Utilizing a Comparative Policy Resource from the WORLD Policy Analysis Center Covering Constitutional Rights, Laws, and Policies across 193 Countries for Outcome Analysis, Monitoring, and Accountability

Abstract Historically, a lack of comparable data on the laws and policies that shape health, education, poverty, and other outcomes has hindered researchers’ ability to provide rigorous evidence on the effectiveness of different policy designs. This article describes public-use downloadable data built by the WORLD Policy Analysis Center to fill this gap. Over 2,000 quantitatively comparable measures of national laws and policies across 193 countries were assembled. This open-access data source provides a tool for monitoring the adoption of evidence-based laws and policies, identifying policy gaps, and rigorously evaluating how policies shape outcomes across different regions and socioeconomic contexts.


Introduction
Laws and policies shape the environments in which people live and work, affecting outcomes ranging from the extent of education to the quality of employment, from experiences of discrimination and marginalization to the safety and adequacy of living conditions, from the risks of getting sick to the likelihood of receiving quality care during illness or after injury. Comparative policy analysis seeks to learn from the full range of countries' experiences in making policy design decisions to inform future decision-making and improve outcomes.
Experimental methods have enabled researchers to rigorously evaluate the impacts of pilot programs and other small-scale initiatives. While these can be important sites for innovation and can provide insights into the improvement of social conditions, action by national governments is essential for taking effective solutions to scale. Identifying and assessing these actions requires collecting data on national laws and social policies in areas that affect education, health, poverty, and other measures of well-being, monitoring the implementation of those policies, and rigorously evaluating their impacts on outcomes. These steps are critical for translating research evidence into measurable improvements in people's lives.
To date, a lack of existing comparable data on laws and policies has hampered the ability of researchers to conduct rigorous studies on the effectiveness of different policy solutions. In cases where policies have been shown to be effective, this lack of data has also limited the global community's ability to monitor progress and policymakers' ability to make evidencebased decisions on how best to improve population well-being. Comparative analyses have often faced trade-offs between the depth of topics of information systematically obtained and the number of countries analyzed because of limited data availability. Moreover, approaches to building internationally comparable indexes have at times suffered from challenges due to insufficient transparency or conceptual limitations, among others (Hoffmann and Dooren 2017). Efforts to bring together policy in individual areas have often been limited by the number of countries, comparability of measures, or comprehensiveness (Marmor et al. 2005).
To fill this gap, since 2000 we have been collecting data on laws and social policies that have the potential to advance health, equity, and well-being. In 2004, we published our first findings from this initiative, which examined work-family policies in 168 countries (Heymann et al. 2004). In the years since, we have continuously expanded and updated our databases to capture a wide range of laws and policies that are fundamental to health, equity, and well-being. Today, WORLD Policy Analysis Center (WORLD), our institute based at University of California, Los Angeles (UCLA), captures well over 2,000 health, social, economic, and environmental policy and program indicators for all 193 United Nations member states. As of 2020, WORLD's databases contain legal and policy data on education, adult labor and working conditions, family, migration, disability, prohibition of discrimination, child labor, child marriage, climate adaptation, constitutional rights, social protection, and poverty reduction in select years from 2007 to 2020. To address problems highlighted in the existing literature, we apply a consistent framework capturing actionable, key policy details based on legal text.
WORLD is the largest globally comparative data resource focused specifically on laws, policies, and constitutional rights that impact health, equity, and well-being. Across databases, emphasis is placed on whether legal rights and protections extend coverage to all. Quantitative measures are designed to capture key aspects of policy design decisions to support actionable research that can inform evidence-based decision-making to improve education, health, income security, equality, and other outcomes.
In this article, we first summarize our approach to constructing detailed, globally comparative policy indicators in a wide range of social policy areas. We then provide an illustrative example of how our data in one area, paid parental leave policies, are constructed and analyzed. Finally, we illustrate how these data can be used to achieve the central goal of comparative policy analysis of contributing to better policy (Radin and Weimer 2018) by highlighting in this paper some ways that the data has been used.
Many factors, including social norms, politics, and windows of opportunity, influence policy decisions in simultaneous and complex ways. Research evidence can play an important role in policymaking. At the same time, policy change is not possible without the efforts of many individuals and organizations working at different levels. The goal of this article is to share information on publicly available data that researchers and analysts can use to generate evidence on which policies and laws are effective, how countries' legal frameworks compare to their commitments and actions taken by comparable nations, and how they can be more effective.

Construction and Content
In deciding what measures to include in our databases, we prioritize those that research evidence demonstrates are important to health, equity, or well-being across a range of settings, and/or those for which there is widespread global consensus on their importance. To determine which areas meet these criteria, we undertake systematic literature reviews identifying policies that create or remove barriers to equal opportunities; analyze widely ratified international treaties, as well as global commitments such as the Sustainable Development Goals, to identify priority areas established by national governments; and consult with civil society organizations, intergovernmental organizations, policymakers, and other stakeholders.

Data Sources
To ensure accuracy, whenever possible data are coded from primary national legal sources, such as original legislation and constitutions. We have obtained full-text legislation primarily through the International Labour Organization's NATLEX database; Lexadin World Law Guide; United Nations Educational, Scientific and Cultural Organization's (UNESCO)

Comparative Policy Resource across 193 Countries 315
Observatory on the Right to Education; United Nations High Commissioner for Refugees' Refworld; and the Constitute Project. We have supplemented these resources with legislation available through national government websites; the World Legal Information Institute; the World Bank's Women, Business and the Law; and UNESCO's Planipolis. When unable to obtain copies electronically, we have searched for hard copies of legislation through university libraries, including those at McGill University, Harvard University, UCLA, and the Swiss Institute for Comparative Law.
Although we use primary data whenever possible, we also use secondary sources when information on particular countries is unclear or missing. We give priority to secondary sources that are comparable across many countries, especially national reports on laws and policies submitted to the UN or other official global or regional bodies. We use this information to clarify, complement, or corroborate information coded from primary sources.

Coding Framework
We develop coding frameworks to capture differences in legislative approaches consistently across countries. We prioritize a framework that allows us to capture the richness and variety of approaches taken by different countries while simultaneously presenting a meaningful comparison of policy variation across countries. The development of each coding framework proceeds through an iterative process, which allows us to refine and improve measures as we learn more about the range of existing approaches.
First, we identify the essential features of each law, policy, or constitutional right based on research evidence and global treaties and agreements. Next, we review legal text from 20 to 30 countries selected across regions and income levels to develop categories for the essential features. We revise these categories to capture the full variety of approaches taken by countries. We then apply this framework to an additional 10 to 20 countries before coding all 193 countries.

Coding Process
A detailed set of coding rules are systematically developed from the framework described above. The coding manual includes the questions, valid response options, and illustrative examples to clarify coding for complex cases. Coding is carried out by a team of policy analysts fluent in multiple languages, including most UN official languages and several supplementary ones. A dedicated coding team is trained and applies the approach across all countries to ensure that countries are accurately and consistently coded in the databases. Each policy analyst codes independently, reading legislation in full, in its original language whenever possible. Each country is independently coded by at least two policy analysts who then compare their answers to minimize human errors. The coding team convenes regularly to discuss difficult cases and ensure consistency in coding across policy analysts. Questions and decisions made in these team coding meetings are recorded and the coding rules are updated to reflect the decisions from the meeting and clarify or update existing coding rules. In rare cases, it is necessary to modify rules in the coding framework as novel approaches are discovered during coding. When these additions occur, previously coded countries are reviewed so they can be analyzed fully according to the expanded framework.

Quality Assurance
Our goal is to ensure that these data are accurate and up-to-date. We take three main approaches to minimizing errors. First, we double-code all countries to reduce the potential for human error or differences in interpretation. Second, once coding is complete, we conduct systematic random checks and verify outliers on particular variables, using additional sources where necessary. Finally, we update databases periodically, systematically reviewing all available sources.

Data Created
Using the methods described above, WORLD has created more than 2,000 national-level indicators of laws, policies, and constitutional rights. The available data are outlined in Table 1. Because the largest investment of time is building the initial database and this data is meant to be an ongoing tool to monitor country progress towards enacting stronger laws and policy, we aim to update existing databases every 2-3 years. These updates are also an opportunity to expand the indicators included to address feedback from civil society, policymakers, and other researchers.

Applying This Approach
In this section, we illustrate the application of this approach in one of the areas covered by the data resource: paid parental leave. Paid parental leave was selected because a wide range of studies have demonstrated its importance for health and economic outcomes . Moreover, international treaties and agreements from the UN and the International Labour Organization reflect a commitment from all countries to provide paid leave, protect health, and advance gender equality in employment (United Nations 1966, 1979, 1990International Labour Organization 2000).

Identifying Sources
To begin building a global database of paid parental leave policies, we identified labor and social security legislation through the ILO's NATLEX, and supplemented this source with information from the ILO's Maternity Protection database, Social Security Programs throughout the World, the International Review of Leave Policies and Related Research, and the Mutual Information System on Social Protection.

Identifying Key Policy Features
In coding, initial key features identified for paid parental leave were the duration, wage replacement rate, whether leave was available to all genders, and how the cost of leave was covered. Specific details from the relevant global instruments, alongside the state of the research evidence at the time, further informed our development of indicators and subsequent analysis. For example, the ILO Maternity Protection Convention outlined that all countries should have at least 14 weeks of paid leave paid at 66 per cent of wages (International Labour Organization 2000). Existing research had identified an association in high-income countries between the duration of paid maternal leave and infant mortality (Ruhm 2000;Tanaka 2005 Raub et al. research was beginning to discuss the importance to gender equality of men having access to paid parental leave as well as women (Haas 1990;Haas and Hwang 2000;Tanaka and Waldfogel 2007;Nepomnyaschy and Waldfogel 2007). Further indicators were added in response to policymaker and civil society inquiries regarding approaches to covering the informal economy.

Developing Comparable Measures
Our process of developing indicators for paid parental leave illustrates how seemingly simple questions often become complex when trying to create comparable measures across 193 countries. For example, how do you make durations of paid maternity leave comparable across countries when the unit of measure varies? Are durations given in calendar days or working days? Is it a five-or six-day work week? These are challenges common to comparative analysis (Hoffmann and Dooren 2017). For paid parental leave, rather than recording only a converted summary measure in the database, we captured in numeric fields both the duration and unit of time to minimize human error in coding and allow for greater flexibility and quality checks in analysis later.

Variable Construction
Across areas, we use the final databases to construct actionable indicators that can be used to rapidly highlight where countries lack policies entirely, as well as answer more detailed questions about the important decisions that policymakers face that can disproportionately have impacts on marginalized individuals and their families. We prioritize creating separate indicators that can answer detailed questions over creating composite measures or rankings, recognizing that rankings can obscure important policy decisions. Moreover, a low ranking does not rapidly identify for decision-makers where policies could be strengthened to markedly expand coverage. For example, in the case of paid parental leave, we separately analyzed the following factors: the duration, the wage replacement rate, whether leave is provided through employers or social security systems, whether there are tenure requirements to access leave, whether there is job protection while on leave, whether there are minimum employer size requirements for leave eligibility, whether self-employed workers are entitled to benefits, whether different forms of vulnerable and informal employment are covered, whether there are extensions to paid leave due to health complications, whether leave can be taken before birth, and whether leave is available in the case of adoption. Detailed questions were answered about mechanisms to promote fathers' use of leave, availability to single parents, same-sex parents. The full set of indicators is described in Table 2.
Variables for all policy areas are constructed in Stata 14. In many cases, constructed variables will pull from many aspects of the raw database in order to ensure that the diverse range of approaches countries take are captured accurately in the final comparative measures in the dataset. Emphasis is placed on creating measures that can be rapidly understood without losing key features that might matter based on international agreement or research evidence. For example, in the case of paid parental leave, information on the duration of paid maternal leave includes both leave reserved for mothers and leave that can be taken by either parent. For data visualization and public use datasets, these data are further categorized by weeks of leave. Categorization is informed by international standards including the minimum duration of paid maternal leave established by the ILO Maternity Protection Convention (14 weeks), as well as the duration of leave that may promote better health outcomes by enabling exclusive breastfeeding for at least six months, as recommended by the World Health Organization. This approach allows the rapid identification of where countries are failing to meet standards outlined in international agreements, as well as where countries could extend leave to promote better health outcomes, at the same time as ensuring detailed data are also available.

Utility and Discussion
This data resource is a powerful tool to inform the research, policy, and monitoring efforts of a wide range of stakeholders. Specifically, researchers and international organizations can use the data to understand where the world stands in a given policy area, and monitor the adoption of evidence-based policies over time; researchers can further strengthen the evidence by linking the global policy data with household-level data on health, education, economic, or other outcomes; and policymakers and civil society can leverage the data to advocate for effective policy reforms. Each of these functions is explored in greater detail below.

Monitoring the Adoption of Evidence-Based Policies
In areas where effective approaches to improving outcomes are known, this resource can enable researchers, policymakers, civil society, and international organizations to monitor the adoption of evidence-based laws and policies and quickly identify policy gaps. A range of studies have demonstrated these capabilities. In terms of monitoring effective policy approaches, researchers analyzing the data have evaluated what share of the world's countries have enacted laws to provide paid maternal leave, highlighting that only eight countries globally lack paid leave for mothers of infants. In-depth studies have also examined how countries approach questions of policy design to ensure that coverage of policies extends to include everyone, including groups that have been historically marginalized. For example, studies have covered detailed questions of eligibility based on tenure requirements (De Guzman Chorny et al. 2019), affordability of benefit levels , and equality in access to paid leave for same-sex couples and single-parent families (Wong et al. 2019). In addition to providing in-depth overviews of where the world stands in particular policy areas, these types of evaluations can be used to assess countries' progress towards fulfilling international human rights agreements .

Linking Policies to Outcomes to Identify What Works
In areas where the most effective approach is unknown, this resource can enable researchers to learn whether specific policies improve health and social conditions across countries, and for

Comparative Policy Resource across 193 Countries 321
which population sub-groups. To strengthen the available evidence on effective approaches and to better understand the relationships among different policy strategies, researchers use these data to analyze the link between policies and outcomes. For example, by the early 2000s, studies of paid maternal leave in high-income countries had found that an increase in paid leave was associated with lower infant mortality rates (Ruhm 2000;Tanaka 2005). Yet there were real questions about whether the same would hold true in lower-income settings where more women work in the informal economy or the implementation challenges might be greater. Building on this work, the global cross-sectional data were first used to identify an association between paid maternal leave and lower infant mortality rates, as well as higher vaccination rates. These studies were suggestive of a relationship, but caution was important in interpreting the results given the difficulty of controlling for the confounding factors among countries and the associational nature of the study. The development of longitudinal policy data opened the door to more rigorous analysis on the impact of policies on outcomes. WORLD and McGill University's PROSPERED Project have collaborated to track not only current policies, but also how legal protections have evolved over time in key areas, including paid parental leave (PROSPERED 2020). The longitudinal data can be merged with existing, publicly available household survey data to use multilevel modelling to rigorously analyze the impact of changes in policies on changes in outcomes at the individual level while controlling for individual, household, and national-level factors. Findings from such analyses offer a more powerful foundation for informing evidence-based decisionmaking by providing information about the true short-and long-term benefits of policy reforms to weigh against short-term costs.
These data made possible the first quasi-experimental study on paid maternity leave in lowand middle-income countries, which merged the longitudinal policy data with data on infant health outcomes from 300,000 live births across 20 countries, while controlling for sex, maternal age, maternal education, birth spacing, urban/rural residence, household socioeconomic status, presence of a skilled birth attendance, country GDP per capita, female labor force participation, per capita total health expenditure, and per capita government health expenditure (Nandi et al. 2016). The study demonstrated that each additional month of paid maternity leave was associated with a 13 per cent relative reduction in infant deaths. Moreover, subsequent studies have been able to illustrate mechanisms for this reduction by quantifying the impact of expanding paid maternity leave on vaccination rates, breastfeeding rates, and diarrheal disease (Hajizadeh et al. 2015;Chai et al. 2018Chai et al. , 2020. These data can also be used to address questions of how laws and policies can shift norms and values to advance more equitable outcomes for all, or whether norms and values need to change first to facilitate the full implementation of better laws and policies. For paid parental leave, individual country studies have suggested that leave policies structured to encourage men to take leave more equally could advance gender equality, but questions remained about selection bias and whether norms were driving policy changes. To answer these questions, a recent study using the comparative policy data examined the relationship of policy changes to attitudes towards women's work from 1995 to 2018. It compared changes in attitudes in countries that adopted a policy to incentivize or encourage fathers to take paid parental leave with countries that did not have a policy change, while controlling for confounding factors, finding that changes to parental leave policies that encourage fathers to take parental leave led to more egalitarian attitudes towards women working (Omidakhsh et al. 2020). Another recent study found that paid parental leave policies improved gender equality in household decision-making, with positive effects on women's and their children's health outcomes .
Additionally, the comparative legal data provide a tool that researchers can use to identify where implementation gaps are occurring. While the limited current availability of household survey data on income sources across lower-income countries limits measuring implementation directly for paid parental leave until better data become available, the extent of implementation for other policy areas can be mapped using these data. For example, data on the legal minimum age of child marriage can be merged with Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Surveys (MICS) data on age at first marriage to understand whether girls are being married due to a lack of implementation or legal loopholes, such as where legislation permits early marriage with parental consent or under religious or customary law. The data can also be used to analyze whether, where, and under what circumstances laws closing loopholes have been successfully implemented.

Supporting Intergovernmental Organizations
UN agencies, intergovernmental organizations, and other international organizations have used this data resource to understand where gaps persist in laws and policies that matter to education, health, economies, well-being, and equal opportunities (Heymann and McNeill 2012;Mahon and Heymann 2012;Latz et al. 2013;. For example, the United Nations Children's Fund (UNICEF) used the Data Center's data in a series of publications to highlight legal gaps in paid parental leave and the evidence on why these policies matter. These included a 2019 report on Family-Friendly Policies spotlighting research evidence on the importance of paid leave for health, which also used the global policy data to highlight how few countries ensure fathers have access to paid parental leave (UNICEF 2018). WORLD data has also been used by the United Nations Department of Economic and Social Affairs (UNDESA), the African Union, reports monitoring progress on the Sustainable Development Goals, and to support the ongoing work of UN human rights committees (African Union 2018; UN Department of Economic and Social Affairs 2018).

Informing Policy Debates at the National Level
Beyond its demonstrated utility for researchers, this data resource has provided actionable information for policymakers and civil society working to enact or strengthen evidence-based policies at the country level. For example, in support of a bill to double maternity leave from 60 to 120 days, two senators in the Philippines cited WORLD's data showing that their country provided relatively little maternity leave compared to international standards and a collaborative study that built on these data to demonstrate that longer leave policies reduced infant mortality. In October 2018, the House of Representatives in the Philippines approved a similar bill guaranteeing 105 days of leave, which was signed by the president in February 2019 (CNN Philippines 2019). In Ireland, a 2015 report by the global civil society advocacy organization Equality Now used findings from this database in a report highlighting that Ireland was one of the few high-income countries that did not provide paid paternity leave (Equality Now 2015). This finding informed debates in Parliament, which in 2016 enacted a law providing fathers with two weeks of paid paternity leave (Irish Statute Book 2020). These findings have also been shared with policymakers and advocates working to advance state-

Comparative Policy Resource across 193 Countries 323
based reforms in the United States. For example, when Washington state was considering passing a new paid leave policy, policymakers concerned about the economic feasibility of different approaches consulted the policy data from other high-income countries to better understand the options for coverage and financing. In particular, policymakers took note of data from other OECD countries highlighting that no other country exempts small businesses entirely from paid leave policies and illustrating the full range of financing mechanisms used to share costs among employers, employees, and government. Washington went on to pass the most generous paid leave policy in the US, which notably excluded the exemptions based on firm size that have been a common feature of the American policy landscape.

Complementary Initiatives
Monitoring national action towards advancing health, equity, and well-being is enhanced when there are initiatives based in both university and intergovernmental settings. An example of a university-based complementary initiative with tremendous depth in constitutional provisions is the Comparative Constitutions Project (2020). An example of a complementary intergovernmental organization-based initiative that looks at legal barriers to women's economic opportunities across a wide range of areas is the World Bank's Women Business and the Law (World Bank 2020).
WORLD's data is unique in four main ways. First, WORLD bridges constitutional rights, laws, and policies, enabling the examination of whether the fundamental rights embedded in countries' constitutions have been realized in national legislation and policies. For example, do gender disparities persist in laws despite a constitutional guarantee of gender equality? Second, WORLD looks across a wide number of substantive areas, recognizing that policies in multiple different areas are likely to impact equal opportunities. For example, children's school attendance may be impacted not only by whether education is tuition-free and compulsory, but also by whether children are protected from work that is likely to interfere with schooling and whether girls are protected from early marriage. Third, WORLD links data on constitutional rights, laws, and policies to individual-and household-level outcomes to understand not only the impacts of policy choices on economic or health outcomes overall, but also any variation in impact for the poorest and most marginalized groups. Fourth, WORLD focuses on policies that cover at least the majority of the population. Our general approach is to capture the lowest level of protection when laws vary sub-nationally, rather than the policy applying to, for example, a specific city or subnational unit. This approach also pays special attention to approaches countries take to extending labor and social security protections to workers in vulnerable and informal employment.

Strengths and Limitations
This source provides the most comprehensive quantitative data on the legal and policy approaches countries are taking to improving social conditions, equal rights, and equal opportunity. There are few other data efforts that allow for quantitative comparisons of laws and policies. Moreover, existing quantitative policy data have generally not offered the level of detail needed to inform evidence-based decision-making. This data resource, when combined with existing global data on outcomes at the individual and national levels, is a unique tool to identify the most effective policy approaches to both improving individual well-being and health and enabling countries to thrive socially and economically. WORLD's policy datasets are easy to harmonize with existing global datasets on health outcomes, such as the data from the World Health Organization's Global Health Observatory, UNESCO's Education Data, and the World Bank's World Development Indicators.
There are three main limitations of these data. First, they describe whether countries have adopted laws and policies, but do not capture implementation. The data do, however, make it possible to carry out studies on implementation when policy data are linked with survey data. Second, in most cases, these data do not include regulations or case law. Although court decisions may strengthen legal guarantees in some countries, case law may be more subject to change than guarantees that are enshrined in the text. Finally, while data for some policy areas are longitudinal, others are cross-sectional. Where longitudinal data are available in WORLD's databases or together with the McGill University's PROSPERED Project, causal modelling using quasi-experimental approaches is feasible. Our crosssectional data can be used for monitoring studies, for associational studies, or as foundations for longitudinal databases created by independent researchers.

Accessing the Data
The open-access data generated by WORLD are available online (WORLD Policy Analysis Center 2020). The WORLD's website is designed to be highly accessible and useful for both researchers and the general public, with the goal of making information on policies affecting health, well-being, and equity widely available and actionable.
At WORLD, researchers can download public-use datasets in thematic policy areas by simply filling out a short form that collects their name, e-mail address, institution, and country. With users' contact information, we are able to inform them when databases have been updated and identify whether data is being used by a range of different groups working to advance health, well-being, and equity. Data are available in both Excel and Stata formats, and documentation is included with all data downloads. WORLD also welcomes collaboration on research projects. These requests can be submitted to world@ph.ucla.edu.
Visitors to the website can access a range of data tools and ready-to-share data visualizations for each database. These include global policy maps; graphs illustrating how the prevalence of particular policies varies by region, income group, and treaty ratification status; charts demonstrating how the global prevalence of specific laws and policies has evolved over time; and custom tables comparing the laws and policies of two countries selected by the user. The global policy maps include functionality to readily export static image versions of the maps for sharing, and to imbed interactive, linked versions on other webpages. All of these resources are available through WORLD's main page. In the past five years, more than 350,000 users from over 190 countries have visited the website and users from 123 countries and territories have downloaded public use datasets more than 2,000 times.

Conclusions
Comparative policy analysis supports evidence-based policymaking that improves population health and addresses the social conditions that shape people's daily lives, Comparative Policy Resource across 193 Countries 325 opportunities, and environments. Quantitative, globally comparable law and policy data that capture national action to advance health, well-being, and equity provide a powerful tool for understanding the steps countries are taking as well as the gaps that remain. Combined with outcome data, policy data also enable researchers to build the evidence on which policy approaches are effective across countries and socioeconomic contexts. By capturing well over 2,000 health, social, economic, and environmental policy and program indicators for all 193 UN member states, WORLD aims to advance these goals and provide a public-use data resource of broad utility to researchers, policymakers, and civil society alike.