the opportunities for power holders to deflect blame, with preserved credibility in the eyes of the public, deteriorate.
By bringing in the perspective of the non-elite blame-makers, our findings enable us to further the theoretical discussion on the relationship between anticipatory and reactive forms of blame-avoidance behaviours (Hinterleitner and Sager 2017) and to discuss the democratic implications of blame-games in both the short and long run. Our analysis indicates that successful non-elite blame-making at one point in time can actually lower the chances of successful blame-making in the future. We show how clear and transparent arrangement of formal accountability at one point in time may force power holders to be attentive to grassroot activists' demands. However, this very same contextual factor may also cause power holders to take anticipatory actions to hamper similar harmful blameattributions in the future. In other words, the same factor that in the short run enables effective blame-making can in the longer run encourage power holders to use anticipatory strategies to avoid future blame.
The paper proceeds as follows. First, we outline how previous research has discussed policy consequences of blame-games, contextual factors that may impact blame-makers' possibilities to attribute blame, and how power holders respond to blame-attributions. From this discussion, we suggest a mechanism by which blame-making can lead to policy changes, and point to the importance of powerful and timely media narratives, as well as clear and transparent institutional mechanisms of accountability when non-elites use blame as a strategy to reach policy influence. Second, we present our two cases and discuss the rationale behind the case selection as well as the practice tracing methodology of the study. Third, the blame-games and policy consequences generated by the 2005 and 2017 campaigns are analysed separately, and then compared and elaborated in relation to the blamegame literature in the discussion section that follows. We conclude the paper by summarizing the argument and discussing the implications of the findings for our understanding of the relationship between blame-making and democratic accountability.

When non-elites attribute blame-rationales, contextual factors, and blame-avoidance responses
In this theory section of the paper, we build on previous research from the blame-game literature, and try to adapt these findings to non-elite groups. We portray non-elite blamegames as a two-player blame-game in which (1) non-elite groups with limited resources have a choice of potential strategies to achieve their policy objectives, (2) power holders' blame-avoidance responses are conditioned by both non-institutional and institutional factors, and (3) policy consequences result from their interaction.
As stated in Introduction, blame-making as a non-elite strategy to change policy is an understudied phenomenon in the blame-game literature. While social movement studies have demonstrated that blame-making can be one of several strategies for non-elite groups to exert influence over policies (Klandermans et al. 1999), these insights have not properly been explored in the blame-game literature. Here, the rationale behind attributing blame to someone in power has hitherto foremost been described as a way for political elites to attempt to capitalize on crises, scandals and other negative messages about their political opponents in order to increase their popularity with voters (Brändström and Kuipers 2003;Hinterleitner and Sager 2015;Hinterleitner 2018;Resodihardjo 2020). However, an exeption is found in an article about contemporary American politics, where R. Kent Weaver (2018) discusses policy change as an alternative rationale for political elites to employ blame-making strategies. He suggests that sometimes blame-makers' (he calls them perpetrators) objective with attributing blame to power holders (called targets) is: to change the behavior of the target of blame […] by intimidating the target into shifting their "zone of acceptable outcomes" to make it more consistent with those of the perpetrator. Targets are threatened with the possible alienation of their supporters if they do not change their behavior' (Weaver 2018, 269).
The mechanism he describes is one of indirect pressure from blame-makers to power holders through influencing perceptions among the power holders' supporters. By publicly framing a policy as a failure and attributing blame to particular power holders for that failure, blame-makers hope to attribute such negative connotations to power holders that they feel forced to change their articulated policy preferences.
This description of the mechanism behind blame-making for policy reasons seems equally applicable to non-elite groups as to political elites. However, one aspect of the mechanism described above is likely to be more demanding for none-elites than for elite players. Blame-attribution for policy reasons can only be effective if the targeted power holders believe that there is broad support among the electorate for changing policies in accordance with the blame-makers' preferences. This means that non-elite groups need to show power holders that they can mobilize broad and visible support among the electorate for their political demands in order to have enough leverage to pressure power holders to change policies. Without the organizational resources to mobilize a large section of the electorate, which elite players are more likely to possess, non-elite groups have to rely on mediators to spread their message. It is well known that blame-makers with communicative skills to produce blame messages that feed into established media narratives will have greater changes to put pressure on power holders than actors who lack such skills (Hinterleitner and Sager 2015; Weaver 2018). If the media narratives already are sympathetically framed towards the group that the activists are seeking to help, and critical towards the power holders, the blame message is more likely to resonate with the public's general views and beliefs (Benford and Snow 2000).
Besides the challenge of mobilizing broad and visible support through mass media, the success of blame-making strategies in altering policies is determined by a variety of contextual factors. The literature has pointed to several non-institutional factors as important explanations for the outcome of blame-games, for example power holders' positions, resources, and capacities to avoid and deflect blame as well as in which arenas blamegames unfold (Hinterleitner and Sager 2015;Weaver 2018). In addition, policy characteristics have been put forward as an important determining factor for blame-game outcomes. Hinterleitner claims that 'distant-visible policies ' (2018, 236), that is, policies with high salience but low familiarity to the general public, are most likely to result in successful blame-making. The reason for this is that the general public cares about the issue, but has little personal experience of it.
Despite there being a number of relevant contextual aspects influencing blame-games, the most decisive factor found in previous blame-game research seems to be the institutional context (Hinterleitner andSager 2015, 2017;Figenschou and Thorbjørnsrud 2018). Figenschou and Thorbjørnsrud studied how Norwegian Ministers responded to public blame-attributions and concluded that 'future studies should include the institutional constraints as a key condition to understanding communication in various public sector organisations ' (2018, 229). Providing a framework for systematic comparative research, Hinterleitner and Sager made a similar claim when stating that 'a comprehensive analysis of BAB [blame-avoidance behaviour] environments needs to take into account the various accountability mechanisms institutionalized in the political system ' (2015, 157). In a later paper, the same authors (Hinterleitner and Sager 2017, 598) specify the importance of the institutional context by writing about its relationship to perceived responsibility, that is, the ease with which the public can assign responsibility to power holders given institutional arrangements. Blame-avoidance strategies thus need to correspond with, at least vaguely, how formal power and responsibility within the policy sector are allocated. Policy-specific factors determining perceptions of responsibility are related to the complexity of the collaborative structure within a policy sector as well as to what degree political actors are involved in decision-making. For example, a policy sector characterized by a clear and transparent collaborative structure with large involvement of political actors will decrease the opportunities for power holders to deflect blame in a credible manner, as these characteristics make it clear to outsiders how responsibility is allocated. In contrast, a policy sector with a complex collaborative structure and little involvement of political actors will make it difficult for the general public to understand who is responsible for what decisions.
Previous studies in the blame-game literature have thus described several reasons why actors engage in blame-making and which contextual factors that are likely to determine the effectiveness of blame-making to influence policy. But how do power holders respond to being blamed? Apart from anticipatory strategies to avoid blame even before it has been attributed, the literature points to a variety of possible reactive strategies power holders may employ to avoid blame when attributed responsibility (Weaver 1986;Hood 2002Hood , 2007Hood , 2011Hering 2008;Resodihardjo et al. 2016;Hansson 2018b). Although many different attempts to categorize blame-avoidance strategies have been made, for the purpose of our analysis we only need to differentiate between reactive blame-avoidance strategies that seek to undermine the problem component of the blame-attribution by reframing the problem, and strategies that seek to undermine the agency component by deflecting responsibility (Hood et al. 2009;Resodihardjo 2020). Reframing strategies could range from simply denying the problem presented by blame-makers to diminishing it or justifying the problem. Deflection strategies could involve scapegoating, delegating, distributing responsibility, or offering excuses and apologies.
The literature on blame-games further suggests that reframing strategies are used as the first avoidance strategy and only when this option is exhausted and fails to work, deflection strategies come into play (Hood et al. 2009;Hinterleitner and Sager 2015;Resodihardjo 2020). This suggests that blame-games are dynamic processes, where blame-makers use rhetorical strategies to try to increase the level of blame on power holders, while power holders respond with blame-avoidance strategies in order to lower the level of blame attributed to them. Although not empirically studied in previous research, this indicates that depending on of the nature of blame-avoidance strategies available for power holders in a given situation, non-elite blamemakers can put greater or lesser pressure on power holders to meet political demands. Given the lack of research on this topic, we will now turn to our empirical exploration of how nonelite initiated blame-games can put pressure on power holders and the various consequences that emanated from such initiatives.

Research design
Given the purpose of this study, we want to investigate cases situated in comparable policy environments and initiated by non-elite blame-makers with similar political demands. However, to reach variation in blame-game dynamics and to explore the role of perceived responsibility, we need cases which differ in terms of institutional arrangements of accountability.
The two Swedish pro-migrant campaigns we selected meet these qualifications. They are situated within the same political system and national policy-making environment, which in the Swedish case is marked by corporatism, consensus, and rationalism (Petersson 2015), and both campaigns were initiated and driven by grassroot activists who took to the streets to demand regularization for denied asylum seekers. Moreover, both campaigns managed to gain high visibility in mainstream media, making both campaigns' focus on asylum-a highly salient, yet very distant, policy area for most Swedes. Consequently, both campaigns had similar chances to pressure power holders by creating high levels of visibility for their demands.
However, in terms of institutional arrangements of accountability, there are differences between the two cases. During the 2005 campaign, the Swedish asylum system was structured in a hierarchical fashion, which gave the Government large discretionary power and responsibility over asylum determinations. The Government was the highest instance of appeal in the asylum system, and thereby steered the application of law in asylum determinations by issuing guiding decisions in individual cases. These guiding decisions often gained attention in the media, hence articulating to the Swedish public that the Government was in control of the asylum system.
In 2017, when the second regularization campaign started, the institutional arrangements of accountability were very different. In 2006, the Swedish asylum system underwent a profound reform which deprived the Government of its possibility to issue guiding decisions and to steer the determinations of asylum. In its place, independent appeal courts, called Migration Courts, were introduced into the asylum system. This reform created a more fragmented allocation of power, as responsibilities for different parts of the system were now shared between the Swedish Migration Agency, the Migration Courts, the Government, and the Swedish parliament (Johannesson 2017;Johannesson and Weinryb 2019). In the empirical analysis, we explore how this difference in institutional arrangements of accountability made a difference for how the blame-game dynamics evolved, and the various consequences that emanated from them.
As in all comparative studies, there are limitations to this research design. In this case, there is at least one important contextual factor which complicates the possibilities to single out the explanatory power of our chosen variable (peceived responsibility), that is, the Swedish official public discourse on migration. We will keep this contextual factor in mind when drawing conclusions from this comparative study in the discussion section of the paper. Nevertheless, we believe that these two cases share enough relevant similarities with regard to what has previously been concluded about contextual factors in the blame-game literature to make a comparison of them useful for advancing our knowledge of non-elite blame-making and its consequences on policy.

Methodology
In order to meet the methodological challenge of assessing policy consequences of activist campaigns, we rely on the method of practice tracing (Pouliot 2015), which combines interpretive analysis with elements of process tracing. An interpretive approach to policy processes means viewing how formal policy goals are transformed into meaningful practices in a local context. These meaningful practices are understood as 'regular forms of action within a given social context' with 'causal power in the sense that they make other things happen' (Pouliot 2015, 241). We here view blame-making as a meaningful practice which 'make other things happen'-in this case blame-avoidance responses and policy changes.
As a methodology, practice tracing shares with process tracing the commitment to find an observable link between a set of initial conditions and a particular outcome. However, practice tracing does not view these observable phenomena as causal mechanisms per se, but aims to produce analytical general insights by using the logics of abstraction from empirical observations. The objective of practice tracing is therefore not to produce general claims of causality that could be validated through testing in other empirical settings, but to produce claims of causality that could be 'useful (or not) in making sense of messy arrays of practices' (Pouliot 2015, 239). We find this methodological approach preferable to process tracing, as it allows for comparative research that combines systematic sensibility with a high degree of context sensitivity, something that has been demanded by previous research on blame-games (Hinterleitner and Sager 2015, 139ff).
In order to assess the impact of blame-making practices on policies in our two cases, we traced the campaigners' demands for regularizations through the various stages of the policy processes, from the launch of the campaigns through the agenda-setting and policy adoption phases, into the parliamentary negotiations, and finally all the way to the passing of Government bills. We have divided the empirical analysis into three consecutive phases: blame-making, blame-avoidance responses, and policy consequences.
Blame-making. To capture the activists' choice of blame strategies, we relied on oral and written data from activists during the two campaigns. The oral data consisted of two public seminars to which we invited twelve leading activists during the two campaigns (see Appendix 1). The seminars were used to document the modern oral history of asylum activism in Sweden. They lasted for about 3 h each and were recorded, transcribed and made publicly available (Johannesson and Weinryb 2020a, b). During and after the public seminars, we received documents that had been produced by campaigns, such as leaflets, op-eds, letters to politicians, and notes from meetings. We also searched the Swedish media archive Retriever for articles with public statements from campaign spokespersons. 1 One common definition of blame-making is that it is an 'act of attributing something considered to be bad or wrong to some person or entity' (Hood 2011, 6). Blame-making thus requires both the construction of harm or loss and someone responsible for that harm or loss, hence both a problem and an agency component. Leaning on our various forms of data, the coding process began by us identifying the problem and agency components of the campaigners' blame-attributions.
Blame-avoidance responses. The political responses to the activists' demands were found in official records from the Swedish Parliament documenting the political reactions to the campaigns' political demands. These included party motions, Parliament committee reports, minutes from chamber debates, and Government bills. We also found political responses in the media material we had collected. The responses from targeted power holders were coded according to which type of strategy (reframing or deflection) they employed.
Policy consequences. To trace the consequences of the campaigners' blame-making, we followed the policy processes from agenda-setting phase to policy adoption and implementation (Weaver 2018). More specifically, we followed how the campaigners' political demands were picked up by different political actors, and later transformed and moulded into politically acceptable solutions which eventually, to various degree, became implemented. The interviews with activists also helped us trace the impact of the campaigners' demands throughout the policy processes. However, as we know from other studies (Basok and Rojas Wiesner 2017), activists tend to exaggerate their influence, and therefore, we triangulated their statements about policy impact against other sources of information, such as media material and political documents.

Empirical analysis
We will now present each case as a chronological sequence, starting with the blame-making attempts, which are then followed by the power holders' responses to these attributions. We conclude the presentation of each case by discussing the policy consequences of these blame-games.

The refugee amnesty campaign in 2005
In the autumn of 2004, it became clear that the administrative reform (described above) would be implemented in March 2006 (Government bill 2004/05:170). This one-and-ahalf-year timeframe between the passing of this reform and its implementation created a strategic window of opportunity for pro-migrant activists to push for a regularization programme for denied asylum applicants who had been assessed according to the old asylum system (Interviewee  Flyktingamnesti 2005]. Although the two campaigns were organized separately and under two different names, they were initiated by grassroot activists in close collaboration with one another. Most of these activists had for a long time worked and volunteered with irregular migrants in local communities throughout Sweden. For both campaigns, the core activity was to collect name signatures for a petition in support of the amnesty demand to hand over to the Government. In addition, the campaigns organized large protest marches, and activities also included theatre plays, art exhibitions, music recordings, and the publication and distribution of a newsletter.

Blame-making
We identified three different problem definitions in the 2005 campaign. The first regarded the difficult situation for families and children with denied asylum applications who stayed in Sweden. This problem definition tied into an already established public debate on 'apathetic children' that had been growing in Swedish media since the mid-1990s. The apathetic children signified asylum-seeking minors who had developed severe depressive symptoms that plunged them into apathetic conditions. The debate among experts and the public about the causes and range of this problem was huge and resulted in various interpretations of why these children developed such serious and life-threatening conditions (Eastmond and Ascher 2011).
A second, and related, problem definition focused on the Swedish national self-image of being a 'moral superpower' (Dahl 2006), and how this image was impossible to uphold when children were suffering under the Swedish asylum system. Both problem definitions are illustrated in the following quote from the public 'Christmas letter' which the most widely known representative of the Easter Petition campaign, the Archbishop of the Swedish Protestant Church, sent to the Government in December 2004: Many colleagues in churches and other civil society organizations who work close to asylum-seeking families have during the autumn reported about their sorrow, resignation and frustration. They are often ashamed of our society and of the fact that we are beginning to get used to doing away with compassion, empathy and solidarity with the people who cross our path.
In this quote, the problem of disgraceful treatment of children and their families refers to the debate about the apathetic children, but the second problem about the Swedish selfimage is also articulated in the concern about how Swedish society as a collective 'we' has begun to lose compassion with other humans. This problem's moral relevance was emphasized and strengthen when articulated by an actor with such high moral authority as the Archbishop.
The third problem that the activists presented was less common in the campaign material, but was mentioned in the interview material as an important strategic problem definition. It was a problem that targeted the administrative challenges connected to implementing the administrative reform in 2006. The interviewees expressed that they for 'practical' (Interviewee 2005-3) reasons advocated a general amnesty for all asylum seekers as a solution which could 'clear the table' and give the new system a 'fair chance of working' (Interviewee 2005-1), without having to inherit old cases which would protract the proceedings and cause delays in the system.
The blame strategy that the campaigners used was coherent in the sense that responsibility for all three problems was unanimously attributed to the Government. In a leaflet that the campaigners handed out during demonstrations and other public activities, the Government's responsibility was clearly articulated in the following declaration: 'The responsibility for the situation therefore lies with the Government, who choose to view condemnations from the UN's Committee Against Torture, repeated reports about apathetic children and suicide among asylum seekers as work accidents.' Although it is stipulated in the Swedish Constitution that the Government makes decisions as a collective and that ministerial rule is forbidden (SFS 1974:152, chap 7;chap 12), the Minister for Migration, Barbro Holmberg, was repeatedly attributed personal responsibility for the problems in the asylum system. This indicates that the perceived responsibility was attributed to her in person, as the most visible representative of the Government in this issue. Our interviewees described the minister's reactions to these accusations as oblivious to the human suffering of the apathetic children. She was described as acting 'like a parrot', meaning that she 'repeated the same things over and over again about legal certainty, predictability and so forth as arguments against this amnesty' (Interviewee 2005-6). One interviewee in particular remembered a meeting with the Minister for Migration, which took place in May 2005 in the Minister's office when the activists handed over binders containing the 157,251 signatures in support of the refugee amnesty to her in person. He found the Minister's reaction to the protest signatures to be insensitive to the activists' critical approach: So, [the binders] were handed over to Barbro Holmberg and I remember so well when we were standing there and she emerged from a low door from the interior of the room, and came out looking very embarrassed, and began with the very surprising line 'thanks for the support'. (Interviewee 2005-1) The Minister's reaction to being blamed was productive for the activists in the sense that it confirmed the problem definition that the campaign had formulated about the Government's ignorance and lack of compassion. It also stressed the rhetorical gap the activists had formulated between the Swedes who supported the amnesty demands, and thereby still lived up to the collective self-image of solidarity and compassion, and the political elite, who had lost their capacity to express compassion. In conclusion, the campaigners succeeded to tap into the prevailing media narrative about ignorant power holders who do not care about ordinary people.

Blame-avoidance responses
In the 2005 case, the political response to being blamed consisted of different types of reframing strategies. This is not surprising, given that the asylum system until the administrative reform in 2006 was deliberately designed to be within the Government's executive discretion, thus making it difficult for the Government to counteract any claims about being the highest responsible actor in the system. The only blame-avoidance response available for the Government was therefore different types of reframing strategies.
The Minister for Migration repeatedly tried to challenge the problem definitions concerning the apathetic children and the lack of compassion in the Swedish asylum system. She mainly used two arguments. Firstly, the Minister tried to redefine the problem in the asylum system as a problem of weak legal certainty and unpredictability. She repeatedly stated that the most important thing was to have individual and thorough assessments of asylum claims. If unpredictability was the main problem with the old asylum system, then a general amnesty for all denied asylum seekers would only increase the sense of unpredictability and uncertainty in the system, the Minister argued (Prot. 2004/05:97, addr. 8).
Secondly, the Minister for Migration used a 'justification' strategy (Hood 2011, 49) by trying to reformulate what compassion and humanitarianism meant in the Swedish asylum system. In an article in one of the major Swedish newspapers, she stated that a general amnesty 'might seem humane, but in reality, it is a form of misdirected humanitarianism' as it signals to asylum seekers that 'a no is not really a no, because if one waits and hides, it could be transformed into a yes' (Holmberg 2005). In a response to her critics in the parliament, the Minister for Migration elaborated the same justification strategy by stating that: We see that there are signs that there is enormous pressure on children in the asylum process. The child becomes the key to the parents' intense desire to obtain a residence permit in Sweden. (Prot. 2004/05:71, addr. 136) In this statement, and in other similar ones, the Minister for Migration justified denying apathetic children and their families asylum due to a belief that they were simulating the illness conditions. By granting them asylum, this harmful behaviour would only continue and even increase, the Minister argued.

Policy consequences
Already from the start, the amnesty campaign received support from the five smaller parties in the Swedish Parliament. On several occasions during the spring of 2005, these five parties presented policy proposals that clearly built on the campaigners' demands. However, the two large parties in Parliament-the Social Democratic Party and the conservative Moderate Party-voted down the proposals each time. Each time the demands were voted down, the mobilization around the campaigns continued to grow in strength and by the summer 2005, the public pressure on the Government to stop deporting apathetic children and their families had grown considerably. In July, the Government exercised its discretionary power by issuing a new guiding decision in an asylum case with a child that suffered from a severe apathetic condition. The family received a permanent residence permit due to humanitarian reasons, and this became precedential for about a hundred similar cases of children with severe apathetic symptoms.
In spite of this partial victory for the campaign, the demand for a general amnesty for all denied asylum seekers in Sweden continued to have broad support in Parliament. When the Social Democrats and Moderates again, in September 2005, voted down a parliament proposition for a regularization programme, the Green Party and the Left Party-in their capacity as supporting parties to the Government-changed strategy. Instead of trying to win a majority in Parliament, they used the Government's dependence on them to pass the annual state budget by threatening to block that budget if the Government did not agree to the amnesty demand. Over the following weeks, intense negotiations between the Government and the supporting parties took place. The close alliance that had developed between the Left and Green parties and the campaigners was strategically used in the negotiations to increase the pressure on the Government. Our interviewees described how they were in direct phone communication with party members from the Green and Left parties during the budget negotiations, offering party representatives their immediate opinions on various suggestions that were on the table (Interviewee 2005-5).
After weeks of negotiations, the Social Democratic Party accepted a temporary law with a modified version of the amnesty demand. The temporary law gave families with children who had been residing in Sweden without permission, as well as denied asylum applicants where the deportation could not be enforced, a new opportunity to get their asylum claims assessed. A total of 17,300 residence permits of the 30,000 lodged were granted due to this temporary law (Borevi 2012, 72).
The activists called this compromise an 'amnesty light' (Interviewee 2005-1) as not everyone who applied received residency. It was, however, clear that the Government had been forced to implement parts of the campaign's demand despite trying to resist it for a long time. In the second campaign case in 2017, the story seemed to repeat itself as the Social Democratic Party once again became pressured by the Green Party to accept a regularization demand. The difference this time was that the institutional arrangements of accountability changed to the disadvantage of the asylum activists and to the benefit of the power holders. In the following section, we describe how this 2017 blame-game evolved differently due to the changing institutional conditions.

The refugee amnesty campaign in 2017
Among the large influx of asylum seekers to Europe in the autumn of 2015, over 160,000 persons sought protection in Sweden. Of the total number of asylum applicants in 2015, about 35,000 were unaccompanied minors, and the majority of these were young men of Afghan origin (Swedish Migration Agency 2016), although many of them had been living in Iran for the greater portion of their lives.
In the aftermath of the 2015 crisis, the Swedish Migration Agency assessed Afghanistan as safe in certain provinces, and therefore began deporting asylum applicants of Afghan origin to these so-called safe provinces. In 2017, 63% of all applicants of Afghan nationality were denied asylum and among unaccompanied minors of Afghan origin the rejection rate was 22% (Swedish Migration Agency 2018). The deportations evoked a lot of grievances among the unaccompanied minors as well as among the people who had helped them through the asylum process (Turunen and Weinryb 2020). Campaigns began to stop the deportations to Afghanistan and to offer amnesty for Afghans in Sweden. Campaign activities included digital petitions, demonstrations outside the deportation centres, theatre performances, live-streaming on social media platforms, protest marches, and candlelight vigils. The most publicly visible campaign activity was a 58-day long sit-in protest in several major cities in Sweden, which lasted from August to October 2017.

Blame-making
We found two central problem definitions in the 2017 campaign, both formulated around the unaccompanied youths of Afghan origin facing deportation to Afghanistan. One problem definition targeted the issue of safety in Afghanistan. The activists claimed Afghanistan was a dangerous country, and that when Sweden deported people to the presumed safe provinces of Afghanistan, they were essentially sending them to war and death. Media photographs from the sit-in protests show signs with messages such as 'Stop deportations to death' and 'We want to live' (Stiernstedt 2017). One activist, interviewed in a regional newspaper during a sit-in protest in a southern town in Sweden, proclaimed that 'I do not know why the Migration Agency thinks Afghanistan is a safe country. We can see people being killed there every day. Ordinary people, not soldiers with guns' (Berg Eidebo 2017).
The other problem definition constructed deportations of Afghan youths as a loss of human capital and thereby tied onto a discourse constructing migrants as contributing members of society (Fujiwara 2005). During the seminar, the activists claimed that Afghans in Sweden could become a 'good resource' in society and repeatedly used the figure of a 'street child' (Interviewee 2017-2) to stress the future which awaited Afghan youths if they were not given residency in Sweden. The image of a street child did not only function as a symbol of victimhood but was also intentionally used in the staging of the sitin protests. One of the interviewees, who initiated the sit-in protest, described how he was certain that in Sweden the authorities would never allow 'a child to sit on the street a whole night', but as the sit-in protest continued night after night, he found himself to be mistaken (Interviewee 2017-3).
As in the 2005 campaign, the 2017 protesters attributed blame to power holders, but due to the complex and ambiguous delegation of accountability within the asylum system, the blame-making was not as coherent as in the 2005 campaign. It was difficult for the campaigners to attribute personal responsibility for the problems to anyone in particular. The interviews demonstrate how the changed institutional arrangements of accountability led to ambiguous perceptions of responsibility. Interviewees described how politicians and public officials were 'blaming each other all the time' (Interviewee 2017-3) and that 'everything is so diffused' (Interviewee 2017-1). One of the interviewees described how activists were being pushed around between the politicians, the Migration Agency and the Migration Courts, 'all the time they kicked us around like a ball' (Interviewee 2017-4).
The confusion that arouse as a consequence of the 2006 reform is clearly illustrated in the following sequence, which one interviewee described (Interviewee 2017-3). The activists had asked politicians to come to the sit-in protest to talk to them , which many also did initially. However, no representatives from the responsible agencies, such as the Migration Agency or the Government came to talk to the protestors. But suddenly, one day in the middle of the sit-in protest, the Social Democratic Minister for Migration, Heléne Fritzon, paid a visit. The activists became excited and thought that this was a sign that the Government would put a stop to the deportations. But the message from the Minister was that she only came to show her support, not to negotiate with the activists. The activists became puzzled by her behaviour. Why did she not meet our demands if she supported us, they wondered. Events like this augmented the activists' confusion about who they should demand changes from and who was responsible for the decisions.
Due to the complexity of the institutional arrangements of accountability, the activists started to attribute blame further down the chain of public authority. The ones that received most personalized blame from the activists were the Director General and the Head of Legal Affairs at the Swedish Migration Agency, as they in legal opinions had assessed the situation in Afghanistan as safe. The activists sent a public letter to the Director General of the Swedish Migration Agency stating that 'it is time for you to live up to your role. You can contribute to a legal opinion calling for the stop the deportations to Afghanistan. Our lives are in your hands' (Ledel 2017). In an interview in one of the largest daily newspapers in Sweden, one of the spokespersons for the sit-in protest turned directly to the Head of Legal Affairs at the Migration Agency and proclaimed that 'you could write a legal opinion about Afghanistan. You have done it before. Why can you not do it now?' (Stiernstedt 2017)

Blame-avoidance responses
As a consequence of the reform in 2006, which led to a more complex and unclear delegation of accountability for asylum determinations, the targeted power holders in 2017 had more blame-avoidance strategies to employ than what the power holders had in 2005.
As will be evident in the following analysis, both reframing and deflection strategies were used by the Government and the Migration Agency, albeit in slightly different ways.
The responses from the Migration Agency targeted the activists' problem definition about Afghanistan not being a safe country. Despite the request from the activists at the sit-in protest, the Director General declined to meet them at the square. Instead, he used a reframing strategy that tries to win the argument through justifications, in this case of the Agency's assessment of safety in Afghanistan. For example, the Director General wrote a public letter, published in several newspapers, attempting to justify why the Migration Agency had a different assessment of the safety situation in Afghanistan than the activists (Eskilstuna-Kuriren 2017) In contrast to the Migration Agency, the Government focused on deflection strategies. Notably, the Minister for Migration responded to the campaign's criticism by referring to the constitutional order of Government in Sweden, for example in this quote from a parliamentary debate about the amnesty demand: Sweden has one of the world's fairest systems, because there is one responsible public agency that takes the decision, which can also be appealed in court. This decision-or this assessment-is an issue that neither I, nor the Government, can or should decide. Nor do I think it would be desirable that such a decision would be taken by politicians, the Riksdag [the Swedish Parliament] or any political body. (Prot. 2017/18:3, addr. 2) In this quote, the Minister for Migration delegated responsibility downwards to the Migration Agency and the Migration Courts. The reason she could do this with credibility was, as noted earlier, that the administrative reform in 2006 had relieved the Government of its decision-making discretion and thereby spread the formal responsibility for the assessment of asylum claims across several actors in the system.
As the protests intensified during the autumn of 2017, the Prime Minister became active in the amnesty debate, defending the Government's position to reject any form of regularization of denied asylum seekers. He used similar deflection strategies as the Minister for Migration, referring to the reformed asylum system, now beyond the Government's control: It is good that we have moved this issue from the politicians to a legally secure procedure where the authorities may perform assessments and where there are opportunities for appeals to be made in court. (Prime Minister Stefan Löfven, quoted in Holmqvist 2017) In sum, the Government's use of deflection strategies and the Migration Agency's use of reframing strategies together created a powerful arsenal of blame-avoidance rhetoric that undermined the activists' blame strategies.

Policy consequences
Due to the reform in 2006, it had become more difficult for activists to attribute personal blame to the top players of the asylum system, at the same time as targeted power holders had broadened their spectrum of possible blame-avoidance strategies. The activists' opportunities to pressure power holders into meeting their demands had thereby decreased.
Nevertheless, according to the activists themselves, their campaign activities were decisive for the later policy changes that were subsequently passed in parliament (Interviewee 2017-1; Interviewee 2017-2; Interviewee 2017-6). They did, however, point to other activities than the blame-making as the crucial components which made politicians attentive to their demands. Foremost, they emphasised the close alliance between the activists and some members of the Green Party, who succeeded to force the party leadership to take actions to promote an amnesty proposal in parliament. The Green Party, which since the 2005 refugee amnesty campaign had fostered close ties with the pro-migrant movement in Sweden, had in 2017 entered a coalition Government with the larger Social Democratic Party. This factor is probably the main reason why the amnesty demand in 2017 made an impact in policies a year later. Both our interviewees and reports in news media (Rosén 2017;Lönegård and Delling 2017) describe how members of the Green Party, who also participated as activists in the amnesty campaign, succeeded to get a majority at the Party congress for an amnesty proposal. This decision forced the party leadership to pursue the amnesty demand in political negotiations with the Social Democratic Party.
After weeks of negotiations between the two parties, the Green Party managed to persuade the Social Democrats to present a bill which was passed in Parliament with the support of the Left Party and the Centre Party and came into effect in July 2018 (Prot. 2017/18:125). It opened up a temporary possibility for Afghan youths with deportation orders to, under certain conditions, apply for prolonged temporary residence in Sweden (Government bill 2017/18:252). 2 However, this law did not satisfy the immediate demand of the 2017 campaign: to stop the deportations to Afghanistan. Swedish authorities continued during 2019 to deport people back to Kabul under the internal flight directive.

Discussion
In the introduction of this paper, we asked how perceptions of responsibility determine the pressure non-elite blame-makers can put on power holders, and which implications that have for our knowledge about the conditions for and consequences of blame-games. In the following discussion, we will try to provide answers to these questions by elaborating the most important general insights from the empirical observations of this study.
Firstly, the two cases demonstrated that blame-making can be an effective strategy for non-elite groups to pressure power holders into making changes in policies. Both campaigns used sophisticated protest strategies to mobilize support for their political demands, and they had sufficient communicative skills to gain widespread media coverage of their blame-making activities. In both cases, it became evident that power holders felt pressured, not only to listen to the activists, but also to show the public that they did so. In the 2005 case, the Minister for Migration met the activists in person to collect the name signatures from the petition, and in 2017, the Minister for Migration made a public appearance at the sit-in protest to 'show her support', as she called it. It seems plausible to assert that the power holders believed that the campaigns had mobilized sufficient support among the electorate to make it strategically sound for them to publicly show that they took the amnesty demands seriously. This finding strengthens Weaver's (2018) suggestion that blame-making can lead to policy change through the mechanism of indirect pressure from blame-makers to power holders through supporters. Secondly, and highly related to the first point, our comparative approach was designed to study the effects on perceived responsibility of differences in the institutional arrangements of accountability within a policy sector. We found, in accordance with previous studies (Hinterleitner and Sager 2017;Figenschou and Thorbjørnsrud 2018) that clear and transparent arrangements of accountability decreased power holders' possibilities to deflect responsibility with preserved credibility. However, through our primary focus on blame-makers instead of blame-avoiders, we also found that activists' possibilities to attribute personal blame to power holders in a credible way decreased when arrangements of accountability became complex and diffused. In 2005, the Swedish asylum system was structured to give the Government unambiguous responsibility for asylum determinations and the system required direct and reoccurring political involvement in decision-making. These factors created a situation in which the Government was held responsible in public for individual decisions on asylum, and the Government did not have the opportunity to deflect this perceived responsibility to any other actors in the system. Their only option was therefore to use reframing strategies to avoid blame. The blame-makers thereby gained credibility in the eyes of the public, and this helped the activists to receive support for their political demands. In 2017, the asylum system had been rearranged into a more complex accountability structure, and the discretionary room for political decision-making had been replaced by judicial powers. This gave the Government more room to use deflection strategies to avoid blame, at the same time as it created confusion among the blame-makers regarding whom to assign responsibility and for what. The 2017 campaign could therefore not use personalized blame narratives with the same confidence and credibility as the 2005 campaign could. The result was a weaker pressure on the power holders in 2017 than in 2005 to change policies in accordance with the demands of the campaigners.
However, our two cases are insufficient to conclude that the blame-making in and of itself was sufficient to cause policy changes. Instead, our two cases support the claim, found in social movement studies, that several different but mutually reinforcing activist strategies are needed to create policy changes (Cress and Snow 2000;Pralle 2003;Amenta et al. 2010;Kirchhoff et al. 2018). Both our cases showed that blame-making strategies together with close alliances between activists and elite players (in this case the Green Party) in politically central positions were crucial for the campaigns' policy influence. In 2005, the Green Party possessed power over the state budget, but in 2017, they possessed direct governmental power as one of two parties in a coalition Government. This means that at the same time as one institutional factor (arrangements of accountability), which determines the strength of blame-making, changed to the disadvantage of the activists, another non-institutional factor (the political power of allies), not directly related to the strength of blame-making strategies but to the broader possibilities for activists to influence policy, changed to their advantage. This accentuates that blame-making is only one out of several strategies which non-elite groups can employ to influence policy.
Moreover, our study cannot exclude other contextual factors, such as the public discourse on migration, from playing a role in determining how effective blame-making strategies will be. It is for example highly plausible that the positive official discourse on migration in 2005 amplified the credibility of the activists' blame-making in the media and among the general public, which in turn, increased the pressure on the Government to listen to the activists' demands. Between 2005 and 2017, the number of immigrants increased in Sweden at the same time as a anti-immigration party-the Sweden Democrats-entered and later dominated the political scene as regards the migration issue. The official public discourse on migration was thus generally more positive and less polarized in 2005 than in 2017 (Bevelander and Hellström 2019). Based on this, it is difficult to determine the relative importance of the blame-making itself in pressuring the power holders, compared to the relevance of altered political discourses on migration. It is highly possible that the relatively positive official political discourse on migration in 2005 facilitated a fertile atmosphere for pro-migrant campaigns which did not exist to the same extent in 2017 when the official discourse was more polarized. This intervening factor can however not explain why the activists' blame-making became less personalized and coherent in 2017 or why targeted power holders used more blame-avoidance strategies to deflect the blame attributed to them in 2017 than in 2005. We argue that the most plausible conclusion to draw from these changes in blame-attributions and responses is that the perceptions of responsibility had changed as a consequence of the reform in 2006.
Thirdly, the findings from this study have implications for the scholarly discussion on anticipatory and reactive blame-avoidance behaviour. Scholars have carved out the distinct nature of reactive and anticipatory blame-avoidance behaviours (Hinterleitner and Sager 2017), but less emphasis has been put on showing how the behaviours are related. We think that the sequentiality of our two cases can shed light on the link between reactive and anticipatory blame-avoidance behaviours.
Anticipatory blame-avoidance strategies differ from reactive strategies because they are utilized to hamper blame-attributions before they are articulated (Hood 2011). In addition, they address the institutional arrangements of accountability in a policy sector instead of engaging in rhetorical struggles. The mechanism that has been theorized to trigger anticipatory blame-avoidance strategies is a calculation from the power holders' side of the risk of being blamed in the future. If that risk is considered high, power holders are likely to engage in anticipatory blame-avoidance strategies (Hinterleitner and Sager 2017;Resodihardjo 2020). How do power holders then calculate that risk? One plausible way of doing it would be for power holders to learn from the past. If blame-attribution has been a reoccurring phenomenon in the past, the likelihood of continued blame-attribution in the future could be calculated as high.
However, from the perspective of the blame-makers, this calculation of risk means that blame-making strategies can undermine themselves in the longer run. From this study, we cannot stipulate that the administrative reform in 2006 was intended to avoid blame, but power holders have clearly capitalized on its result, namely the ambiguities regarding the formal allocation of responsibility that exists in the reformed asylum system. Hinterleitner and Sager (2017) make the claim that anticipatory blame-avoidance behaviour can 'translate into blame-deflecting institutional arrangements and policy design that may alter the institutional set-up of policy sectors and influence the effectiveness of policies' (Hinterleitner and Sager 2017, 601). Our study sketches a related, albeit longer chain of cause-andeffects: blame-making can lead to reactive blame-avoidance strategies, which can trigger anticipatory blame-avoidance strategies, which can 'translate into blame-deflecting institutional arrangements' (ibid), which, in turn, can put formerly successful non-elite blamemakers in less favourable positions to exert policy influence in the future.

Conclusions
It is a democratic problem that some interest groups in society lack the material and political resources to exert influence over political questions that concern them. Our analysis shows that this problem can temporarily be reduced when these groups manage to use blame-making strategies to get their political demands heard. This brings forth a hitherto neglected democratic aspect of blame-games, namely their potential of providing marginalized interest groups-such as asylum seekers-with a powerful tool to force power holders to be attentive to their demands.
At the same time, our two cases also reveal that blame-making, in a longer perspective, can bring about other, less democratically valuable consequences. If non-elite groups repeatedly use blame-making to force power holders to take policy actions, they might also trigger power holders to change the institutional arrangements of accountability in such a way that it becomes harder to attribute blame to power holders in the future. In situations where power holders calculate a high risk of future blame-attribution, they are likely to engage in anticipatory blame-avoidance behaviours, which may hamper one of the foundations of a democratic system, namely the possibility for the electorate to hold power holders accountable for their actions (Thompson 1980;Hood 2007;Bovens 2010). The reason for this is that anticipatory blame-avoidance behaviours, such as spreading responsibility across several actors or increased protocolization in decision-making, often makes it difficult for the electorate to trigger informative and productive discussions with power holders about their previous actions. However, Hood (2014) also reminds us that blame-avoidance, under particular circumstances, actually can support democratic accountability. Blame-avoidance activities can result in clearer and more transparent procedures and guidelines for best practices, and also result in more adequate individual or organizational attribution of responsibility for certain actions and decisions. In these cases, blame-avoidance activities will improve democratic accountability mechanisms instead of hampering them. Nevertheless, these cases are exceptions in a world marked by accountability cultures preoccupied with sanctions (Mansbridge 2014), and therefore it is more likely that anticipatory blameavoidance behaviour weakens democratic accountability mechanisms. Our study indicates that blame-making can be an important strategy for increased democratic participation of marginalized groups in the short run, but the effect on democracy in the longer run seems to be more dubious.
By zooming in on non-elite blame-makers, we have in this study explored a new route of inquiry for the blame literature to engage in. We hope that the insights from this comparative study are useful for understanding the phenomenon of non-elite blamemaking, and that they may have relevance also for settings beyond Swedish asylum policy. In addition, we hope that future studies can pursue this route even further, perhaps by bringing insights from social movement studies into the blame-game literature's work on non-elites' strategic attempts to change policies. Social movement studies give us deep knowledge on non-elites' struggles and strategies to achieve policy change, while blame-game research has developed important insights about what happens at the receiving end of blame-attributions, namely how political elites respond to being blamed. We think that integrating, for example, elements from the rich work on political opportunity structures into the blame-game literature (cf. Kitschelt 1986), and bringing insights from blame-game research into social movement studies, would advance and broaden our knowledge about the function of blame in political life. In sum, by looking at blame beyond political elites' struggle for power within the institutionalized political arenas we can start to formulate new insights about the broader functions of blame in modern democratic societies. We look forward to the elaboration of these ideas in future research.