Explaining the killing of journalists in the contemporary era: The importance of hybrid regimes and subnational variations

Abstract

Explanations for an increase in the killing of journalists worldwide over the last quarter century include the dangers of war coverage and unsafe conditions in formally democratic countries. Analyzing 1812 killings of journalists from 1992 to 2016, we find that countries with hybrid political regimes mixing liberal and illiberal elements create by far the most dangerous context for journalists, while a particular spatial configuration within those countries, subnational authoritarianism, clarifies the logic of the killings. In short, the study finds that most journalists died in countries where formal democratic norms and practices at the national level encourage investigative reporting in local arenas where powerholders have incentives to violently suppress critical press coverage. On a theoretical level, the synthesis of spatial analysis, comparative politics and journalism studies opens a fruitful path for theorizing anti-press violence and journalist safety. In terms of policy, international action should increase costs to national governments that overlook local misbehavior in return for political support. Discourse spotlighting local journalists’ contributions to human rights and democratic accountability may also be helpful.

Keywords

Anti-press violence journalist safety,hybrid political regimes spatial analysis subnational authoritarianism

Introduction

In the quarter century since killings of journalists worldwide have been systematically documented, 1812 have died and the average annual rate of killings has increased – from 48 annually in 1992–2001, to 72 per year in 2002–2011, and 91 per year in 2012–2016, according to data we independently verified.¹ There are two alternative explanations for this phenomenon in the academic literature – more journalists killed covering combat and institutional weaknesses in a growing number of formally democratic countries. The former hypothesis emerges from studies of foreign war correspondents and their local employees (‘fixers’), especially following the US invasion of Iraq (Dusterhoft, 2013; Murrell, 2010; Saul, 2008). Nuanced versions of this account emphasize dangers in a wide variety of combat theaters (Tumber, 2002). In contrast, recent academic literature suggests the later explanation is more accurate. There is something about unevenly performing democracies that create lethal danger for journalists (Asal et al., 2016; Bjørnskov and Freytag, 2016; Hughes et al., 2017). Neither explanation has been fully tested with global data nor have the mechanisms encouraging lethal threats been comprehensively developed.

Our systematic review of journalists’ killings across the last 25 years finds that troubled democracies, or ‘hybrid regimes’, – where electoral competition is combined with authoritarian practices that repress critics and significantly limit access of the opposition to the political process – are responsible for the vast majority of journalists’ deaths worldwide. Furthermore, we find that the most lethal hybrid systems are those with territorial pockets where electoral accountability is spotty and human rights are more routinely abused. Within these territories, we locate by far the largest number of journalist murders in the last quarter century. In contrast, the only time when journalists were more at risk in countries with open armed conflict was in Iraq in 2006 and 2007. Regimes classified as more authoritarian have not been the most lethal since the early 1990s, when political liberalization was beginning to challenge social controls in some of those systems.

In the remainder of this article, we review the literature on hybrid regimes, subnational authoritarianism and threats to journalists and then test a number of hypotheses drawn from that literature. We conclude with an empirically supported explanation for why spatially uneven hybrid regimes have become the most dangerous for journalists.

Hybrid regimes and threats to journalists

Political scientists typically conceptualize regimes as national units. According to Dahl (1973), regime types vary across two dimensions: public contestation and inclusiveness. They differ in access to political power (degree of political competition) and in their relationships with citizens (including the responsiveness of elected representatives to the population). Academics distinguish two ideal types of authority patterns: democracy and autocracy. In practice, however, many contemporary states fall somewhere between full-fledged democracies and autocracies. The procedural definition of democracy revolves around governmental institutions guaranteeing free and fair participation in politics, typically through elections (Dahl, 1973), and those that ensure political and civil liberties by placing constitutional limits on executive action (O’Donnell et al., 1986).

By the turn of the century, a wave of political liberalization had produced numerous national political regimes with free and fair elections, but significant shortcomings in other aspects of democracy, defined more broadly. Neither fully democratic nor openly authoritarian, ‘hybrid’ regimes (Diamond, 2002) or ‘illiberal democracies’ (Zakaria, 1997), practice ‘electoral authoritarianism’ (Schedler, 2006) and ‘competitive authoritarianism’ (Levitsky and Way, 2010). Many contemporary regimes combine formal democratic features with non-democratic practices. They may hold elections and possess other formal democratic institutions, but incumbents resort to informal practices that undermine political rights, civil liberties, and constitutional restraints.²

Emerging evidence suggests that hybrid regimes are especially dangerous for critical journalists. Bjørnskov and Freytag (2016) show that the murder of journalists is prevalent in societies where relatively high press freedom is combined with weak rule of law and widespread corruption. Similarly, Asal et al. (2016) argue that middle-level democracies are the most dangerous because they are incapable of protecting journalists or punishing their attackers. In hybrid political regimes, the state is less capable or willing to enforce the rule of law, but accountability processes are still embedded in the political system, if weakly. Furthermore, at least some journalists in hybrid democracies believe their role as professionals is to monitor powerful actors and report abuses (Cottle et al., 2016; Hanitzsch et al., 2019). This creates a context in which unscrupulous politicians and criminals have incentives to suppress critical journalists, so exposés will not launch wider accountability processes (Waisbord, 2000).

The global increase in hybrid political regimes and concurrent diffusion of liberal norms in journalism provides one possible explanation for the simultaneous rise in the killing of journalists, but the mechanisms at work need illumination. National executives cannot kill journalists willy-nilly without risk to their rule, given national oppositional forces and numerous organizations that monitor journalists’ safety internationally. The national executives mentioned in Reporters Without Borders’ ‘Predators Gallery’ overwhelmingly intimidate rather than murder to tame the press.³ Spatial analysis afforded by subnational authoritarian theory may help clarify the process.

Subnational authoritarianism and repression of dissent

Durability of hybrid regimes by the 1990s led comparative political scientists to focus on the uneven territorial reach of their democratic institutions. O’Donnell (1993) intuitively identified subnational areas where a disjuncture between formal rules and actual practices extremely attenuated the rule of law and civil liberties. Later studies confirmed that many contemporary electoral democracies exhibit what Behrend and Whitehead (2016a, 2016b) describe as local constellations of ‘illiberal structures and practices’, which differ from their federal democratic frameworks. The concept of ‘subnational authoritarianism’ sheds light on these serious shortcomings in some democracies, including territorial units where elected and de facto powerholders operate with wide discretion.

Viewed from the spatial vantage point of subnational authoritarian theory, these democracies feature uneven quality in elections and protections for civil rights across territorial units. Informal rules and practices give officials discretionary authority in some areas, while in others the legally designated authorities are weaker than or collaborate with de facto powerholders, such as ‘traditional rulers, faction leaders, business elites, or indeed criminal bands’ (Behrend and Whitehead, 2016b: 6). Powerholders in illiberal areas may manipulate elections, violate civil rights, engage in corruption and capture local institutions, such as political parties, legislatures, courts, police, and news media. These practices and structures may consolidate into enduring subnational regimes (Gibson, 2013) or fall short of institutionalization (Behrend and Whitehead, 2016b: 5).

Threats to discretionary power are worrisome to local autocrats because subnational authoritarianism is ultimately limited by a national framework that can be activated to restrain local rulers (Behrend, 2008; Gibson, 2013). Gibson (2013) details the ‘boundary work’ local elected authoritarians engage in to contain politics and scandal locally, so that national authorities do not intervene. Local media compliance is a crucial component of boundary protection because it limits the participation of the local political opposition and circumscribes information about abuses within the local sphere, where elected authoritarians and other de facto powerholders have the advantage (Abbott, 2011; Durazo Herrmann, 2016; Gibson, 2013).

Therefore, critically inclined local journalists may face danger in countries where the fairness of elections, protection of rights, and effectiveness of the rule of law vary significantly across national territory, especially if they cover politics, corruption, crime, or human rights (Aguilar et al., 2014; Bartman, 2018; Brambila, 2017, 2018; Del Palacio and Olvera, 2017; Garcés et al., 2019). Critical journalists working in illiberal regions do not enjoy the same protections as their compatriots in regions where accountability mechanisms and the rule of law are stronger. Local powerholders use anti-press violence to silence criticism or support for opponents.

Hypotheses and methods

Attributing responsibility for a journalist’s death is often difficult, but the simultaneous expansion of hybrid democratic–authoritarian regimes and subnational authoritarian politics suggests a number of testable hypotheses, the first five of which are incremental.

H1. Most journalists killed on the job were intentionally murdered, rather than killed during military combat or on dangerous assignments.

H2. Most journalists killed were domestic journalists, not international correspondents.

H3. Most journalists died in countries with hybrid national political regimes, and the importance of hybrid regimes has increased over time.

H4. Most journalists were killed outside of national capitals.

H5. Most journalists who were killed covered newsbeats, such as politics, corruption, human rights, or crime, that could expose the wrongdoing of local politicians and/or other de facto powerholders.

H6. Most journalists died in countries where subnational authoritarianism is indicated by the uneven quality of elections and rights protections across their territories.

H7. Associations between the murder of journalists and indicators of regime hybridity and subnational authoritarianism are significant and run in directions predicted by subnational authoritarian theory.

We tested these hypotheses using Committee to Protect Journalists’ (CPJ) ‘killed’ database, which records journalist deaths worldwide from 1992 to 2016 and provides the widest, longest, and most consistent international coverage. We verified the data using online reports from news sites, journalist associations, and other human rights groups. CPJ defines a journalist as anyone who covers ‘news or comment on public affairs through any media – including in print, in photographs, on radio, on television, and online’. They provide information about lethal attacks on journalists in formal and informal labor situations, including ‘staff journalists, freelancers, stringers, bloggers, and citizen journalists’.⁴

The verified database contains information on 1812 deaths, including 1229 deaths confirmed as work-related, 489 deaths suspected as work-related, and 94 media worker deaths. We used data on confirmed and suspected work-related journalist deaths because, in some countries, judicial systems are not able or willing to clarify the circumstances. We included the deaths of media workers because they were targeted for supporting editorial functions. We report significant differences between the groups where relevant.

In addition to confirming the CPJ data, we added information about each journalist’s work and place of death by cross-referencing the cases with the online reports. We also coded each death according to the type of political regime and level of territorial unevenness in elections and civil liberties in the country and year in which it occurred. Regime classification data came from Polity IV and the territorial unevenness data from the V-Dem database. The Polity score is a quantitative measure of a country’s political institutions and processes, including the types and effectiveness of constraints on executive authority. The V-Dem unevenness variables capture subnational variation in election quality and protections for civil rights.

A trained undergraduate honors student verified the CPJ codes and added the new codes. The second author verified the work by double-coding approximately 10 percent of the cases for location, type of death, principle newsbeat, and whether the journalist was domestic or foreign. The level of agreement was national capital (95%), type of death (97%), and local or foreign (99%). We also checked whether the two coders agreed on whether each journalist covered war or politics, since journalists who did not cover war tended to cover many beats. Of the 32 journalists, the second coder recorded as covering war, 30 cases matched (94%), and of the 29 journalists, the second coder recorded as covering politics, 25 matched (86%). Finding these results acceptable, we tested the hypotheses.

Findings

H1. Most journalists were intentionally murdered

To test this hypothesis, we used the CPJ database classification of manner of death. The categories are (1) deaths due to crossfire, or in a battlefield or military context; (2) deaths during an unexpectedly dangerous assignment, such as a demonstration, riot, clashes, or mob event; and (3) deaths due to murder as a reprisal for news work or to prevent such work. We identified 18 deaths in which the journalists perished on assignment in a combat zone and were classified as ‘murdered’. We reclassified these as death due to crossfire or exposure to military contexts.⁵ We also identified 20 instances in which the manner of death was unresolved, and we assigned two unclassifiable deaths as ‘other’. We used the same classification to code media workers and suspected work-related deaths for journalists.

Table 1 reports results. Far more journalists have been intentionally murdered for their work outside of combat zones: they account for 72 percent of all deaths over nearly 25 years. Nineteen percent were killed by crossfire or in combat-related ways, 9 percent died on dangerous assignments, and 1 percent disappeared without resolution.⁶ This hypothesis is supported.

Table 1.

Type of death.

Type of death	Count	%
Murder	1281	72
All combat related	343	19
Dangerous assignment	155	9
Disappeared with no resolution	11	1
Total	1790	100

The type of death was unresolved in an additional 20 instances and two more deaths were classified as ‘other’. Percent adds to more than 100 due to rounding.

H2. Most killed were domestic journalists

To test Hypothesis 2, we coded whether the journalist was a native or foreigner in the country of death. A journalist with an international news outlet, who was from the country where his or her death occurred, was coded as domestic, since many such journalists also work in domestic journalism, and, unlike international correspondents, face continual danger since they live in the country. Data strongly support this hypothesis. Fully 90 percent of journalists killed on the job over the last 25 years were domestic journalists rather than international correspondents.

We then examined whether domestic journalists were intentionally murdered. Table 2 reports results and includes a z-test for differences in column proportions. If subscript letters for subsets of journalists do not match, this indicates that the proportions are significantly different at the p = .05 level. The analysis found that domestic journalists were murdered at more than twice the rate of international journalists, while international journalists died covering combat at three times the rate of domestic journalists. There were no significant differences in the proportions of domestic or international journalists who died on dangerous assignments or disappeared.

Table 2.

Manner of death.

	International	Domestic	All
All combat related	53%^a	16%^b	19%
Dangerous assignment	11%^a	8%^a	9%
Murder	35%^a	75%^b	72%
Disappeared with no resolution	1%^a	1%^a	1%
Total (%)	100	100	100
Total count	171	1619	1790

Each superscript letter denotes a subset of categories whose column proportions do not differ significantly from each other at the p = .05 level.

H3. Most journalists died in countries with hybrid political regimes and the importance of hybrid regimes increased over time

To test this hypothesis, we coded each case using the Polity IV regime type classification for the country and year in which death occurred. We collapsed Polity IV scores into three categories: +3 indicates a regime with strongly democratic features, +2 a hybrid political system with characteristics associated with both democratic and authoritarian regimes, and +1 indicates a regime with strongly authoritarian features. The regime code covered only 1563 deaths because 248 deaths occurred in countries where Polity scores were not recorded due to open, armed conflict between supporters of contesting political regimes. In addition, one death was not coded because of the country’s small population (the Maldives).

The overwhelming majority of journalists died in countries with hybrid political regimes: 67 percent or 1053 cases compared to 31 percent or 489 cases in authoritarian countries and only 1 percent or 22 cases in democracies. Even if we include deaths in countries that were not coded due to open conflict between supporters of different regimes, such as Afghanistan between 2001 and 2014, democratic–authoritarian hybrids remain the most lethal: 58 percent of deaths occur in hybrid systems, 27 percent in authoritarian systems, 14 percent in countries with armed conflict between supporters of different regimes, and 1 percent in democratic regimes. Hybrid regimes were most dangerous whether or not deaths could be confirmed as work related, although it was harder to confirm the cause of death within such systems because the quality of justice may be poorer.

We ran cross-tabulations on manner of death and political regime of the country in which death occurred. Journalists who died in countries with hybrid political regimes were more often intentionally murdered: at a rate of 58 percent of 1809 deaths. By comparison, murders comprised 27 percent of deaths in authoritarian regimes, 1 percent in democratic regimes, and 14 percent in countries where Polity IV recorded open, armed conflict between regime supporters. Z-tests found that journalists were murdered at proportionally higher rates within countries with either democratic or hybrid political regimes. The higher proportion of murdered journalists in hybrid political regimes meets the expectations of literature on subnational authoritarianism. A review of the fewer cases of journalists who died in democracies suggests other causes, including coverage of organized crime. By contrast, higher proportions of journalists suffered combat-related deaths within countries with authoritarian political regimes and, unsurprisingly, in those that remained uncoded because of open conflict.

Table 3 reports these findings. The analysis supports the hypothesis: journalists are murdered in far greater numbers in hybrid regimes and proportionally are more likely to be murdered than to die on the job in a different way within hybrid and democratic regimes.

Table 3.

Manner of death by political regime.

	Most authoritarian	Hybrid regimes	Most democratic	Conflict between regimes in polity 4 data
All combat related	33%^a	8%^b	0%^b	42%^a
Dangerous assignment	11%^a	9%^a	14%^a	1%^a
Murder	55%^a	83%^b	86%^b	57% ^a
Disappeared with no resolution	1%^a	1%^a	0%^a	0%^a
Total deaths	483	1045	22	237

Each superscript letter denotes a subset of categories whose column proportions do not differ significantly from each other at the p = .05 level.

Has the danger of hybrid regimes become more important over time? We analyzed changes across different types of political regimes for the 25-year period for which we have data. Year-to-year chi-square analysis presented in Table 4 provides evidence that hybrid political regimes have indeed become increasingly dangerous in the global context, especially across the last 10 years in the period. Adjusted standardized residuals for year-to-year chi-square tests show that journalist killings have increased without interruption in hybrid regimes every year since 2008.⁷ From 1992 to 2015, the killings of journalists in hybrid regimes increased by 91 percent (from 24 to 64), before declining to 48 deaths in 2016 (although, given a decline in deaths overall in 2016, still slightly higher than the number predicted as shown by the standardized residual).⁸ Absolute number of deaths in authoritarian regimes remained relatively stable, but their share among all killings decreased from 50 to 39 percent during the time period. Highly performing democracies account for a tiny 1 percent of journalists’ deaths and prove to be the safest settings for journalists. Except for 2006–2007, when journalists’ deaths in wartime Iraq climbed sharply, hybrid regimes have been the most dangerous contexts for journalists in the 21st century.

Table 4.

Changes in regime type and killing of journalists across time.

		1992–1996	1997–2001	2002–2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	Total
Most authoritarian	Count	35	9	8	13	2	14	12	14	42	38	28	33	31	486
	% within year	50	20	10	12	3	14	15	16	40	38	31	33	39	27
	Adjusted residual	5.08	–0.9	–2.88	–3.8	–4.6	–3.1	–2.5	–2.3	3.0	2.5	0.8	1.4	2.5
Hybrid regimes	Count	24	31	33	44	49	80	64	68	63	62	63	64	48	1047
	% within year	42	72	52	39	71	79	80	79	59	62	69	64	61	58
	Adjusted residual	–2.78	1.72	–1.54	–4.2	2.20	4.4	4.0	4.0	0.3	0.8	2.2	1.2	0.5
Most democratic	Count	1.4	1.8	0.2	1.0	0.0	0	1.0	0.0	0.0	0.0	0.0	3.0	0.0	22
	% within year	3	4	0	1	0	0	1	0	0	0	0	3	0	1
	Adjusted residual	0.9	1.64	–0.72	–0.3	–0.9	–1.2	0.0	–1.1	–1.2	–1.1	–1.1	1.7	–1
Not coded because of open conflict between regimes	Count	1	2	29	54	18	7	3	4	1	0	0	0	0	245
	% within year	1	4	36	48.2	26.1	7	4	5	1	0	0	0	0	14
	Adjusted residual	–2.84	–1.86	6.18	11.0	3.1	–2.0	–2.6	–2.5	–3.9	–4.1	–3.9	–4.1	–3.6
Total	Count	61	44	69	112	69	101	80	86	106	100	91	100	79	1800
	% within year	100	100	100	100	100	100	100	100	100	100	100	100	100	100

H5. Most journalists died outside of national capitals

To test this hypothesis, we coded whether each journalist was killed outside a national capital. Data show that 72 percent of journalists who died perished outside of national capitals, making up 1276 of the 1783 deaths for which we have information. This proportion held for journalists who were murdered, died in crossfire, or disappeared with no resolution. The proportion of journalists who died outside of capital cities in countries with hybrid political regimes was higher than in countries with other governing systems: 84 percent of the 1046 journalists who died in countries with hybrid regimes were killed outside of the national capital; as were 67 percent of the 21 journalists who died in higher functioning democracies; and 55 percent of the 469 journalists who died in countries with authoritarian regimes. Most of the journalists killed outside capitals in hybrid regimes were murdered: 722 out of 881 or 82 percent.

The hypothesis was supported: Most journalists were murdered away from the center of their countries’ political life, in places where subnational authoritarian theory predicts accountability of local politicians will be weaker and political incentives to suppress critical coverage will be stronger.

H5. Most killed journalists covered politics, corruption, human rights, or crime

To test this hypothesis, we cross-checked the newsbeats journalists covered and created dummy variables for different beats since, with the exception of war correspondents, many journalists covered multiple beats. Table 5 reports these findings. Politics was the most common newsbeat (n = 628), covered by 35 percent of all 1802 journalists killed about whom we have information. The second most common newsbeat was war, which 29 percent of killed journalists covered. The next most common were corruption (286, 16%), human rights (265, 15%), and crime (241, 13%). It was far less dangerous to cover business, sports, or culture, although some killed journalists did specialize in those beats.

Table 5.

Newsbeats of killed journalists.

	Counts of beats		% murdered of all killed, by beat
	All killed	Murdered	% murdered of all killed, by beat
Politics	628	453	72
War^—	517	214	41
Corruption+	286	271	95
Human rights	265	151	57
Crime+	241	222	92
Business	54	40	74
Sports alone or sports + culture	25	14	56
n =	2049	1378

Newsbeats add to more than killed journalists because many covered multiple beats.

Denotes beats where distribution of those murdered in countries with hybrid political regimes was significantly higher than those who died in more uniformly authoritarian or democratic regimes (p = .05).

—

Denotes newsbeat where distribution was significantly lower (p = .05).

Examining murdered journalists’ coverage beats sheds further light on the potential dangers of critically covering subnational authoritarians. Rather than dying during a dangerous assignment or while covering combat, 95 percent of the dead journalists who covered corruption, 92 percent of those who covered crime, and 72 percent of those who covered politics were murdered. Covering these newsbeats was more dangerous in hybrid national political systems: 84 percent of killed journalists who covered corruption died in countries with hybrid regimes, as did 88 percent of those who covered crime. In contrast, while 60 percent of journalists who covered politics died in countries with hybrid political regimes, the distributions of political journalists killed inside and outside hybrid systems were not significantly different.

We find support for this hypothesis. Newsbeats associated with coverage of subnational authoritarian leaders were more dangerous. The most dangerous newsbeat in terms of numbers was politics. Other risky newsbeats included corruption, crime, and human rights, all of which might spotlight political wrongdoing, including collusion between state actors and criminals. The relationship between deaths and those newsbeats is even stronger when we examine murders alone.

H6. Most journalists died in countries with uneven elections quality and civil liberties protections across their territories

To test this hypothesis, we created a database of journalists killed in 173 countries between 1995 and 2016, coded by year. This produced 3749 data points, or country-years, ranging from 0 to 55 journalists killed per country-year. To each case, we added the V-Dem measures of territorial unevenness. These are ordinal variables, coded on three levels according to degree of protection of civil liberties or quality of elections: significant, moderate or equal protections and elections quality across territories. We then tested differences in the distributions of journalists killed in each category, using a non-parametric Kruskal–Wallis test, after checking that the distribution shapes in each category were roughly equivalent. We also tested whether there was a trend in the differences in the means for each category, using the Jonckheere–Terpstra test. These results are reported in Table 6.

Table 6.

Means and standard deviations of journalists killed in countries with differing levels of subnational variation.

Level of territorial unevenness in a country	Civil liberties protections	Quality of elections
All killed^jt
High	0.89**** (2.66)	1.58**** (5.01)
Moderate	0.66**** (3.14)	0.38**** (1.35)
Low or none	0.05**** (2.21)	0.14**** (1.43)
Murdered journalists^jt
High	0.713**** (2.09)	1.16**** (3.80)
Moderate	0.430**** (2.15)	0.301**** (1.15)
Low or none	0.040**** (0.280)	0.069**** (0.48)

N = 3749 country years, 174 countries, and 1812 deaths.

****

Kruskal–Wallis test for differences across categories significant at p = .000 level.

Jonckheere–Terpstra test for a trend in categories significant at p = .000 level.

The mean number of journalists killed was .885 in countries with high levels of unevenness in civil liberties, .662 in countries with moderate levels of unevenness across territories, and .047 in countries where respect for civil liberties was even across the country. The Kruskal–Wallis test found significant differences in the distributions (H = 328.147, p < .001). Jonckheere’s test revealed a significant trend: as respect for civil liberties across national territories became less even, the number of journalist killings in a country increased, T_JT = 1,872,299, z = −18.11, p = .000. Tests were re-run using only journalists who had been murdered and similar proportions with significant differences and trends were found.

We then repeated the analytical procedures for the variable unevenness in elections across territories. The mean number of killed journalists was 1.58 in countries with high levels of electoral unevenness, .38 in countries with moderate levels, and .14 in countries where election quality was even. The Kruskal–Wallis test found significant differences in the distributions of journalists killed in countries with varying levels of electoral quality across territories (H = 301.378, p < .001). Jonckheere’s test revealed a significant trend: as electoral quality across national territories became less even, the number of journalist killings in a country went up, T_JT = 1,768,314, z = −16.48, p = .000. We again examined the relationship between the number of journalists murdered and the same patterns held.

The evidence supports the hypothesis: greater unevenness in protections for civil liberties and the quality of elections across a country’s territories is positively associated with more killings of journalists.

H7. Associations between murdered journalists and indicators of regime hybridity and subnational authoritarianism are significant and run in directions predicted by subnational authoritarian theory

To test this hypothesis, we analyzed whether there were statistically significant relationships between murders over the last 25 years (1 = murder, 0 = another manner of death) and other variables that met expectations from the literature on hybrid political regimes and subnational authoritarian politics. We used the Phi coefficient for variables that were discrete and dichotomous (coded as 1 = condition present, 0 = not present). We used Spearman rank-order correlations for scale variables measuring subnational unevenness in election quality and respect for civil liberties, which we recoded as 0 = even quality and respect across territories, 1 = some territorial unevenness, and 2 = great territorial unevenness. Table 7 reports coefficients and their significance.

Table 7.

Associations between assassinations and indicators of regime hybridity and subnational authoritarianism.

	Murdered	Died covering combat	Domestic journalist	Died in national capital	Beat = war	Beat = politics	Beat = corruption	Beat = crime	Beat = human rights	Elections uneven	Civil liberties uneven	Autocracy	Hybrid regime	Democracy
Murdered	1.000	–.744***	.265***	–.015	–.424***	.012	.233***	.180***	–.134***	.139***	.088***	–.232***	.288***	.037
Died covering combat	–.744***	1.000	–.255***	–.021	.494***	–.120***	–.205***	–.186***	.068**	–.102***	–.132***	.214***	–.315***	–.053*
Domestic journalist	.265***	–.255****	1.000	–.059*	–.300***	.055*	.099***	.088***	–.041	.083***	.061*	–.126***	.195***	–.033
Died in national capital	–.015	–.021	–.059^*	1.000	.038	.045	–.138^***	–.139^***	.041	–.013	–.257^***	.220^***	–.334^***	.012
Beat = war	–.424***	.494***	–.300***	.038	1.000	.002	–.196***	–.211***	.140***	–.070***	–.172***	.272***	–.402***	–.059*
Beat = politics	.012	–.120***	.055*	.045	.002	1.000	.057*	–.050*	.227***	–.040	.006	.067***	.033	–.017
Beat = corruption	.233***	–.205***	.099***	–.138***	–.196***	.057*	1.000	.316****	.026	.047	.091***	–.161***	.226***	.090***
Beat = crime	.180***	–.186***	.088***	–.139***	–.211***	–.050*	.316****	1.000	–.043	–.022	.070***	–.198***	.240***	.149***
Beat = human rights	–.134***	.068**	–.041	.041	.140***	.227***	.026	–.043	1.000	–.139***	–.030	.156***	–.074***	–.003
Elections uneven	.139***	–.102***	.083***	–.013	–.070***	–.040	.047	–.022	–.139***	1.000	.097***	–.496***	.226***	–.162***
Civil liberties uneven	.088***	–.132***	.061*	–.257***	–.172***	.006	.091***	.070***	–.003	.097***	1.000	–.150***	.361***	–.176***
Autocracy	–.232***	.214***	–.126***	.220***	.272***	.067***	–.161***	–.198***	.156***	–.496***	–.150***	1	–.718***	–.068**
Hybrid regime	.288***	–.315***	.195***	–.334***	–.402***	.033	.226***	.240***	–.074**	.226***	.361***	–.718***	1	–.131***
Democracy	.037	–.053*	–.033	.012	–.059*	–.017	.090***	.149***	–.003	–.162***	–.176***	–.068**	–.131***	1

Significance levels: ***.001 (two-tailed), **.01 (two-tailed), and *.05 (two-tailed).

Findings met expectations about manner of death and domestic versus international journalists. Murder and death while covering combat are strongly and negatively correlated (−.744, p = .001), while being a domestic journalist is moderately and positively associated with assassination (.265, p = .001). If a journalist was murdered, she or he was much less likely to have died while covering military combat and slightly more likely to be a domestic journalist.

The relationships between assassinations and type of political regime in which deaths occurred were as expected for hybrid regimes and autocracies. There was a weak to moderate positive correlation between assassinations of journalists and hybrid political regimes (.288, p = .001) and a significant and slightly weaker negative correlation between murders of journalists and autocracies (−.232, p = .001). There was no significant relationship between democracy and the assassination of journalists. These findings show that a journalist who was intentionally murdered, rather than having died on a dangerous assignment or in a combat zone, is slightly more likely to have died in a country in which the political practices and structures that govern accountability, civil liberties, and the rule of law are mixed and slightly less likely to have died in a country whose political practices and structures are more uniform and authoritarian.

There are significant relationships between journalists being assassinated and covering war (−.424, p = .001), corruption (.233, p = .001), crime (.180, p = .001) or human rights (−.134, p = .001), but not politics. The first three relationships support theoretical expectations that the assassination of journalists is associated with coverage that could expose wrongdoing in areas where political and de facto powerholders act with little restraint. The lack of a significant relationship between murder and the coverage of politics was unexpected, as was the negative correlation between assassinations and covering human rights. Given the large number of journalists who died and who covered politics, perhaps most of the dangers journalists face are political in origin, and covering politics exposes journalists to myriad physical dangers, including unexpectedly dangerous assignments involving political factions and open combat resulting from the failure of peaceful political processes. Covering human rights and politics were, comparatively, strongly and positively correlated (.227, p = .001), even while these newsbeats were weakly or not at all correlated with assassinations. An explanation might be that the analysis fails to capture the time lag between journalist assassinations and the onset of human rights abuses when national leaders perceive the possibility of political breakdown, as found by Gohdes and Carey (2017). In contrast, covering corruption or crime was positively associated with assassination and was the strongest associations between newsbeats and assassination. Furthermore, covering corruption and crime were positively correlated with each other (.316, p = .001), dropping slightly to .287 (p = .001) when controlling for type of death (murdered). This correlation was among the strongest in the analysis. These findings suggest that crime and corruption coverage are related to each other and to the murder of journalists, but these newsbeats are different from covering human rights and politics, which are also less associated with purposeful assassination. These findings support some prior findings from Latin America that the most dangerous contexts for critical journalists are those involving corrupt state actors and their collusion with organized crime, especially at the subnational level (Garcés et al., 2019; Hughes et al., 2017).

The assassination of journalists was also associated with uneven elections quality and respect for civil liberties across areas in a country. There were statistically significant positive correlations between assassination and territorially uneven elections quality (.139, p = .001) as well as uneven respect for civil liberties (.088, p. = 001), although these relationships were relatively weak, especially with regard to the protection of civil liberties. The unevenness measures were not strongly correlated with each other, either (.097, p. = 001), contrary to expectations, but were strongly correlated (in different directions) with autocratic and hybrid political regimes. Partial correlations suggest regime-type intervenes in the relationship between assassination and uneven operation of elections and rights protections. When the type of political regime is controlled for, the directions of association and strengths of the correlations between journalist assassinations and territorial unevenness are more in line with theoretical expectations. Partial correlations for journalist assassination and territorial unevenness in elections controlling for types of regime in countries where journalists died are: autocracies (−.473, p = .01), hybrid regimes (.251, p = .01), and democracies (−.154, p = .01). Partial correlations are the following: journalist assassinations and territorial unevenness with respect to civil liberties when types of regimes are controlled are the following for autocracy (−.147, p = .01), hybrid regimes (.353, p = .01), and democracy (−.173, p = .01).

Whether a journalist was killed outside of a capital city and whether the journalist was murdered compared to another type of death had no statistically significant relationship, contrary to expectations. However, regime type also played a mediating role in this relationship. The relationship was significant but weak when the variable hybrid regime was controlled for (.092, p = .001) and was slightly stronger when hybrid regime and covering corruption or crime were controlled for (.109, p = .001).

Bivariate tests of association between the assassination of a journalist, national regime hybridity and newsbeats that are risky in less-restrained subnational contexts produced evidence to support the hypothesis. There was also evidence that regime type was an intervening factor in the importance of measures of subnational unevenness in elections and respect for civil liberties, which provides further support for the hypothesis. Contrary to expectations, covering politics is dangerous across many contexts, perhaps because open conflict and unexpectedly dangerous assignments are politics turned violent (or the failure of politics to keep the peace). The relationship between being murdered outside of capital cities, compared to other types of death, was not strong. However, this analysis has limitations. Most of the associations are relatively weak in strength, except for those showing that, among journalists who were killed on the job, international correspondents and those covering war are qualitatively different from domestic journalists, whose assignments include risky newsbeats that do not involve covering armed conflict.

Conclusion

A systematic analysis of the killings of journalists over the last quarter century, viewed through a spatial lens, finds evidence that the increase in journalist deaths disproportionately responds to the growing presence of mixed political regimes with uneven electoral accountability and rights protections across territories. The combination of national-level hybrid regime with subnational authoritarian enclaves was especially dangerous. We believe this is because institutionalized democratic norms at the national level explicitly guarantee press freedom and condemn illiberal practices, thus encouraging journalists, many of whom formed professionally during periods of political liberalization, to monitor power and denounce abuses. However, exposing corruption, failed policies or other abuses become hazardous where accountability is weak and local authorities have incentives to attack critics to silence them. The collusion of criminal, business, and political interests often flourishes where local governments and security forces are unrestrained (Behrend and Whitehead, 2013; Von Holdt, 2016b), and correlations between coverage of these topics and the killing of journalists were stark evidence of the peril this creates for journalists.

Table 8 maps lethal attacks across regimes and levels of subnational variation in democratic norms and practices. The most dangerous contexts for contemporary journalists are hybrid regimes with high levels of territorial unevenness in respect for rights and middle levels of unevenness in electoral accountability, followed closely by hybrid contexts with moderate variation in rights protections and high levels of electoral variability. A second tier of dangerous contexts are authoritarian regimes with variability in rights, which may indicate that authoritarian social controls are unstable, as Gohdes and Carey (2017) suggest. In contrast, findings confirm democracies with greater electoral accountability and rights protections provide greater safety for journalists. The 22 lives lost in these democracies exceed quantitative measurement, as do all the deaths. Each casualty represents loss for society and democracy, as well as the most gross violation of individual rights. Nevertheless, the mapping shows that conditions for journalism in democracies and democratic hybrids vary enormously across the globe, in ways that profoundly affect the professional performance and personal wellbeing of journalists.

Table 8.

Mapping of assassinations by regime type and territorial variability.

		Territorial unevenness
		Low		Medium		High
Democracy	Elections	15	0.9%	1	0.1%	0
	Rights	14	0.8%	1	0.1%	1	0.1%
Hybrid	Elections	30	1.8%	401	24.1%	391	23.5%
	Rights	41	2.5%	329	19.8%	452	27.2%
Autocracy	Elections	98	5.9%	115	6.9%	26	1.6%
	Rights	54	3.3%	166	10.0%	19	1.1%
Open conflict between regimes	Elections	1	0.1%	6	0.2%	126	7.6%
	Rights	0		132	7.9%	1	0.1%
Shaded by intervals of 90:			0–90	91–180	181–270	271–360	361–452

N = 1661. Chi-square tests for independence between regime types and levels of territorial unevenness in elections or civil liberties protections were statistically significant for all interactions at p < .001 level.

The percentage refers to the number of deaths in the subcategory as a percentage of the total number of deaths. Percentages sum over 100 because each case may be counted more than once.

The spread of hybrid political regimes as liberalization stalled – or backslid – across the last quarter century involved spatial variation in the performance of democratic institutions, including supports for a monitorial press. However, findings from this study and others suggest journalists sometimes resist encroachment on democratic norms (Høiby, 2019; Hughes and Márquez-Ramírez, 2018; Salazar, 2019). Resistance carries potentially far-reaching consequences for politics, g and rights. Studies in some of the most dangerous countries for journalists find that exposés can activate accountability mechanisms, for example, by significantly affecting elections’ results (Enikolopov et al., 2011), providing the opposition with issues to organize around (McFaul, 2005), or empowering the judiciary or investigative functions of congress (Waisbord, 2000). Local media can also help overcome illiberal subnational practices. A process of local democratization can ensue when journalists’ investigations are backed by mobilization of local civic organizations or opposition parties (Behrend and Whitehead, 2016b), which in turn may increase journalists’ resolve to monitor power and signal abuse (Salazar, 2019). Thus, safeguarding journalists may well safeguard liberalism, or at least forestall further erosion, especially since most countries today retain some democratic features (Levitsky and Way, 2010).

At a broader level, the study suggests that integrating journalism studies, comparative politics and spatial analysis offers a basis for theorizing in journalist safety research, which tends to remain descriptive given its newness and policy relevance. Policy implications of the findings include spotlighting local authoritarians and raising the cost to national governments that overlook local abuses in return for political support. Finally, shaping international discourse to highlight local journalists’ contributions to protecting rights and democratic accountability could also be powerful.

Footnotes

Acknowledgements

The authors acknowledge and thank Dylan Ceder for exceptional research assistance during this project. We also thank two anonymous reviewers for recommendations that greatly improved this manuscript.

Funding

The author(s) received no external financial support for the research, authorship, and/or publication of this article.

ORCID iD

Sallie Hughes

Notes

Author biographies

Sallie Hughes is an Associate Professor of Journalism and Faculty Director of the Institute for Advanced Study of the Americas, both at The University of Miami. Her academic work on issues related to journalism, journalists and risk is widely published.

Yulia Vorobyeva is a Post-doctoral Fellow at the Latin American and Caribbean Center at Florida International University. Her research interests include gang violence, organized crime, and rule of law in Latin America.

References

Abbott

(2011) Electoral authoritarianism and the print media in Malaysia: Measuring political bias and analyzing its cause. Asian Affairs: An American Review 38: 1–38.

Aguilar

Mendoza

Candelaria

ALK

(2014) Keeping the state at bay: The killing of journalists in the Philippines, 1998-2012. Critical Asian Studies 46(4): 649–677.

Asal

Krain

Murdie

, et al. (2016) Killing the messenger: Regime type as a determinant of journalist killing, 1992–2008. Foreign Policy Analysis 14(1): 24–43.

Bartman

(2018) Murder in Mexico: Are journalists victims of general violence or targeted political violence? Democratization 25: 1093–1113.

Behrend

(2008) Democratic Argentina and the ‘closed game’ of provincial politics: Protest and persistence. PhD Thesis, Department of Politics and International Relations, Oxford University, Oxford.

Behrend

Whitehead

(2016a) The struggle for subnational democracy. Journal of Democracy 27(2): 155–169.

Behrend

Whitehead

(2016b) Setting the comparative agenda. Territorially uneven democratic processes in large federations. In: Behrend

Whitehead

(eds) Illiberal Practices: Territorial Variance within Large Federal Democracies. Baltimore, MD: Johns Hopkins University Press, pp. 1–19.

Bjørnskov

Freytag

(2016) An offer you can’t refuse: Murdering journalists as an enforcement mechanism of corrupt deals. Public Choice 167(3–4): 221–243.

Brambila

(2017) Forced silence: Determinants of journalist killings in Mexico’s states, 2010–2015. Journal of Information Policy 7: 297–326.

10.

Brambila

(2018) Reporting dangerously in Mexico: Capital, risks and strategies among journalists. PhD Thesis, University of Leeds, Leeds.

11.

Cottle

Sambrook

Mosdell

(2016) Reporting Dangerously: Journalist Killings, Intimidation and Security. New York: Springer.

12.

Dahl

(1973) Polyarchy: Participation and Opposition. New Haven, CT: Yale University Press.

13.

Del Palacio

Olvera

(2017) Acallar las voces, ocultar la verdad. Violencia contra los periodistas en Veracruz. [Quiet the voices, hide the truth: Violence against journalists in Veracruz]. Argumentos 30(85): 17–35.

14.

Diamond

(2002) Thinking about hybrid regimes. Journal of Democracy 13(2): 21–35.

15.

Durazo Herrmann

(2016) Social heterogeneity, political mediation, and subnational illiberalism: Oaxaca and Puebla, Mexico. In: Behrend

Whitehead

(eds) Illiberal Practices: Territorial Variance within Large Federal Democracies. Baltimore, MD: Johns Hopkins University Press, pp. 229–260.

16.

Dusterhoft

(2013) The protection of journalists in armed conflicts: How can they be better safeguarded. Merkourios: Utrecht Journal of International and European Law 29: 4.

17.

Enikolopov

Petrova

Zhuravskaya

(2011) Media and political persuasion: Evidence from Russia. The American Economic Review 101(7): 3253–3285.

18.

Garcés

Jaramillo

Arroyave

, et al. (2019) Libertad de prensa y conflicto armado en Colombia: Un análisis desde la autonomía profesional percibida por los periodistas colombianos [Freedom of the press and armed conflict in Colombia: An analysis from the perspective of Colombian journalists’ perceived professional autonomy]. Saber, Ciencia y Libertad 14(1): 21–34.

19.

Gibson

(2013) Boundary Control: Subnational Authoritarianism in Federal Democracies. New York: Cambridge University Press.

20.

Gohdes

Carey

(2017) Canaries in a coal-mine? What the killings of journalists tell us about future repression. Journal of Peace Research 54(2): 157–174.

21.

Høiby

(2019) Covering Mindanao: The safety of local vs. non-local journalists in the field. Journalism Practice. Epub ahead of print 8 May 2019. DOI: 10.1080/17512786.2019.1598884.

22.

Hanitzsch

Hanusch

Ramaprasad

, et al. (eds) (2019) Worlds of Journalism: Journalistic Cultures around the Globe. New York: Columbia University Press.

23.

Hughes

Márquez-Ramírez

(2018) Local-level authoritarianism, democratic normative aspirations, and antipress harassment: Predictors of threats to journalists in Mexico. The International Journal of Press/Politics 23(4): 539–560.

24.

Hughes

Mellado

Arroyave

, et al. (2017) Expanding influences research to insecure democracies: How violence, public insecurity, economic inequality and uneven democratic performance shape journalists’ perceived work environments. Journalism Studies 18(5): 645–665.

25.

Levitsky

Way

(2010) Competitive Authoritarianism: Hybrid Regimes after the Cold War. Cambridge: Cambridge University Press.

26.

McFaul

(2005) Transitions from postcommunism. Journal of Democracy 16(3): 5–19.

27.

Murrell

(2010) Baghdad bureaux: An exploration of the interconnected world of fixers and correspondents at the BBC and CNN. Media, War & Conflict 3(2): 125–137.

28.

O’Donnell

(1993) On the state, democratization and some conceptual problems: A Latin American view with glances at some post-communist countries. World Development 21: 1355–1369.

29.

O’Donnell

Schmitter

Whitehead

(eds) (1986) Transitions from Authoritarian Rule: Comparative Perspectives, Vol. 3. Baltimore, MD: Johns Hopkins University Press.

30.

Salazar

(2019) Strategic allies and the survival of critical media under repressive conditions: An empirical analysis of local Mexican press. The International Journal of Press/Politics 24: 341–362.

31.

Saul

(2008) The international protection of journalists in armed conflict and other violent situations. Australian Journal of Human Rights 14(1): 99–140.

32.

Schedler

(2006) Electoral Authoritarianism: The Dynamics of Unfree Competition. Boulder, CO: Lynne Rienner.

33.

Tumber

(2002) Reporting under fire: The physical safety and emotional welfare of journalists. In: Zelizer

Allan

(eds) Journalism After September 11. Oxford: Taylor & Francis, pp. 247–262.

34.

Von Holdt

(2014) On violent democracy. The Sociological Review 62: 129–151.

35.

Waisbord

(2000) Watchdog Journalism in South America: News, Accountability, and Democracy. New York: Columbia University Press.

36.

Zakaria

(1997) The rise of illiberal democracy. Foreign Affairs 76(6): 22–43.