our methods

The Keystone Index is a 50-state comparative index designed to assess livability, competitiveness, resilience, and long-term prosperity across the United States, with a particular focus on Pennsylvania’s performance relative to the other states. The Index provides policymakers, community leaders, researchers, and the public with a transparent, empirically grounded framework for understanding the complex conditions that shape state performance.

The final Keystone Index framework consists of seven domains, 26 subdomains, and 92 indicators, organized hierarchically to allow users to examine state performance from individual measures through broader dimensions of economic, social, environmental, fiscal, and workforce conditions.

The Index draws primarily from federal, nonprofit, and other nationally comparable data systems and emphasizes methodological transparency, reproducibility, and empirical validation. Principal Component Analysis (PCA) is used through the domain level to evaluate statistical structure and derive empirical weights based on relationships among indicators and subdomains. The seven resulting domain scores are then equally weighted to construct the overall Keystone Index, ensuring that each domain contributes one-seventh of the final overall score.

The Keystone Index is designed around seven domains that represent distinct but interconnected dimensions of state performance. Together, the domains provide a multidimensional framework for examining the economic, social, environmental, fiscal, and workforce conditions that contribute to long-term prosperity and quality of life.

Each domain is divided into conceptually related subdomains, which organize individual indicators into meaningful components of state performance.

Business Climate & Future Readiness

Examines the conditions that support economic competitiveness, innovation, workforce capacity, and future economic growth.

• Economic Dynamism

• Talent Infrastructure

• Innovation & Capital

Conservation & Environment

Assesses environmental quality, access to natural resources and outdoor opportunities, hazard exposure, and the economic and recreational importance of natural assets.

• Outdoor Access

• Outdoor Participation

• Environmental Quality

• Outdoor Economy

• Hazard Exposure

Economic Vitality & Jobs

Measures economic performance through indicators of output and growth, employment and wages, and entrepreneurial activity.

• Output & Growth

• Employment & Wages

• Entrepreneurialism

Fiscal Health & Government Capacity

Evaluates the fiscal conditions affecting a state’s long-term obligations, current fiscal condition, and capacity to withstand fiscal stress.

• Long-Term Obligations

• Fiscal Condition

• Fiscal Resilience

Household Well-Being

Examines the economic and social circumstances of households through measures of income and mobility, housing and wealth, health and longevity, and child and family well-being.

• Income & Mobility

• Housing & Wealth

• Health & Longevity

• Child & Family

Livability

Assesses conditions that contribute to everyday quality of life, including material security, social connectivity, access to care infrastructure, and personal safety.

• Material Security

• Social Connectivity

• Care Infrastructure

• Personal Safety

Workforce, Education & Talent

Measures the development, participation, and inclusion of human capital through educational quality and credentials, employment conditions, and access to economic opportunity.

• K–12 Quality

• Credentials

• Employment Conditions

• Opportunity & Inclusion

This structure allows users to move from individual indicators to broader subdomains and domains, helping distinguish the specific factors contributing to a state’s relative strengths and areas warranting further attention.

Indicators included in the Keystone Index were selected to provide conceptually meaningful, nationally comparable measures of state performance. The Index draws primarily from federal statistical agencies and other credible national data sources that provide consistent information across states.

Indicator selection was guided by several criteria:

• Conceptual relevance to the domain and subdomain being measured

• Availability for all or nearly all 50 states

• Use of credible and preferably public data sources

• Comparability across states

• Availability of reasonably current data

• Clear interpretation

• Suitability for repeated updates

Where possible, the Index uses measures that directly represent the concept of interest. In some cases, proxy indicators are used when a nationally consistent direct measure is not available.

Because national data systems are updated on different schedules, indicators may represent different reference years. The Index generally uses the most recent comparable data available at the time of each update. This approach prioritizes comparability across states while allowing the Index to incorporate updated information as national data sources are released.

Candidate indicators were reviewed for data quality, conceptual fit, and statistical performance before inclusion in the final framework. The resulting 92 indicators represent the measures retained following this review process, including Principal Component Analysis (PCA) and reliability testing.

Before index construction, indicators were reviewed and prepared to support meaningful comparison across states and measures. Because the underlying indicators are reported in different units and scales—such as percentages, rates, ratios, dollar amounts, and index values—they were standardized before being combined into subdomain and domain scores.

Indicators were also directionally aligned so that higher index scores consistently represent more favorable performance. For indicators where a higher raw value represents a less favorable condition—such as poverty, crime, fiscal liabilities, or other adverse outcomes—the direction of the measure was reversed for scoring purposes. This ensures that subdomain, domain, and overall Index scores have a consistent interpretation: higher scores represent stronger relative performance.

Index construction is based on continuous standardized measures. Principal Component Analysis (PCA) is applied to these continuous measures rather than to the simplified 1–5 performance categories displayed on the dashboard. This preserves the underlying variation among states during statistical analysis and index construction.

After the continuous scores are calculated and oriented, states are ranked nationally and results are translated into five performance categories based on their relative position among the 50 states:

• 5 — Top 20% of states

• 4 — Above average

• 3 — Middle 20%

• 2 — Below average

• 1 — Bottom 20% of states

These categories are intended to make results easier to interpret and communicate. They represent relative performance among the 50 states rather than an absolute standard of success or failure.

The Keystone Index uses a hierarchical Principal Component Analysis (PCA) framework for index construction and statistical validation. PCA is used to examine whether groups of indicators and subdomains form coherent statistical structures, evaluate dimensionality and factor loadings, and identify measures or structures requiring additional review.

At the subdomain level, PCA is applied to groups of conceptually related continuous indicators. The first principal component (PC1) is used to summarize the common variation among the indicators, with PCA loadings reflecting the strength of each indicator’s relationship to that component. These results provide the basis for constructing continuous subdomain scores.

At the domain level, PCA is applied to the continuous subdomain scores. The resulting first principal component summarizes the shared statistical structure of the subdomains and provides the basis for constructing each continuous domain score.

PCA therefore serves both a diagnostic and an aggregation function within the Keystone Index. It provides empirical evidence about whether indicators and subdomains behave as expected while allowing the statistical relationships among measures to inform their contribution to subdomain and domain scores.

All final production scores are directionally oriented so that higher scores consistently represent more favorable relative performance. Because the mathematical sign of a principal component is arbitrary, the direction of an individual PCA loading should not by itself be interpreted as indicating whether an indicator is favorable or unfavorable.

Armor’s Theta is used alongside PCA to evaluate the internal consistency and reliability of the Index’s component structures. Armor’s Theta is a reliability statistic and should not be interpreted as a measure of state performance.

The following thresholds are used as a general interpretive guide:

• Below 0.60 — Weak reliability

• 0.60–0.70 — Minimal or questionable reliability

• 0.70–0.80 — Acceptable reliability

• 0.80–0.90 — Strong reliability

• 0.90–0.95 — Excellent reliability

• Above 0.95 — May indicate redundancy

PCA and reliability testing are considered together with the conceptual meaning of the indicators. Statistical results inform construction of the Index but do not substitute for substantive judgment about whether measures appropriately represent the concepts being assessed.

The overall Keystone Index is constructed differently from the subdomain and domain scores. Rather than using PCA-derived domain loadings as overall weights, the seven oriented continuous domain scores are assigned equal weight. Each domain therefore contributes one-seventh of the overall Keystone Index.

This final equal-weighting step preserves the empirically derived PCA structure within the domains while ensuring that no individual domain receives greater influence over the overall Index solely because of its statistical relationship with the other domains.

The overall Keystone Index brings together the seven domains into a single comparative measure of state performance. Each domain represents a distinct dimension of the Index and contributes equally to the final overall score.

The overall Index is calculated from the seven directionally aligned continuous domain scores, with each domain assigned an equal weight of one-seventh.

This equal-weighting approach ensures that each of the seven domains has the same influence on the overall Keystone Index. It also provides a transparent and readily interpretable framework in which no single dimension of state performance is given greater importance than another in determining the overall score.

The overall Index should therefore be interpreted as a broad summary of relative state performance across the seven domains. The domain, subdomain, and individual indicator results remain essential for understanding the specific strengths and challenges underlying a state’s overall position.

Keystone Index results are presented through an interactive Tableau dashboard designed to make state comparisons accessible and easy to explore. Users can examine the overall Keystone Index, the seven domains, individual subdomains, and the underlying indicators that contribute to state performance.

The dashboard presents results using national rankings and standardized 1–5 performance categories. Rankings show a state’s position among all 50 states, with 1 representing the highest-ranked state and 50 representing the lowest-ranked state.

The 1–5 performance categories summarize each state’s relative position using quintiles:

• 5 — Top 20% of states

• 4 — Above average

• 3 — Middle 20%

• 2 — Below average

• 1 — Bottom 20% of states

These categories represent relative performance rather than absolute standards. A state receiving a 5 is performing relatively well compared with other states on that measure; it does not necessarily mean that the state has achieved an ideal level of performance. Similarly, a category of 1 indicates placement in the bottom 20% of states rather than failure to meet a predetermined standard.

The dashboard allows users to move beyond headline rankings to examine the domains, subdomains, individual indicators, and underlying data contributing to a state’s performance. Rankings and performance categories are therefore best used as a starting point for further investigation rather than as standalone measures of success or failure.

The Keystone Index is designed as a comparative benchmarking and diagnostic tool rather than a definitive measure of every aspect of state performance. As with any composite index, its results should be interpreted in the context of several methodological limitations.

Indicators are drawn from multiple national data systems that are released on different schedules. As a result, the Index incorporates the most recent comparable data available, but individual indicators may represent different reference years.

Some concepts cannot be measured directly using nationally consistent state-level data. In these cases, the Index uses proxy measures selected for their conceptual relevance, comparability, and data quality. The availability and quality of national data therefore influence how some concepts are represented within the Index.

State-level measures can also mask substantial variation within states. Economic, social, environmental, fiscal, and workforce conditions may differ considerably among regions, counties, communities, and population groups. State rankings should not be interpreted as describing the experience of every place or population within a state.

Rankings and 1–5 performance categories simplify continuous differences among states. States with very similar underlying scores may occupy different ranking positions or fall on opposite sides of a quintile boundary. Rankings and categories should therefore be interpreted as indicators of relative position rather than precise measures of the magnitude of differences between states.

Finally, statistical relationships among indicators do not establish causal relationships. Strong or weak performance in a domain, subdomain, or indicator should be interpreted as a benchmarking signal that can help identify areas for further investigation—not as evidence that a particular factor, condition, or policy caused the observed result.

The Keystone Index is most useful when viewed as a starting point for understanding state performance. Users are encouraged to examine the domains, subdomains, individual indicators, and underlying data together rather than relying on any single score or ranking.

Methodological transparency is a core design principle of KIP. The project emphasizes publicly available data sources, reproducible workflows, empirically derived weights, explicit reliability testing, and transparent documentation of limitations.

No arbitrary or subjective weighting schemes are applied within the index construction process.

Future iterations of KIP will incorporate PA region and PA county-level expansion, PA local level data, and expanded dashboard analytics.

The project framework is intentionally modular and designed to support future expansion and refinement.