Handbook of Organization Measurement

Leader Forecasting Accuracy

construct.leader_forecasting_accuracy

A leader's ability to predict their own organization's measures (e.g., engagement scores, attrition rates, operational outcomes) in advance of actuals. Posited as a measurable, expected, LEARNABLE leadership attribute (Tetlock-adjacent: cf. the judgmental-forecasting / superforecasting literature on forecasting as a trainable skill). Two distinct subcomponents, registered as child constructs: (a) point accuracy — closeness of point estimates to actuals — and (b) calibration — interval honesty, i.e., whether stated uncertainty intervals contain actuals at their nominal rate. Proposed PA-original construct (provenance: pa-original); psychometric properties unestablished.

§1How it’s measured

This is a behavioral construct — an event recorded in HR systems, not a self-report. It is measured as an operational metric with a formula, not a questionnaire; a Likert scale here would be a category error.

Leader Forecast Accuracy Scorebehavioralindividual

Point-estimate accuracy of a leader's forecasts of their own organization's measures, scored against recorded actuals. Proposed (pa-original); no normative benchmarks established.

mean absolute (or scale-relative) error of point forecasts vs recorded actuals · score

Leader Forecast Calibration Scorebehavioralindividual

Interval honesty of a leader's forecasts: whether stated 80%/90% intervals contain actuals at their nominal rates, scored with the interval-score (width penalty + miss penalty) / Brier family. Proposed (pa-original); no normative benchmarks established.

interval score (interval width + coverage-miss penalty) + 80/90 interval hit rates vs nominal coverage · score

§3The math on what we know

Synthesized priors from the live registry: pooled r, studies (k), total N, top study grade, and heterogeneity. Rows flagged for heterogeneity or single-study evidence are directional, not settled — read them as a central tendency.