What Happens After the Storm? Investigating Three Conditions Under Which Local Governments Change Integrity Policy After Scandals

Integrity violations often generate political scandals, but what conditions determine whether this eventually leads to the improvement of integrity measures and policies? This article analyzes integrity scandals of 17 Dutch municipalities in the period 2006–2010 using qualitative comparative analysis. Three conditions are taken into consideration: media attention, presence of external audits, and involvement of civil servants in the violation. First, the results describe that in nearly half the cases nothing changed at all after a scandal. Second, this study explains why integrity policies changed—or remain the same. The analysis shows that changes are more likely when other civil servants are involved in a scandal and that the presence of an external audit helps to create a window of opportunity for change. On the other hand, more intense media coverage was not a sufficient condition for change in integrity policy. The study concludes that a scandal alone is not enough to create a window of opportunity for change, but that particular conditions need to be present—most notably external audits and involvement of other civil servants in the violation. This study uses past integrity cases (2006–2010) and investigates whether they have led to organizational changes. To uncover causal patterns, cases were analyzed using qualitative comparative analysis (QCA). Before going into detail about what QCA is, the various instruments that were employed to collect data are explained.

conditions preceding the scandal (Hoekstra & Kaptein, 2014;Maesschalck, 2002Maesschalck, , 2005. However, little is known about the conditions under which scandals actually cause changes in organizational integrity measures. To fill this dearth in the literature, this article focuses on three specific conditions-two external and one internal-that may cause change in the integrity policy of a local government organization. The external conditions that are investigated are the influence of media attention and external audits. Literature on policy change posits that these external pressures can create focusing points that bring together politics, policies, and problems (Kingdon, 2003) or punctuate otherwise stable policies (Baumgartner & Jones, 1993). For instance, if media pressure mounts on a municipality and an external audit is commissioned by the local political council, there is a great deal of pressure for the organization to make changes accordingly (e.g., Thompson, 2000). On the internal side, an important condition is the involvement of civil servants in a scandal, because this hints at more systemic misbehavior (Dubnick, 2003). If an integrity violation is more than an "isolated" political scandal, the involvement of multiple civil servants makes it more likely to be a systemic problem that needs to be addressed, and improvements in integrity measures are probably needed to address this. Therefore, the third condition relates to the involvement of other civil servants in an integrity violation in an organization.
Of course, it should be acknowledged there are other conditions that may play a role in the effect of political scandal on integrity changes in local government. For instance, there might be an organizational culture (Johnson & Cox, 2004) or (a lack of) ethical leadership (e.g., Bai & Morris, 2014), which can both influence organizational responses after a political scandal. However, in this study-being a first step to better understanding the consequences of political scandal for organizational change-only a limited set of conditions is taken into consideration. By looking at conditions from the institutional environment (media and external audits) and the organization itself (involvement of multiple civil servants), this article focuses on three "most likely" conditions. In other words, these conditions are likely to play a role, although it is unknown in what combination and under what circumstances. If even these conditions are not relevant, future research can more safely ignore them and focus on other aspects. Yet because as yet so little is known, providing empirical evidence on a set of "most likely" conditions is valuable to improve our understanding of what happens after an integrity scandal. The following question is central to our endeavor: Under what conditions do perceived integrity violations lead to change in the integrity and ethics policy of local government?
This article does not simply assess conditions and their effect on change separately, but also examines whether particular configurations of conditions have an effect. For instance, media hype after an integrity violation may not by itself change anything in the organization, but when there is both media attention and civil servant involvement, there may be reason for decisionmakers to adjust and tighten integrity standards. Therefore, this article assesses the combination of internal and external organization dynamics that are triggered by integrity violations to understand what happens "after the storm." Seventeen cases were analyzed in which Dutch local aldermen resigned due to an integrity violation. In the Netherlands, aldermen are appointed officials of the executive board of the municipality. They take care of day-to-day decisions and management, and are, practically speaking, the most important political officials in the municipal government. In order to investigate this, a document analysis and a small-n survey were used to gather rich and focused data: rich in the sense that there was much room for qualitative, descriptive information in the survey, WHAT HAPPENS AFTER THE STORM? and focused because real integrity cases were used to ask practitioners (city clerks) what the consequences of this particular scandal were for their organization. Moreover, the data were complemented with information from open sources, such as websites and newspapers, to gather background information concerning each case. The data, gathered from both surveys and open sources, were analyzed using crisp-set qualitative comparative analysis (QCA) (Ragin, 1987(Ragin, , 2000. One of the advantages of QCA is that it makes it possible to assert the effects of combinations (called configurations) of conditions. QCA is a research method that provides an instrument for systematically analyzing medium-n samples and enables analysis of the contribution of combinations (configurations) of conditions to a certain phenomenon. QCA has not been used frequently in the field of integrity issues within political systems (but see Six & Lawton, 2013, for an exception).

THEORETICAL FRAMEWORK
When studying integrity violations, many scholars of political and legal science have focused on corruption (e.g., Klitgaard, 1988;Rose & Heywood, 2013;Rose-Ackerman, 1999;Treisman, 2007). However, a focus on corruption when trying to capture integrity is somewhat narrow and leaves out important questions regarding honest and ethical behavior (Lasthuizen, Huberts, & Heres, 2011;Menzel, 2005;West & Berman, 2003, pp. 384-385). More recently, public administration scholars have opted for a broader approach to integrity. In this article, too, integrity is viewed as a broader phenomenon which is not limited to corruption. For this reason, the definition of integrity used is as given by Lasthuizen et al. (2011, p. 387): "the quality of acting in accordance with relevant moral values, norms and rules in society." This definition pertains to integrity as behavior that is in accordance with relevant norms and values in society, which means that there is an innate sense of subjectivity when assessing whether behavior is in accordance with norms, values, and rules, especially since norms and values change over time and differ from organization to organization. This article, therefore, examines perceived integrity violations: cases in which a behavior by a politician is generally perceived to be a deviation from relevant norms, values, or rules in society.
This study focuses on changes relating to integrity policy. These changes may prevent deviant behavior in the future, which in itself can be related to the aim of (re)gaining public legitimacy. Literature on ethics management shows what kind of approaches and practices are identified and how these lead to ethical behavior. Menzel (2005) argues that there is no "one shot instrument" to improve ethics. A variety of instruments and reforms are necessary in order to achieve an organization of integrity. This study, therefore, incorporates a broad range of integrity policies, both on the formal and the informal side (Berman, West, & Cava, 1994). Formal-"hard" control-mechanisms are the rules and regulations concerning the detection and sanctioning of integrity incidents. In addition, one could think of adoption of organizational rules that promote ethics, such as disclosure of outside activities. Informal-"soft" control-mechanisms are the investments on a human level, such as moral competence building, training, and ethical leadership, that can be beneficial for a decent working climate. Moreover, informal strategies involve positively influencing behavior; for example, trying to influence behavior through role models or improved communication.
But why would perceived integrity scandals trigger change? Literature on agenda setting is useful to better understand this. Kingdon's (2003) streams model postulates that crises can create a window of opportunity to change existing institutional structures. According to Kingdon, there are three streams in politics that are relatively independent, and when these come together, change is possible. One stream consists of (perceived) policy problems, such as enduring misbehavior in a local government. A second stream is the policy stream in which solutions for the problem "float around." Third, there is the political stream, in which the governmental agenda is formed. This stream consists mostly of visible political actors, such as political leaders and high-ranking civil servants. These three streams meet when there is a window of opportunity. A window of opportunity opens, for instance, when there is a sudden shift of mood. In a similar vein, Baumgartner and Jones's (1993) theory on punctuated equilibrium states that under normal circumstances, policies are relatively stable and only change marginally, because of (among other things) vested interests and existing institutional cultures. Real changes in policy will mostly occur-are punctuated by-changes in conditions, such as shifts in society and government. Integrity scandals can be serious threats to organizational legitimacy and can therefore be expected to create windows of opportunity or punctuate integrity policy.
Case study research applying these theories on integrity policy has been carried out by some authors (e.g., Hoekstra & Kaptein, 2013, 2014Maesschalk, 2002). Kaptein (2013, 2014) analyzed the development of Dutch national integrity policy over time and found that integrity breaches are an important condition for national policy change. In their study on the development of the Dutch central government's integrity policy from 1990 to 2010, they conclude: "The potential for fundamental policy change is created particularly when breaches of integrity occur that policy and decision makers, rather than treating them as isolated incidents, take seriously and seize to address structural weaknesses from a broader organizational perspective" (Hoekstra & Kaptein, 2013, p. 24). Maesschalck (2002Maesschalck ( , 2005 applies a combination of the theories of Kingdon and of Baumgartner and Jones in an in-depth case study. In doing so he shows that integrity-related scandals can function as wake-up calls for organizations, while he points out that to really understand the effect of a scandal or integrity breach, one should concentrate on the context and not place too heavy a focus on the scandal itself. This conclusion fits neatly with Kingdon's window of opportunity, where the integrity scandal can function as a window to bring together problems (structural weakness in organizational integrity), solutions (strengthening integrity policy), and political actors that address this. This also fits with the idea that the outrage occurring after a scandal punctuates integrity policy and leads to change.
The news media are an important external actor that plays an important role in exposing integrity violations of local governments and thus affecting their legitimacy. Media influence the political context within which government leaders act, through their focus on and treatment of particular issues, and as such can be a window of opportunity (Kingdon, 2003;McCombs & Shaw, 1972;Thompson, 2000). Active media that exert pressure on municipalities may increase the feeling that organizational change is necessary. The relationship between media and change has been investigated, for example, in the case of government transparency. Empirical studies have indicated that increased media pressure is associated with more transparent websites and stronger freedom of information Laws (Grimmelikhuijsen & Welch, 2012;Relly & Sabharwal, 2009, p. 149). Likewise, in response to a great deal of media coverage, local governments may feel forced to improve their integrity measures.
The second condition in the institutional environment of (local) government is external auditing that investigates and reports on integrity violations. Regularly recurring external audits are common in many areas of the public sector (e.g., Power, 1994), but in the case of integrity violations external audits often are on an ad hoc basis. When a violation is discovered, the local council may retain an external bureau to gather factual evidence of what precisely happened. The reports shed light on "what happened," but often also provide certain recommendations for the organization under investigation. Again, this could contribute to setting the agenda such that organizational integrity policies can be improved (e.g., Hoekstra & Kaptein, 2013).
Next to these external institutional conditions, the internal organization matters too. An integrity violation that triggers change can be thought of as the act of a single politician, but this individual "blame" may be the tip of the iceberg. Policy change may occur if a breach is viewed not as an isolated case but as something which addresses structural weaknesses from an organizational perspective (Dubnick, 2003;Hoekstra & Kaptein, 2013). For example, if multiple civil servants within the municipal bureaucracy knew about a violation or have been actively involved in it, this implies that misconduct is more systemic and transcends the isolated violation of a single politician. In such cases, solutions or changes in standards to improve integrity conduct are needed. In terms of Kingdon's streams model discussed above, this is the problems stream; if the perceived problem is stronger, it is more likely that actors in the political stream will act and change integrity policy.
In sum, the three conditions identified here are: (a) media attention, (b) an external audit, and (c) involvement of multiple actors that are involved in the violation as a representation of the integrity culture in an organization. Qualitative comparative analysis is used to distinguish and analyze configurations of conditions. One might expect that a certain combination of presence/ absence of variables could be predictive of a probable outcome. The following configurations are expected to have an effect on integrity-related policy change: Hypothesis 1: Presence of a great deal of media attention and external audits will lead to changes in integrity-related measures.
As argued above, media attention can be an important focusing event for government. If the news media heavily report on an integrity issue, while at the same time an external audit has been carried out, the pressure to make changes, from the institutional environment of a local government, is likely to be very high, which most certainly will bring various streams together in a window of opportunity (Kingdon, 2003). Therefore, if both conditions are present, this is expected to lead to changes in integrity-related measures.
Hypothesis 2: Presence of heavy media attention and civil servant involvement will lead to changes in integrity-related measures.
Hypothesis 3: Presence of external audits and civil servant involvement will lead to changes in integrity-related measures.
In a local government organization in which multiple civil servants have been involved in a violation, there is relatively high need for improvement of integrity-related measures. However, without external pressures, such as an audit or media attention, this involvement may continue by covering up misbehavior. Either external audits or media attention is needed to reveal such misbehavior and so to force the local government to adopt acceptable norms and values. External pressure and civil servant involvement are expected to amplify the perception that there are systemic integrity problems in an organization, increasing the chance of an open window to change integrity policy (e.g., Hoekstra & Kaptein, 2013).

METHOD Design
This study uses past integrity cases (2006)(2007)(2008)(2009)(2010) and investigates whether they have led to organizational changes. To uncover causal patterns, cases were analyzed using qualitative comparative analysis (QCA). Before going into detail about what QCA is, the various instruments that were employed to collect data are explained.

Phase 1
First, a list of all political leaders in the Netherlands (excluding mayors) who resigned between the years 2006 and 2010 was consulted to select cases of resignation that were reported to be related to integrity violations. This list was compiled by the Dutch journal for local government politicians (Binnenlands Bestuur) and provides a comprehensive annual overview of all resignations. The classification by Lasthuizen et al. (2011) was used to identify cases of integrity violation. This classification is comprehensive and provides clear definitions for an array of unethical behaviors, including corruption (bribing and favoritism), fraud and theft of resources, conflicts of interest (through gifts, services, assets, or promises taken, or through sideline activities), improper use of authority, misuse and manipulation of information, indecent treatment of colleagues or citizens, waste and abuse of organizational resources, and misconduct in private time.
Although the definitions mentioned by Lasthuizen et al. provide quite clear guidelines on how to classify a violation, both authors were involved in the coding process to ensure intercoder reliability. The following procedure was used to improve intercoder reliability: First, five cases were coded in by both in conjunction. Information in publicly available sources (newspapers, web pages) was obtained about the violation. The coding of these five cases showed that there was consensus about how to code each type of integrity violation. Furthermore, a set of criteria was developed during the coding of the first five cases on how to code each condition. The exact criteria for the conditions are explained below in the section on operationalizations. Next, each author coded six cases separately which meant that all 17 cases were coded. After this, these six cases were reviewed by the other author to cross-validate the codings. In the end, the cross-validation did not lead to substantial adjustments.

Phase 2
A survey was designed with both open-ended and closed-ended questions for each selected case. These surveys were distributed among city clerks and were used to cross-validate the information collected from news websites and Binnenlands Bestuur. In the survey, respondents were first shown a short case description to make sure they remembered the past situation. In the questionnaire, respondents were asked about the context of the violation and the resignation that followed. In addition, respondents were asked to comment on changes that had occurred as a result of the integrity violation.

Operationalization
The theoretical framework hypothesized that the following conditions facilitate change: media attention, external audits, and involvement of other civil servants. All variables were assigned a score of 0 (not present) or 1 (present), based on information provided by respondents in the survey and information from available open sources. By using a combination of sources, a more reliable coding of the variables was established.

Media Attention
The extent of media attention was coded using two sources. Respondents were asked, "Did the integrity violation attract attention from local and/or national news media?" Next, the researchers used the newspaper search engine LexisNexis to cross-validate what kind of media attention occurred during the time of the scandal. In cases where media attention by national newspapers was present, variables were coded as 1, and as 0 in cases where only local attention was present.

Audit Reports
The presence of audit reports was asserted based on two sources. Respondents were asked, "Was a report written by an independent investigation bureau?" These types of reports are often part of the discussions held in the open meetings of a local council. Because such reports are an important part of the council's deliberations, they are often disclosed to the public. Therefore, the existence of reports was checked by searching for them on municipal websites and Google to cross-validate respondents' answers. Cases where an external report was present were coded 1; cases where a report was absent were coded 0. To be coded as 1, a report had to be available on the web or be mentioned in articles.

Organizational Involvement (0 ¼ No Involvement of Other Civil Servants, 1 ¼ Involvement of Other Civil Servants)
The authors used the following question to assess this item: Respondents were asked, "To what extent were civil servants of the administration involved in the violation?" Answers could vary from none to passive (e.g., permitting but not initiating violation) or active involvement (initiating and/or actively participating in violation). The variable was coded as 0 if no civil servants were involved, and 1 if a civil servant was involved in a negative way, thus contributing or permitting a violation.
Respondents were asked what measures were implemented or adjusted in the aftermath of the violation (e.g., code of conduct for politicians, code of conduct for civil servants, regulation for handling confidential information, changes in organizational structure/culture, new screening tools for employees). If one or more of these changes occurred, this variable was coded 1; if no changes were reported, it coded a 0.

Sample
Questionnaires were administered to the council clerk of each municipality that had encountered integrity-related resignations between 2006 and 2010. In total, 33 questionnaires were distributed to these respondents. City clerks were targeted as respondents because they are regarded as "knowledgeable outsiders" in Dutch municipal political processes. A council clerk in the Netherlands assists the local council but is not part of the council, and, as such, is not part of decision-making processes. Furthermore, the council clerk is up-to-date about what is happening in the political leadership but has no formal or subordinate relation with it. In none of the cases were there indications of a personal involvement of the clerk in the integrity violation under study. Other data sources (e.g., websites, search engines) were used to cross-check the survey outcomes where possible (see the preceding section on Operationalization for details).
Every respondent was approached individually by phone, and after one or more reminders, 24 individuals (72.3%) responded, which is a satisfying response rate compared to other survey response rates, which tend to be from 20% to 30% (e.g., Sheehan, 2001). From these 24 cases, eventually 17 were selected. In the seven cases that were omitted, the respondents claimed that in their case the resignation was not related to an integrity violation. These claims were then checked using Websites and search engines, such as Google and LexisNexis. These cases all regarded the misuse of information; for instance, informing the local council too late. This was considered a political-not an integrity-issue, and therefore these seven respondents were withdrawn. Of the 17 respondents, 77.8% were male, with an average age of 50 years. On average, the city clerks had spent 6.8 years on their current job and 22 years in public service.

QCA Analysis
Qualitative comparative analysis (QCA) was developed by Ragin (1987Ragin ( , 2000 and is a suitable technique in medium-n case analysis. QCA was developed to facilitate and explicate the dialogue between theory and empirical evidence (Ragin, 2000). By using QCA, the iterative nature of qualitative study becomes more transparent. However, the most promising advantage of QCA, as compared to other analytical techniques, is that it acknowledges equifinal causal patterns; namely, that different sets of causal conditions can lead to the same effect/outcome. The so-called "truth table" is a key element in any QCA analysis. It lists all logical possibilities of conditions. In this study, three conditions were distinguished, which means there were 2 3 ¼ 8 possible configurations. Next, the researchers used the selected cases to investigate which of these configurations led to the outcome (i.e., change in ethics and integrity measures) and which combinations did not. In doing so, one can assess which combination of conditions (in other words, which configuration) leads to change.

RESULTS
The results are presented in two steps. First, the descriptive results are provided about the nature of the policy changes that occurred after the political resignations. Second, the relation between WHAT HAPPENS AFTER THE STORM? the conditions and the potential for change in organizational integrity and ethics policies is analyzed.

Descriptive Results
For the first step, respondents were asked whether the specific integrity violation had led to specific changes in the organization or the organizational policy on integrity, such as introduction of codes of conduct, additional screening of political leaders, or organizational changes. Figure 1 shows the frequency of each reported change.
Seventeen cases of integrity-related resignations were identified. These has resulted in a variety of organizational and/or integrity policy changes. Some cities reported more than one change, so the total number of changes does not add up to 17. Figure 1 shows that eight times no changes were implemented, but in one of these cases the respondent thought a change was necessary nevertheless. Most frequently, codes of conducts were improved (five times) or employees were moved to other jobs (three times). The revision of tasks of employees was another response to integrity violations that was mentioned two times. In addition, cities reported to have introduced an obligation to disclose side jobs and to have improved screening of newly appointed administrators were mentioned two times. Under "other," respondents entered "softer," less tangible, changes. For instance, a city clerk highlighted that "the coalition now has more attention for administrative leadership." In conclusion, both hard and soft measures were implemented, such as increasing awareness about integrity issues, increasing attention to administrative management by the political leadership, and making arrangements about information provision by politicians to the local

GRIMMELIKHUIJSEN AND SNIJDERS
council. The type of violations and the kind of change this caused in individual municipalities was also investigated (Table 1). Table 1 shows that five cities implemented more than one change after an integrity violation. Moreover, it gives an overview of the main types of violation that caused a politician to resign. Five times misconduct in private time, such as being caught for drunk driving, was the main cause. Favoritism occurred three times; information misuse occurred five times.

QCA Analysis
The second step in our analysis is to move from description to an analysis of the relationships between the conditions (media, audit, organizational involvement) and the potential for policy and organizational change. To do so, the complexity of the cases was reduced by constructing binary variables-for instance, whether a condition was present or not (1 or 0) and whether a change was present or not (1 or 0). This resulted in a so-called raw data matrix.
Looking at the raw data in Table 2, various combinations of conditions and outcomes can be observed. There are 17 cases in which an alderman resigned because of a perceived integrity violation. Looking at the overall occurrence of the conditions, Table 2 shows that there was national media attention in ten cases, an external audit was conducted in nine municipalities, and seven times there was organizational involvement in the perceived violation. In eight cases, this led to some change in the organizational integrity policy. To see which conditions in which combinations led to change, it is useful to make a configuration table (Table 3). Table 3 shows that from all the logically possible configurations (2 3 ¼ 8), one configuration did not occur in the data, namely a combination of media attention, no audit, and involvement of others (configuration 1-0-1).
In four cases all conditions were present (1-1-1). These are often "high-profile organizational integrity violations" that attract a great deal of attention from media, both locally and nationally. As one might expect, such cases lead to changes in the organization to prevent such negative exposure in the future. However, in one of these cases no change was reported. In this case (case 13 in Table 2), the external audit was studied in more detail, and this report showed that the organization had sufficient rules and regulations regarding integrity already, but that they were not sufficiently enforced. Furthermore, the report made no recommendations in the direction of   stricter enforcement, which could explain why no changes were implemented by the organization.

TABLE 3 Configurations of Conditions and Outcomes
A second configuration might be qualified as "high-profile political integrity violation" (1-1-0). These cases attract a great deal of media attention and involve an external audit of the violation, yet do not involve other civil servants. In this sense, it is an isolated integrity violation by one politician. In two cases, changes in the organization were considered necessary by respondents. However, in one case (case 15) no organizational change was carried through despite the existence of an audit report. However, just as in the earlier-mentioned case 14, no recommendations were made in this report. 1 The third configuration (1-0-0) is a variation of the high-profile political case, as it involves no other civil servant and there is much media attention. The difference here is that no external audit was commissioned. Remarkably, despite the heavy media exposure, these municipalities did not seek to change their integrity policies. Furthermore, two singular configurations (0-1-1 and 0-1-0) were found. In both configurations, there was a change in organizational integrity policy. In both configurations, external audits played some role. This suggests that external audits can be an important lever for the administrative leadership to force changes in the organization. Looking at the hypotheses, this refutes Hypothesis 2 (heavy media attention and involvement of other civil servants), and provides support for Hypothesis 3 (external audits and involvement of other civil servants). This is different in the next type of configuration, which one might call "silent integrity organizational violations": perceived violations in which a politician steps down in relative silence. In two cases, there was organizational change despite the absence of media attention and external audits (0-0-1). These two cases showed that perceived violations in which other organizational members were involved, but which do not attract much media attention, lead to change. As opposed to what was hypothesized, it sometimes seems to help if the external environment is silent.
Finally, there are three "silent political integrity violations" (0-0-0), which have little media attention, no external audit, and no involvement of others. In such cases there has been a clear offense by a politician but it is seen as an isolated individual violation. Typically in these cases, it entails behavior that is outside the political sphere, but is considered so condemnable that it reason to step down. Cases in this configuration do not lead to changes in integrity-related measures.

CONCLUSION
What can be concluded from these results? First, there are configurations with limited media exposure, and in these cases, changes only occur if there are also other organizational members involved in the violations. Hence, in cases where there are multiple civil servants involved, hardly any external pressure is needed to force organizational change. Second, what seems to be crucial in most cases is the presence of an external audit report. In these cases, where there is organizational involvement and an external audit, change is instilled, and the degree of media attention seems to be of minor importance. This is also highlighted by the qualitative data that were provided by respondents in the free text of the survey, which helps to better interpret the findings from the QCA analysis. The WHAT HAPPENS AFTER THE STORM? media condition was found to be an important external pressure, but often not enough for change. According to respondents, media attention was mainly focused on the period preceding the resignation of the alderman. A city clerk highlighted this: "The local press paid much attention to the incident, which led to enormous pressure from society, which forced the alderman to step down, even before the local council could give it a vote." In another municipality a city manager reported that there was elaborate regional media attention about how the alderman in a case "arranged business" for a shady entrepreneur. After the resignation, attention is often limited to an announcement in a newspaper, as the same respondent says: "After that [the initial report], there were some short articles about 'possible resignation' and 'alderman resigns.'" Other respondents also indicate that there has been moderate to heavy media attention, mostly by local newspapers or broadcasters, and that this pressure often contributes to resignation. After resignation, attention quickly wanes. In conclusion, media attention has an ambivalent role: it does not necessarily lead to change directly, but does seem to catalyze resignation, which in the end can be a driver for change.
The external audit report was found to be a more decisive condition. These reports often provide recommendations to the organization on to improve integrity safeguards and organizational structures. The following statement by a city clerk illustrates how reports from independent investigators can be important for changes in the organization. "The report and the process through which this was written … positively contributed to clearing the air. As a result, the recommendations [from the report] were used to work on improvement of the internal relations and information provision." External audits are indeed a decisive stimulus for organizational change to occur.
The respondents also provide insightful examples of how organizational involvement is related to change. One example regards a case in which a politician transferred public money to befriended organizations without following proper procedures. A respondent said: "A civil servant … sent the letter [to the befriended organizations] that was signed by one alderman only, without reporting that basic laws and procedures were violated." The incident led to increased screening and an (improved) code of conduct. Further, in another case in which a politician misinformed the council, civil servants also had a role: "Mistakes were made in advising the coalition, and they may have been an insufficient critical attitude toward the alderman." This, in the end, led to an arrangement about how the coalition should inform the local council about their activities.
This brings out a paradoxical situation. The first type of configuration (with organizational involvement) indicates that in more serious violations, it is sometimes better if there is not too much interference from the external environment. In other words, that there are not too many others who are "looking over the shoulders" of the local government. On the other hand, an external audit seems to be an important force in organizational change. The implications of these findings are discussed in the next section.

DISCUSSION
In this article, the following research question was posed: Under what conditions do perceived integrity violations lead to changes in the integrity and ethics policy of local government?
Two dominant configurations that contributed to a positive outcome were found. The first configuration is a combination of civil servant involvement and either limited media attention or heavy media attention. This means that the involvement is paramount, and not the magnitude of media attention. This partly supports the second hypothesis, which stated that a combination of heavy media attention and civil servant involvement leads to integrity-related changes. The second configuration shows a different pattern: civil servant involvement leads to change when there is an external audit. These findings support the first and third hypotheses, stating that a combination of media attention with an external audit (Hypothesis 1) and broader organizational involvement combined with an external audit (Hypothesis 3) are both conditions that cause change in integrity-related measures.
Based on this finding the following is concluded: First, integrity-related changes occur only if the violation is related to integrity violations in which multiple civil servants were involved, which may embody an organizational culture in which such a violation could occur. Second, changes do not always occur after heavy media attention and can also be fostered in cases where there is relative silence-where there is no big media spectacle, but only some attention by local news media. Third, organizational change can occur in the face of strong media attention, but only if there is a strong external force prescribing specific changes; only in case there is an external audit report will an organization adapt its integrity policies. Fourth, a general finding is that in more than half of the cases, no change occurred. In most of these cases, multiple conditions, such as an external audit, were absent. This means that a political scandal, indeed, does not automatically lead to changes in integrity policy or measures in local government (Maesschalck, 2002).
What are the implications of these findings for the academic debate on public integrity? Corruption and integrity scandals often cause politicians to resign, and many researchers have devoted attention to understanding the scandals themselves (e.g., Kaptein & Wempe, 2002;Kolthoff, 2007;Menzel, 2007;Van den Heuvel et al., 2010). Furthermore, some authors have already identified integrity scandals as important focusing points in integrity policy changes (Hoekstra & Kaptein, 2014;Maesschalck, 2002). This study systematically investigated what happens "after the storm": Is the scandal seized as an opportunity to improve integrity policy of the municipality, and if so, under what conditions? Remarkably, although others have noted integrity breaches as important conditions for change (Hoekstra & Kaptein, 2013, 2014, this study found that there is often "silence after the storm"; in eight integrity-related resignations, nothing changed. However, this is not always necessary, for instance, if the scandal is an isolated incident. That said, this indeed shows that a scandal does not create the window of opportunity that is needed for change, by definition. This was already noted by Maesschalck (2002Maesschalck ( , 2005, who subsequently speculated about whether Kingdon's theory was also able to explain the numerous cases where scandal was not followed by comprehensive policy changes. Based on various conditions that are linked with this theory a preliminary answer to this question is found. This study shows that a scandal alone is not enough to create a window of opportunity, but that particular conditions-most notably external audits and the existence of a systemic integrity problem-need to be present.
The findings have implications for practitioners too. To start with, focused attention after a resignation, like the presence of an external audit, facilitates organizational change. This means that keeping the topic of integrity on the agenda after a resignation is crucial to addressing the issue structurally and making changes in the organization. Nonethical behavior might not stop after a resignation if it was only the "tip of the iceberg." Specifically asking and looking for concrete recommendations to improve organizational integrity keeps the topic on the table, in the end leading to improved ethical behavior in the public sector.
This study is based on a limited number of cases, and therefore the conclusions should be interpreted with some caution. There are various features of this study that require further research in order to infer more firm conclusions about when scandals lead to integrity-related changes in an organization. First of all, only integrity-related cases in which a political leader actually resigned were selected. However, there could be cases in which there is unethical behavior but the political leadership is not blamed for this or does not resign. Another issue is the limited set of conditions included in the analysis. This was necessary because the total number of cases was low (N ¼ 17). Because each new condition in the analysis would mean an exponential accumulation of possible configurations, this would lead to too many possible configurations and a limited number of cases. As a result, it would have become very hard to distill any meaningful patterns from the data. Future research could include a broader or different set of conditions, such as organizational culture (Johnson & Cox III, 2004) and ethical leadership (Bai & Morris, 2014;Maesschalck, 2002, p. 188), to see if this affects local governments' responses after a scandal.
In doing so, scholars could either broaden the set of cases and include more contextual conditions, or narrow down the set of cases and do a more in-depth qualitative comparative case analysis. With regard to the latter, a condition such as media attention could be fleshed out more by assessing more than one dimension of attention. For instance, in-depth analysis could assess not only if media coverage is local or national, but also if it is short-lived or prolonged, of a personal nature or not, and so on.
Finally, this study was conducted in the Netherlands, which also potentially affects the results. In general, there are not that many integrity-related scandals in the Netherlands, which ranks as the eighth least corrupt country worldwide (Transparency International, 2014). This may have a profound impact on the ability to deal with scandals. For instance, a scandal in the Netherlands-even a relatively small one-may attract attention and lead to changes, whereas countries where unethical conduct is more endemic to the public sector may need different and much stronger conditions. Cross-national comparisons are necessary to determine whether our conclusions hold in other cultural contexts too.
This study is the first systematic exploration into the contextual conditions under which scandals lead to integrity-related change in local government. Corruption and integrity scandals can be detrimental to public trust in and legitimacy of government (Bull & Newell, 2003;DellaPorta, 2000;Van der Meer, 2009), and to prevent this from happening, firm integrity policies are needed. This study identified some potential conditions that may foster the improvement of these policies.