Lethal Negotiation: The Effect of Violence on the Likelihood of State-Organization Negotiation


I have investigated a few of the underlying preconditions for state ethno-political organization negotiation. Specifically, I explored the effect of organizational violence on the likelihood of reaching the bargaining table. Organizations and states act in a rational and strategic framework. Negotiation is most likely to occur when it is in the state’s best interest. Negotiation is in the state’s best interest when organizations operate at the extremes of violence, either very low or very high levels. Therefore, I argue that groups using intermediate levels of violence are less likely to experience negotiation with the state. I used a logistic regression on the MAROB dataset from 1980-2004 to test this causal relationship. Although I found no support for my argument regarding the effect of violence on negotiation, my data confirm other claims made in the literature. I close this study with a discussion of my findings and suggestions for further research.

Table of Contents: 


    Ethno-political groups vary both in the extent to which they use violence as a tactic and the degree to which they are able to successfully negotiate their grievances with their host states. On April 10, 1998, the major political parties of Britain, Ireland, and Northern Ireland sat down at the bargaining table and walked away having crafted the Good Friday Agreement which ostensibly ended the Troubles – an exhausting period of ethno-nationalist conflict marked by widespread violence. In 2009, the Tamil Tigers were defeated following a protracted civil war. While there are incontrovertible differences between the two conflicts, they share some fundamental similarities. Both cases involve state and sub-state actors and experienced extreme organizational violence. The fact that some ethno-political conflicts end with negotiated settlements while others do not presents an interesting dichotomy. Many ethno-political organizations embrace a policy of violence to further their causes even though most states have a firm policy against negotiating with violent groups, often deeming them “terrorists.” This begs the question:Does the level of violence that an ethno-political group uses have an important effect on the likelihood of negotiations? Some organizations are certainly not concerned with negotiations, but for those seeking concessions from the state, does violence have a significant impact? There has been little research examining the preconditions of negotiation, specifically the consequences of organizational violence. The goal of this study is to further clarify the complexities of this issue and to provide a new perspective on an old question.

    I argue that states and ethno-political organizations each act strategically to reach their goals. When an ethnic group mobilizes into a political organization, it seeks the strategy that provides the best chance to realize its ambitions. For some, violence is strategically the best option while for others it is not. An organization must examine its circumstances and perform a cost-benefit analysis on the utility of violence. Similarly, a state must make the strategic decision whether to negotiate or respond with violence. I argue that generally, the level of violence that a group employs has a notable effect on the strategic decision-making process used by the state.

    This paper will begin with a comprehensive survey of the relevant literature to build a foundation of the knowledge on the possible causal relationship between level of violence and negotiation. The second section is an explication of my theory. This is followed by an explanation of the empirical model. Finally, I discuss the results of the statistical analysis and offer some suggestions for further research.

    Literature Review

    There has been an abundance of work on both civil conflict and negotiation, but significantly fewer studies that investigate the relationship between the two. Those studies that have looked at the two phenomena in conjunction are most often concerned with areas such as the conditions that favor conflict or the peace duration after negotiated settlements (Downes 2004; Duyvesteyn and Schuurman 2011; Licklider 1995; Mason et al. 2011; Walter 2004). These studies are useful in advancing the field, but there is one piece of the puzzle that is missing from them−the conditions that must be met in order for negotiation to occur. Several scholars have identified an increasing need for further study in negotiation and civil conflict (Mandel 2007; Miller 2011; Sanchez-Cuenca and de la Calle 2009; Walter 2006, 2009). This study is necessitated by the apparent gap in the literature of negotiation and conflict. The current literature tends to focus on case studies and is often circumstantial rather than broad and generalizable. In this cross-national study, I argue that the level of violence of a group is a good indicator of the prospects for such negotiation

    Conditions for Conflict

    Many scholars have sought to explain the onset of civil conflict (Collier and Hoeffler 2004; Collier, Hoeffler, and Rohner 2009; Fearon and Laitin 2003; Gurr 2000; Regan and Norton 2005; Sambanis 2001; Walter 2004). Several theories regarding conflict onset have competed over the years, yet no universally accepted explanation has emerged. A large distinction exists between theories espousing motivational causes and feasibility causes (Collier, Hoeffler, and Rohner 2009). Motivational explanations include greed (Collier and Hoeffler 2004; Regan and Norton 2005) and grievance (Gurr 2000; Sambanis 2001), while feasibility explanations include conditions that favor insurgency (Fearon and Laitin 2003) and violent relapse occurring after previous civil conflict (Walter 2004). Each competing theory seems sufficient and valid, meaning scholars cannot reach a general consensus on the cause of civil conflict. Since scholars cannot determine any combination of practical indicators, they cannot accurately predict or prevent the outbreak of civil conflict. Therefore, authors must focus on conflict resolution. Specifically, scholars must concentrate on negotiation rather than partition or military victory because it has generally been found to be preferable in terms of genocide, politicide, and human rights (Licklider 1995).

    Negotiated Settlements versus Military Victory

    One section of the negotiation literature contrasts peace duration following negotiated settlements and military victories. Several studies have concluded that military victories provide a more durable peace than negotiated settlements (Downes 2004; Duyvesteyn and Schuurman 2011; Licklider 1995; Mason et al. 2011; Walter 2004). These authors argue that decisive victories lead to a stronger peace because the winning side is able to forcibly disarm its opponent, eliminating any opportunity for a relapse into violent conflict. Negotiated settlements, however, are unable to mitigate the possibility of recurring violence because neither party secures a monopoly on force.

    The aforementioned conclusion has gained widespread traction and is referred to as “conventional wisdom” (Mason et al. 2011, 173). Nevertheless, recent studies present a strong counterargument (Hartzell and Hoddie 2007; Mason et al. 2011; Mattes and Savun 2010). These studies claim that negotiated settlements may produce a peace just as stable as one arising from military victory. New findings indicate that while peace generated by a negotiated settlement is relatively fragile in the first five to ten years after negotiation, it is much more stable than peace stemming from military victory after ten years (Mason et al. 2011). Another theory is that negotiated settlements are most likely to succeed when information imbalances are eliminated through the use of third-party mediators (Mattes and Savun 2010). In this theoretical model, when both parties in a civil conflict make complete information available, the agreements significantly diminish the risk for relapse into violence. Many of the studies critical of the peace-duration produced by negotiated settlements were generated using data from post-Cold War civil conflicts. Therefore, it is reasonable to assume that the data was simply not mature enough to produce accurate results.

    Spoilers, Institutions, and Power-sharing

    Several scholars have examined one of the greatest impediments to the peace process: spoilers (Greenhill and Major 2006; Kydd and Walter 2002; Stedman 1997). Spoilers are generally considered to be splinter-groups that disagree with the organization’s choice to engage in negotiation with the opposition. Their primary goal is to undermine the peace process by using violence to create mistrust (Stedman 1997). Spoilers present a considerable obstacle to negotiations because actors cannot proceed in the peace process if near-complete trust is not attainable (Kydd and Walter 2002).

    Obstructions to the peace process, such as spoilers, have necessitated the formulation of various solutions. Many studies have suggested that the best way to curb these issues is by establishing certain political institutions at the bargaining table (Hartzell 1999; Hartzell and Hoddie 2007; Hartzell, Hoddie, and Rothchild 2001; Walter 1997; Walter 2002). For instance, some authors argue that the key to mitigating these obstacles is the formation of power-sharing measures that allow all parties some influence (Hartzell 1999; Hartzell and Hoddie 2007). Others maintain that extensively institutionalized negotiated settlements function best (Hartzell 1999; Hartzell, Hoddie, and Rothchild 2001; Walter 1997). These arguments assert that negotiated settlements must be drawn to include tenable guarantees that persuade both parties to disarm, as well as security assurances for all parties involved. Some authors argue that the best way to assure the credibility of the guarantees is through a third party, such as an international organization or superpower (Hartzell and Hoddie 2007; Hartzell, Hoddie, and Rothchild 2001; Walter 2002). There is, however, one clear caveat in the literature: negotiating settlements that are highly institutionalized is challenging, especially after protracted conflicts where mistrust is high (Downes 2004; Hartzell and Hoddie 2007). Taking this difficulty into account, some scholars have theorized that partition may be a more advantageous strategy in these situations (Downes 2004).

    Separatists and Secessionists

    Negotiation with ethno-political organizations that seek some measure of autonomy or adhere to separatist ideology is fundamentally different than other types of negotiation. Scholars have gathered that it may be prudent for a government to go to war rather than negotiate with separatist or autonomy-seeking groups in order to maintain its territory and establish a deterrent for future groups to rebel (Huth 1999; Walter 2006). Clearly, states strategically choose whether to negotiate or grant concessions, yet the reasoning remains obscured. The level of violence used by a group must play some role in this strategic decision-making, but to what extent is still unclear.

    Strategic Decisions in Civil Conflict

    Several authors have written about the strategic decision of whether or not to negotiate (Greig 2012; Licklider 1995; Sisk 2009; Walter 2006, 2009; Zartman 1995). One reason often cited as rationale against negotiating is that it may strengthen the legitimacy of the organization (Licklider 1995; Zartman 1995). States do not want to “reward” violent organizations with concessions because it may incentivize violence, thereby engendering future attacks (Sisk 2009). By this logic, it appears that states should exclusively refuse to negotiate and suppress the group violently. One study, however, indicates that when a state negotiates, it is more often with a group that poses the greatest threat (Walter 2009). Logically, groups that pose great threats are likely to operate at high levels of violence. Thus, states must occasionally negotiate with violent groups in spite of conventional wisdom. Clearly, there is a debate in the scholarly literature on the strategic decision made by a state to negotiate.


    In my study, I argue that the level of violence that an ethno-political organization employs has an important effect on the likelihood of negotiations with the state. For the purposes of this work, I will define an ethno-political organization as any organization that “makes explicit claims to represent the interests of one or more ethnic groups and/or the organization’s members are primarily members of a specific ethnic minority” and “is political in its goals and activities” (Asal, Pate, and Wilkenfield 2008, 2). I argue that while both states and ethno-political actors make very calculated and influential decisions at different junctures in their interaction, negotiation is most likely to occur when it is in the state’s best interest to negotiate.

    The most critical moment for an ethno-political organization with any sort of grievance is the decision regarding whether or not to use violence. An organization will always act in its best interest while deciding whether or not to use violence, using a cost-benefit analysis familiar in rational actor theory. Groups that have the means to enact high levels of violence are likely to garner more attention than groups unable to properly militarily challenge the state. The state will recognize the threat posed by violent groups and swiftly seek resolution often in the form of a negotiated settlement (Walter 2009). It is possible that a group using violence as a tactic may solely be seeking to overthrow the state. If the organization’s primary goal is to supplant the existing leadership, the organization will shun negotiation because it does not fall in line with its central aim. Groups may additionally simulate negotiation in order to regroup or gain the upper-hand in the conflict (Duyvesteyn and Schuurman 2011). Violence, however, may alienate conservative members of the organization, leading to fragmentation and weakening of the group. Conservative associates of the group may feel that violence is not in the group’s best interest, and they can potentially opt to leave the organization to form one that embraces their preferred tactics. An organization may change strategies if violence as a tactic does not pay off. If an organization begins with violence and moves to seeking negotiation, I believe that the state will be less likely to reciprocate with negotiation. While the organization’s strategic choice to use violence plays a large part in determining whether negotiation will occur, there is not a simple causal link between the two. Rather, when an organization makes the decision whether or not to use violence, it gives the state the information it needs to decide on the prospects of negotiation.

    States must use strategic decision-making to determine whether to negotiate with organizations, and I argue that the state’s decision weighs most heavily in determining whether or not negotiation occurs. States will generally act in their best interest to make this decision, and it may be different on a case by case basis. The likelihood of negotiation is most heavily dependent on whether negotiation is in the state’s best interest. When making the decision, the state must identify the organization’s level of violence1as the most critical element. In other words, organizational violence profoundly influences a state’s decision by changing what is in its best interest, and a state’s best interest controls the likelihood of negotiation.

    Negotiation will be in the state’s best interest if the organization presents a viable challenge to the state’s rule. This would mean that a state is more likely to negotiate with a strong group than a weak one (Walter 2009). A state will only choose to negotiate if the organization presents a real threat to the state either in the present or future. If this is not the case, the state has no reason to concede. If the organization is seeking territory or autonomy, the state is less likely to offer concessions because it considers these things valuable. Thus, unless the organization is able to impart heavy costs on the state, it may be strategically pragmatic to go to war. The state, however, must be careful when deciding to open negotiation with a violent organization. If a state negotiates with an organization that uses violence, it may appear weak, possibly leading to an increase in both civil and interstate conflict. Also, a state deciding to negotiate may incentivize the use of violence as a tactic (Sisk 2009). If a state negotiates and ultimately concedes to a violent organization, it could engender future groups to adopt a similar strategy (Licklider 1995; Zartman 1995). On the other hand, if a state violently suppresses organizations that use violence as a tactic, it may stop future groups from resorting to violence. While different situations may produce different results, states will always act in their best interest when deciding whether to negotiate with ethno-political groups.

    I argue that negotiation is most likely to occur in two different situations, and they both occur when negotiation is in the best interest of the state. First, if a state recognizes an organization operating at a low level of violence with potential to become more violent, it will likely go to the bargaining table. A state may identify a group’s potential to grow more violent by considering its rhetoric and tracking its leadership. Organizations using threatening or aggressive language may be preparing to intensify any violent strategies. A change in leadership can cause a change in a group’s strategy regarding violent conflict. For example, a more radical leader may be more prone to using violence as a tactic to achieve the organization’s goals. Negotiation will be in the state’s best interest in this situation because it prevents increased violence and offers a settlement favorable to military victory in terms of violent conflict and human rights. Another benefit from the state perspective is that, by negotiating, it gains publicity from its citizens or the international community. Although the state may not offer bona fide concessions, the general public may observe negotiation and consider it a sign of legitimacy, justice, or decency. The state also does not have to worry about appearing weak or incentivizing violence because the group in this situation operated at a low level of violence.

    Generally, groups that use higher levels of violence as a tactic are strong groups that could potentially challenge the state militarily. Rather than risk military defeat, it is in the state’s best interest to negotiate with the organization. When the state is confronted by an ethno-political organization operating at a high level of violence, it will identify the organization’s potential to defeat the state in conventional warfare. The state will then recognize that the pursuit of a negotiated settlement is a far more favorable alternative than chancing its rule in a full-scale war. I argue that because a state will always act in its best interest, it will choose to negotiate rather than fight a war with potential to strip it of all power; in every case, some power is better than none. While the risk of incentivizing violence is high in this scenario, negotiation is preferable to the prospect of military defeat. Although organizations operating at a high level of violencewill have a better chance of reaching negotiations with the state than some groups, I argue that they will have a lesser chance than organizations operating at a low level of violence.

    On the other hand, organizations operating at intermediate levels of violence find themselves in a precarious position in terms of reaching negotiations with the state. Since these organizations do not use peaceful tactics or low levels of violence, the state is unlikely to negotiate in fear that it may incentivize violence as a tactic. Because organizations will mimic others that were successful in reaching the bargaining table, the state must only negotiate with violent groups as a last resort. Since groups using intermediate levels of violence do not present an immediate military threat, it is not in the state’s best interest to negotiate; the organization does not have enough strength to challenge the state in conventional war, so the state is not at risk of losing power. Groups operating at intermediate levels of violence also lose out on the possibility of the state negotiating solely for publicity. While I argue that states may negotiate with groups using little to no violence simply as a show, they will not negotiate with violent groups that do not present an equal military threat. Organizations using intermediate levels of violence fall in a unique spot; the organizations are too violent for the state to negotiate without incentivizing violence but not powerful enough to militarily challenge the state. For these reasons, I argue that organizations using intermediate levels of violence are less likely to reach negotiations with the state than any other organization.

    Hypothesis 1: Organizations using a high level of violence are more likely to reach negotiations with the state than organizations using intermediate levels of violence, but less likely than organizations using low levels of violence.

    I argue that timing and sequencing also play a large role in determining the likelihood of state-organization negotiations. As previously mentioned, negotiation is most likely to occur when it is in the state’s best interest to negotiate. There is, however, a certain dynamic change that occurs following the first negotiation. After the two parties reach the bargaining table for the first time, the state is more inclined to entertain negotiations a second time. After the second iteration of negotiations, the two parties are even more likely to experience a third negotiation. The likelihood of future negotiation continues to increase each time the state and organization begin discussions. The increase in the chances for negotiation stems from the fact that the state has reached some measure of comfort in negotiating with the group. Since the first negotiation occurred, it must have been in the state’s best interest to negotiate. After the first iteration of bargaining, regardless of how successful negotiations were, the state is predisposed to accept negotiation a second time. Each negotiation simply increases the trust between the two parties, encouraging further talks.

    Hypothesis 2: Each time negotiation occurs between the state and an ethno-political organization, the likelihood for future negotiations increases.

    This second hypothesis follows the framework of Gent and Shannon (2011) who argue that previous negotiation engenders future negotiation, likely under the control of a third-party adjudicator or mediator. I believe that the same theoretical structure applies to organizations that have previously reached negotiations with their host states.

    States and ethno-political organizations act in a framework of rational-actor and strategic decision-making when they deal with negotiations. In this very specific context, both the group and the state act in their best interests in order to achieve their goals. Negotiation is most likely to occur when it is in the state’s best interest to negotiate, which could begin to explain why negotiated settlements do not always end civil conflicts.

    Research Design

    To test my hypotheses, I investigated instances of negotiation and non-negotiation between states and ethno-political organizations in the Middle East and Northern Africa from 1980 to 2004 at the organization-year level of analysis. The data that I used was taken from the Minorities at Risk: Organizational Behavior (MAROB) dataset, an extension of the Minorities at Risk dataset. The MAROB examines the organizational characteristics of groups that make “explicit claims to represent the interests of one or more ethnic groups and/or the organization’s members are primarily members of a specific ethnic minority” and are “political in [their] goals and activities” (Asal, Pate, and Wilkenfield 2008, 1). To qualify for inclusion in my dataset, groups must also have been active for at least three consecutive years from 1980 to 2004.

    I have chosen to use this dataset for three primary reasons. First, MAROB includes data on organizations that form in the most contentious region in the world─the Middle East and Northern Africa. The Middle East and Northern Africa are home to a myriad of ethnic groups which makes this area perfect for studies on ethnic conflict. Also, because of the large variance in the levels of violence over the region, the sample will include an appropriate distribution of groups using different tactics. Some organizations operate peacefully while others resort to violence to accomplish their goals. The stark differences in the level of violence over the region will provide a good basis for generalization. Thus, the region delivers a large sample that I argue is representative for any state where civil conflict occurs.

    The second reason that the MAROB data is optimal for this study is the temporal component of the dataset. It contains some of the most recent data on violent civil conflict at the organizational level of analysis. By using more recent data, I was able to create a more accurate statistical model. While it would be impossible to create a perfect model, given the complex and circumstantial essence of behavior in states, recent data aids in the creation of a more useful approximation.

    Finally, MAROB includes both violent and nonviolent groups, adding an interesting dynamic to this study. Many studies of conflict neglect organizations that pursue their goals using peaceful tactics. This approach is fundamentally flawed because it ignores many groups that use legitimate strategies other than violence to achieve its aims. Excluding peaceful groups also significantly lowers the sample size, potentially skewing results. Including nonviolent groups also adds another level of violence to my analysis, which increases the specificity and scope of my study.

    Dependent Variable: Negotiation

    The dependent variable in this study was negotiation. I define negotiation simply as the onset of negotiation in any given year. For example, if an organization experienced any talks with the state addressing an organizational grievance, it was coded as having negotiated in that year. I also treated cases of mediation under the aegis of any third-party actor as negotiation for this study. While I did not sum the negotiations if they occurred in separate years, I did address the impact of multiple iterations of negotiation in Hypothesis 2 and with another independent variable.

    Negotiation can be fairly easily measured. For the purposes of this project, I treated negotiation as a dichotomous variable: either negotiation occurred, or it did not. I used the variable ORGSUCCESS from the MAROB dataset to represent the phenomenon of negotiation. ORGSUCCESS is an ordinal variable ranging from -1 to 4. I use groups coded as either 1, 2, or 3 in my study as groups that have successfully reached the bargaining table in that year. Codes 1, 2, and 3 represent the following outcomes, respectively: “state has negotiated with the organization in this year”; “state has made some concessions to the organization but not to its primary goal in this year”; “state has conceded to the organization its primary goal in current year” (Asal, Pate, and Wilkenfield 2008, 9). I recoded the variables coded as 1, 2, and 3 as signifying successfully reaching negotiation or 1, while -1 and 0 represent a failure to achieve negotiation, or 0. Codes of 4 represent an organization that has achieved its primary goal via state concession in a previous year, so I remove all observations where ORGSUCCESS is a 4. I choose to convert this ordinal variable to a dichotomous variable because it does not seem appropriate to rank negotiation in an ascending order. In this study, my objective was to investigate some of the causes of organizations and states reaching the bargaining table. Therefore, this study did not require an exploration of the events that occur after the two sides agree to begin discussions. Ranking the value of negotiation based on its outcome would fundamentally alter my argument.

    Independent Variables

    This study focused on a group’s level of violence as the primary substantive independent variable. I defined violence as the use of military force resulting in the death of any state authorities including military targets, state-affiliated police, or political officials. I used the MAROB variable ORGREB to represent the level of violence that the organization used against the state. This is an ordinal variable running on a 0-7 scale, with the codes corresponding to the following levels of violence, respectively: none reported; political banditry; small-scale campaigns of violence; local rebellion; small-scale guerrilla activity; intermediate guerrilla activity; large-scale guerrilla activity; and civil war (Asal, Pate, and Wilkenfield 2008, 24). Each of these categories contains specific parameters that must be fulfilled to be coded as such.2 I argue that negotiation will be most likely at the extremes of violence, low and high, while the likelihood of negotiation for groups using intermediate levels of violence will be the lowest. I chose to use the categories given by the ordinal variable because it captures the most information available from the dataset. For this study, I have chosen a measure that accurately depicts the tactics of an organization in seeking its goals.

    In order to account for the proposed curvilinear relationship between organizational violence and negotiation, I also included the squared term of the variable ORGREB. The inclusion of this variable allowed me to build a statistical model that accurately represented my theory that negotiation is more likely to occur when groups act at the extremes of violence toward the state. Another option was to recode organizations into arbitrary categories of violence, but this idea was flawed. If I were to regroup the ethno-political organizations into broader classifications, I would be sacrificing valuable specification and information. I choose to include the squared term to conserve information while still producing a model conducive to my theoretical relationship.

    To address Hypothesis 2, I have created a variable that counts the number of previous negotiations with the state that an organization has experienced. For example, if an organization reached the bargaining table with the state in 1980, it would have one previous negotiation with the state in 1981. I argue that, as the number of previous negotiations increases, the likelihood that negotiation will occur again increases. By counting the number of previous negotiations and including this count in my statistical model, I was able to determine whether earlier negotiation had a substantial effect on the likelihood of future negotiation.

    Control Variables

    In order to produce an accurate model, this study must account for several factors that may also play a role in determining whether or not negotiation between an ethno-political group and a state occurred. Territorial control has been shown to be a salient issue in interstate negotiations (Gent and Shannon 2011; Gochman and Leng 1983; Hensel 1996; Vazquez 1993). This has led me to believe that an organization controlling territory in its host state may have some explanatory impact. A state may respect a group that controls land because it necessarily has some measure of sovereignty and autonomy that detracts from the state’s legitimacy. This increased respect may then lead to an increase in the likelihood for negotiation. To control for this effect, I used a dichotomous variable from the MAROB dataset called ORGST9. Scholars have argued that the strength of leadership in an international conflict plays an important role in determining whether negotiation will occur (Rothstein 2007). Thus, I argue that the strength of an organization’s leadership structure has some impact on the chances of state-organization negotiation. A state is more likely to negotiate with a group that has an established leadership structure because the leadership must effectively convey any negotiated terms to its members. If a state were to negotiate with a group with weak or ineffective leadership, there is a great chance that the group’s members may disregard any provisions reached at the bargaining table. I have created a dichotomous control variable that codes organizations as having strong or weak leadership. In the creation of this variable, I have modified the LEAD variable from MAROB. An intuitive contributing factor is the legality of the organization. Some states have strict laws on congregations and groups in general, meaning that they are less likely to negotiate with a group that is breaking the law simply by existing. I use the dichotomous variable ORGLEGAL from the MAROB dataset as a control in my model. When observing negotiation, organizational popularity may have some measure of explanatory force. States may be more inclined to negotiate with groups that have the dedicated support of an ethnic group as a whole. To control for this effect, I use the ordinal variable ORGPOP from the MAROB dataset. ORGPOP separates groups into three categories: fringe organization; one of many organizations with support from an ethnic group; and dominant organization (Asal, Pate, and Wilkenfield 2008). Finally, regime type likely has an impact on the likelihood of negotiation. An oppressive, autocratic regime is less likely to negotiate with organizations than a democratic leadership. To account for this in my model, I created a dichotomous variable that determines whether an organization’s host state was democratic. I use the Polity2 variable from the Polity IV dataset, in which states coded as greater than or equal to 6 are considered democratic and all others are considered non-democratic.


    I have identified several important factors when deciding which type of regression was appropriate for this study. First, I have recognized the curvilinear nature of my theorized relationship between violence and negotiation as fundamentally important. Because of this curvilinear relationship, simple linear models and ordinary least squares regressions are not applicable. Another critical factor in my decision-making process is the nature of my dependent variable. The dichotomous nature of my dependent variable necessitates a model designed to handle a binary variable. Therefore, I have chosen to test my hypotheses using a logistic regression (Bonney 1987). The logistic regression allows for a curvilinear relationship and is used for non-continuous variables. I argue that it provides the best opportunity for statistical inference and prediction.

    Following convention, I use a threshold of p < .05 to determine significance and subdivide this into varying measures of significance using p < .01 and p <.001. To ensure the quality of my statistical analysis, I have chosen to use robust standard errors for all of my variables. For my hypothesis regarding level of violence and negotiation, in which I argue for a curvilinear relationship, I reject the null hypothesis only if the violence variable and the squared term of violence are statistically significant. For my second hypothesis, concerning prior negotiations, I reject the null hypothesis if the previous negotiation variable is statistically significant.


    Throughout this study, I argued that negotiation was most likely at the extremes of violence, very low levels and very high levels. I tested this hypothesis using a violence variable and the squared term of that variable in my logistic regression. The findings of this analysis are shown in Table 1. Both violence and the squared term of violence were not found to have statistically significant effects. The most likely reason that Hypothesis 1 was not supported by the data lies in the data itself. Fundamentally, the data does not lend itself well to this study because of the disparity between the number of cases where negotiation occurred and those where it did not. Cases that did not experience negotiation outnumbered those that experienced negotiation by more than a ten-to-one ratio. Therefore, the statistical model may not completely capture the crux of my argument because of the presence of so many cases that did not encounter negotiation.

    Hypothesis 2 called for a positive relationship between previous occurrences of negotiation and the likelihood of new negotiation. I found support for this hypothesis as demonstrated in Table 1. The statistical significance gives credence to my claim that states are more likely to negotiate with organizations with which they have previously negotiated. To some extent, my findings confirm previous studies that identify a history of negotiation as a salient factor in the likelihood or success of future negotiation.

    Two other variables, territorial control and legality, were also found to be statistically significant. Territorial control was likely significant because states generally will take organizations more seriously if they can conquer and control land. Organizations that control land are also likely to be strong enough to challenge the state militarily, which follows my theory regarding organizations using high levels of violence. The significance of legality can be intuitively understood. If ethno-political organizations are legal in a given state, they are seen as legitimate political institutions. In a state that classifies ethno-political organizations as illegal, they are not recognized as legitimate or valid groups. Thus, negotiation is logically much more likely when ethno-political organizations are legal.

    Discussion and Conclusion

    Although I was unable to find support for my hypothesis regarding the effect of violence on the likelihood of negotiation, I would like to continue to research the relationship between the two phenomena. There are several plausible explanations for the disconnect between my theory and the statistical results. First, the data used for this study may not be appropriate for the subject. The MAROB dataset is primarily structured to investigate the different ways that ethno-political groups organize. Although it is not ideal for studying negotiation behavior specifically, it is the best data available for organization-level analysis. The distribution of cases is also not conducive to negotiation research. Since there were so few cases of organizations reaching negotiations with the state, the results of this study may be skewed. In other words, the great disparity in the number of cases of negotiation versus non-negotiation may have diminished the effects in the logistic regression. Data that is more appropriate to the study of negotiation may produce a different outcome. It is likely that my hypothesis regarding the relationship between organizational violence and the likelihood of negotiation failed to achieve statistical significance due to inappropriate or incomplete data.

    My hypothesis regarding the effect of previous negotiation on the likelihood of future negotiation produces some interesting implications. I argue that the state is more likely to open talks with an organization with which it has previously negotiated because each side has reached a level of comfort and trust. Logically, this implies that the first negotiation is the largest obstacle to prolonged talks. If the first iteration of negotiation engenders a future of negotiation, it may be in the state’s best interest to attempt negotiation with more groups. If a state deems negotiation to be in its best interest, it is likely that talks will progress past the first meeting. Therefore, if a state is more willing to open dialogue with an ethno-political organization, it may lead to an increase in the number of civil conflicts that end in negotiated settlements rather than military victory.

    Overall, this study has produced intriguing results. Although my assertion that violence plays a large role in predicting negotiation was unable to gain support from the statistical model, I argue that this is a design flaw. Further research with more appropriate data may produce different results. This project did, however, confirm literature that cites the importance of previous negotiation and creates interesting implications for the future of ending civil conflict.


    1. For the purposes of this study, I separate levels of violence into three categories for ease of discussion: low, intermediate, and high. These categories are derived from the variable ORGREB in the MAROB dataset, where I refer to low as ORGREB = 0-1; intermediate as ORGREB = 2-5; and high as ORGREB = 6-7. Further explanation on the meanings of the codes can be found in the research design of this study.
    2. 0: None reported; 1: Political banditry- (fewer than 6 violent attacks); 2: Small scale campaigns of violence- (6 or more violent attacks); 3: Local rebellion- (Armed attempt to seize control over locale, such as a town or a village); 4:Small-scale guerrilla activity- All of the following must exist: a) fewer than 1000 armed fighters; b) sporadic armed clashes (less than six reported per year); c) attacks in a small part of the area occupied by the organization, or in one or two other locales. 5: Intermediate guerrilla activity- Has one or two of the defining traits of large-scale activity and one or two of the defining traits of small-scale activity; 6: Large-scale guerrilla activity- All of the following must exist: a) more than 1000 armed fighters; b) frequent armed clashes (more than 6 per year); c) attacks affecting a large part of the area occupied by the organization; 7: Civil war- (regional division of country) Fought by rebel military units with base areas (Asal, Pate, and Wilkenfield 2008).
    3. Out of the 1,513 organization-year cases, only 150 were coded as experiencing negotiation.


    • Asal, Victor, Amy Pate and Jonathan Wilkenfeld. 2008. Minorities at Risk Organizational Behavior Data and Codebook Version 9/2008
    • Bonney, George E. 1987. “Logistic regression for dependent binary observations.” Biometrics 43 (4): 951-73.
    • Collier, Paul, and Anke Hoeffler. 2004. “Greed and grievance in civil war.” Oxford Economic Papers 56 (4): 563-595.
    • Collier, Paul, Anke Hoeffler, and Dominic Rohner. 2009. “Beyond greed and grievance: feasibility and civil war.” Oxford Economic Papers 61 (1): 1-27.
    • Downes, Alexander B. 2004. “The Problem with Negotiated Settlements to Ethnic Civil Wars.” Security Studies 13 (4): 230-279.
    • Duyvesteyn, Isabelle, and Bart Schuurman. 2011. “The Paradoxes of Negotiating with Terrorist and Insurgent Organisations.” The Journal of Imperial and Commonwealth History 39 (4): 677-692.
    • Fearon, James D, and David D Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” 97 (1): 75-90.
    • Gent, Stephen E., and Megan Shannon. 2011. “Decision Control and the Pursuit of Binding Conflict Management: Choosing the Ties that Bind.” Journal of Conflict Resolution 55 (5): 710-734.
    • Gochman, Charles S., and Russell J. Leng. 1983. ‘‘Realpolitik and the Road to War: An Analysis of Attributes and Behavior.’’ International Studies Quarterly 27 (1): 97-120.
    • Greenhill, Kelly M, and Solomon Major. 2006. “The Perils of Profiling.” International Security 31 (3): 7-40.
    • Greig, J Michael. 2012. “Rebels at the Gates: Civil War Battle Locations, Movement, and Openings for Mediation.” unpublished manuscript, University of North Texas.
    • Gurr, Ted Robert. 2000. “Chapter 3.” In Peoples versus States, Washington, DC: United States Institute of Peace Press, 65-95.
    • Hartzell, Caroline. 1999. “Explaining the Stability of Negotiated Settlements to Intrastate Wars.” Journal of Conflict Resolution 43 (1): 3-22.
    • Hartzell, Caroline, and Matthew Hoddie. 2007. Crafting Peace: Power-Sharing Institutions and the Negotiated Settlement of Civil Wars. University Park, PA: Pennsylvania State University Press.
    • Hartzell, Caroline, Matthew Hoddie, and Donald Rothchild. 2001. “Stabilizing the peace after civil war: An investigation of some key variables.” International organization 55 (1): 183-208.
    • Hensel, Paul R. 1996. ‘‘Charting a Course to Conflict: Territorial Issues and Militarized Interstate Disputes, 1816-1992.’’ Conflict Management and Peace Science 15 (1): 43-73.
    • Huth, PK. 1999. “Deterrence and international conflict: Empirical findings and theoretical debates.” Annual Review of Political Science 2 (1): 25-48.
    • Kydd, Andrew, and Barbara F Walter. 2002. “Sabotaging the Peace: The Politics of Extremist Violence.” International Organization 56 (2): 263-296.
    • Licklider, Roy. 1995. “The Consequences of Negotiated Settlements in Civil Wars, 1945-1993.” American Political Science Review 89 (3): 681-690.
    • Mandel, Robert. 2007. “Reassessing Victory in Warfare.” Armed Forces & Society 33 (4): 461-495.
    • Mason, T. David, Mehmet Gurses, Patrick T. Brandt, and Jason Michael Quinn. 2011. “When Civil Wars Recur: Conditions for Durable Peace after Civil Wars.” International Studies Perspectives 12 (2): 171-189.
    • Mattes, Michaela, and Burcu Savun. 2010. “Information, agreement design, and the durability of civil war settlements.” American Journal of Political Science 54 (2): 511-524.
    • Miller, Carl. 2011. “Is it Possible and Preferable to Negotiate with Terrorists?” Defence Studies 11 (1): 145-185.
    • Regan, Patrick M., and Daniel Norton. 2005. “Greed, Grievance, and Mobilization in Civil Wars.” Journal of Conflict Resolution 49 (3): 319-336.
    • Rothstein, Robert L. 2007. “The Timing of Negotiations: Dueling Metaphors.” Civil Wars 9 (3): 262-281.
    • Sambanis, Nicholas. 2001. “Do Ethnic and Non-Ethnic Civil Wars Have the Same Causes? A Theoretical and Empirical Inquiry (Part I).” Journal of Conflict Resolution 45 (3): 259-282.
    • Sánchez-Cuenca, Ignacio, and Luis de la Calle. 2009. “Domestic Terrorism: The Hidden Side of Political Violence.” Annual Review of Political Science 12 (1): 31-49.
    • Sisk, Timothy D. 2009. International Mediation in Civil Wars: Bargaining with Bullets. New York: Routledge.
    • Stedman, Stephen John. 1997. “Spoiler problems in peace processes.” International Security 22 (2): 5-53.
    • Vasquez, John A. 1993. The War Puzzle. Cambridge, UK: Cambridge University Press.
    • Walter, Barbara F. 1997. “The Critical Barrier to Civil War Settlement.” International Organization 51 (3): 335-364.
    • Walter, Barbara F. 2002. Committing to peace: The Successful Settlement of Civil Wars. Princeton, NJ: Princeton University Press.
    • Walter, Barbara F. 2004. “Does conflict beget conflict? Explaining recurring civil war.” Journal of Peace Research 41 (3): 371-388.
    • Walter, Barbara F. 2006. “Building Reputation: Why Governments Fight Some Separatists but Not Others.” American Journal of Political Science 50 (2): 313-330.
    • Walter, Barbara F. 2009. “Bargaining Failures and Civil War.” Annual Review of Political Science 12 (1): 243-261.
    • Zartman, I. William. 1995. Elusive peace: Negotiating an End to Civil Wars. Brookings Institution Press.

    Table 1: Logit Analysis: Likelihood of State-Ethnopolitical Organization Negotiation (1980-2004)

    Variables Coefficient
    (Robust Standard Error)
    Violence 0.556
    Violence Squared -0.081
    Previous Negotiation 0.377***
    Territorial Control 1.095***
    Legalitya 1.057**
    Strength of Leadership a 0.433
    Organizational Popularity 0.114
    Democracya 0.067

    n=1513     *p<.05; **p<.01; ***p<.001