Jump to ContentJump to Main Navigation
The Digital Origins of Dictatorship and DemocracyInformation Technology and Political Islam$

Philip N. Howard

Print publication date: 2010

Print ISBN-13: 9780199736416

Published to Oxford Scholarship Online: September 2010

DOI: 10.1093/acprof:oso/9780199736416.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 23 May 2017

(p.203) Appendix A: Countries in This Study

(p.203) Appendix A: Countries in This Study

Source:
The Digital Origins of Dictatorship and Democracy
Publisher:
Oxford University Press

The full dataset of all variables in the causal recipes described in the Conclusion is available at www.pitpi.org, as are the technical scripts for secondary solution sets not described here, the calibration points for specific membership sets. The fuzzy sets analyzed here were composed from the data presented in the chapters in the book, and replication data is available on the website. For more on fuzzy set calibrations see the codebook for the fs/QCA 2.0 software and Ragin (2009).

Preparing data for treatment as fuzzy set required several steps. First, I computed indices for causal attributes analyzed in each chapter, and then I computed the indices for additional context variables often recommended by the literature on democratization in the developing world. Then I calibrated the indices, a process that evens out the distribution of cases between the thresholds for full inclusion in each set, full exclusion from the set, and the crossover point at which cases go from being partially in the set to being partially out of the set.

Fuzzy Calibrations

Calibrating the fuzzy set membership of a group of countries requires judgments for the threshold of full membership in the set, full exclusion from the set, and the transition midpoint at which a country is neither in nor out of the set. For example, among the 75 countries there are a few very populated countries and many countries with a small population. Figure A.1 reveals a skewed curve that comes from organizing countries by population. India is at the top of this set, and obviously helps define the category of “populated country.” In fact, India has such a large population that if the set were left uncalibrated, Indonesia and Pakistan would be barely in the set, and most of the countries would be fully out of the set. Yet the important attribute is that some countries are comparatively more populated than others, so calibration makes the differences between the populous countries more comparable (p.204) to those between smaller countries. The very populated countries still define the set by being almost full members, while the rest of the cases get indexed by their degrees of membership in the set. In this case, the threshold value for full membership in the set of populated countries is established just below the actual population of India. At the lowest points in the curve are countries such as the Maldives, Brunei, Western Sahara, and Suriname, and these are definitely not very populated countries. So the threshold for full exclusion is set at 470,000 people because these countries have even smaller populations than that. The crossover threshold has been set at 7 million people, which roughly splits countries into two groups. Since Tajikistan and Bulgaria have slightly larger populations, these two countries are just barely in the category of “populated country.” The recalibration around these thresholds allows for fuzzy set values that more meaningfully reveal the degree to which each country can be included in the theoretical set of populated countries.

Appendix A: Countries in This Study

Figure A.1 Membership in the Set of Populated Countries: Uncalibrated

Source: World Bank, 2010.

(p.205)
Appendix A: Countries in This Study

Figure A.2 Membership in the Set of Populated Countries: Calibrated

Source: See Figure A.1.

Calibrating the Infrastructural Conditions

For membership in the category of countries with well-developed state information infrastructure, the threshold for full membership is defined as improvement in government communications infrastructure, with points allotted for the number of telecommunications sector reforms, whether or not the state had a privacy policy, whether or not the state had a public spectrum allocation policy, the number of executive agencies with internet portals, and the e-government score determined by West (2008). Tanzania and Jordan are full members of this category. The threshold for full non-membership excludes the West Bank and Western Sahara from the category. The transition point barely excludes Montenegro and barely admits Oman to the category.

For membership in the category of countries that experienced significant improvements in the information infrastructure of political parties, with points allotted for having a high ratio of political parties with internet portals, whether those political parties maintained their information infrastructure (p.206) in-country, and the gigabytes of content found at party websites. Turkey and Bulgaria are full members of this set, while the threshold for full non-membership effectively excludes a number of countries where political parties are effectively banned. The transition is defined at a point that barely excludes Togo, and barely admits Bosnia to the category.

For membership in the category of countries that experienced significant improvements in the information infrastructure of news media, the threshold for full membership is set to include Pakistan, Russia, and India, based on the number of online news websites serving people in each country. The threshold for full non-membership excludes Guinea-Bissau and the Ivory Coast. The transition is defined by Tanzania, which is barely out of the set, and Libya, which is barely in the set of countries that experienced significant improvements in digital news media.

For membership in the category of countries that experienced significant improvements in the information infrastructure of civil society, the full membership is defined by Singapore and Israel, based on a technology distribution index described in the Introduction and Chapter 5. Non-membership is defined by Iraq and Liberia, while Togo is barely out of the set and Pakistan barely in the set.

For membership in the set of countries using ICTs to censor political content and manage cultural identity, a variety of sources on censorship were consulted. The well-respected sources on censorship, even when assembled together, provided comparative information on only two-thirds of the cases. So to help determine membership in the set of countries that censor their population's internet use, some general information about press censorship was taken from a Reporters Without Borders index. Still, the membership values for the majority of cases are heavily weighted by data specifically about internet freedoms. Iran and Turkmenistan define full membership, while Israel, Cyprus, Ghana, and Mali define non-membership. Sierra Leone is barely excluded from the category, while Afghanistan is barely in the category of countries that use ICTs for censorship.

Calibrating Contextual Conditions

For membership in the category of wealthy countries, the threshold for full membership is set to include the UAE and Qatar, where the average annual income was US$50,000 in 2008. Liberia, Guinea-Bissau, and Somalia define full non-membership in the category of wealthy countries, because in these countries the average annual income is below US$500. The transition is defined by Cameroon and Yemen, where average annual income is just over US$2,000 a year, and these countries are barely admitted to the category of wealthy countries.

(p.207) For membership in the category of countries with income inequalities, the threshold for full membership is set to include Bosnia, the Central African Republic, and Sierra Leone, where the gini coefficient is above 0.56. Albania, Cyprus, Montenegro, and Ethiopia define full non-membership in the category of countries with income inequalities, because in these countries the gini coefficient is below 0.30. The transition is defined by Mauritania and Malawi, where the gini coefficient is just over 0.39, and these countries are barely admitted to the category of countries with income inequalities.

For membership in the category of countries with an educated population, the threshold for full membership is set to include Bahrain, Uzbekistan, Qatar, and Bulgaria, where close to 100 percent of the population has completed secondary education. Niger, Guinea-Bissau, Somalia, Liberia, and Western Sahara define full non-membership in the category of countries with an educated population because less than 11 percent of the population has completed secondary education. The transition is at the 50 percent mark, with Ghana just outside the category and Tanzania barely in the category.

For membership in the category of countries where fuel exports are an important part of the economy, Algeria and Nigeria define the set because fuel exports are almost 100 percent of the merchandise exports for these countries. Full non-membership is defined by Ethiopia and Malawi, and the transition to partial membership is marked by Gambia and Somalia. The former is barely excluded from the set, while the later is barely included.

For membership in the category of Muslim countries, the threshold for full membership is set to include the 11 countries where more than 98 percent of the population self-identifies as being practicing Muslims. Full non-membership is defined by countries such as Georgia and Kenya, where Muslims make up less than 20 percent of the population. The transition is defined by Albania and the Sudan, countries that are barely in the set because just over 70 percent of the population are practicing Muslims.

As described in the example above, the threshold for full membership in the set of populated countries is defined by India, easily the largest country in the set. Full non-membership is defined by countries such as the Maldives and Brunei, where there are fewer than 500,000 people. The transition is defined by Tajikistan and Bulgaria, countries that are barely in the set with a population of just over 7 million.

Calibrating Institutional Outcomes

The Polity IV index provides a good starting point for comparing democratization trends. The original dataset extends to 2007, and the author has made some adjustments to the original dataset and has extended coverage to 2008. The size of the Muslim community is based on the proportion of the (p.208) population practicing Islam, according to the World Bank World Development Indicators. The size of the Muslim community in each country is judged as a small minority if it is between 10 and 49 percent (), majority if it is between 50 and 89 percent (●) and a totality if it is more than 90 percent () of the total. Information and communication technology distribution is based on an index of the distribution of information technology resources, weighted by economy size, as described in the introduction. Index values were computed using data from the World Bank World Development Indicators for mobile phones, internet users, internet hosts, personal computers, internet bandwidth, and internet subscribers, and then averaged. Technology diffusion is judged as slow () if the country is in the bottom third of the index, medium (□) if the country is in the middle third of the index, or rapid () if it is in the top third of the index.

Multiple sources were used to inform the decisions on if and when a Muslim country experienced a political transition, whether the transition was for better or worse, and how the degree of shift might compare to other countries. The primary source was the Polity IV dataset, which tracks regime change up to 2007 and includes most of the states with large Muslim communities. The dataset ranks countries on a 21-point index from -10 (completely authoritarian) to +10 (completely democratic), with a 0 score representing neither an authoritarian nor democratic regime, but a chaotic state of transition usually involving war, anarchy, or state collapse. The index itself has several components, including assessments of the competitiveness of political participation, the openness and competitiveness of executive recruitment, and the constraints on the chief executive, whether president, prime minister, or monarch (Marshall and Jaggers 2008). Working from this index, countries that experienced a democratic transition are those for which experts find a 3-point transition toward the positive democratic end of the scale. Countries that experienced one or two points of improvement, if they already ranked as strong democracies, are categorized as experiencing entrenchment. The remaining countries are ones that continued to be ruled by authoritarian regimes, or experienced only temporary regime transitions. Crisis states include those embroiled in the chaos of war, invasion, or state failure. While interesting examples from these states are presented in chapters on specific types of political actors, these states are ultimately dropped from fuzzy set analysis. Additional sources, regional experts, and fieldwork allow for rankings to be updated to 2008 and for some adjustments to the Polity IV rankings. These adjustments are described in the tables below.

Thirteen countries experienced democratic entrenchment, and steadily received a democracy score of 6 or higher over the period of study. A country is consistently somewhat democratic if it received a score of between (p.209) 1 and 5 over the period of study. A country is consistently authoritarian if it was steadily scored between -6 and -10. A country is consistently somewhat authoritarian if it received a score of between -1 and -5. A country is a crisis state if it suffered from an extended period of lawlessness, civil war, foreign invasion, or other form of state collapse. Territories with unresolved status, Western Sahara, the West Bank, and Gaza strip, are included as crisis states.

Table A.1: Proportion of Population Practicing Islam, Rate of ICT Diffusion, and Countries Experiencing a Democratic Transition

Country

Muslim Population

Rate of Technology Diffusion

Transition Years, Magnitude and Notes on Author Adjustments to Polity IV Ranking

Albania

2001–2005; 4 points.

Algeria

+

1994–2004; 9 points.

Bahrain

1992–2002; 3 points.

Bosnia

2006–2008; 6 points. Bosnia has become more democratic, with recent elections, significant reduction in EU troop presence, and the changing mission of these troops from peacekeeping to civil policing.

Comoros

+

2003–2006; 5 points.

Djibouti

1998–1999; 8 points.

Egypt

2004–2005; 3 points. Transitioned from being authoritarian to being only somewhat authoritarian, but in author's judgment still a democratic transition.

Georgia

1994–2004; 3 points.

Ghana

2000–2004; 6 points.

Indonesia

+

1998–1999; 11 points.

Kenya

2001–2002; 10 points.

Kuwait

2006–2008; 4 points.

Kyrgyzstan

2004–2006; 7 points.

Lebanon

2004–2008; 7 points. Lebanon has been plagued by violence, war, and foreign interference. Between 2005–2006 there were signs of democratic stability, then in 2007 more political violence, and in 2008 a successful round of elections.

Liberia

+

2002–2006; 6 points.

Macedonia

2001–2002; 3 points.

Maldives

2005–2008; 5 points. Transitioned from being authoritarian to being only somewhat authoritarian, but in author's judgment still a democratic transition. Maldives has long been a dictatorship, but has recently transitioned to being less authoritarian with the legalizing of political parties and the drafting of a constitutional document.

Mauritania

2004–2006; 3 points. Transitioned from being authoritarian to being only somewhat authoritarian, but in author's judgment still a democratic transition.

Montenegro

2006–2008; 4 points. Montenegro, in recent years, has appealed for independence from Serbia, held a referendum and chose independence, and successfully elected a head of state.

Niger

+

1998–1999; 11 points.

Nigeria

+

1997–1999; 10 points.

Senegal

1999–2000; 9 points.

Sierra Leone

+

2000–2002; 5 points.

Suriname

2005–2008; 3 points. Suriname has been governed by a coalition of political parties since 1991, and the coalition expanded in 2005.

Tanzania

1999–2005; 3 points. Tanzania has become more democratic, with the two‐term president not challenging the constitutional ban on a third term, and 2005 elections being judged by international observers as having some cases of intimidation and logical irregularities, but on the whole free and fair.

Uganda

2004–2005; 3 points. Transitioned from being authoritarian to being only somewhat authoritarian, but in author's judgment still a democratic transition.

(p.210) (p.211)

Table A.2: Proportion of Population Practicing Islam, Degree of ICT Diffusion, and Countries Experiencing Democratic Entrenchment

Country

Muslim Population

Technology Diffusion

Democracy Status

Bangladesh

+

Consistently democratic.

Benin

Consistently democratic.

Bulgaria

Consistently democratic.

Cyprus

Consistently democratic.

Guinea‐Bissau

Had a non‐linear transition: Guinea‐Bissau was slightly democratic until 2002 when it became slightly authoritarianism for two years (‐6), then returned to be a stronger democracy (+7).

India

Consistently democratic.

Israel

Consistently democratic.

Malawi

+

Had a non‐linear transition: Malawi has been democratic, but briefly became somewhat democratic between 2001–4 (‐2), after which it returned to being more democratic (+2).

Malaysia

Consistently somewhat democratic. Even though Malaysia does not rank as a strong democracy up to 2007, successful elections in 2008 are evidence of democratic entrenchment (+2).

Mali

+

Consistently democratic.

Mauritius

Consistently democratic.

Mozambique

Consistently democratic.

Turkey

Consistently democratic.

(p.212)

Table A.3: Proportion of Population Practicing Islam, Degree of ICT Diffusion, and Countries That Remained Authoritarian

Country

Muslim Population

Technology Diffusion

Authoritarian Status

Azerbaijan

+

Became more authoritarian.

Brunei

Consistently authoritarian. Brunei is consistently authoritarian, run as a constitutional monarchy with some recent improvements, but not enough to call a transition.

Cameroon

+

Consistently somewhat authoritarian.

CAR

+

Became more authoritarian.

Chad

+

Consistently somewhat authoritarian.

Eritrea

Consistently authoritarian.

Ethiopia

+

Remained authoritarian. Consistently scored +1 in Polity IV over the period of study, but in author's judgment this country is not in the set of transition or entrenchment countries.

Gambia

Consistently authoritarian.

Guinea

+

Consistently somewhat authoritarian.

Iran

Had a non‐linear transition: Iran became slightly less authoritarian in 1996–1997 (+3) but returned to authoritarianism in 2004 (‐3).

Jordan

Consistently somewhat authoritarian.

Kazakhstan

Became more authoritarian.

Libya

+

Consistently authoritarian.

Morocco

Consistently authoritarian.

Oman

Consistently authoritarian.

Pakistan

Became more authoritarian.

Qatar

+

Consistently authoritarian.

Russia

Experienced a +3 point transition towards democracy according to Polity IV, but in author's judgment this country is not in the set of transition or entrenchment countries. Russia has had difficulty with executive turnover and has developed a strong presidentialist regime.

Saudi Arabia

Consistently authoritarian.

Singapore

Consistently somewhat authoritarian.

Sudan

+

Experienced a +3 point transition towards democracy according to Polity IV, but in author's judgment this country is not in the set of transition or entrenchment countries.

Syria

+

Consistently authoritarian.

Tajikistan

Consistently somewhat authoritarian.

Togo

Consistently somewhat authoritarian.

Tunisia

Consistently somewhat authoritarian.

Turkmenistan

+

Consistently authoritarian.

UAE

Consistently authoritarian.

Uzbekistan

Consistently authoritarian.

Yemen

+

Consistently somewhat authoritarian.

(p.213)

Table A.4: Proportion of Population Practicing Islam, Degree of ICT Diffusion, and Countries Experiencing an Extended Period of Interruption from Foreign Powers, Interregnum, or Anarchy

Country

Muslim Population

Technology Diffusion

Crisis Status

Afghanistan

+

Burkina Faso

+

Had a non-linear transition: Burkina Faso became slightly democratic 1999–2001 but returned to chaos in 2002.

Iraq

+

Ivory Coast

+

Had a non-linear transition: Ivory Coast became a slight democracy between 1998–2001 (+10), but has been in anarchy since 2002.

Somalia

West Bank

+

Western Sahara

+

For membership in the category of countries that had experienced entrenchment, the threshold for full membership is defined as an 11-point improvement in democratic institutions during the period of transition, which uses Indonesia and Niger to define full membership in the category. The threshold for full non-membership is defined by Pakistan and the Central African Republic, which experienced a -13 and -6 drop in their democracy score over the study period. Many other countries had a consistent (p.214) regime with no transition in either direction. The transition point between categories was set at a 1-point improvement, which barely excludes Guinea-Bissau and barely admits Oman into the category.

For membership in the category of countries that had experienced democratic entrenchment, the threshold for full membership is defined as at least 36 years of durability as a democratic regime, which sets Malaysia, Mauritius, India and Israel as full members of the category. The threshold for full non-membership is defined as at least 50 years of durability as an authoritarian regime, which includes Oman, Libya, and Saudi Arabia. The transition point between categories was set at one year as a democratic regime, which barely excludes Pakistan and barely admits the Comoros into the category.

Several other kinds of calibrations and manipulations help get the data into shape for fuzzy set analysis. For example, the fuzzy category of countries experiencing a democratic transition includes 34 countries that, relying on Polity IV data alone, are neither in nor out of the set. Both Yemen and Turkey are neither in nor out of this category, but for different reasons, so it makes sense to adjust the membership values of each so that they “lean” in a sensible direction. Thus Turkey was recoded to be barely in the set of countries experiencing a democratic transition, and Yemen recoded to be barely out of the set.

Western Sahara and the Palestinian Territories of the West Bank and Gaza are treated as countries in some datasets but not others. Given their prominent role in international Islamic politics, they are treated as units of analysis here. Since data on these two units of analysis are patchy, the fuzzy set variables were patched with additional sources, or by hand with information about neighboring countries. Depending on the variable, Mauritania was often a reference point for the Western Sahara, while Egypt or Lebanon served as reference points for the West Bank and Gaza. For example, the fuzzy set of countries with large online civil societies had missing data in these instances, so given the author's knowledge of these cases, Western Sahara was coded as having a slightly smaller online civil society than Mauritania, and the West Bank and Gaza was graded with a slightly larger online civil society than Egypt. Since the crisis states are unlikely to teach us much about the recipe for democratization, they have been removed from the analysis.

Selecting Causal Sets

As in all social science, the researcher must make difficult decisions about what to include in a causal explanation that covers most of the cases and for (p.215) which most of the cases are consistent. The fs/QCA program renders complex, parsimonious, and intermediate solutions. In the concluding chapter, the parsimonious solutions were presented because they reveal which conditions are essential to distinguishing between the positive and negative cases of democratic transition. Parsimonious solutions tend to sacrifice some set-theoretic consistency and include cases with missing data, but yield better case coverage overall. However, only the West Bank and Western Sahara were missing data, and these cases were dropped in the final analysis. Moreover, the consistency cutoff was set to be relatively high. The complex solutions provide a more complete account of the causal mechanisms behind democratization, but supplement without contradicting the key findings. For these reasons, the parsimonious explanations are offered in the conclusion.

Removing the crisis states left 68 countries in the comparison set. With 12 causal variables, there are 4,096 possible combinations of ingredients for the democratization recipe. However, only a much smaller number of these combinations actually describe real cases; a few combinations explain many cases, some combinations only explain one unique case, most combinations have no real examples. So fs/QCA identifies a series of minimal solutions. If two solutions differ in only one causal condition but have the same outcome, then the causal condition that distinguishes the two solutions can be removed to create a simple, combined expression. The concluding chapter discusses the most prominent causal recipes for democratic transition as determined by the number of cases covered and the degree to which cases are consistent with the solutions.

For the sufficient causes of institutional change, cases are noted if they have greater than 0.5 membership in the outcome for that specific causal configuration. This may mean that some of the noted cases are not fully consistent with other causal conditions. Some cases can be explained by several recipes, so the countries that best fit with membership in the outcome are listed in descending order of fit. Since these set relations are fuzzy, there are additional cases that would be explained, though perhaps with less consistency. The noted cases have the same causal conditions and also share, to varying degrees, membership in the outcome. The necessary conditions are shared by the cases that have democratized.

The Solution Sets for Sufficient Causes

The solution set for sufficient causes of democratic transition had a consistency cutoff of 90 percent, meaning that cases with full membership in the outcome had to be at least 90 percent consistent. There were 41 causal combinations containing real cases of democratic transition. Table 7.1 presents the two parsimonious solutions with the highest case coverage, though there (p.216) were 38 parsimonious solutions all in all that covered 96 percent of the cases with 69 percent consistency.

The solution set for sufficient causes of democratic entrenchment had a consistency cutoff of 95 percent. There were 41 parsimonious causal combinations containing real cases of democratic entrenchment. Table 7.1 presents the two parsimonious solutions with the highest case coverage, though there were 28 solutions all in all that covered 82 percent of the cases with 77 percent consistency.

Membership in the outcome of democratic transition caused by having a comparatively active online civil society with a comparatively small population includes the following countries, in order of consistency (with consistency and coverage for each case in brackets): Brunei (0.95, 0.75), Cyprus (0.95, 0.5), Mozambique (0.93, 0.75), Bahrain (0.89, 0.69), Kuwait (0.89, 0.75), Mauritania (0.88, 0.5), Pakistan (0.87, 0.39), Lebanon (0.83, 0.85), Maldives (0.81, 0.69), Niger (0.8, 0.54), Turkey (0.79, 0.39), Bosnia (0.78, 0.85), Sierra Leone (0.78, 0.5), Georgia (0.77, 0.69), Sudan (0.69, 0.69), Kyrgyzstan (0.67, 0.89), Jordan (0.65, 0.5), Israel (0.62, 0.54), Albania (0.62, 0.75), and Djibouti (0.52, 0.91). Membership in the outcome of democratic transition caused by having a comparatively active online civil society with a comparatively well-educated population includes: Sierra Leone (0.94, 0.5), Brunei (0.94, 0.75), Bulgaria (0.94, 0.5), Cyprus (0.93, 0.5), Turkey (0.92, 0.39), Israel (0.92, 0.54), Kuwait (0.9, 0.75), Bahrain (0.89, 0.69), Mozambique (0.88, 0.75), Mauritania (0.88, 0.5), Pakistan (0.87, 0.39), Qatar (0.87, 0.62), Lebanon (0.85, 0.85), Tunisia (0.84, 0.5), Bosnia (0.83, 0.85), Georgia (0.81, 0.69), Maldives (0.81, 0.69), Russia (0.81, 0.39), Niger (0.8, 0.54), and Kazakhstan (0.76, 0.16).

Membership in the outcome of democratic entrenchment caused by having a state with comparatively well-developed information infrastructure and an economy not dominated by imports includes: Jordan (0.93, 0.25), Mauritania (0.93, 0.9), Tajikistan (0.93, 0.66), Malawi (0.9, 0.67), Bangladesh (0.83, 0.7), Liberia (0.68, 0.51), and Ethiopia (0.53, 0.66).

Membership in the outcome of democratic entrenchment caused by having a state with a comparatively well-educated population and an economy not dominated by fuel exports includes: Israel (0.92, 0.96), Jordan (0.9, 0.25), Mauritania (0.9, 0.9), and Lebanon (0.85, 0.53).

The Solution Sets of Necessary Causes

The two solution sets for the sufficient causes of democratic transition each suggest three variables that should be tested for their possible role as necessary causes. Indeed, each of the causes appear in many of the other solution sets not presented here, so these are certainly among the most (p.217) important ingredients for democratization. For each institutional outcome—transition or entrenchment—the candidates were tested on their own, in pairs, and all together. The necessary causes with the greatest case coverage were ones that had little to do with information technology diffusion, though the factors that did concern technology diffusion had only slightly lower levels of case coverage. Putting all three causes together yielded a necessary combination of causes with excellent consistency and some sacrifice in coverage.

Whereas the best two sufficient causes were selected on the basis of case coverage, the necessary conditions were chosen in a different way. It is important to distinguish between trivial and non-trivial necessary causes, because a solitary condition may appear in almost all causal recipes, whether or not these cases display democratic outcomes. Indeed, every one of the causal conditions could be tested for necessity, but the most theoretically useful ones have high case coverage, may also double as sufficient conditions, and in combination with other conditions have good consistency. For example, a well-educated population may be a necessary condition for democratization, but if well-educated populations are always present then the absence of an educated population will never be found to constrain democratic outcomes. Alone it is a somewhat trivial finding. But in combination with other necessary causes, the cases that exhibit democratic outcomes have much greater consistency. It turns out that having an active online civil society and a small, well-educated population is a non-trivial causal condition for democratic transitions. Moreover, having a well-developed state information infrastructure, an economy not dominated by fuel exports, and a relatively well-educated population is a non-trivial necessary cause of democratic entrenchment. Rather than list all seven causes and causal combinations, the conclusion highlights the necessary causes with the greatest coverage and greatest consistency. (p.218)