Transparent by design.

PathLens combines many indicators into a single 0–100 score. The key idea is: compare each country to the global distribution of that indicator for the relevant year, then aggregate with weights.

• Winsorization: per indicator/year, values are clamped to the 2nd–98th percentile to reduce outlier distortion.
• Scale to 0–100: min-max scaling using the winsorized min/max across countries for that year.
• Directionality: indicators where “lower is better” are inverted after scaling.
• Missing values: dimension scores compute from remaining indicators. If coverage is below 60% in a dimension, that dimension’s weight is reduced proportionally and the result is marked lower confidence.
• Confidence badge: reflects the “effective weight sum” after missing-data down-weighting. Higher means more complete coverage.
• Value year & staleness (Strict vs Latest ≤ year): some datasets are lagged or non-annual (notably Global Findex). In Latest ≤ year mode, we use the most recent available datapoint at or before the selected year and surface the underlyingvalue year.

Important: the “Scenario weighting (custom)” panel computes a score client-side from normalized indicators (and supports Latest ≤ year). The headline score on the Country Lens page is a precomputed scores row for the selected scenario/year.

• If an indicator is updated irregularly, strict-year comparisons can look “missing” even when a country has a recent datapoint.
• Use Latest ≤ year for a stable view when combining datasets with different update cadences.
• Always check the value year badge to understand how current the underlying data is.