Internet,Political Regime and Terrorism: A Quantitative Analysis

Abstract

The Internet provides a medium for the rapid mobilization of dissatisfied citizens and potentially contributes to various forms of political instability, including terrorism. However, the spread of the Internet may not lead to a higher intensity of terrorist attacks because direct perpetrators rely on close personal offline ties, and the national security agencies derive symmetrical benefits from Internet development as terrorists. In addition, the number of connections proxies a general level of country development, which is associated with less terrorist activity. We analyze the relationship between the number of Internet connections and the intensity of terrorist attacks using time-series cross-sectional data from the Global Terrorism Database from 1970 to 2018. Estimation of negative binomial regression models demonstrates an inverse relationship between Internet proliferation and the number of terrorist attacks, which holds for democracies and is absent for autocracies. Our results suggest that Internet proliferation is not a decisive factor in terrorism activity. Its impact on terrorism depends on the type of political regime and the level of socio-economic development.

Keywords

internet terrorism destabilization political regimes media

Introduction

The rapid development of communication technologies has been noted for most countries worldwide in the last 20 years. In 2020, 58.7% of the world’s population had used the Internet. The growth of Internet usage in 2000–2020 is estimated from 1.9% in Asia to 592% in Europe. The penetration rates range from 39.3% in Africa to 94.6% in North America (Internet World Stats, 2020). Because the World Wide Web has significantly increased the volume, diversity, and speed of information, several researchers have suggested a direct relationship between the Internet and different forms of socio-political destabilization: anti-government demonstrations (Ruijgrok, 2017), riots (Amorim et al., 2018; Fuchs, 2012), and terrorist attacks (Conway, 2017; Schumacher & Schraeder, 2019). However, there is a lack of time-series cross-sectional studies focusing on Internet proliferation and terrorist activity in countries with different political regimes. We address this research gap by analyzing how the relative number of Internet connections affects the intensity of terrorist attacks in 196 countries from 1970 to 2018.

Contrary to the previous literature distinguishing between two broad categories of democracies and autocracies, we expand analysis with a more nuanced classification of political regimes. Our study provides empirical evidence of an inverse relationship between Internet penetration and terrorist activity, driven by the sample of democratic countries, especially partial democracies with factionalism. For the latter category, the magnitude of the effect is the highest compared to other types of political regimes.

The following section provides a literature overview of how the Internet affects terrorist activity. Then, the article proceeds with the hypotheses, methodology and empirical analysis. The final section discusses the results and offers conclusions.

Literature

Internet and Socio-Political Instability

Kris Ruijgrok (2017, 499) identifies cyber utopians and cyber pessimists among researchers of the Internet and socio-political instability. Cyber utopians argue that the World Wide Web provides citizens with broad access to alternative political information and space to discuss politics (see Shirky, 2009; Weber et al., 2003). The higher Internet penetration rate contributes to greater public awareness of domestic and international events and creates a potential for non-violent protest mobilization. At the same time, the suppression of non-violent protests by the government can provide incentives to the radical groups for a terrorist activity to deliver political messages and draw attention to specific problems.

On the contrary, cyber pessimists point out the inefficiency of the Internet as a tool to increase citizen engagement in political life. Thus, most of the population, especially in authoritarian countries, prefers to use the World Wide Web for non-political purposes (Morozov, 2011). In addition, the Internet contributes to the between-country inequality in a potential impact of the population on politics due to the different Internet coverage rates. Moreover, there is a within-country inequality between the younger and older generation (Margolis, 2007). Younger people interested in politics can quickly mobilize and spread their messages in an online environment because they constantly use the Internet. In contrast, older people do not have the same mobilization potential because they spend less time online than younger generations.

More recent literature has abandoned the dichotomy of cyber optimists versus cyber pessimists, paying more attention to how various actors and social contexts interact with the World Wide Web. In line with this trend, Internet proliferation in combinations with other factors such as education, economic development and political regime have different effects on violent and non-violent forms of political destabilization. Thus, the expansion of social networks and new digital media increases the potential for non-violent mobilization of the dissatisfied population. Nevertheless, the Internet access itself might not affect the possible transformation of non-violent protests into violent ones. Except for coup d’états and terrorist attacks, actions associated with socio-political instability are likely to become violent only after harsh government responses (McCormick & Giordano, 2007, 309–311). Consequently, Internet proliferation is likely to directly affect non-violent forms of political mobilization (general strikes and anti-government demonstrations) as a medium for disseminating information. However, the spread of the World Wide Web might not cause higher levels of violent destabilization (riots, terrorism, and guerilla warfare) because national security agencies employ effective defense strategies.

Internet, Terrorist Organizations, and Individual Radicalization

The previous research has focused on the relationship between the Internet, activities of terrorist groups and individual radicalization. Because of the fast spread of information and anonymity of users, the World Wide Web provides terrorists with an appropriate environment for propagating extremist views, mobilizing supporters, and attracting funding (Britz, 2010; Von Behr et al., 2013). Specialized terrorist websites operate as online libraries of ideological texts, platforms for recruiters and forums for information sharing (Zeman et al., 2017). In addition, photo and video materials, games, training aids and technical instructions prepared by terrorist groups contribute to the radicalization of their supporters (see Holt et al., 2015). The “electronic jihad” declared by Al-Qaeda aimed to spread the ideas of jihad among the Muslim population of Western Europe is a prominent example of how terrorist groups actively use the Internet (Rudner, 2017).

The Internet promotes the self-radicalization of people by providing opportunities to get more relevant information on thematic forums and social media. At the same time, organized terrorist groups search for radicalized individuals, as they can be recruited, trained, and sent to perform tasks at relatively low cost (Simon, 2013, 20). In turn, lone-wolf terrorists tend to communicate with like-minded people online if they have more offline social interactions (Gill & Corner, 2015). Right-wing extremists who plan terrorist attacks and recruit potential supporters are more likely to engage in online training than people of similar views but not involved in terrorist activity (Gill et al., 2017). Even though the Internet provides broad opportunities for the expansion of terrorist organizations and radicalization of individuals, it does not mean that the World Wide Web itself contributes to a higher intensity of terrorist attacks because of two reasons. At first, radicalization depends on interactions between potential terrorists irrespective of offline or online mode of communication. Secondly, direct perpetrators of terrorist attacks rely more on personal offline interactions.

David Benson (2014) has proposed an alternative explanation of why the Internet spread is not related to the higher intensity of terrorism. According to this view, the state security agencies use the Internet with at least the same degree of efficiency as terrorist groups, offsetting the advantages of the World Wide Web in terms of anonymity and speed of information (Benson, 2014, 293). The comparison of Al-Qaeda terrorist attacks and countermeasures of security agencies did not provide empirical evidence that the Internet was an additional advantage to the terrorist group. Furthermore, most of the successful Al-Qaeda terrorist attacks after 2005, when the group began to use the Internet for coordinating transnational attacks, were organized at the local level through personal communication of the perpetrators without using digital technologies (Benson, 2014).

Freedom of the Online Press, Political Regime and Terrorism

The media, including online sources, make a separate contribution to the potential terrorist activity. On the one hand, terrorists are trying to achieve maximum publicity for delivering their messages to a broad audience (Wilkinson, 2006, 152). On the other hand, the media resources compete for the public, which determines their desire to spread the news as quickly as possible. Supporters of the publicity argument believe that media freedom in democracies encourages terrorists to select them as targets for attacks (Chenoweth, 2013, 352; Hoffman, 2006). Nevertheless, with the development of the Internet, terrorist groups can be widely publicized in countries with authoritarian political regimes and lose incentives to attack more democratic countries. In autocracies, official media tend to underreport the incidences of terrorist attacks. Consequently, the increase of Internet users and the corresponding access to alternative information potentially contributes to a greater intensity of terrorism in countries with authoritarian political regimes since more terrorist attacks begin to be made public. At the same time, the expansion of the Internet in democracies may not lead to the more prominent terrorist activity since democracies had relatively high press freedom even before the Internet era.

We acknowledge the importance of studies that offer theoretical mechanisms for a direct relationship between the spread of the Internet and the intensity of terrorist attacks. However, we allege that the relationship between the two phenomena at an aggregated level is more complex, depending on the accompanying political, economic, and social factors. Thus, countries with various types of political regimes experience different levels of terrorist activity. Numerous studies reveal that the highest intensity of terrorism is inherent in hybrid regimes which combine democratic and authoritarian institutional elements (Aksoy et al., 2012; Chenoweth, 2013). For instance, partial autocracies conduct regular elections, but they are not always fair and competitive. In turn, unconsolidated autocracies allow opposition parties, but the appointed character of legislatures limits their political participation. As a result of the inconsistent instructional structure, countries with hybrid regimes are more prone to socio-political instability than consolidated democratic and authoritarian countries.

Hypotheses

In general, we expect a positive relationship between Internet penetration and terrorist activity. The Internet provides opportunities for terrorist groups to deliver their messages to a broad audience, find new supporters, gain financial support, and radicalize individuals. In other words, the World Wide Web plays a role of a medium for organizational activities of terrorists. At the same time, the Internet increases the amount of information available to people. Consequently, digital media can report more terrorist attacks compared to the traditional media before the massive spread of the World Wide Web. In addition, incumbents have less chances to control information as much as before the Internet underreporting terrorist incidents to prevent reputational costs.

The relative number of Internet connections is associated with a higher intensity of terrorist attacks.

Nevertheless, the effect of Internet proliferation on the intensity of terrorist attacks can be heterogenous in different political regimes. Thus, autocratic leaders control all aspects of the political, social, and economic realm, including access to independent media and freedom of speech on the Internet (Guriev & Treisman, 2019; King et al., 2013; Rød & Weidmann, 2015). As a result, they suppress benefits of the World Wide Web as a medium for mobilization of dissatisfied citizens and organizational activities of terrorist groups. Hypothesis 2 below follows this reasoning.

The relative number of Internet connections is associated with a lower intensity of terrorist attacks in countries with authoritarian political regimes.

At the same time, we do not expect to find a robust relationship in countries with democratic political regimes since they had experienced relative media freedom before the Internet era. In countries with democratic institutions, governments have not controlled all information channels, so some media could report terrorist attacks, even if state media underreported them:

There is no relationship between the relative number of Internet connections and the number of terrorist attacks in countries with democratic political regimes.

Methodology

Data

We employ the Global Terrorism Database (START, 2021) to test the hypotheses about the relationship between Internet proliferation and terrorist activity in countries with different political regimes. The original database reports more than 200 thousand terrorist incidents in 214 countries and territories from 1970 to 2018. A terrorist attack is defined as a “the threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation” (START, 2021, 10). The database captures terrorist incidents, if they are intentional, “entail some level of violence or immediate threat of violence”, and are perpetrated by sub-national actors (START, 2021, 10–11). In addition, an incident is classified as a terrorist attack if it meets at least two out of three additional criteria: the aim to attain a political, economic, religious, or social goal, the intention of the perpetrators to convey a specific message to a larger audience than the immediate victims of an attack, and the commission of an incident outside the context of legitimate warfare activities (START 2021, 12). The main dependent variable of our interest is the number of terrorist attacks in a country-year. It should be mentioned that the authors of data made significant methodological changes in 2012 (Jensen, 2013). They significantly increased the range of primary media sources to derive information about terrorist attacks, started to use natural language processing and re-organized coding team by specific domains and groups of variables. As a result, the exponential growth of detected terrorist attacks after 2011 can be explained not only by the massive wave of socio-political destabilization due to the Arab Spring, but also with the changes in methodology. We address this issue by employing negative binomial models, accounting for a time-series cross-sectional structure of data with regional and time fixed effects and including a set of socio-economic and political control variables.

In a robustness check, we test the models for an alternative measure of terrorist activity: the number of victims of terrorist attacks (START, 2021).

The main independent variable is the number of Internet connections per 100 people (World Bank, 2020). The indicator reflects the spread and intensity of Internet usage in 196 countries and includes 4898 observations from 1960 to 2018.

To address the differences in how Internet proliferation is connected to the intensity of terrorist attacks in countries with various political regimes, we employ a categorical variable, following the classification by Goldstone et al. (2010, 194–198). Their measure of political regime is based on two indicators from the Polity database (Center for Systemic Peace, 2021): the openness of executive recruitment and the competitiveness of political participation, which correspond to the two dimensions of Dahl’s forms of government: contestation and inclusiveness (Dahl, 2008 (1971)). Political regimes include the following categories (Goldstone et al., 2010, 195–196):

1. Full autocracies: absent contestation and absent inclusiveness

2. Partial autocracies: absent contestation and relatively high inclusiveness or relatively high contestation and absent inclusiveness

3. Partial democracies with factionalism: factionalism¹ with high contestation and limited inclusiveness or limited contestation and high inclusiveness

4. Partial democracies without factionalism: high contestation and limited inclusiveness or limited contestation and high inclusiveness

5. Full democracies: high contestation and high inclusiveness

Countries in the database have different values of political regimes across time. Appendix 1 illustrates breakdown of countries by political regime in the latest available year, 2018, according to the methodology by Goldstone et al. (2010). It should be noted that classification based on two Polity dimensions is not without caveats. For example, Russian Federation in 2018 belongs to the category “partial democracies without factionalism”, despite authoritarian elements in an institutional structure. Nevertheless, the classification scheme provides a clearer breakdown of countries than the arbitrary split of countries according to the aggregated Polity index.

To consider socio-economic characteristics which are relevant in the context of terrorist activity, we include several control variables in regression models: the logarithm of the population, unemployment rate (according to national statistical offices, in percent), and the logarithm of inflation (consumer price index relative to the previous year, in per cent) from the World Bank database (2020). Moreover, we control for the logarithm of GDP per capita (Pemstein et al., 2020) and the share of the urban population (The United Nations Population Division, 2015).

On average, more populated countries experience more terrorist attacks because terrorist groups can recruit more potential supporters (Piazza, 2006). In addition, more populated countries provide a broad audience for intimidation and delivery of terrorist messages and the additional costs of counterterrorist measures because they might suppress citizen rights.

Higher unemployment contributes to citizens’ dissatisfaction with the economic situation in the country and can lead to increased terrorist activity if the non-violent forms of protest do not change the status-quo. Terrorist organizations can mobilize new supporters who are dissatisfied with their economic situation and are ready to resort to radical measures to draw attention to the problem (Caruso & Schneider, 2011). Consistent with this reasoning, higher unemployment rate is associated with a higher number of terrorist attacks (Adelaja & George, 2020).

Inflation is potentially associated with the intensity of terrorist attacks in the same way as unemployment, contributing to the dissatisfaction of radicalized individuals with an economic situation (Shahbaz, 2013). The increase in inflation, particularly for food products, is expected to be associated with the growth of violent and non-violent forms of socio-political destabilization.

The level of economic development, proxied by the logarithm of GDP per capita, is most likely curvilinearly associated with terrorist activity (Korotayev et al., 2019). Countries with an average level of economic development are most susceptible to terrorist activity. The potential benefits of terrorist attacks in the least developed countries are relatively small. In contrast, the most developed countries can sustain an effective system of counterterrorist agencies.

Urbanization, expressed in the proportion of the urban population, can be associated with a higher terrorist activity, as terrorists choose the most populous places to carry out attacks, which are more typical for cities than for rural settlements (Campos & Gassebner, 2013; Tavares, 2004). Indirectly, a higher proportion of the urban population can be a proxy for economic development since the most productive sectors of the economy (the manufacturing sector in developing countries and the services sector in developed countries) concentrate in urban areas.

Furthermore, we control for a set of political and cultural variables that can affect the terrorist intensity: ethnic fractionalization, Polity5 score, political participation, and domestic instability. The index of ethnic fractionalization, based on the methodology of Alesina (2003), corresponds to “the probability that two randomly drawn individuals within a country are not from the same ethnic group” (Drazanova, 2019). The measure of regimes, alternative to classification by Goldstone et al. (2010), is the Polity5 index of Political Regime Characteristics and Transitions (Center for Systemic Peace, 2021). To account for different dimensions of political regimes, we employ an index of political participation in a robustness check (Freedom House, 2021). Finally, we account for general domestic instability with the aggregated index of socio-political destabilization, normalized by million people, from the Cross-National Time-Series Data Archive (Banks & Wilson, 2021).

The regression models include regional fixed effects to account for shocks specific to different geographic regions resulting from various forms of regional political violence and unrest. We control for 12 regions from the Global Terrorism Database (START, 2021): Eastern Europe, Western Europe, South Asia, Southeast Asia, East Asia, Central Asia, Middle East and North Africa, Sub-Saharan Africa, Australasia and Oceania, North America, South America, and Central America and Caribbean. Appendix 2 presents the breakdown of countries by region.

Table 1 presents the descriptive statistics for the main dependent, independent and control variables. The number of terrorist attacks ranges from 0 to 3774 cases recorded by the Global Terrorism Database. The standard deviation and variance (squared standard deviation) significantly exceed the mean of the dependent variable, providing a reason for using negative binomial regression models. Descriptive statistics of independent and control variables indicate the heterogeneity of observations and the need to include fixed effects.

Table 1.

Descriptive Statistics.

Variable	N	Min	Max	Mean	Sd
Terrorist attacks	9630	0	3774	18.543	113.306
Internet connections	4904	0	99.702	23.04	28.512
Political regime	11257	1	5	2.794	1.524
Population (log)	14009	−0.446	14.169	8.1	2.314
GDP per capita (log)	5354	6.079	11.656	9.087	1.2
Unemployment	3408	0.05	57	8.154	6.289
Urban share	14009	1.7	100	46.813	25.22
Inflation (log)	7250	−5.073	10.076	1.686	1.288
Ethnic fractionalization	7636	0.001	0.89	0.425	0.275
Polity	16303	−10	10	0.21	6.171
Political rights	8650	1	7	3.738	2.237
Political instability	10725	0	427397.3	347.531	6077.921

Methods

We use the negative binomial regression models appropriate for a count dependent variable with a non-normal distribution to test our hypotheses. In our case, the distribution resembles a general negative binomial form with a concentration of values around zero, meaning the absence of terrorist attacks in a specific country-year (see Figure 1). In this case, the ordinary least squares estimates are biased and ineffective since they assume a normal distribution of the outcome. We do not choose the Poisson general linear models because they require equality between the variance and the mean of the dependent variable. In turn, the negative binomial model allows the variance to exceed the mean with an additional parameter theta (Allison & Waterman, 2002, 250). We include region and time fixed effects to address the unobserved heterogeneity and the omitted variable bias.

Figure 1.

Distribution of the number of terrorist attacks (logged).

The negative binomial regression formula for the full sample models can be presented with the following equation

\begin{array}{l} l o g (n_t e r r o r_a t t a c k s_{i t}) = β_{0 i t} + β_{1} i n t e r n e t_c o n n e c t i o n s_{i t} + β_{2} (p o l i t i c a l r e g i m e_{i t} = 1) + β_{3} (p o l i t i c a l r e g i m e_{i t} = 2) \\ + β_{4} (p o l i t i c a l r e g i m e_{i t} = 3) + β_{5} (p o l i t i c a l r e g i m e_{i t} = 4) + β_{6} (p o l i t i c a l r e g i m e_{i t} = 5) + (β_{7} \dots β_{k}) {c o n t r o l s_{i t}} + ϵ \end{array}

where i denotes the country, t – the year, k – the number of predictors, $ϵ$ – the random error term.

Furthermore, we divide the entire sample into two sub-samples: authoritarian countries (full autocracies and partial autocracies) and democratic countries (full democracies, partial democracies with factionalism and partial democracies without factionalism). The sample split allows checking whether the relationships between the predictors and the outcome are similar for all observations or driven by a specific subset. Instead of mid-level regime measurement by Goldstone et al. (2010), we employ aggregated Polity5 index (Center for Systemic Peace, 2021) with a range of values from −10 (full autocracy) to 10 (full democracy). The model specifications for separate authoritarian and democratic countries are as follows

\log (n_t e r r o r_a t t a c k s_{i t}) = β_{0 i t} + β_{1} i n t e r n e t_c o n n e c t i o n s_{i t} + β_{2} (p o l i t i c a l r e g i m e_{i t}) + + (β_{3} \dots β_{k}) {c o n t r o l s_{i t}} + ϵ

where i denotes the country, t – the year, k – the number of predictors, $ϵ$ – the random error term.

Results

Main Models

The Pearson correlation coefficient between the number of Internet connections per 100 people and the intensity of terrorist attacks has a low negative value, indicating the need to include the proposed control variables and split the sample in a more detailed assessment of the relationship between the two variables (Table 2). However, due to the relatively high values of Pearson correlations between GDP per capita and the number of Internet connections per 100 people and GDP per capita and the share of urban population regression models may encounter the problem of multicollinearity (the presence of the relationship between explanatory variables). We address this issue with the robust standard errors clustered by region.

Table 2.

Pearson Correlations between the Variables.

	Terrorist Attacks	Internet Connections	Polit Regime	Population (Log)	GDP Pc (Log)	Unempll	Urban Share	Inflation (Log)	Ethn. Frac	Polity	Pol Rights
Terrorist attacks
Internet connections	−0.088
Polit regime	−0.117	0.325
Population (log)	0.236	−0.115	−0.268
GDP pc (log)	−0.129	0.631	0.513	−0.213
Unempl	−0.044	−0.086	0.040	−0.114	−0.030
Urban share	−0.120	0.458	0.318	−0.143	0.745	0.043
Inflation	0.104	−0.386	−0.212	0.062	−0.368	−0.029	−0.228
Ethn. Frac	0.155	−0.111	−0.184	0.140	−0.294	−0.012	−0.155	0.119
Polity	−0.024	0.220	0.879	−0.174	0.348	0.109	0.186	−0.090	−0.100
Pol rights	0.133	−0.321	−0.883	0.259	−0.514	−0.126	−0.348	0.181	0.133	−0.890
Instability	0.076	−0.127	−0.110	−0.138	−0.129	0.061	−0.08	0.110	−0.013	−0.042	0.073

Table 3 and 4 present the main regression results. Table 3 provides initial estimates. The coefficients demonstrate how the expected log count of terrorist attacks changes for a one-unit increase in a predictor. We report exponentials of initial estimates in Table 4 to understand the relative incident rate ratios of terrorist attacks. The exponential of each original value is calculated by raising a mathematical constant e to a power of the corresponding value²: exp(x) = e^x, where x is the initial coefficient from the Table 3. The resulted values in Table 4 indicate the incident rate ratios of the dependent variable with the one-unit increase of the value of the independent variable. For example, the value of internet connections in Model 1 (Table 3) is -0.018. It corresponds to the exponential of the value in Table 4: e^-0.018=0.982. This is an incident ratio rate of the number of terrorist attacks when the number of internet connections increases by one unit

\frac{the number of terrorist attacks when the number of Internet connections = x + 1}{the number of terrorist attacks when the number of Internet connections = x}

Table 3.

Internet Connections and Terrorist Attacks by Regime Type.

	Number of Terrorist Attacks
	Full Sample	Full Sample	Democracies	Autocracies
	(1)	(2)	(3)	(4)
Internet connections	−0.018^**	−0.020^**	−0.018^*	0.013
	(0.008)	(0.008)	(0.010)	(0.052)
Partial autocracy	0.851
	(0.676)
Part. dem. with fact	1.768^***
	(0.666)
Part. dem. without fact	0.951
	(0.689)
Full democracy	0.967
	(0.739)
Polity		0.054	−0.023	0.234
		(0.048)	(0.045)	(0.157)
Population (log)	0.971^***	1.002^***	0.938^***	1.426^***
	(0.108)	(0.102)	(0.112)	(0.344)
GDP per capita (log)	0.403^**	0.384^*	0.061	1.109^***
	(0.205)	(0.211)	(0.223)	(0.427)
Unemployment	0.039^**	0.032	0.066^***	0.072^***
	(0.019)	(0.021)	(0.018)	(0.016)
Inflation (log)	−0.010	0.009	−0.017	0.072
	(0.078)	(0.077)	(0.097)	(0.333)
Urban share	−0.027^***	−0.024^*	−0.031^***	−0.024
	(0.010)	(0.012)	(0.012)	(0.032)
Political instability	0.004^***	0.005^***	0.004^***	0.002^***
	(0.0003)	(0.0003)	(0.0005)	(0.001)
Ethnic fractionalization	1.688^***	1.715^***	1.752^**	−1.210
	(0.339)	(0.524)	(0.747)	(1.134)
Region fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes
Observations	1323	1330	1075	248
Log likelihood	−3200.127	−3239.919	−2525.989	−637.545
Theta	0.345^*** (0.017)	0.333^*** (0.017)	0.380^*** (0.022)	0.374^*** (0.043)
Akaike Inf. Crit	6448.254	6521.838	5093.978	1311.089

Negative binomial models with robust standard errors, clustered at the region level. Baseline regime category: full autocracy. Standard errors are reported in parentheses.

p-values significance: * p<0.1; **p<0.05; ***p<0.01.

Table 4.

Internet Connections and Terrorist Attacks by Regime Type (Incident Rate Ratios).

	Number of Terrorist Attacks
	Full Sample	Full Sample	Democracies	Autocracies
	(1)	(2)	(3)	(4)
Internet connections	0.982^**	0.980^**	0.982^*	1.013
	(0.008)	(0.008)	(0.010)	(0.052)
Partial autocracy	2.341
	(1.582)
Part. dem. with fact	5.860^***
	(3.905)
Part. dem. without fact	2.589
	(1.784)
Full democracy	2.630
	(1.942)
Polity		1.056	0.978	1.264
		(0.051)	(0.044)	(0.199)
Population (log)	2.642^***	2.723^***	2.556^***	4.162^***
	(0.286)	(0.279)	(0.287)	(1.432)
GDP per capita (log)	1.497^**	1.469^*	1.063	3.032^***
	(0.306)	(0.310)	(0.237)	(1.293)
Unemployment	1.040^**	1.033	1.068^***	1.075^***
	(0.019)	(0.022)	(0.019)	(0.018)
Inflation (log)	0.990	1.009	0.983	1.075
	(0.077)	(0.078)	(0.095)	(0.358)
Urban share	0.973^***	0.976^*	0.970^***	0.976
	(0.010)	(0.012)	(0.011)	(0.032)
Political instability	1.004^***	1.005^***	1.004^***	1.002^***
	(0.0003)	(0.0003)	(0.0005)	(0.001)
Ethnic fractionalization	5.406^***	5.554^***	5.767^**	0.298
	(1.835)	(2.908)	(4.307)	(0.338)
Region fixed effects	Yes	Yes	Yes	Yes
Time fixed effects	Yes	Yes	Yes	Yes
Observations	1323	1330	1075	248
Log likelihood	−3200.127	−3239.919	−2525.989	−637.545
Theta	0.345^*** (0.017)	0.333^*** (0.017)	0.380^*** (0.022)	0.374^*** (0.043)
Akaike Inf. Crit	6448.254	6521.838	5093.978	1311.089

Negative binomial models with robust standard errors, clustered at the region level. Exponentials of coefficients from Table 3. Standard errors are calculated by multiplying exponentials of coefficients by standard errors for initial non-exponentiated coefficients from Table 3. Standard errors are reported in parentheses. Baseline regime category: full autocracy. p-values significance: * p<0.1; **p<0.05; ***p<0.01.

Thus, it is possible to extract the resulting value from 1 and get the incident rate change with a one unit increase in the predictor.

For a total sample (Models 1 and 2), the number of Internet connections per 100 people is statistically significant with a p-value less than 0.05. The coefficient value less than 1 indicates lower incident rates of terrorist attacks with an increase in Internet proliferation. More precisely, with a one-unit increase of Internet connections per 100 people, the number of terrorist attacks decreases by 1.8% ((1-0.982)*100%). The exact magnitude and significance of the relationship between the main independent variable and outcome are present for a sample of democracies (Model 3). For a sample of autocracies (Model 4), Internet proliferation is not a statistically significant predictor of terrorism.

Model 1 includes estimates for separate categories of political regimes, classified with methodology from Goldstone et al. (2010). The baseline category is full autocracies. The regression coefficients for four categories of political regimes mean the estimated number of terrorist attacks relative to full autocracies. The estimate for partial democracies with factionalism is statistically significant: countries from this group experience 5.86 times more terrorist attacks per year than full autocracies, given that all other variables are equal to zero. In turn, the coefficients for three other categories of political regimes (partial autocracies, partial democracies without factionalism and full democracies) are not statistically significant, meaning that countries from these groups do not have significantly different levels of terrorist activity in comparison to consolidated autocracies.

Models 2–4 include an alternative indicator of political regimes: the aggregated Polity5 score. The corresponding estimates are not significant, suggesting the need for more disaggregated measures of regimes, like one provided by Goldstone et al. (2010), to trace the differences in terrorist activity. The Polity5 score includes a lot of sub-indicators, which provide a complex estimation of the political regime but do not allow checking the possible relationships between components and terrorism. Goldstone et al.'s measure addresses this limitation of the Polity5 index by reducing the number of dimensions to two and capturing one of the essential regime characteristics.

Among socio-economic control variables, the logarithm of population and share of the urban population are significant for all models. More populated and less urbanized countries have higher terrorism incident rates, taking all other variables equal to zero. Coefficients for unemployment are positive and statistically significant for all regression specifications except Model 2. GDP per capita is significant for the whole sample and autocracies, while inflation is not associated with terrorism in all regression specifications.

Domestic instability demonstrates statistically significant positive coefficients among political control variables for all models. A country with a general non-stable socio-political environment will likely experience a higher terrorist incidents rate because terrorists face fewer obstacles in organizing attacks. In turn, the index of ethnic fractionalization is positive and statistically significant for models on a total sample and sub-sample of democracies. Many domestic political conflicts have an ethnic nature, and the association is likely to be driven by partial democratic countries with factionalism. Often, ethnic and religious motivations drive terrorism activity in multi-national polarized societies. A prominent example is ETA (Euskadi ta Askatasuna, “Euskadi and Freedom”) in the Basque region in Spain (Zulaika & Murua, 2017).

To better understand the magnitude of effects for the primary independent variable, the number of Internet connections per 100 people, we calculated the predicted number of terrorist attacks disaggregated for the types of political regimes and plotted them in Figure 2. The graph shows the negative trend for all categories, but the slope is much greater for partial democracies with factionalism than other regimes.

Figure 2.

Predicted values by regime type.

In general, we have reasons to reject all three hypotheses. More Internet connections per 100 people are associated with fewer terrorist attacks in a country year. The effect remains statistically significant on a sub-sample of democracies, while it loses significance on a sub-sample of autocracies. Among all types of political regimes, identified by methodology from Goldstone et al. (2010), partial democracies with factionalism have the maximum number of terrorist attacks and the most profound effect of Internet connections.

Robustness Checks

To check the robustness of our empirical results, we run a set of alternative regression models. The first robustness check includes an alternative dependent variable: the number of casualties from terrorist attacks (START, 2021). We exclude GDP per capita from the model on a sample of democracies because it is most correlated with other controls. For a model on a sub-sample of autocracies, we exclude the logarithm of inflation due to the small number of complete observations. Table 5 reports the results of the first robustness check. The coefficients for the Internet connections are not statistically significant in Model 1 (total sample) and Model 2 (democracies). At the same time, an estimate of Internet proliferation is positive and statistically significant for autocracies. Inconsistency in the relationships between the number of internet connections per 100 people and different measures of terrorist activity can be explained by the different aspects of terrorism spread measured by an alternative dependent variable. While the main outcome of interest (the number of terrorist attacks) corresponds to the quantitative aspect of terrorist activity, the alternative indicator (the number of casualties) measures the severity of the attacks.

Table 5.

Internet Connections and Terrorist Attacks Casualties (Incident Rate Ratios).

	Number of Terrorist Attacks Casualties
	Full Sample	Democracies	Autocracies
	(1)	(2)	(3)
Internet connections	1.010	1.010	0.971^**
	(0.013)	(0.012)	(0.013)
Partial autocracy	1.067
	(0.841)
Part. dem. with fact	2.250
	(1.715)
Part. dem. without fact	1.145
	(0.485)
Full democracy	0.485
	(0.473)
Polity		0.702^***
		(0.059)
Political rights			1.225
			(0.164)
Population (log)	2.485^***	2.997^***	3.358^***
	(0.461)	(0.437)	(0.476)
GDP per capita (log)	0.599^*		1.206
	(0.174)		(0.241)
Unemployment	1.017	0.985	1.151^***
	(0.050)	(0.043)	(0.039)
Inflation (log)	1.206	1.287^*
	(0.183)	(0.182)
Urban share	0.982^**	0.983	0.981
	(0.007)	(0.017)	(0.015)
Political instability	1.005^***	1.010^***	1.003^***
	(0.0004)	(0.001)	(0.0004)
Ethnic fractionalization	1.899	3.474^*	3.406^***
	(1.837)	(2.525)	(1.332)
Observations	1323	1090	293
Log likelihood	−2699.012	−2068.686	−808.803
Theta	0.150^*** (0.009)	0.118^*** (0.008)	0.148^*** (0.015)
Akaike Inf. Crit	5446.025	4155.371	1635.606

Negative binomial models with robust standard errors, clustered at the region level. Exponentials of the initial coefficients. Standard errors are calculated by multiplying exponentiated coefficients by standard errors for initial non-exponentiated coefficients. Standard errors are reported in parentheses. Baseline regime category: full autocracy. p-values significance: * p<0.1; **p<0.05; ***p<0.01.

Since the main regression results show the different signs of an association between the Internet spread and terrorist attacks in democracies and autocracies, we consider an alternative sample split to check the validity of the results. We divided the sample into full autocracies and other regimes. The direct relationship becomes significant in the most model specification for full autocracies (Supplemental Table S1) and remains inverse for countries with democratic and hybrid political regimes (Supplemental Table S2). These findings support our main results. Countries with different political regimes experience the heterogenous effects of Internet proliferation on the intensity of terrorist attacks.

Finally, we check whether the main results hold for sub-samples of countries divided by the level of socio-economic development rather than political regime (Supplemental Table S3). The samples include countries with GDP per capita values below the fifth quantile (Model 1) and above or equal the fifth quantile (Model 2). In addition, we used the same principle to split samples for Models 3 and 4 into countries with the various average number of years of adult education (Pemstein et al., 2020). The negative effect of the number of Internet connections on the number of terrorist attacks is 7.5 times higher for less economically developed countries compared to more economically developed countries and 6.3 times higher for countries with a relatively low level of adult education compared to countries with a relatively high level of education. In this way, the third robustness check supported the inverse relationship between Internet proliferation and terrorist attacks. Moreover, it illustrated the difference in magnitudes of effect depending on the level of socio-economic development.

Discussion and Conclusion

The results of regression analysis provide evidence against our hypotheses. Contrary to hypothesis 1, we have found an inverse relationship between the number of Internet connections per 100 people and the number of terrorist attacks in a country year. This inverse relationship is present for a sub-sample of democracies and absent for authoritarian countries. The magnitude of the negative effect is significantly higher for partial democracies with factionalism.

We suggest three explanations of these empirical results. First, direct perpetrators of terrorist attacks may rely more on personal offline connections between each other rather than online medium because the Internet does not provide the same safety as in-person interactions while preparing a terrorist attack (Benson, 2014, 307). The cost of miscalculation when preparing a terrorist attack is fatal, so perpetrators minimize it by choosing a more secure mode of interaction with peers. The terrorists should know each other to organize a successful attack. Furthermore, anonymity and information available on the Internet are not so helpful to terrorists as others might think. There is no complete anonymity of users. Websites and Internet providers constantly trace the actions of users and can detect suspicious activity. In turn, redundant online information does not mean that radicalized individuals know how to use it properly. It is hard to choose relevant knowledge from multiple sources with various reliability and transform it into action (Benson, 2014, 306).

At the same time, government security agencies have many more resources and capacities than terrorists to extract benefits from Internet proliferation (Benson, 2014). They use the World Wide Web to monitor and detect suspicious activities and prevent potential attacks at the first stages of preparation. Cybersecurity becomes one of the main priorities for governments that invest budget resources to build a reliable protection system against the violent mobilization of dissatisfied citizens.

Finally, Internet penetration can be interpreted as a proxy for the general socio-economic development of a country. The quantitative spread of the Internet may be inversely associated with terrorist incidents because countries with more Internet connections per 100 people have a higher level of socio-economic development which means that countries have more capacities to organize adequate anti-terrorism protection.

The main limitations of our empirical results are potential multicollinearity and a relatively limited number of observations due to data unavailability. We addressed these problems by robust standard error clustered at the regional level and robustness checks with an alternative dependent variable and different sub-samples. Future research can focus on the causal mechanisms and the relationship between Internet penetration and terrorist activity in various regions.

Supplemental Material

Supplemental Material - Internet, Political Regime and Terrorism: A Quantitative Analysis

Supplemental Material for Internet, Political Regime and Terrorism: A Quantitative Analysis by Nikita Khokhlov and Andrey Korotayev in Cross-Cultural Research

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Russian Science Foundation; Project No. 18–18–00254.

Authors’ Note

This article is an output of a research project implemented as part of the Basic Research Program at HSE University in 2022 with support by Russian Science Foundation (Project No. 18–18–00254).

Data Availability

START (National Consortium for the Study of Terrorism and Responses to Terrorism). (2020) Global Terrorism Database. www.start.umd.edu/gtd/. Pemstein, D., Marquardt, K. L., Tzelgov, E., Wang, Y. T., Krusell, J., and Miri, F. (2018). The V-Dem measurement model: latent variable analysis for cross-national and cross-temporal expert-coded data. V-Dem Working Paper, 21. https://www.v-dem.net/en/data/data-version-10/. World Bank. (2020) World Development Indicators Online.

Software Information

R Development Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL . The custom code may be obtained from the corresponding author.

ORCID iDs

Nikita Khokhlov

Andrey Korotayev

Supplemental Material

Supplemental material for this article is available online.

Notes

Appendix 1

Countries by Political Regime in 2018 according to Goldstone et al., 2010

Political Regime	Country
Full Autocracy	Afghanistan
Full Autocracy	Bahrain
Full Autocracy	China
Full Autocracy	Cuba
Full Autocracy	Korea, Democratic People’s Republic of
Full Autocracy	Lao People’s Democratic Republic
Full Autocracy	Mauritania
Full Autocracy	Qatar
Full Autocracy	Saudi Arabia
Full Autocracy	Swaziland
Full Autocracy	Syrian Arab Republic
Full Autocracy	United Arab Emirates
Full Autocracy	Uzbekistan
Full Autocracy	Viet Nam
Partial Autocracy	Sudan
Partial Autocracy	Algeria
Partial Autocracy	Angola
Partial Autocracy	Azerbaijan
Partial Autocracy	Bangladesh
Partial Autocracy	Belarus
Partial Autocracy	Burundi
Partial Autocracy	Cambodia
Partial Autocracy	Cameroon
Partial Autocracy	Chad
Partial Autocracy	Comoros
Partial Autocracy	Congo
Partial Autocracy	Congo, the Democratic Republic of the
Partial Autocracy	Djibouti
Partial Autocracy	Egypt
Partial Autocracy	Equatorial Guinea
Partial Autocracy	Eritrea
Partial Autocracy	Ethiopia
Partial Autocracy	Fiji
Partial Autocracy	Gambia
Partial Autocracy	Iran, Islamic Republic of
Partial Autocracy	Jordan
Partial Autocracy	Kazakhstan
Partial Autocracy	Kuwait
Partial Autocracy	Morocco
Partial Autocracy	Oman
Partial Autocracy	Papua New Guinea
Partial Autocracy	Rwanda
Partial Autocracy	Somalia
Partial Autocracy	Tajikistan
Partial Autocracy	Thailand
Partial Autocracy	Togo
Partial Autocracy	Turkey
Partial Autocracy	Turkmenistan
Partial Autocracy	Uganda
Partial Autocracy	Venezuela, Bolivarian Republic of
Partial Democracy with Factionalism	Belgium
Partial Democracy with Factionalism	Bolivia, Plurinational State of
Partial Democracy with Factionalism	Central African Republic
Partial Democracy with Factionalism	Colombia
Partial Democracy with Factionalism	Cote d'Ivoire
Partial Democracy with Factionalism	Gabon
Partial Democracy with Factionalism	Guinea
Partial Democracy with Factionalism	Iraq
Partial Democracy with Factionalism	Israel
Partial Democracy with Factionalism	Kyrgyzstan
Partial Democracy with Factionalism	Lebanon
Partial Democracy with Factionalism	Madagascar
Partial Democracy with Factionalism	Malawi
Partial Democracy with Factionalism	Pakistan
Partial Democracy with Factionalism	Sri Lanka
Partial Democracy with Factionalism	Tanzania, United Republic of
Partial Democracy with Factionalism	Ukraine
Partial Democracy with Factionalism	United Kingdom
Partial Democracy with Factionalism	United States
Partial Democracy with Factionalism	Zambia
Partial Democracy with Factionalism	Zimbabwe
Partial Democracy without Factionalism	Armenia
Partial Democracy without Factionalism	Haiti
Partial Democracy without Factionalism	Nicaragua
Partial Democracy without Factionalism	Albania
Partial Democracy without Factionalism	Argentina
Partial Democracy without Factionalism	Benin
Partial Democracy without Factionalism	Bhutan
Partial Democracy without Factionalism	Botswana
Partial Democracy without Factionalism	Brazil
Partial Democracy without Factionalism	Bulgaria
Partial Democracy without Factionalism	Burkina Faso
Partial Democracy without Factionalism	Croatia
Partial Democracy without Factionalism	Czech Republic
Partial Democracy without Factionalism	Dominican Republic
Partial Democracy without Factionalism	Ecuador
Partial Democracy without Factionalism	El Salvador
Partial Democracy without Factionalism	Estonia
Partial Democracy without Factionalism	Georgia
Partial Democracy without Factionalism	Ghana
Partial Democracy without Factionalism	Guatemala
Partial Democracy without Factionalism	Guinea-Bissau
Partial Democracy without Factionalism	Guyana
Partial Democracy without Factionalism	Honduras
Partial Democracy without Factionalism	India
Partial Democracy without Factionalism	Indonesia
Partial Democracy without Factionalism	Jamaica
Partial Democracy without Factionalism	Kenya
Partial Democracy without Factionalism	Korea, Republic of
Partial Democracy without Factionalism	Kosovo
Partial Democracy without Factionalism	Latvia
Partial Democracy without Factionalism	Lesotho
Partial Democracy without Factionalism	Liberia
Partial Democracy without Factionalism	Macedonia, the former Yugoslav Republic of
Partial Democracy without Factionalism	Malaysia
Partial Democracy without Factionalism	Mali
Partial Democracy without Factionalism	Mexico
Partial Democracy without Factionalism	Moldova, Republic of
Partial Democracy without Factionalism	Montenegro
Partial Democracy without Factionalism	Mozambique
Partial Democracy without Factionalism	Myanmar
Partial Democracy without Factionalism	Namibia
Partial Democracy without Factionalism	Nepal
Partial Democracy without Factionalism	Niger
Partial Democracy without Factionalism	Nigeria
Partial Democracy without Factionalism	Paraguay
Partial Democracy without Factionalism	Peru
Partial Democracy without Factionalism	Philippines
Partial Democracy without Factionalism	Romania
Partial Democracy without Factionalism	Russian Federation
Partial Democracy without Factionalism	Senegal
Partial Democracy without Factionalism	Serbia
Partial Democracy without Factionalism	Sierra Leone
Partial Democracy without Factionalism	Singapore
Partial Democracy without Factionalism	Solomon Islands
Partial Democracy without Factionalism	South Africa
Partial Democracy without Factionalism	Suriname
Partial Democracy without Factionalism	Timor-Leste
Partial Democracy without Factionalism	Tunisia
Full Democracy	France
Full Democracy	Australia
Full Democracy	Austria
Full Democracy	Cabo Verde
Full Democracy	Canada
Full Democracy	Chile
Full Democracy	Costa Rica
Full Democracy	Cyprus
Full Democracy	Denmark
Full Democracy	Finland
Full Democracy	Germany
Full Democracy	Greece
Full Democracy	Hungary
Full Democracy	Ireland
Full Democracy	Italy
Full Democracy	Japan
Full Democracy	Lithuania
Full Democracy	Luxembourg
Full Democracy	Mauritius
Full Democracy	Mongolia
Full Democracy	Netherlands
Full Democracy	New Zealand
Full Democracy	Norway
Full Democracy	Panama
Full Democracy	Poland
Full Democracy	Portugal
Full Democracy	Slovakia
Full Democracy	Slovenia
Full Democracy	Spain
Full Democracy	Sweden
Full Democracy	Switzerland
Full Democracy	Taiwan, Province of China
Full Democracy	Trinidad and Tobago
Full Democracy	Uruguay

Appendix 2

Regions of the World according to the Global Terrorism Database

Region	Country
Eastern Europe	Albania
Eastern Europe	Belarus
Eastern Europe	Bosnia and Herzegovina
Eastern Europe	Bulgaria
Eastern Europe	Croatia
Eastern Europe	Czech Republic
Eastern Europe	Czechoslovakia
Eastern Europe	Estonia
Eastern Europe	German Democratic Republic
Eastern Europe	Hungary
Eastern Europe	Kosovo
Eastern Europe	Latvia
Eastern Europe	Lithuania
Eastern Europe	Macedonia, the former Yugoslav Republic of
Eastern Europe	Moldova, Republic of
Eastern Europe	Montenegro
Eastern Europe	Poland
Eastern Europe	Romania
Eastern Europe	Russian Federation
Eastern Europe	Serbia
Eastern Europe	Slovakia
Eastern Europe	Slovenia
Eastern Europe	Ukraine
Eastern Europe	Yugoslavia
Western Europe	Andorra
Western Europe	Austria
Western Europe	Belgium
Western Europe	Cyprus
Western Europe	Denmark
Western Europe	Finland
Western Europe	France
Western Europe	German FR
Western Europe	Germany
Western Europe	Greece
Western Europe	Holy See (Vatican City State)
Western Europe	Iceland
Western Europe	Ireland
Western Europe	Italy
Western Europe	Liechtenstein
Western Europe	Luxembourg
Western Europe	Malta
Western Europe	Monaco
Western Europe	Netherlands
Western Europe	Norway
Western Europe	Portugal
Western Europe	San Marino
Western Europe	Spain
Western Europe	Sweden
Western Europe	Switzerland
Western Europe	United Kingdom
South Asia	Afghanistan
South Asia	Bangladesh
South Asia	Bhutan
South Asia	India
South Asia	Maldives
South Asia	Mauritius
South Asia	Nepal
South Asia	Pakistan
South Asia	Sri Lanka
Southeast Asia	Brunei Darussalam
Southeast Asia	Cambodia
Southeast Asia	Indonesia
Southeast Asia	Lao People’s Democratic Republic
Southeast Asia	Malaysia
Southeast Asia	Myanmar
Southeast Asia	Philippines
Southeast Asia	Singapore
Southeast Asia	Thailand
Southeast Asia	Timor-Leste
Southeast Asia	Viet Nam
East Asia	China
East Asia	Japan
East Asia	Korea, Democratic People’s Republic of
East Asia	Korea, Republic of
East Asia	Mongolia
East Asia	Taiwan, Province of China
Central Asia	Armenia
Central Asia	Azerbaijan
Central Asia	Georgia
Central Asia	Kazakhstan
Central Asia	Kyrgyzstan
Central Asia	Tajikistan
Central Asia	Turkmenistan
Central Asia	Uzbekistan
Middle East and North Africa	Algeria
Middle East and North Africa	Bahrain
Middle East and North Africa	Cyprus: Turkish Sector
Middle East and North Africa	Egypt
Middle East and North Africa	Iran, Islamic Republic of
Middle East and North Africa	Iraq
Middle East and North Africa	Israel
Middle East and North Africa	Jordan
Middle East and North Africa	Kuwait
Middle East and North Africa	Lebanon
Middle East and North Africa	Libya
Middle East and North Africa	Morocco
Middle East and North Africa	Oman
Middle East and North Africa	Palestine, State of
Middle East and North Africa	Qatar
Middle East and North Africa	Saudi Arabia
Middle East and North Africa	Syrian Arab Republic
Middle East and North Africa	Tunisia
Middle East and North Africa	Turkey
Middle East and North Africa	United Arab Emirates
Middle East and North Africa	Yemen
Middle East and North Africa	Yemen North
Middle East and North Africa	Yemen People’s Republic
Sub-Saharan Africa	Angola
Sub-Saharan Africa	Benin
Sub-Saharan Africa	Bophutswana
Sub-Saharan Africa	Botswana
Sub-Saharan Africa	Burkina Faso
Sub-Saharan Africa	Burundi
Sub-Saharan Africa	Cameroon
Sub-Saharan Africa	Central African Republic
Sub-Saharan Africa	Chad
Sub-Saharan Africa	Comoros
Sub-Saharan Africa	Congo
Sub-Saharan Africa	Congo, the Democratic Republic of the
Sub-Saharan Africa	Cote d'Ivoire
Sub-Saharan Africa	Djibouti
Sub-Saharan Africa	Equatorial Guinea
Sub-Saharan Africa	Eritrea
Sub-Saharan Africa	Ethiopia
Sub-Saharan Africa	Gabon
Sub-Saharan Africa	Gambia
Sub-Saharan Africa	Ghana
Sub-Saharan Africa	Guinea
Sub-Saharan Africa	Guinea-Bissau
Sub-Saharan Africa	Kenya
Sub-Saharan Africa	Lesotho
Sub-Saharan Africa	Liberia
Sub-Saharan Africa	Madagascar
Sub-Saharan Africa	Malawi
Sub-Saharan Africa	Mali
Sub-Saharan Africa	Mauritania
Sub-Saharan Africa	Mozambique
Sub-Saharan Africa	Namibia
Sub-Saharan Africa	Niger
Sub-Saharan Africa	Nigeria
Sub-Saharan Africa	Rwanda
Sub-Saharan Africa	Senegal
Sub-Saharan Africa	Seychelles
Sub-Saharan Africa	Sierra Leone
Sub-Saharan Africa	Somalia
Sub-Saharan Africa	South Africa
Sub-Saharan Africa	South Sudan
Sub-Saharan Africa	Sudan
Sub-Saharan Africa	Swaziland
Sub-Saharan Africa	Tanzania, United Republic of
Sub-Saharan Africa	Togo
Sub-Saharan Africa	Transkei
Sub-Saharan Africa	Uganda
Sub-Saharan Africa	Venda
Sub-Saharan Africa	Zambia
Sub-Saharan Africa	Zimbabwe
Australasia and Oceania	Australia
Australasia and Oceania	Fiji
Australasia and Oceania	Kiribati
Australasia and Oceania	Marshall Islands
Australasia and Oceania	Micronesia, Federated States of
Australasia and Oceania	Nauru
Australasia and Oceania	New Zealand
Australasia and Oceania	Palau
Australasia and Oceania	Papua New Guinea
Australasia and Oceania	Samoa
Australasia and Oceania	Solomon Islands
Australasia and Oceania	Tonga
Australasia and Oceania	Tuvalu
Australasia and Oceania	Vanuatu
North America	Canada
North America	Mexico
North America	United States
South America	Argentina
South America	Aruba
South America	Bolivia, Plurinational State of
South America	Brazil
South America	Chile
South America	Colombia
South America	Ecuador
South America	Guyana
South America	Paraguay
South America	Peru
South America	Suriname
South America	Uruguay
South America	Venezuela, Bolivarian Republic of
Central America and Caribbean	Antigua and Barbuda
Central America and Caribbean	Bahamas
Central America and Caribbean	Barbados
Central America and Caribbean	Belize
Central America and Caribbean	Costa Rica
Central America and Caribbean	Cuba
Central America and Caribbean	Dominica
Central America and Caribbean	Dominican Republic
Central America and Caribbean	El Salvador
Central America and Caribbean	Grenada
Central America and Caribbean	Guatemala
Central America and Caribbean	Haiti
Central America and Caribbean	Honduras
Central America and Caribbean	Jamaica
Central America and Caribbean	Netherlands Antilles
Central America and Caribbean	Nicaragua
Central America and Caribbean	Panama
Central America and Caribbean	Saint Kitts and Nevis
Central America and Caribbean	Saint Lucia
Central America and Caribbean	Saint Vincent and the Grenadines
Central America and Caribbean	Trinidad and Tobago

Author Biographies

Nikita Khokhlov is a PhD student in Politics and International Relations at Dublin City University and research assistant at the International Center for Study of Institutions and Development of the HSE University. For communication with the author: nvkhokhlovhse@gmail.com.

Andrey Korotayev is a Doctor of historical sciences, Professor, Head of the Laboratory for Monitoring the Risks of Socio-Political Destabilization of the HSE University; Leading Research Fellow, Institute for African Studies of the Russian Academy of Sciences. For communication with the author: akorotayev@gmail.com

References

Adelaja

George

(2020). Is youth unemployment related to domestic terrorism? Perspectives on Terrorism, 14(5), 41–62. https://www-jstor-org-443.web.bisu.edu.cn/stable/26940038.

Aksoy

Carter

D. B.

Wright

(2012). Terrorism in Dictatorships. The Journal of Politics, 74(3), 810–826. https://doi.org/10.1017/s0022381612000400.

Alesina

Devleeschauwer, A., Easterly W., Kurlat, S. & Wacziarg R. , et al. (2003). Fractionalization. Journal of Economic growth, 8(2), 155–194. https://doi.org/10.1023/A:1024471506938.

Allison

P. D.

Waterman

R. P.

(2002). Fixed–effects negative binomial regression models. Sociological Methodology, 32(1), 247–265. https://doi.org/10.1111/1467-9531.00117.

Amorim

Costa Lima

Sampaio

(2018). Broadband Internet and Protests: Evidence from the Occupy Movement. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2764162.

Banks

A.S.

Wilson

K.A.

(2021). Cross-national time-series data archive. Databanks international. Israel. http://www.databanksinternational.com.

Benson

D. C.

(2014). Why the internet is not increasing terrorism. Security Studies, 23(2), 293–328. https://doi.org/10.1080/09636412.2014.905353.

Britz

(2010). Terrorism and technology: Operationalizing cyberterrorism and identifying concepts. In Holt

(Ed.), Crime on-line: Correlates, causes, and context. Durham: Carolina academic, (pp. 193–210).

Campos

N. F.

Gassebner

(2013). International terrorism, domestic political instability, and the escalation effect. Economics & Politics, 25(1), 27–47. https://doi.org/10.1111/ecpo.12002.

10.

Caruso

Schneider

(2011). The socio-economic determinants of terrorism and political violence in Western Europe (1994–2007). European Journal of Political Economy, 27(suppl 1), S37–S49. https://doi.org/10.1016/j.ejpoleco.2011.02.003.

11.

Center for Systemic Peace (2021). Polity5 Annual Time-Series, 1946-2018. http://www.systemicpeace.org/inscrdata.html.

12.

Chenoweth

(2013). Terrorism and democracy. Annual Review of Political Science, 16(1), 355–378. https://doi.org/10.1146/annurev-polisci-032211-221825.

13.

Conway

(2017). Determining the role of the internet in violent extremism and terrorism: Six suggestions for progressing research. Studies in Conflict & Terrorism, 40(1), 77–98. doi: 10.1080/1057610x.2016.1157408.

14.

Dahl

R. A

(2008). Polyarchy: Participation and opposition. Yale University Press.

15.

Drazanova

(2019). Historical index of ethnic fractionalisation dataset. https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/4JQRCL.

16.

Freedom House (2021). Freedom in the World. https://freedomhouse.org/report/freedom-world.

17.

Fuchs

(2012). Social media, riots, and revolutions. Capital & Class, 36(3), 383–391. https://doi.org/10.1177/0309816812453613.

18.

Gill

Corner

(2015) Lone-actor terrorist use of the internet and behavioural correlates. In Jarvis

L.L.

Macdonald

Chen

(Eds.), Terrorism Online: Politics, Law, Technology and Unconventional Violence (pp. 35–54). Routledge.

19.

Gill

Corner

Conway

Thornton

Bloom

Horgan

(2017). Terrorist use of the Internet by the numbers: Quantifying behaviors, patterns, and processes. Criminology & Public Policy, 16(1), 99–117. https://doi.org/10.1111/1745-9133.12249.

20.

Goldstone

J. A.

Bates

R. H.

Epstein

D. L.

Gurr

T. R.

Lustik

M. B.

Marshall

M. G.

Woodward

. (2010). A global model for forecasting political instability. American Journal of Political Science, 54(1), 190–208. https://doi.org/10.1111/j.1540-5907.2009.00426.x.

21.

Guriev

Treisman

(2019). Informational autocrats. Journal of Economic Perspectives, 33(4), 100–127. https://doi.org/10.1257/jep.33.4.100.

22.

Hoffman

(2006). Inside terrorism. Columbia University Press.

23.

Holt

Freilich

J. D.

Chermak

McCauley

(2015). Political radicalization on the Internet: Extremist content, government control, and the power of victim and jihad videos. Dynamics of Asymmetric Conflict, 8(2), 107–120. https://doi.org/10.1080/17467586.2015.1065101.

24.

Internet World Stats (2020). World internet usage and population statistics. www.Internetworldstats.com/stats.htm.

25.

Jensen

(2013). Discussion Point: The Benefits and Drawbacks of Methodological Advancements in Data Collection and Coding: Insights from the Global Terrorism Database (GTD). https://www.start.umd.edu/news/discussion-point-benefits-and-drawbacks-methodological-advancements-data-collection-and-coding.

26.

King

Pan

Roberts

M.E

(2013). How censorship in China allows government criticism but silences collective expression. American Political Science Review, 107(2), 326–343. https://doi.org/10.1017/s0003055413000014.

27.

Korotayev

Vaskin

Tsirel

(2019). Economic Growth, Education, and Terrorism: A Re-Analysis. Terrorism and Political Violence, 33(3), 572–595. https://doi.org/10.1080/09546553.2018.1559835.

28.

Margolis

(2007). In Dalton

R.J.

H-Klingemann

(Eds.), E'Government and DemocracyThe oxford handbook of political behavior. (pp. 765–782). Oxford University Press.

29.

McCormick

G.H.

Giordano

(2007). Things come together: symbolic violence and guerrilla mobilisation. Third World Quarterly, 28(2), 295–320. https://doi.org/10.1080/01436590601153705.

30.

Morozov

(2011). The net delusion: The dark side of internet freedom. Public Affairs.

31.

Pemstein

Marquardt

K. L.

Tzelgov

Wang

Y. T.

Krusell

Miri

(2020). The V-Dem measurement model: latent variable analysis for cross-national and cross-temporal expert-coded data. V-dem Working Paper, 21. http://doi.org/10.2139/ssrn.3167764.

32.

Piazza

J. A

(2006). Rooted in poverty?: Terrorism, poor economic development, and social cleavages. Terrorism and Political Violence, 18(1), 159–177. https://doi.org/10.1080/095465590944578.

33.

Rød

E. G.

Weidmann

N. B

(2015). Empowering activists or autocrats? The Internet in authoritarian regimes. Journal of Peace Research, 52(3), 338–351. https://doi.org/10.1177/0022343314555782.

34.

Rudner

(2017). “Electronic Jihad”: The Internet as Al Qaeda's Catalyst for Global Terror. Studies in Conflict & Terrorism, 40(1), 10–23. https://doi.org/10.1080/1057610x.2016.1157403.

35.

Ruijgrok

(2017). From the web to the streets: Internet and protests under authoritarian regimes. Democratization, 24(3), 498–520. https://doi.org/10.1080/13510347.2016.1223630.

36.

Schumacher

M. J.

Schraeder

P. J

(2019). Does domestic political instability foster terrorism? Global Evidence from the Arab Spring Era (2011–14). Studies in Conflict & Terrorism, 44(3), 1–25. https://doi.org/10.1080/1057610X.2018.1538124.

37.

Shahbaz

(2013). Linkages between inflation, economic growth and terrorism in Pakistan. Economic Modelling, 32(1), 496–506. https://doi.org/10.1016/j.econmod.2013.02.014.

38.

Shirky

(2009). Here comes everybody: How change happens when people come together. Penguin.

39.

Simon

J. D

(2013). Lone wolf terrorism: Understanding the growing threat. Prometheus Books.

40.

START (National Consortium for the Study of Terrorism and Responses to Terrorism) (2021). Global Terrorism Database. www.start.umd.edu/gtd/.

41.

Tavares

(2004). The open society assesses its enemies: shocks, disasters and terrorist attacks. Journal of Monetary Economics, 51(5), 1039–1070. https://doi.org/10.1016/j.jmoneco.2004.04.009.

42.

The United Nations Population Division (2015). Revision of World Population Prospects. www.un.org/en/development/desa/population/.

43.

Von Behr

Reding

Edwards

Gribbon

(2013). Radicalisation in the Digital Era: The Use of the Internet In 15 Cases of Terrorism and Extremism. www.rand.org/content/dam/rand/pubs/research_reports/RR400/RR453/RAND_RR453.pdf.

44.

Weber

L. M.

Loumakis

Bergman

(2003). Who participates and why? An analysis of citizens on the Internet and the mass public. Social Science Computer Review, 21(1), 26–42. https://doi.org/10.1177/0894439302238969.

45.

Wilkinson

(2006). Terrorism versus democracy: The liberal state response. Routledge.

46.

World Bank (2020). World Development Indicators Online. http://data.worldbank.org/indicator.

47.

Zeman

Břeň

Urban

(2017). Role of internet in lone wolf terrorism. Journal of Security & Sustainability Issues, 7(2). https://doi.org/10.9770/jssi.2017.7.2(1).

48.

Zulaika

Murua

(2017). How terrorism ends–and does not end: the Basque case. Critical Studies on Terrorism, 10(2), 338–356. https://doi.org/10.1080/17539153.2017.1282666.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.05 MB