Evidence Domain

Data Science

What it studies

The extraction of valid insight from data: statistics, measurement theory, analytical methods, machine learning and — most relevantly for workplace evidence — the disciplines of validity, bias, uncertainty and inference that separate finding from artefact.

Why it matters

Workplace strategy now runs on quantitative evidence streams, and every one of them can mislead: samples skew, sensors miscount, averages hide, correlations masquerade as causes. Data science supplies the immune system — the methodological discipline that keeps triangulation honest, flags false precision, and distinguishes a demand signal from a measurement artefact. Without it, more data means more confident error.

Questions it answers
  • Is this finding real, or an artefact of how it was measured?
  • What uncertainty does this number carry, and does the decision respect it?
  • When is team-level analysis valid, and when does small-cohort noise masquerade as signal?
  • What can and cannot be inferred from observational workplace data?
Evidence sources
  • Statistical and measurement methodology
  • The workplace's own data with its documented limitations (every stream has them)
  • Validation studies: sensor accuracy, survey reliability, reported-versus-observed calibration
  • Causal inference literature — the discipline of what observational data cannot say
Design and policy implications

Confidence levels attached to every quantitative input (the Calibration discipline); triangulation as standard practice, single-stream conclusions as flagged exceptions; aggregation thresholds respected analytically, not just legally; AI-supported analysis treated as acceleration of interpretation, never replacement of it.

Related methods
Common misuse

Precision theatre: dashboards reporting utilisation to decimal places from sensors with ±10% error, correlations presented as causes, and 'data-driven' invoked as authority rather than method. The most dangerous workplace analytics are the confident ones.

Further research

Measurement validity in workplace sensing — what each technology actually measures versus what it is sold as measuring — deserves far more independent study than it receives.