Casualties and Support for Violent Conflict in Civil Wars

ABSTRACT

The casualty effect is a widely studied explanation of public support for war in the context of overseas military operations, yet work on the effect of casualties on support for intrastate war is scant. This paper examines the impact of local casualties on support for using military action as a conflict resolution method for intrastate war, using data from two public opinion surveys, collected in Turkey in the absence and presence of large-scale violence, and an original dataset for the local casualties. We find that local-level casualties on average increase the support for military action in ethnic wars.

View correction statement:

Correction

Acknowledgments

We would like to thank Prof. Banu Baybars Hawks, Prof. Mustafa Aydın, Prof. Hasan Bülent Kahraman and Prof. Osman Zaim at Kadir Has University for allowing us to use the dataset from the research titled ‘Social and Political Values in Turkey,’ and Lerna Yanık for helping us get access to the dataset. The two anonymous reviewers provided us with very useful input, and we acknowledge their guidance and contributions.The data can be requested from the authors but its release is subject to approval by the researchers who managed the survey study.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. ‘Military solution’ refers to the use of all kinds of violence to terminate a conflict, and it is the term used in public discussions as well as in the surveys used. We use ‘support for military solution,’ ‘support for military intervention,’ and ‘support for war’ interchangeably for the purposes of this paper.

2. The conflict between the Turkish state and the PKK rebels has continued since 1984, with two major periods of ceasefire in between (1999–2004 and 2013–2015). Important definitional differences notwithstanding, we use ‘intrastate war,’ ‘internal armed conflict,’ and ‘civil war’ interchangeably for the purposes of this paper. The Turkish government’s use of violence is referred to as ‘counterinsurgency’ in academic and journalistic literature, which is why we use the same concept in identifying the use of violence by the state’s security forces.

3. For the purposes of this paper, we use the general public’s and the mentioned surveys’ characterization of military actions by state agents as ‘counterterrorism’ or ‘counterinsurgency.’

4. For studies of public opinion on overseas military operations in Great Britain and Canada, see: Boucher (2010), Gribble et al. (2015).

5. They also show that temporally and spatially proximate casualties provide additional explanation over and above cumulative casualties. More recently, reporting results from six experiments, Gartner (2008) argues that support for war is affected by recent casualties, overall casualty trends and the interaction of the two.

6. In addition to the disaggregated effect of local-level casualties, Karol and Miguel (2007) suggest that national-level casualties have a localized effect, i.e., US states respond differently to news of casualties.

7. Kriner and Shen (2010) take into consideration the possibility that local casualty news produce greater impact as a result of emotionally charged local news reporting, but the local effect is found significant even when they provide the respondents in their experiments with identically worded local and national news.

8. They find the overall effect of a permanent increase in Palestinian fatalities on the preference for moderate attitudes negative but not statistically significant (Jaeger et al. 2012).

9. It is worth noting that while most of these works operate within the cost-benefit analysis, Berinsky (2007) puts into question the validity of the cost-benefit paradigm, suggesting that citizens often have limited information about the costs of war, and consequently, elite cues are a much more important determinant of citizens’ views than their own estimations of the costs. For instance, the American public is casualty-insensitive in the Afghanistan conflict, but elite division may be reason for this insensitivity (Wells 2015).

10. For studies of public opinion on overseas military operations in Great Britain and Canada, see: Boucher (2010), Gribble (2015).

11. The three-decade-long conflict remains the longest-lasting Kurdish uprising in the history of the Republic of Turkey, but the state’s failure to recognize the Kurdish language and culture, as well as Kurdish demands for local autonomy have fueled violent conflict since the 1920s. While official figures on combatant and civilian casualties are sketchy, it is estimated that at least 30,000 people were killed between 1984 and 2015, not to mention the conflict’s enormous human rights toll (Bozarslan 2001). The Turkish state managed to weaken the PKK in the late 1990s and capture its leader in 1999, but the absence of a political settlement resulted in the reinvigoration of the PKK in 2004. Peace proposals accompanied fighting in the 1990s and 2000s, to no avail.

12. Political elites’ and ordinary citizens’ narratives on the renewed violence provide clues on the country’s ideological polarization. Pro-government circles and Turkish nationalists blame the PKK for using the ceasefire to pursue a strategy of expanding its power in cities and town centers, in preparation for renewed fighting and an eventual declaration of local autonomy. Critics of the government point to the AKP government’s increasingly indifferent attitude toward the peaceful resolution of the Kurdish conflict and Kurdish demands, especially in the wake of the June 2015 elections in which the political party representing the Kurdish political movement, called Peoples’ Democratic Party (HDP in Turkish acronym), effectively denied the AKP a parliamentary majority for the first time in its 13-year rule. In addition, the HDP’s campaign promise to never let Erdoğan introduce a presidential system earned them the president’s hostility. According to the government’s critics, Erdoğan and the AKP leadership reignited the war with the PKK to force the voters to choose stability and security by associating the absence of one-party AKP government with instability and war. Added to these domestic factors, the resurgence of violence is often attributed to the spill-over effects of the Syrian civil war, especially the Turkish government’s unwillingness to cooperate with the PKK-affiliate PYD in northern Syria.

13. Between the 7 June 2015 general election and October 11 (when the PKK declared a three-week ceasefire before the November 1 general election), a total of 694 casualties was reported; more than 200 were noncombatant civilians. In addition, bomb attacks against rallies and activities organized by the HDP and HDP affiliates in Diyarbakır, Suruç and Ankara resulted in over 140 deaths and many wounded.

14. Individuals in other urban areas face the risk of suicide bombings, most of which come from Islamist groups like ISIS rather than the PKK and its affiliates (though PKK affiliates claimed responsibility for one suicide bombing that took place in Ankara in early 2016).

15. According to a report prepared by the Grand National Assembly of Turkey Human Rights Investigation Commission, the total number of civilian casualties in the fight with the PKK between 1984 and 2012 amounts to 5,557, with additional disappearances estimated to be more than 1,000 individuals (TBMM Human Rights Investigation Commission 2013).

16. The sample is representative of Turkish population, and the survey is conducted on adults residing in the city centers İstanbul, Ankara, Konya, Bursa, Kocaeli, İzmir, Aydın, Manisa, Tekirdağ, Balıkesir, Adana, Antalya, Hatay, Zonguldak, Samsun, Kastamonu, Kayseri, Kırıkkale, Trabzon, Gaziantep, Diyarbakır, Mardin, Malatya, Bitlis, Erzurum ve Ağrı.

17. The data has limitations as it was collected only from a select number of provinces, and the two waves were conducted one year apart on different samples though comparable. However, we believe that the data is still highly precious given the scarcity of studies of similar nature from Turkey and given the significance of the context. The political developments in Turkey in the recent years could help political science to explore intricate matters as Turkey hosts extremely complex and intertwined political issues. So, every reliable data from the Turkish context is valuable, and inheres in critical stories, which could contribute to myriad literatures within political science. Hence, we trust that the data is reliable and instrumental for our purpose. Bearing in mind its limitations we resort to complex statistical tools to overcome some of the limitations and to extract solid answers.

18. We believe funerals are the strongest media to move people’s sentiments. As the impact is probably greatest in the home province, we coded SFCs by province of birth.

19. It is indeed common in the literature to encounter conflation of insurgency with terrorism. Terrorism is laden with political undertones, and is usually opted for by the governments. Quite a few insurgent groups often studied in the civil war literature such as the Kurdistan Workers’ Party (PKK) in Turkey, Shining Path (PCP-SL) in Peru, National Liberation Army (ELN) or the Revolutionary Armed Forces of Colombia (FARC_EP) in Colombia, and Tamil Tigers (LTTE) in Sri Lanka are all listed as terrorist groups by the EU and the US due to their strategies of combat and record of targeting civilians (Stanton 2013).

20. We do not include the rebel force casualties in our analysis as it may confound the mechanism. Security force casualties will certainly influence the majority in-group members yet the report of guerilla fighters, may be equated with the idea of ‘success over the enemy’ (for the majority group members) rather than with human cost of the war. Also dehumanization of the rebel fighters is an oft-encountered psychological process, hence the death of an ‘enemy’ may not fuel any attitudinal reaction. We added ethnicity of the respondents as a covariate to capture any possible different reaction from the minority group.

21. This paper does not claim to adjudicate the debate on the validity of all these causal mechanisms across all types of war and in all context. Nonetheless, it cautions against the portrayal of (national or local) casualties as producing non-positive effects on the support for war only. Recent local-level casualties heighten individuals’ awareness of, and/or reactions to, a conflict, but this reaction may take the form of seeking a military solution to the conflict at hand.

22. We ran the model with province-level fixed effects, and the results are similar.

23. Ideological divisions across provinces are easy to encounter: certain provinces are known for having population majorities that are Turkish nationalists, or sympathetic to the PKK insurgency. Likewise, support for war or non-violent resolution of conflict may vary across geography. Therefore, there is a strong theoretical rationale to address the multilevel nature of the data.

24. In any case, this category does not alter the results. Results with the full nominal categories are available upon request.

25. Since 2015 was general election year, the survey uses a more straightforward question compared to the 2014 survey: ‘who did you vote for in the previous election?’ The discrepancy in the distribution of party choice also stems from this fact (Table 1). Because the survey is conducted in December, the 2015 survey responses are for the November election where the AKP gained 49% of the votes (rather than %40 in June). In the 2014 sample, the AKP has only 36.5% but it is very close to the estimations back then, and it is closer to the June election results. The 2015 survey captures the switch of the votes after the June election, and hence the AKP voter base is more diluted as compared to the 2014 sample, which only shows the hardcore supporters. We need to keep this in mind when interpreting the results.

26. The options for the length of school years are: 1, 3, 5, 7, 8, 10, 11, 13, 15, and 17. Since 1997, eight years of education is compulsory in Turkey, but those who were schooled before then may have dropped out earlier. The first four categories capture such drop-outs.

27. Gartner and Segura (2000) suggest that the race and ethnicity of citizens might affect their casualty sensitivity. We examined whether there is an interaction between ethnicity and casualties yet did not find any evidence for it. Casualties affect both ethnicities (Turks and Kurds) in the same way (see Model I1 in Table A1 in Appendix).

28. The effect of elections could be confounding mostly for the casualty effect because it is operationalized as a binary variable, indicating its absence in 2014, and its presence in 2015.

29. The ministries that are assigned a different minister are: Ministry of Labor and Social Security, Ministry of Family and Social Policies, Ministry of Finance, Ministry of National Defense, Ministry of Agriculture and Village Affairs, Ministry of Customs and Trade, Ministry of Economy, Ministry of Transportation and Communication, Ministry of Environment and Forestry, Ministry of Culture and Tourism, Ministry of Energy and Natural Resources.

30. Because the Kurdish conflict started out as an ethnic secessionist war, conventional wisdom may suggest that the minorities would not perceive secession as a threat and this question would only capture the Turkish majority’s attitudes. However 34.42% of ethnic Kurds answered this question ‘Yes’, and this ratio is 50.80% for ethnic Turks. Hence, not every Kurd perceives secession favorably, and a sizeable portion of them associates it with threat to the nation.

Additional information

Notes on contributors

Şule Yaylacı

Şule Yaylacı is a postdoctoral research fellow in the department of sociology at the University of British Columbia.

Onur Bakıner

Onur Bakıner is an assistant professor in the department of political science at Seattle University.