Law by BogdanovaThe Encyclopedia of Law and Legal Studies

RESEARCH DOSSIER

Quantitative Comparative Law

Abstract. Quantitative comparative law is the systematic construction and analysis of numerical observations concerning legal rules, institutions, practices, outcomes, and experiences across jurisdictions or time. Its intellectual promise is to make propositions about legal variation explicit, testable, and revisable. Its central methodological burden is that legal meaning and institutional operation must be reconstructed before numerical equivalence can be assumed. This monograph develops a comprehensive source-grounded account of units of comparison, historical evolution, legal coding, construct validity, composite indicators, expert and population surveys, judicial statistics, missingness, statistical uncertainty, selection, longitudinal change, and causal inference. It compares the World Justice Project Rule of Law Index, the World Bank Worldwide Governance Indicators, the World Bank Business Ready initiative, the European Commission's EU Justice Scoreboard, Council of Europe judicial statistics, and United Nations access-to-justice indicators as different research instruments rather than interchangeable league tables. Historical legal-origin studies and their subsequent re-coding are examined as an unusually instructive test of whether quantitative legal conclusions survive changes in classification. Original analytical contributions consist of a unified comparison of inferential targets, worked calculations with explicitly hypothetical data, a source-provenance audit of operational choices, and reasoned conclusions concerning when indicators inform doctrinal or institutional judgments. The analysis establishes that an indicator is most reliable when its legal object, observation process, coding decision, aggregation rule, reference population, uncertainty, and intended inference remain traceable. Numerical comparison enables rigorous legal inquiry when those conditions are satisfied and when substantive interpretation accompanies estimation.

Keywords: Quantitative comparative law; empirical legal studies; comparative legal methodology; legal indicators; rule of law; legal origins; construct validity; coding protocol; measurement invariance; composite indices; World Justice Project; Worldwide Governance Indicators; Business Ready; CEPEJ; EU Justice Scoreboard; access to justice; missing data; legal statistics; causal inference; reproducibility

Introduction — The Subject and Problem of Quantitative Comparison

Quantitative comparative law examines variation among legal orders through systematically defined and numerically coded observations. Its subjects include the existence and design of rights, procedures, remedies, and institutions; the performance of officials and courts; the behavior and experiences of litigants; and the distribution of outcomes resulting from legal processes. Its questions range from descriptive comparisons—how different systems regulate disclosure or process civil claims—to explanatory inquiries concerning relationships among legal arrangements, institutional capacity, and economic or social results. What links these studies is the passage from legal evidence to a structured dataset and from a dataset to a constrained statement about a legal population.

A numerical variable becomes a comparative legal variable only after a concept has been translated into an observation rule that preserves the legal meaning relevant to the research question. An ordinary count of statutes, for example, may register legislative production without revealing the rights actually exercisable by residents. A count of filings may track demand, recording practices, incentives to litigate, or procedural architecture. A low court-case rate can be consistent with accessible informal resolution, prohibitively expensive litigation, or both. Legal interpretation determines what the number means.

Comparative law's classical commitment to institutional context remains indispensable. Zweigert and Kötz (1998), in their introduction to comparative legal methods, examine legal families and functional comparison; Siems (2022) situates comparative legal methodology within an interdisciplinary and global field. Numerical comparison adds the requirements of transparent sampling, documented coding, estimable uncertainty, and a clear inferential target. These obligations reinforce contextual legal analysis by exposing the assumptions that prose can leave implicit.

The present study distinguishes five questions that are often compressed into a single ranking. What legal object is being represented? How was it observed? How was the observation transformed into a score? What population or period does the score describe? Which conclusion follows from it? Each question can be answered satisfactorily while another remains unresolved. An accurate count of law-on-the-books provisions may still be a poor measure of access to a remedy; a carefully designed survey may still have limited coverage across populations; a reproducible cross-country correlation may still fail to identify a causal effect.

The evidentiary basis includes official methodological publications, source descriptions, institutional data standards, and published scholarship. Institutional sites and scholarly abstracts establish their documented methods and reported findings; restricted books are used for their authenticated bibliographic and publicly described arguments, without invented page-specific quotations. Mathematical examples in this monograph are expressly hypothetical and contain no reported country-level scores or results. The research cutoff is November 21, 2026; dataset releases, underlying observation years, and publication dates are distinguished throughout.

Research Design, Sampling, and Comparison

Questions, target populations, and contrasts

The research question determines the study's observable universe. A descriptive question about civil-justice timeliness can compare first-instance cases resolved during specified calendar years. A question about corporate protection might require individual statutory rights and remedies as coded in law in force on a common reference date. A causal question about the effect of introducing specialized courts might require pre-reform trends, implementation timing, untreated comparison units, and a plausible identification design. All three can be “quantitative comparative law,” yet they cannot be answered by one generic national score.

A reproducible comparative design records inclusion and exclusion criteria before inspecting results. This includes the population of legal systems, the time period, language access, data-quality thresholds, and the method of dealing with regional or institutional units. An international sample composed mainly of systems with accessible official statistics may overrepresent well-resourced administrations. Such selection can systematically relate the availability of data to the legal performance one wishes to investigate.

King, Keohane, and Verba (1994) describe the importance of explicit inference and observation selection in Designing Social Inquiry. Seawright and Gerring (2008), in their study of case-selection techniques, clarify why qualitative follow-up cases should be chosen strategically from a defined comparative universe. A quantitative legal map gains interpretive depth when typical, deviant, similar, and contrasting cases are investigated with documentary legal evidence.

Matched legal problems and scope conditions

Functional comparison is useful when legal terminology differs but the research problem is shared. For example, a legal order may provide a contractual remedy through a statute, another through precedent, and another through administrative enforcement. A coding protocol that records only the presence of one named statute will classify the second and third systems as lacking a remedy, even if functional substitutes exist. The research question must specify whether it measures textual form, legally available remedy, or the remedy's effective operation.

Consider a hypothetical survey of six jurisdictions. If three are chosen because their legal-technology programs are well documented and three because they feature prominently in international reform narratives, the sample permits a carefully described comparative case series. It does not authorize a population estimate for all legal orders. Expanding the sample's number of coded variables does not repair nonrepresentative selection of jurisdictions.

Nested and clustered observations

Legal datasets frequently feature observations nested within countries and years. Statutory rights, court-level outcomes, and expert assessments can share national shocks, legal history, and administrative recording conventions. Standard errors from a model that treats all court observations as independent may be too small. Appropriate methods may include clustered uncertainty estimates, multilevel models with justified assumptions, and separate analyses by source. The choice depends on the inferential target, number of clusters, and possible dependence across time.

A model specification should also state which comparisons have support in the data. A cross-country association cannot be interpreted as a within-country reform effect without assumptions linking these estimands. When the data concern different courts in the same country, the study can identify institutional heterogeneity without claiming that national legal traditions caused it.

Measurement Validity and Cross-System Equivalence

From a concept to an observable claim

Measurement validity concerns whether an observation supports the interpretation placed upon it. Adcock and Collier (2001), in Measurement Validity, articulate a shared methodological vocabulary connecting background concepts, systematized concepts, indicators, and measured scores. Applied to comparative law, this framework requires the researcher to state the legal idea before selecting a convenient dataset column.

“Judicial independence,” for example, might refer to appointment and tenure protections, budgetary autonomy, freedom from improper pressure, perceived impartiality, or case-specific decisional behavior. A statutory safeguard may be a valid measure of formal institutional design. An expert survey can provide evidence of perceptions of interference. Neither alone measures every element of independence. Reproducible arithmetic does not establish that the source question captures the desired legal construct.

Content validity asks whether the measure covers relevant dimensions; construct validity tests whether observed patterns accord with justified expectations; criterion-related validity evaluates relationships with a suitable outside reference, when one exists. For legal constructs, an external “gold standard” may itself be contested, incomplete, or unavailable. Triangulation across legal documents, surveys, administrative records, and judicial decisions can therefore be more informative than correlation with another composite index built partly from the same underlying sources.

Measurement invariance

Cross-system comparison requires that a comparable response or code carry comparable meaning. A response category such as “somewhat independent” may reflect different expectations, social risks, and reference groups. A yes/no question about whether a lawyer is legally available may mean licensed access, free public representation, or practical ability to retain counsel. Cross-language translation can reproduce the words while changing the juridical category.

King and Wand (2007) explain the difficulty of comparing ordinal survey responses across respondents who use scale categories differently, and examine anchoring vignettes as a potential adjustment. In legal surveys, vignettes have a related function: a common factual scenario can help isolate variation in legal standards or perceptions, provided respondents interpret the scenario comparably. The vignette itself may require adaptation and validation across jurisdictions.

Measurement invariance may be examined through item-level behavior, known-group tests, multi-group latent-variable models, cognitive interviewing, translation audits, and sensitivity analyses. A failure of invariance limits what differences in country means can signify. The validity of a within-country trend does not automatically imply the validity of a cross-country rank, and the reverse is equally possible.

Criterion contamination and source overlap

Indicators that appear independent can share expert respondents, commercial risk providers, country narratives, or administrative submissions. Two indices correlated because of overlapping input data are weaker corroboration than two independent measurement processes. Research should document provenance at the variable level where feasible. It should distinguish triangulation across observation mechanisms from repetition of one institutional judgment in multiple products.

An index may also incorporate outcomes whose inclusion makes an explanatory claim circular. If the predictor labeled “effective legal institutions” contains measures closely tied to economic performance, its association with economic outcomes requires special examination. This is a design issue to be addressed by inspecting item definitions, source lists, overlap, and alternative specifications rather than by interpreting a high coefficient as independent confirmation.

Composite Indicators and Index Construction

The path from observations to an index

A composite score maps several variables onto a common summary. The OECD and European Commission Joint Research Centre's Handbook on Constructing Composite Indicators (2008) sets out methodological questions concerning conceptual framework, data suitability, normalization, weighting, aggregation, robustness, and interpretation. These decisions are decisive for comparative legal scores because the relevant domains—rights, transparency, enforcement, access, timeliness—are often normatively heterogeneous.

Suppose three hypothetical dimensions of legal-system performance are coded on a scale from zero to one: access = 0.80, predictability = 0.60, and timeliness = 0.20. An equal-weight arithmetic index is (0.80 + 0.60 + 0.20)/3 = 0.5333. A geometric mean gives approximately 0.458, assuming strictly positive dimensions. The arithmetic formulation permits relatively high values on two dimensions to compensate for a low value on a third; the geometric construction penalizes imbalance. Neither aggregation rule follows automatically from the law. The choice encodes a view about substitutability among legally significant dimensions.

For weights w_j summing to one, the standard weighted arithmetic construction is:

[ I_c = \sum_{j=1}^{J} w_j x_{cj}, \qquad w_j \geq 0,\qquad \sum_{j=1}^{J} w_j = 1. ]

The symbols denote country or jurisdiction c, dimension j, normalized observed value x, and selected weight w. If weights are estimated statistically, their interpretation depends on the model and sample. If they are chosen normatively, the normative premise should be disclosed. “Data-driven” weights are also the product of a choice about the method, covariance structure, and dataset.

Normalization, thresholds, and direction

Normalization maps variables measured in different units onto a common scale. Min-max transformations rely on observed or fixed endpoints and can change when new jurisdictions enter a dataset. Standard-score transformations depend on the sample distribution and may obscure absolute change. Fixed policy benchmarks offer continuity but require a justification of the benchmark.

A rule-oriented indicator should also declare whether a larger number is desirable. More court filings may indicate greater accessibility, greater conflict, economic activity, or recording completeness. More prosecutions may reflect stronger enforcement or a larger incidence of crime. A legitimate analysis can report such variables without assigning an unqualified good/bad direction.

For binary laws, the existence of a provision can be recorded consistently without implying its comparative social value. Whether an outcome such as judicial settlement rate signals fairness, coercive bargaining, or procedural efficiency depends on the institutional context. Index designers should publish a sensitivity interval around any value judgment that materially affects rank ordering.

Robustness and the problem of ranks

A small numerical score difference can produce a large rank change in a tightly clustered distribution. Ranks discard absolute distance; the gap between ranks one and two need not resemble the gap between ranks two and three. When measurement uncertainty or plausible alternative weights overlap, a single definitive ordinal order overstates the empirical information.

A useful sensitivity analysis varies weights, normalization, imputation, source inclusion, and aggregation while tracking the resulting score and rank distributions. It asks whether the principal substantive claims remain stable. If one jurisdiction's place shifts widely among plausible specifications, the scientifically informative result is methodological fragility, not a preferred point ranking selected after inspection.

A composite index can serve exploratory and monitoring functions without being a universal legal-quality ordering. Readers should have access to factor-level data, definitions, coverage conditions, and uncertainty intervals. Disaggregation is especially important where dimensions may stand in legal tension—for example, procedural speed and due-process safeguards, or enforcement intensity and equal treatment.

The World Justice Project and the Measurement of the Rule of Law

Conceptual architecture and field evidence

The World Justice Project's 2025 Rule of Law Index methodology describes eight composite factors and 44 subfactors entering reported scores and rankings. Its conceptual framework includes informal justice as a ninth factor, but the methodology states that this factor is excluded from aggregate scores and rankings to maintain cross-country comparability. That distinction between the conceptual universe and scored universe illustrates why the exact instrument must be examined.

The eight scored domains are constraints on government powers, absence of corruption, open government, fundamental rights, order and security, regulatory enforcement, civil justice, and criminal justice. Their inclusion reflects an articulated conception of the rule of law with legal, institutional, and lived-experience components. An individual factor or subfactor is a more precise target for many research questions than the general aggregate.

The 2025 methodology reports a combination of General Population Poll responses and Qualified Respondents' Questionnaires, more than 550 underlying variables, and coverage of 143 countries and jurisdictions in that edition. Its household surveys combine perception and experience questions, while expert instruments address legal and administrative operation through specialized respondents and common scenarios. These are recorded properties of that edition, not timeless features guaranteed for every WJP release.

Period mismatch and representativeness

A country's displayed “2025” score is a score published in the 2025 index, not necessarily a direct record of what every respondent experienced in 2025. The underlying WJP methodology and survey notes explicitly describe household data gathered in different years across countries; some contributing population polls are substantially older than the index edition. The report also distinguishes nationally representative polls from surveys formerly centered on major urban areas. Both temporal mismatch and changing coverage matter when interpreting cross-country differences and trends.

The index's annual expert data can be updated at a different frequency from some population components. Consequently, a year-to-year score change can reflect new expert inputs, revised household evidence, a shift in measurement, or genuine institutional change. The question is empirical and instrument-specific. Research that claims a discrete legal reform affected the next annual score must establish a plausible connection between reform timing, observation timing, and the indicator items capable of registering the reform.

From sampling error to substantive interpretation

WJP's published methodology uses bootstrap-based uncertainty procedures for certain changes over time and documents sensitivity and validation work. Sampling uncertainty captures only some forms of error. Expert selection, response framing, translation, coverage, and theoretical construct choices also matter. A confidence interval for a score can be mathematically sound while omitting uncertainty about legal meaning or cross-national invariance.

The WJP methodology itself cautions against treating its scores as causal findings. This is a fundamental boundary. A low civil-justice score can identify domains for investigation and comparison; explaining why that score arose requires examining procedural law, funding, dispute patterns, access barriers, and administrative behavior. The most defensible scholarly use combines index diagnosis with primary legal and institutional evidence.

Analytical implication

WJP's great comparative value lies in disciplined articulation of experienced and perceived rule-of-law dimensions across a wide set of jurisdictions. A researcher can compare factors, inspect individual questions, and relate them to local legal records. The indicator's appropriate interpretation is determined by the item, respondent population, fieldwork period, and aggregation rule. Claims about abstract national “legal quality” should be reformulated as explicit statements about the exact measured dimension.

Worldwide Governance Indicators and Latent Perceptions

What the WGI estimates

The World Bank's Worldwide Governance Indicators documentation identifies six governance dimensions: voice and accountability; political stability and absence of violence or terrorism; government effectiveness; regulatory quality; rule of law; and control of corruption. The framework combines information from multiple cross-country sources, including expert assessments and surveys, to estimate latent dimensions of perceived governance.

The WGI rule-of-law measure is therefore neither a statute count nor a complete legal-doctrine inventory. It is a composite measure built from assessments and reports that can bear on perceptions of institutional reliability, enforcement, and respect for rules. A legal scholar examining the right to appeal, for example, should read the underlying law and procedure, then use WGI as a broad contextual indicator if a justifiable link to the question exists.

Kaufmann, Kraay, and Mastruzzi (2010), in their analysis of WGI methodology, explain the motivation for aggregation and the role of uncertainty. The work is historically foundational, but users of current data must consult the applicable revision rather than assuming that earlier scaling and source assumptions remain unchanged.

Revising time comparability

The World Bank describes a 2025 revision of methodology and provides a current documentation framework for its governance estimates and 0–100 absolute-scale reporting. The 2026 WGI data release overview reports its then-current scope and data coverage. These source versions matter: a historical WGI column retrieved after a method revision is not necessarily identical to a column obtained from an earlier archived release with the same country-year label.

In earlier WGI practice, standardized governance estimates had a global mean constrained approximately to zero for each year; the 2025 revision changed the framework so global trends can be examined more directly and introduced an additional fixed-benchmark scale. A researcher making long-run claims needs a stable vintage, documented transformations, and an analysis of whether apparent changes are measurement changes or developments in governance. Raw scores from different published vintages should not be silently spliced.

WGI's reported standard errors serve a particularly important function. Two jurisdictions' point estimates may differ without producing compelling evidence of a difference in the underlying latent score. Interval comparisons must account for the intended confidence level, possible common-source dependence, and the aggregation model. An analyst should distinguish an estimated point difference from a statistically resolved distinction and from a substantive legal claim.

Source dependence and external validation

The WGI provides broad geographic and longitudinal coverage. Its data can help identify where institutional perceptions move together or apart. Yet using WGI rule-of-law scores to “validate” another index that partly draws on related expert information risks circular confirmation. Cross-dataset concordance becomes stronger evidence when the underlying information and observation mechanisms are sufficiently distinct.

A sound comparative legal application could relate a documented procedural reform to WGI indicators as one contextual signal, examine other institutional datasets, and analyze the reform's actual statutory and operational implementation. The broad index supplies a background pattern; the legally specific evidence carries the interpretation.

Business Ready and the Difference Between Rules and Operation

A three-part object of measurement

The World Bank's Business Ready methodology organizes assessment around Regulatory Framework, Public Services, and Operational Efficiency. These are analytically distinct elements of a business environment. Legal rules can enable an activity, services can make compliance and enforcement feasible, and experience-based operational measures can register whether firms actually encounter functioning procedures.

The World Bank's Business Ready 2025 report documents an expanding multi-economy assessment built around topics that follow firm activities. A research report's coverage and scores belong to that publication edition; the methodological program continues to evolve. For research extending into 2026, the edition-specific handbook and source fields should be archived alongside the dataset.

A country could score comparatively well on formal regulatory design and less well on public-service infrastructure. Combining these dimensions into one unexplained “business-law quality” variable would erase a central finding the program is designed to expose. Their divergence can help generate more informative comparative hypotheses about implementation capacity, administrative organization, and the experience of firms.

Expert legal coding and firm experience

Expert questionnaires can encode the legally stipulated sequence of steps, fees, rights, or obligations; enterprise surveys can report operational delays, reliability, and administrative treatment. Each source has different strengths and biases. Legal coding risks disagreement about applicable rules and hypothetical scenarios. Firm surveys may be influenced by industry structure, firm size, locality, experience, selection into business activity, or respondents' understanding of the question.

A comparative study should align the source of a numerical claim with its meaning. A figure based on a hypothetical legally compliant firm cannot be recast as a measured distribution of all firms' actual experiences. Conversely, a sample of firms' reported waiting time may not establish what a statute requires. Interpreting both together makes it possible to identify gaps between enacted rules and their practical operation.

Lessons from the predecessor program

The World Bank's official Doing Business discontinuation statement provides a reminder that institutional data integrity is part of measurement quality. A later program's framework should be assessed in its own terms and by its actual methodology. A new indicator requires independent examination of its governance, data collection, revision practices, source disclosure, and sensitivity to reform incentives.

A causal interpretation of improved B-READY scores demands particular attention to the possibility of strategic responses. When governments know the variable definitions, legal reforms can target the measured procedures without improving unmeasured institutional capacity. This is a general incentive problem for public benchmarks, not a claim of misconduct by any specified jurisdiction or program.

Judicial Statistics and European Comparative Justice

Measuring a court system's flow of cases

Judicial statistics offer comparatively direct measures of institutional activity when their categories and reporting systems are understood. Courts record incoming, resolved, and pending cases; appellate progression; staffing; expenditures; and, in some systems, observed proceeding durations. The Council of Europe's European Commission for the Efficiency of Justice (CEPEJ) describes its judicial evaluation program as a structured process supported by national correspondents and explanatory methodology. The 2024 evaluation cycle primarily reports data for 2022, demonstrating why data year and publication year must be separately identified.

The European Commission's 2026 EU Justice Scoreboard presents indicators of efficiency, quality, and independence and explicitly avoids a single overall ranking. Much of its quantitative judicial information comes through CEPEJ, supplemented by other sources and subject-specific indicators. The Scoreboard's declared use of separate indicator families is instructive: it allows a reader to compare institutional aspects without treating them as numerically substitutable in one overall score.

Judicial caseload data require precise category boundaries. Whether an administrative enforcement proceeding belongs to the judicial system may vary across legal orders. Some systems count small claims separately, some distinguish contested from uncontested proceedings, and some distribute cases among specialized courts. Differences in digital filing, transfer rules, consolidation, and settlement registration may affect the number of reported incoming or resolved cases.

Clearance rate and estimated disposition time

Two common indicators are the clearance rate and disposition time. Let R be cases resolved during year Y, I incoming cases during Y, and P cases pending at the end of Y. Then:

[ \mathrm{ClearanceRate} = 100 \times R/I ]

and

[ \mathrm{DispositionTime} = 365 \times P/R ]

for a 365-day convention and positive denominators. The clearance rate compares flow to flow; the estimated disposition time combines end-of-period stock with annual resolution flow. The EU Justice Scoreboard's methodological description distinguishes this estimated measure from actual observed case durations in some specialized EU-law areas.

Consider a hypothetical court that receives 1,200 cases, resolves 1,000, and ends with 1,500 cases pending. Its clearance rate is 83.3 percent, while its indicative disposition time is 547.5 days. Neither statistic tells us that the median completed case lasted 547.5 days. That would require the distribution of actual durations from filing to disposition, with a defined approach to censoring and the fate of cases still pending.

A second hypothetical court might receive 500 cases, resolve 550, and end with 300 pending. Its clearance rate is 110 percent and its illustrative disposition time about 199.1 days. A clearance rate above 100 percent means more cases were resolved than received during the chosen period; it can reflect backlog reduction. It does not establish that every class of dispute is processed faster or that litigants experience equal access. A distinction between “all cases” and “contested civil and commercial cases” could materially change the comparison.

Recorded duration, survival analysis, and composition

Where case-level data are available, the analyst can examine Kaplan–Meier-type survival curves or other duration models, treating ongoing proceedings as right-censored when appropriate. The entry date, resolution event, competing outcomes, appeal, transfer, and interruptions must be defined. A simple average among completed cases can underestimate the experience of long-running disputes when those disputes remain unresolved at the observation cutoff.

Case complexity matters. A system handling large volumes of routine payment orders may appear faster on a broad measure than one counting only contested cases. Normalizing by number of judges can also be misleading if judges have different subject-matter responsibilities, associate staff, workloads, or assignment structures. The legally meaningful comparison must disclose case mix, staffing assumptions, and court type.

The limits of efficiency as a legal value

Speed is valuable but is not the sole criterion of justice. Adequate time to present evidence and appeal, procedural equality, public reasoning, independence, and enforceability may conflict with a single-minded drive to reduce duration. The EU Scoreboard's parallel attention to efficiency, quality, and independence offers a useful reporting convention. Where a policy recommendation concerns a change in procedure, the analysis should ask which rights, access conditions, and administrative burdens are affected, rather than using a faster clearance score as sufficient proof of a superior legal system.

The United Nations and Access-to-Justice Indicators

Formal measures and people's lived legal problems

The United Nations' SDG 16.3 indicator metadata demonstrates that the phrase “access to justice” can refer to distinct observable events. Indicator 16.3.1 concerns violence victims reporting their victimization to a competent authority or recognized mechanism. Indicator 16.3.2 measures the share of persons in detention who remain unsentenced. Indicator 16.3.3 concerns people who experienced a dispute within a specified period and accessed a formal or informal resolution mechanism, broken down by type.

The three measures have different denominators. The population of violence victims is not the total population; the population of detained persons is not all persons accused of an offense; the population of individuals experiencing a dispute is not all residents. Percentages constructed over different denominators cannot be equated because their numerical ranges match. Each numerator also requires operational rules: which experiences qualify as a report, who counts as detained but unsentenced, what constitutes a dispute, and which mechanisms qualify as recognized resolution.

The UN's 2011 Rule of Law Indicators guide makes a distinctive methodological choice. Its 135-indicator instrument was developed to monitor criminal-justice institutions in conflict and post-conflict settings over time within a country, drawing on four sources: administrative and field records, expert surveys, public surveys, and document review (United Nations 2011). The guide expressly excludes direct country rankings as an intended use. That restriction is legally instructive: an instrument can be sophisticated and valid for internal institutional monitoring while lacking the cross-system standardization required for a multinational ordering. Its defined domain also excludes a comprehensive measurement of informal justice and certain conflict-related accountability mechanisms. The guide's precise observation rules offer a valuable comparative-method lesson, while its official purpose remains within-country monitoring. The combination of survey evidence, administrative records, and local interpretation is especially salient where formal judicial case counts understate the prevalence of legal needs.

The denominator problem

Imagine two hypothetical jurisdictions. In jurisdiction A, 60 of 100 surveyed individuals reporting a qualifying dispute sought assistance from a recognized resolution mechanism; in jurisdiction B, 75 of 150 did so. The corresponding access percentages are 60 percent and 50 percent. These are valid numerical summaries of the hypothetical observed samples if their questions and sampling procedures are comparable. They do not by themselves establish the relative fairness, independence, enforcement success, or affordability of either system.

Further complications arise if individuals with unresolved problems cannot identify their experience as a “legal dispute,” if informal mediation is understood differently across cultures, or if violence victims avoid disclosure because of safety concerns. These mechanisms can bias both the numerator and its estimated reference population. Questions about social trust and legal literacy may help explain differences in reported rates, but they must not be mechanically interpreted as successful enforcement outcomes.

Comparing pretrial detention

The unsentenced share of prison or detention populations is a stock proportion. A high share can be consistent with slow criminal adjudication, detention policy, fluctuating sentenced populations, or other features of criminal process. Identifying its causes requires separate measures of duration, admission, release, bail, sentencing, case mix, and the governing law. Comparing countries solely on this proportion risks confusing an institutional stock measure with an individualized measure of procedural legality.

An original comparative analysis should specify whether detainees awaiting appeal are classified differently from those awaiting first sentence, the reporting date, detention institutions included, and the extent to which national administrative categories match the UN metadata. Classification harmonization is not optional in cross-national criminal-justice statistics.

Missing Data, Uncertainty, and Statistical Inference

Missingness is an institutional phenomenon

Quantitative legal datasets have missing values for substantive reasons. A court may lack electronic records; a statistical authority may use a nonmatching case category; a government may decline to participate; violence victims may not respond to surveys; the law under study may be ambiguous; and certain communities may be absent from the sampling frame. The mechanism generating missingness can be related to exactly the institutional capacity or legal vulnerability the study seeks to measure.

Rubin's foundational article Inference and Missing Data (1976) distinguishes analytically consequential patterns of missing-data generation. Under missing completely at random, missingness is independent of the data; under missing at random, it can depend on observed information, given suitable assumptions; under missing not at random, it can depend on unobserved values even after conditioning on what is observed. These are modeling assumptions, not labels automatically known from a missing-value table.

A legal index that assigns zero whenever data are unavailable equates missing evidence with an observed legal failure. A complete-case analysis may privilege systems with the greatest reporting capacity. Imputing an average score based on neighboring jurisdictions may conceal precisely the distinctive institutions comparative law seeks to understand. The response should be determined by the study's objectives, plausible mechanisms, and available auxiliary evidence.

Multiple imputation and sensitivity analysis

Rubin's Multiple Imputation for Nonresponse in Surveys (1987) develops an approach in which several completed datasets are generated under a specified imputation model and uncertainty is combined across analyses. In comparative law, an imputation model might include documented court type, income level, past data, and administrative capacity, but its adequacy depends on whether these variables make the assumed missingness process plausible.

For a jurisdiction whose court-case category has no reliable analogue, numerical imputation may be conceptually inappropriate until harmonization has been performed. Multiple imputation can address statistical missingness under assumptions; it cannot invent a legal equivalence that the research has failed to establish. The codebook should separate “not available,” “not applicable,” “definition incompatible,” and “source disputed” before statistical treatment.

A sensitivity analysis can compare complete-case estimates, appropriately justified imputation, explicit best/worst bounds, and models of nonignorable missingness. Material changes among these analyses should appear in the substantive findings. When the conclusion depends on the assumption that unreported courts perform like observed courts, the published claim should say so.

Sampling uncertainty, model uncertainty, and legal uncertainty

There are several analytically distinct sources of uncertainty. Sampling uncertainty arises because a sample is observed rather than the complete target population. Measurement uncertainty reflects imprecise indicators and potential error in their recorded values. Model uncertainty arises from competing defensible specifications. Legal uncertainty concerns the interpretation of ambiguous sources, competing authorities, and unclear jurisdictional application.

Confidence intervals can quantify selected statistical uncertainties given their procedures and assumptions. They do not automatically encompass the full range of legal interpretation. For example, a survey's reported confidence interval does not measure the probability that coders misunderstood a rule in a foreign language or that its response category is substantively non-equivalent across countries.

A good comparative legal estimate therefore includes a quantitative uncertainty statement and a source-critical note on what remains unquantified. The former might concern cluster-robust standard errors or bootstrap intervals; the latter might record that one country's recent court reform was not yet reflected in the applicable data vintage. Neither form substitutes for the other.

Small-country samples and false precision

Cross-national regressions often involve dozens rather than thousands of independent legal systems, even when datasets contain many country-years or individual responses. Complex models with numerous controls and clustered dependence can produce unstable coefficients. Confidence in an estimate should reflect the number and structure of independent units, model fit, specification choices, and the influence of unusual observations.

For a hypothetical regression of judicial performance on a legal tradition indicator, adding a dozen covariates to a sample of 25 countries does not guarantee correction for all confounding. The additional controls can consume degrees of freedom, encode consequences of legal history, or introduce measurement error. A more demanding design may require fewer variables, clearly justified adjustments, substantive case analysis, and transparent reporting of sensitivity.

Multiple comparisons and publication selection

If investigators examine 50 legal indicators against 30 outcomes, they generate 1,500 potential associations even before changing specifications or subgroups. Standard significance thresholds can then identify apparent relationships by chance. Preregistration, prespecified primary outcomes, correction where appropriate, and disclosure of the complete specification search are valuable safeguards. The correct response depends on whether the exercise is exploratory mapping or confirmatory inference.

The broader question is whether a dataset can support a particular type of claim. A descriptive atlas of statutory rules needs different controls from a policy-impact analysis. Applying one statistical ritual to both misses the actual logic of inference.

Historical Change, Data Versions, and Replicability

Law changes at more than one date

A legal rule can have a drafting date, enactment date, publication date, commencement date, transitional date, effective date, repeal date, and date of judicial interpretation. Quantitative coding requires an unambiguous temporal rule. For a country-year dataset, one convention is the law legally in force on December 31; another is the rule in force on the date an annual business transaction begins. Both are intelligible if the research question and coding dates are recorded. They can produce different observations during transition years.

Institutional statistics add observation windows. A report labeled 2026 may summarize events from 2024 or earlier and cite survey responses gathered over several years. The 2026 EU Justice Scoreboard, for example, reports judicial data covering the 2014–2024 period in its broad efficiency series. The WJP 2025 methodology records multiple years of household-survey fieldwork underlying the same published index edition. A dataset requires at least a publication-vintage field and an underlying legal or statistical observation-period field.

A reform that enters into force after the statistical observation window may have no opportunity to change a contemporaneously published number. This temporal fact is especially important when public dashboards are used to evaluate short-term government actions. Apparent disagreement between a legislative chronicle and an indicator may reflect different clocks rather than false data.

Versioned panels and breaks in series

A longitudinal dataset can change through new source inclusion, altered item wording, respondent-coverage expansion, retrospective correction, or reconstruction of historical scores. When the World Bank revises the WGI methodological framework, a user should preserve the exact downloaded version and investigate retrospective recalculation before comparing it to earlier versions. Similarly, the V-Dem Dataset v16 was issued in March 2026 with an accompanying codebook and cautionary materials; its version identifier is part of the evidence needed to interpret a country-year variable.

A trend analysis should maintain a data-release manifest containing the dataset title, producing institution, version, download date, geographic units, indicator codes, source vintage, codebook version, and script checksum. If a historical correction changes all years, the researcher should report whether analyses use a consistently revised series or an explicitly harmonized combination. Revision is ordinary scientific maintenance; concealed splicing creates ambiguity about what was actually measured.

Legal change may be abrupt, but measured change may be gradual. Expert respondents can update judgments after delay, administrative records can change as new cases work through a system, and population experience can evolve more slowly. The relationship among reform date, implementation, outcome date, and publication date is part of the causal account.

Reproducible legal evidence and reproducible computation

Reproducibility has at least two stages. Documentary reproducibility permits another researcher to recover the source provision, court rule, or official statistical definition that supported a coding decision. Computational reproducibility permits the regeneration of tables, models, and charts from documented data and scripts. Neither alone suffices for a trustworthy comparative legal result.

A complete record contains source links or stable identifiers, original-language citations, translation rules, jurisdiction dates, coding notes, duplicate decisions, dataset transformation scripts, software dependencies, and alternative specifications. The published article can remain readable while the supporting repository preserves the legal detail needed for audit. Licensing and confidentiality may limit release of respondent-level data; in that case the record should state the restriction and make reproducible metadata, lawful aggregates, or synthetic validation data available where possible.

Reproducibility also requires an explicit audit of changes made after results were inspected. An alternative coding that improves a model's fit is still an alternative coding; its choice must have a legal justification independent of the preferred substantive conclusion. The principle is symmetrical: documentary and statistical judgments should be governed by recorded evidence rather than adjusted to obtain an attractive country ordering.

Computational identification and evidence provenance

An integrated research file can be organized around stable observation identifiers such as jurisdiction–institution–legal-proposition–effective-date, rather than assuming each country-year represents a single coherent legal order. Every numerical output then points to its underlying observations and each observation to its source. This is a practical data-engineering recommendation following ordinary principles of traceability, not a newly named model or substantive theory.

When a law is amended, the updated observation should inherit and distinguish its predecessor's source identity. When an indicator value changes because of a methodological revision, the record should distinguish a revised estimate from newly observed legal performance. These provenance checks prevent historical claims from being silently overwritten by newer files.

Integrated Comparative Case Studies and Analytical Findings

Case study — Legal origins and shareholder-protection coding

The comparative question is whether particular statutory and institutional rights of shareholders systematically vary by legal tradition and predict financial outcomes. The early Law and Finance dataset offered an influential operationalization of such rules. Spamann's 2010 audit of the index supplies an unusual and concrete audit trail for the scientific significance of source checking and code reconstruction.

The first analytical step is to identify the legal proposition to be encoded. Does the rule confer the shareholder's right directly by statute, through mandatory company law, via court enforcement, or only through contingent contractual arrangements? The second step is to verify the legal source, temporal version, exclusions, and factual assumptions. The third is to apply the same code to legally equivalent situations, recording interpretive disputes instead of using country tradition as a proxy for the rule itself.

A fourth step reruns the relevant associations under the original and revised variables. If the estimated association changes, the immediate finding concerns the sensitivity of the empirical result to the treatment variable's construction. One cannot infer from the change alone that a historical legal tradition has no economic influence. That wider causal proposition needs separate evidence and identification. The intellectually defensible conclusion is precise: legal indicators are findings produced by coding, and replication of the coding can change downstream econometric results.

The case also clarifies why a single textual code cannot measure an entire remedial system. An apparently shareholder-protective rule may require expensive litigation, may be limited by standing, or may interact with disclosure, enforcement, and ownership structures. Distinguishing formal entitlement from operational enforcement gives a more useful comparative account.

Case study — Judicial efficiency and false equivalence of duration

The comparative question is whether one judicial system processes comparable disputes more efficiently than another. The CEPEJ evaluation series and EU Justice Scoreboard provide authentic statistical and institutional reporting frameworks, including warnings about category matching and the difference between estimated and observed duration.

Suppose a researcher reads an estimated disposition time of 550 days in hypothetical court A and an observed median duration of 250 days in hypothetical court B. Treating these numbers as a direct comparison of “case length” is invalid because their estimands differ. Court A's 550-day figure may be derived from pending-case stock and annual resolution flow; court B's 250-day figure describes the median among a specified set of observed cases. Harmonization requires either calculating the same stock-flow estimator for both from compatible inventories or obtaining comparable case-level elapsed-time distributions.

Now suppose that both courts have the same reporting category and use the same estimator, but court A includes uncontested matters while court B excludes them. The metrics remain non-equivalent with respect to the study's intended population. The apparent quantitative precision of two decimals does not resolve a legal-category mismatch.

A defensible report therefore shows case classifications, instance, subject matter, procedural exclusions, observation year, incoming flow, resolved flow, pending stock, statistical estimator, and sensitivity to the case mix. Its interpretation can then ask whether the observed difference reflects resources, institutional design, procedural incentives, or changes in demand. Numerical comparison is the beginning of explanation.

Case study — Composite rank reversal under plausible weights

Consider three hypothetical jurisdictions, Alpha, Beta, and Gamma. Each has three invented normalized scores: formal rule clarity, operational access, and procedural speed. These numbers are constructed solely to demonstrate mathematical sensitivity and have no empirical country referents.

Hypothetical jurisdiction Rule clarity Access Speed Equal-weight mean Access-focused weighted mean
Alpha 0.90 0.30 0.60 0.600 0.480
Beta 0.65 0.65 0.65 0.650 0.650
Gamma 0.45 0.85 0.65 0.650 0.730

The access-focused composite assigns weights of 0.20 to clarity, 0.60 to access, and 0.20 to speed. All table entries are exact illustrative inputs except means rounded to three decimals. Under equal weights, Beta and Gamma tie above Alpha. Under the access-focused criterion, Gamma ranks above Beta and Alpha. The change in Gamma's relative standing comes entirely from a disclosed change in valuation of domains, not from newly observed institutional facts.

A researcher who reports only the access-weighted ranking conceals the normative premise that access should dominate. A researcher who reports only the equal-weight ranking makes an equally consequential decision to treat all three dimensions as substitutable in equal proportions. Both indices can be computed transparently; the relevant scientific task is to display the sensitivity and articulate why the chosen weights serve the declared question.

The same principle applies to uncertainty about measurements. If Gamma's access estimate of 0.85 is based on fewer respondents or a narrower sampling frame than Beta's 0.65, one should not equate the apparent 0.20 gap with a statistically resolved or legally complete superiority claim. The weight exercise and a measurement-error exercise answer distinct questions.

Case study — Partial identification with a missing jurisdiction

Assume five hypothetical jurisdictions are intended for equal-weight descriptive comparison of a scalar legal-access outcome between zero and one. Four have observed values 0.40, 0.50, 0.70, and 0.80; the fifth is missing. The mean of observed jurisdictions is 0.60, but the target mean for all five is bounded by:

[ \frac{0.40+0.50+0.70+0.80+0}{5}=0.48 ]

and

[ \frac{0.40+0.50+0.70+0.80+1}{5}=0.68. ]

Without a defensible assumption about the missing jurisdiction, the full-population mean is between 0.48 and 0.68. Assigning the observed mean to the missing case produces 0.60, a single imputed point that depends on an assumption not established by the four values. This elementary bound can be more honest and more informative than a complicated imputation unsupported by legal comparability.

The example also highlights the difference between statistical missingness and conceptual inapplicability. If the fifth jurisdiction organizes access through institutions outside the original indicator's definition, simply imputing its value may conceal a problem in the target concept. The correct first task is to decide whether the variable describes a common institution or whether the research universe needs redefinition.

Case study — Survey precision and cross-cultural meaning

Assume a hypothetical simple random sample of 1,000 independent residents and a reported proportion of 0.50 responding that a dispute-resolution procedure is accessible. The conventional binomial standard error is:

[ SE=\sqrt{\frac{0.5(1-0.5)}{1000}}\approx0.0158. ]

An approximate normal 95 percent sampling interval is 0.50 ± 1.96(0.0158), or roughly [0.469, 0.531]. This calculation assumes the stated sampling model and ignores clustering, nonresponse, weights, item ambiguity, and measurement error. It describes uncertainty about a defined survey response proportion, not the metaphysical probability that the system “has access to justice.”

If residents in two legal cultures interpret the term “accessible” according to different expected fees, distances, linguistic barriers, and reference institutions, their responses may differ even under similar operational conditions. King and Wand's anchoring-vignette analysis supplies methods relevant to diagnosing differences in survey response scales. Contextual legal questions and alternative survey formulations can provide additional evidence concerning measurement equivalence.

The example identifies the central double requirement of comparative research: a narrow confidence interval is useful only when the measured concept and response category are sufficiently understood.

Case study — Cross-program comparison without a false composite

Suppose a jurisdiction's WJP civil-justice subfactor is lower than the peers' corresponding scores, its WGI rule-of-law estimate is near a comparison-group average, and its B-READY operational-efficiency measure is relatively high. There is no necessary contradiction. The instruments use different constructs, populations, data collection, and aggregation. The WJP measurement description, WGI documentation, and B-READY methodological framework explain their respective targets.

One possible research program would examine whether relatively efficient business-facing administrative operations coexist with obstacles experienced by residents resolving civil grievances. That is a hypothesis for primary research, not an inferred fact about any named country. The analyst would inspect source items, administrative categories, jurisdiction-specific court procedures, and possible overlap among providers. It would be methodologically unsound to average the three index scores into a new “overall legal performance” measure without a defensible construct and common scale.

The example shows the value of disaggregated triangulation. Disagreement across instruments may reveal different aspects of a legal order, measurement differences, or incompatible time periods. Investigating the discrepancy can produce more legal knowledge than selecting one index as definitive.

Findings from the integrated analysis

Several findings follow from the sources and worked examples. First, comparative numerical observations are always conditional on a legal definition of what is counted. Second, a shared numerical range does not create comparability across unlike institutional objects. Third, corrections to legal coding can materially change estimates and substantive claims, as documented in the shareholder-rights literature. Fourth, composite scores and ranks depend on choices that combine empirical and normative premises. Fifth, versioned datasets and country-year labels require a documented observation calendar. Sixth, uncertainty about survey sampling, legal interpretation, and causal identification must be distinguished. Seventh, numerical evidence achieves its greatest scholarly value when joined to source-specific doctrinal analysis, not when one ranking stands in for the legal order it summarizes.

These findings are interpretive syntheses of existing methods and evidence. They are offered as conclusions of this monograph, without attribution of a novel general theory or independent empirical effect estimate.

Conclusion

Quantitative comparative law has become a consequential form of legal inquiry because modern legal institutions generate extensive records of rules, cases, experiences, and administrative operation. Its central achievement is methodological visibility. It makes researchers state which legal facts have been selected, what scale represents them, which institutions are compared, and which numerical relationships can be checked against evidence. That achievement is substantial precisely because legal orders differ in the sources, authority, classifications, procedures, and histories through which they operate.

The first conclusion is that a legally meaningful observation must precede numerical comparison. Identical codes should refer to equivalent propositions under a documented coding rule. When the objects differ, harmonization requires substantive legal work. A statutory entitlement, an expert's institutional assessment, a resident's reported legal experience, and an administrative count can each be measured reliably while describing different features of legal reality. They belong in a common analysis when their relationships are explained, not because they can all be displayed on a scale from zero to one.

The second conclusion concerns the epistemic status of composite indicators. The OECD's composite-indicator guidance provides methods for investigating how normalization, weighting, missing values, and aggregation affect outputs. The worked examples in this monograph show why no single country rank should be read as a complete comparative judgment. One plausible weighting can produce a different ordinal outcome from another plausible weighting even when every underlying observation is unchanged. The appropriate response is to publish the constituent dimensions, disclose the chosen aggregation, and test sensitivity to alternatives.

The third conclusion concerns institutional specificity. The WJP Rule of Law Index, Worldwide Governance Indicators, Business Ready methodology, and European judicial evaluation framework are different scientific products. Their constructs, respondents, time periods, statistical models, and operational units differ. WJP offers rule-of-law factors grounded in population and expert evidence; WGI estimates broad governance dimensions through aggregated source assessments; B-READY explicitly distinguishes regulatory design, public services, and operational efficiency; European judicial statistics trace court-system activity within specified categories. Treating each instrument according to its actual scope produces a stronger account of law than collapsing them into a generic cross-national quality score.

The fourth conclusion is that revision and uncertainty are ordinary features of legal evidence. Legal rules change over time, data definitions are refined, surveys incorporate samples from different years, and scores can be recalculated retrospectively. A reproducible study must preserve publication vintages and observation periods together with the sources and interpretation that supported its codes. The international legal-origins literature demonstrates how revised legal coding can affect previously influential findings. The re-examination of shareholder-rights coding is especially instructive because it connects close doctrinal scrutiny to the validity of downstream quantitative conclusions.

The fifth conclusion is that uncertainty must be reported at its proper level. Sampling intervals, model sensitivity, legal-source ambiguity, incomplete population coverage, and alternative coding rules answer different questions. A precise standard error does not cure a non-equivalent legal category. A carefully bounded legal interpretation does not erase sampling error. Missing-data techniques can support inference under stated assumptions but cannot create an absent legal fact or resolve a normative disagreement about the target construct.

The sixth conclusion concerns explanation. The combination of an institutional score and an economic outcome does not identify an effect of law merely because the two are correlated. A causal argument must define its intervention, comparison, counterfactual, confounders, observation period, and relevant mechanism. Historical legal classifications can guide research questions, but the empirical explanation must be established with suitable designs and substantive evidence. In settings where causal identification is weak, careful descriptive and diagnostic findings remain valuable and should be presented as such.

The seventh conclusion concerns the public use of indicators. Quantification shapes the decisions of governments, courts, international institutions, scholars, and affected communities. It can expose injustices and monitor reforms, yet it also creates incentives to optimize measured procedures and overlook unmeasured legal needs. Its legitimate institutional use depends on documentation, transparency, disaggregation, attention to omitted populations, and the preservation of rights alongside efficiency. The UN's SDG 16.3 indicators show why measures of reporting, detention, and dispute-resolution access must retain their separate denominators and purposes.

These conclusions locate quantitative comparative law within the larger scholarly task of understanding how legal norms operate across diverse social and institutional settings. Its method is strongest when legal interpretation establishes what counts as an observation, statistical analysis determines what can be inferred from those observations, and contextual comparison explains the mechanisms and limits of the observed result. The result is an accountable body of comparative knowledge whose numerical claims can be reconstructed, challenged, corrected, and applied with precision.

References

Adcock, Robert, and David Collier. 2001. “Measurement Validity: A Shared Standard for Qualitative and Quantitative Research.” American Political Science Review 95 (3): 529–546. Publisher article.

Angrist, Joshua D., and Jörn-Steffen Pischke. 2009. Mostly Harmless Econometrics: An Empiricist's Companion. Princeton, NJ: Princeton University Press. Publisher book record.

Cohen, Jacob. 1960. “A Coefficient of Agreement for Nominal Scales.” Educational and Psychological Measurement 20 (1): 37–46. Publisher DOI.

Council of Europe, European Commission for the Efficiency of Justice (CEPEJ). 2024. Evaluation of European Judicial Systems: 2024 Evaluation Cycle (2022 Data). Strasbourg: Council of Europe. Evaluation publications and methodology.

Davis, Kevin E., Benedict Kingsbury, and Sally Engle Merry. 2012. “Indicators as a Technology of Global Governance.” Law & Society Review 46 (1): 71–104. Publisher DOI.

Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2003. “Courts.” Quarterly Journal of Economics 118 (2): 453–517. Publisher DOI.

European Commission. 2026. The 2026 EU Justice Scoreboard. Brussels: Directorate-General for Justice and Consumers, Publications Office of the European Union. Official edition.

Hernán, Miguel A., and James M. Robins. 2020. Causal Inference: What If. Boca Raton: Chapman & Hall/CRC; freely available online edition. Author-maintained book.

Kaufmann, Daniel, Aart Kraay, and Massimo Mastruzzi. 2010. “The Worldwide Governance Indicators: Methodology and Analytical Issues.” World Bank Policy Research Working Paper 5430. Authoritative research summary.

King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry: Scientific Inference in Qualitative Research. Princeton, NJ: Princeton University Press. Publisher book record.

King, Gary, and Jonathan Wand. 2007. “Comparing Incomparable Survey Responses: Evaluating and Selecting Anchoring Vignettes.” Political Analysis 15 (1): 46–66. Publisher DOI.

La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny. 1998. “Law and Finance.” Journal of Political Economy 106 (6): 1113–1155. Publisher DOI.

La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer. 2008. “The Economic Consequences of Legal Origins.” Journal of Economic Literature 46 (2): 285–332. Publisher DOI.

Merry, Sally Engle. 2016. The Seductions of Quantification: Measuring Human Rights, Gender Violence, and Sex Trafficking. Chicago: University of Chicago Press. Publisher book record.

Organisation for Economic Co-operation and Development (OECD) and European Commission Joint Research Centre. 2008. Handbook on Constructing Composite Indicators: Methodology and User Guide. Paris: OECD Publishing. DOI.

Rubin, Donald B. 1976. “Inference and Missing Data.” Biometrika 63 (3): 581–592. Publisher DOI.

Rubin, Donald B. 1987. Multiple Imputation for Nonresponse in Surveys. New York: John Wiley & Sons. Book DOI.

Seawright, Jason, and John Gerring. 2008. “Case Selection Techniques in Case Study Research: A Menu of Qualitative and Quantitative Options.” Political Research Quarterly 61 (2): 294–308. Publisher DOI.

Siems, Mathias. 2022. Comparative Law. 3rd ed. Cambridge: Cambridge University Press. Publisher book record.

Spamann, Holger. 2010. “The 'Antidirector Rights Index' Revisited.” Review of Financial Studies 23 (2): 467–486. Publisher DOI.

Spamann, Holger. 2010. “Legal Origin, Civil Procedure, and the Quality of Contract Enforcement.” Journal of Institutional and Theoretical Economics 166 (1): 149–165. Publisher article.

United Nations. 2011. The United Nations Rule of Law Indicators: Implementation Guide and Project Tools. New York: United Nations. Official United Nations guide PDF.

United Nations Statistics Division. 2026. Global Indicator Framework, Sustainable Development Goal 16, Target 16.3: Indicator Metadata. New York: United Nations. Official indicator metadata.

Varieties of Democracy (V-Dem) Institute. 2026. V-Dem Dataset, Version 16, and Associated Documentation. Gothenburg: University of Gothenburg. Versioned dataset and codebook.

World Bank. 2021. “World Bank Group to Discontinue Doing Business Report.” Statement, September 16. Washington, DC: World Bank. Official statement.

World Bank. 2025. Business Ready (B-READY) 2025. Washington, DC: World Bank. Official report record.

World Bank. 2026. Business Ready (B-READY): Methodology. Washington, DC: World Bank. Official methodology.

World Bank. 2026. Worldwide Governance Indicators: Data and Release Information. Washington, DC: World Bank. Official program and data releases.

World Bank. 2026. Worldwide Governance Indicators: Documentation, Methodology, and Revisions. Washington, DC: World Bank. Official documentation.

World Justice Project. 2025. WJP Rule of Law Index 2025: Methodology. Washington, DC: World Justice Project. Official methodology PDF.

Zweigert, Konrad, and Hein Kötz. 1998. An Introduction to Comparative Law. 3rd ed. Translated by Tony Weir. Oxford: Clarendon Press. Bibliographic record.