T407016 // Autumn 2026 // Université de Genève
Comparative
Politics
Session 03. Regimes and how we measure them.
Quiz first
Sessions 01 and 02. Lijphart 1971, Tilly 1985, and Hanson and Sigman 2021.
Ten questions. Single choice, select all that apply, and statements to correct.
Closed book. Phones and laptops away. Write your name at the top of the sheet.
Your best four of the six quizzes count. Pens down when time is called.
Last week, this week
- Tilly on war, protection and extraction as the makers of states
- Hanson and Sigman on measuring state capacity and testing the measure
- Levitsky and Way on regimes that hold real elections and still tilt the field
- Munck and Verkuilen on how democracy indices are built and where they fall short
- One group exercise
Last week we measured what states can do. This week we ask who governs them, and by what rules.
One question before the readings
A country holds regular multiparty elections, and the opposition sometimes wins. Is it a democracy?
What else would you need to know before you answer?
Who draws the line, and with what evidence?
Would two democracy indices give the same answer?
What is wrong with "hybrid"?
Scholars read mixed regimes as incomplete democracies still in transition. Mexico, Senegal and Taiwan did democratize in the 1990s. Azerbaijan and Belarus moved toward authoritarianism, and Malaysia, Russia, Ukraine, Zambia and Zimbabwe stayed put or moved in several directions.
Freedom House rated El Salvador, Latvia and Ukraine "Partly Free" in 1992-93, each with a combined score of six.
- Latvia
- No citizenship for people of Russian descent
- El Salvador
- Human rights violations and no civilian control of the military
- Ukraine
- Civil liberties violated and procedures abused by incumbents
"It may therefore be time to stop thinking of these cases in terms of transitions to democracy and to begin thinking about the specific types of regimes they actually are."
Four minimum criteria
Free elections
Open, free and fair elections choose the executive and the legislature.
Universal suffrage
Virtually all adults have the right to vote.
Rights and liberties
Press freedom, freedom of association and the freedom to criticize the government without reprisal are broadly protected.
Real authority
Elected officials govern free of tutelary control by military or clerical leaders.
Democracies break these rules at times too. In a competitive authoritarian regime the violations are frequent and serious enough to tilt the playing field.
Between two borders
Elections are absent or noncompetitive. Cuba and China ban competition by law, Kazakhstan and Uzbekistan in practice. Façade electoral regimes such as Egypt and Singapore in the 1990s belong here too.
- Rule of thumb
- A president reelected with more than 70 percent
Democratic institutions are the main route to power, and incumbents abuse them often enough to fail the minimum standards. Croatia under Tudjman, Serbia under Milošević, Russia under Putin, Peru under Fujimori.
All four criteria hold. Flawed democracies stay on this side, such as O'Donnell's delegative democracies in Argentina and Brazil in the early 1990s. Peru after Fujimori's 1992 self-coup falls outside.
Following Linz, Levitsky and Way treat these regimes as a diminished form of authoritarianism.
Where the opposition fights back
Bitterly fought and mostly free of massive fraud. Kuchma took 35 percent in the first round in 1999. Opposition forces beat incumbents in Nicaragua in 1990, Zambia in 1991 and Ghana in 2000.
Usually weak, sometimes a focal point. Ukraine's parliament blocked Kuchma's referendum to cut its powers, and an opposition deputy used it to release damaging tapes of the president.
When Russia's Constitutional Court ruled against his 1993 decree, Yeltsin cut its phone lines and took away its guards. Maverick judges still ruled against governments in Slovakia, Serbia and Malaysia.
Independent outlets are legal and often influential. Governments answer with bribes, state advertising, debts, taxes and libel suits. Croatia's major independent papers faced more than 230 libel suits by 1997.
Incumbents prefer "legal" harassment through tax authorities, compliant courts and libel law. Open repression costs more while the rules still matter.
The incumbent's dilemma
Pay with international isolation and domestic conflict.
Risk losing power.
Either way the result is often a regime crisis, as in Mexico in 1988, Russia in 1993, and Peru, Serbia and Ukraine in 2000.
| Outcome | Cases |
|---|---|
| Weathered the storm | Kenya, Malaysia, Russia, Ukraine |
| Held back and lost | Nicaragua 1990, Zambia 1991, Ghana and Mexico 2000 |
| Cracked down, then fell | Peru, Serbia |
"But succession is not democratization."
Croatia, Nicaragua, Peru, Slovakia and Serbia democratized after turnover. New leaders in Albania, Zambia, Ukraine and Belarus kept or intensified the old practices.
Ties to the West
Competitive authoritarian regimes in Latin America that democratized after 1990. The Dominican Republic, Mexico, Nicaragua and Peru did. Haiti did not.
The same count in Central Europe. Croatia, Serbia, Slovakia and Romania did. Albania did not.
Former Soviet republic whose competitive authoritarian regime democratized in the 1990s, Moldova.
Cultural and media influence, elite networks and Western pressure raised the cost of entrenchment. Where Russia or China held sway, these regimes tended to persist or harden.
Three paths, one decade
Economic crisis and outside pressure pushed established autocrats to call multiparty elections. Weak oppositions let many of them keep power. Much of sub-Saharan Africa.
Weak electoral regimes emerged by default after a breakdown. Governments ruled autocratically and lacked the capacity to consolidate. Armenia, Croatia, Romania, Russia, Serbia, Ukraine, and Haiti after 1994.
Elected governments undermined institutions through a self-coup or piecemeal abuse, and lacked the will or capacity to remove them entirely. Peru in the early 1990s, and perhaps Venezuela.
"Pluralism and democratic contestation persisted less because elites wanted them than because elites simply could not get rid of them."
Three challenges, in order
- Task
- Pick the attributes and arrange them by level of abstraction
- Standard
- Avoid maximalist and minimalist definitions, redundancy and conflation
- Task
- Choose indicators and a measurement level, and publish the coding rules
- Standard
- Validity, reliability and replicability
- Task
- Decide whether to combine the scores, how far, and by what rule
- Standard
- Validity, robustness and replicability
"Problems of causal inference have overshadowed the equally important problems of conceptualization and measurement."
Too little or too much
A missing attribute
Polity omits participation, so its data back to 1800 miss the slow spread of the vote. Coppedge and Reinicke, Gasiorowski and Vanhanen omit offices.
Four attributes
Dahl's contestation and participation, plus whether key offices are filled by election and whether elected officials hold agenda-setting power.
Borrowed attributes
Freedom House counts socioeconomic rights, freedom from gross inequality, property rights and freedom from war, which belong to other concepts.
Put markets into the definition of democracy and the link between the two is settled in advance. Some questions are "just too interesting to be resolved by a definitional fiat."
The concept tree
- Leaves are the most concrete attributes. Measurement starts from them.
- Conflation puts an attribute on the wrong branch. Public financing of parties concerns contestation.
- Redundancy lets two attributes on one level cover the same ground. Extent of suffrage repeats the right to vote.
- Polity IV splits two single issues into pairs of attributes. Freedom House lists 9 and 13 components as a "checklist."
From leaves to numbers
Use several indicators and show they mean the same thing across countries. Watch for bias in the record. More reported corruption may only reflect a freer press.
Nominal, ordinal or interval, chosen to keep cases in a class alike with as few distinctions as needed. Bollen called democracy "continuous," and Alvarez and coauthors called that view "ludicrous."
Different coders give the same scores. Only Coppedge and Reinicke and Polity IV used several coders and tested their agreement. "Reliable measures need not be valid ones."
Publish the coding rules, the coding process and the disaggregate data. Freedom House offered no coding rules, vague sources and no disaggregate data, so its scores are "accepted largely on faith."
Add or multiply?
- State how the attributes relate before you choose a rule.
- Equal weight, so add. High participation can make up for low contestation.
- Both necessary, so multiply. A low score on either one drags the total down.
- Either one sufficient, so take the higher score.
| Contestation | Participation | Add | Multiply | |
|---|---|---|---|---|
| Country A | 0.9 | 0.1 | 1.0 | 0.09 |
| Country B | 0.5 | 0.5 | 1.0 | 0.25 |
Addition ranks them level. Multiplication puts B almost three times higher than A.
Freedom House adds its components, so decentralization weighs as much as the power of elected representatives. Most indices push straight to one score, which in Munck and Verkuilen's words "puts the statistical cart before the theoretical horse."
The scorecard
| Index | Strengths | Weaknesses |
|---|---|---|
| Polity IV | Offices and agenda setting, detailed coding rules, intercoder test, wide coverage | Omits participation, redundant attributes, poor aggregation |
| Freedom House | Covers the whole world | Maximalist, conflation, measurement problems, poor aggregation |
| ACLP | Offices, sound conceptual logic, good indicators, detailed coding rules | Omits participation and agenda setting |
| Hadenius | Offices, agenda setting and fairness, good indicators, sophisticated aggregation | Redundancy and conflation, a single year |
| Vanhanen | Clear coding rules, wide coverage, replicable | Omits offices and agenda setting, questionable indicators, poor aggregation |
No index meets all three challenges. The indices with the widest coverage are mostly among the weaker ones, with Polity IV a partial exception.
What high correlations hide
- The indices draw on the same sources, such as Arthur Banks's coded data, so they may share one bias.
- Agreement between indices speaks to reliability. Validity needs evidence of its own.
- Correlations of aggregate scores can hide a second dimension.
| Index, 1973-1990 | Dimension 1 | Dimension 2 |
|---|---|---|
| ACLP | −.93 | −.18 |
| Gasiorowski | .91 | .26 |
| Polity autocracy | −.96 | −.27 |
| Polity democracy | .95 | .25 |
| Freedom House civil liberties | −.57 | .80 |
| Freedom House political rights | −.56 | .81 |
| Share of variance | 69% | 26% |
Freedom House loads heavily on a second dimension that the other indices barely touch. Signs follow each index's coding direction.
Two ways to study regimes
| Levitsky and Way 2002 | Munck and Verkuilen 2002 | |
|---|---|---|
| Question | What kind of regime holds real elections and still tilts the field? | How well do existing indices measure democracy? |
| Move | Define a new regime type with clear borders | Judge nine indices against one framework |
| Evidence | Cases from Africa, Latin America, Asia and postcommunist Eurasia in the 1990s | Codebooks, and a principal components test on six series |
| Complaint | "Partly Free" lumps together regimes that differ | One score can hide several dimensions and a shared bias |
| Earlier sessions | Sartori's stretched concepts, as in calling every regime with elections a democracy | Hanson and Sigman tested for more than one dimension before settling on one score |
Hanson and Sigman's Capacity correlates at .91 with the World Bank's government effectiveness index. What would Munck and Verkuilen make of that evidence?
A development agency sends your group one question
Our rules allow budget support only for democracies. Which of these four countries qualify?
Multiparty elections every five years, and observers found no massive fraud. The president won with 57 percent. His allies own the main TV channels, and tax inspectors audited two opposition newspapers before the vote. The constitutional court struck down one of his decrees.
Free and fair elections with high turnout, a free press and independent courts. A quarter of adult residents belong to a minority denied citizenship at independence, so they cannot vote.
Competitive elections, a free press and turnover in power. The armed forces appoint a fifth of the senate and can veto the defense budget.
The president was reelected with 91 percent. The main opposition party is banned, and international observers were refused entry.
- Test each country against Levitsky and Way's four criteria. Which criterion does each one fail? Which country can be considered as a competitive authoritarian regime following Levitsky and Way's definition?
- Build the agency's index. Score each country from 0 to 1 on contestation and on participation, then combine the two scores by adding and by multiplying. Which rule would you defend, and why?
- Write a three-sentence memo that answers the agency and names one indicator to watch before Country A's next election.
The profiles are invented and describe no real country.
Before Tuesday 6 October
Week 04. Democratization and democratic erosion.
Acemoglu, Daron, and James A. Robinson. 2006. Economic Origins of Dictatorship and Democracy. Cambridge University Press, chapter 2.
Lührmann, Anna, and Staffan I. Lindberg. 2019. "A Third Wave of Autocratization Is Here. What Is New About It?" Democratization 26(7), 1095-1113.
- Week 4 classM2170
- No class
- Quiz 2 opens the class
- Complement on concepts, measurement and validityM2170
Progress saved
Questions?
Office hours Thursdays 14:30 to 16:00 in M4328
Rate each group member's participation in today's exercise. Required before you leave the room.