Scoring

Cross-Validation

A statistical method used to evaluate how well a polygraph scoring algorithm will generalise to independent data sets. Involves training the algorithm on one sample of confirmed cases and testing it on a separate sample. Essential for preventing overfitting and establishing genuine predictive accuracy. The National Academy of Sciences (2003) emphasised that proper cross-validation is critical for all polygraph scoring systems.

What Is Cross-Validation?

Cross-Validation is a polygraph scoring concept. A statistical method used to evaluate how well a polygraph scoring algorithm will generalise to independent data sets. Involves training the algorithm on one sample of confirmed cases and testing it on a separate sample. Essential for preventing overfitting and establishing genuine predictive accuracy. The National Academy of Sciences (2003) emphasised that proper cross-validation is critical for all polygraph scoring systems.

Role of Cross-Validation in Polygraph Scoring

Scoring methodology is one of the most critical factors affecting polygraph accuracy. The evolution from subjective global evaluation to structured numerical scoring and automated algorithms has significantly improved the reliability and reproducibility of polygraph results, as measured by interrater reliability studies.

Modern scoring approaches analyse physiological responses across multiple channels — electrodermal activity (considered the most diagnostically powerful single channel), respiratory patterns, and cardiovascular changes — using validated Kircher features that have demonstrated replicable diagnostic value across decades of research.

Practical Application

APA Standards of Practice require the use of validated scoring methods for all diagnostic and screening polygraph examinations. The choice of scoring system directly affects the cut-scores applied for classification decisions (Deception Indicated">Deception Indicated, No Deception Indicated, or Inconclusive) and the overall balance between false positives and false negatives.

Related Terms

Understanding Cross-Validation in context requires familiarity with several related polygraph concepts:

See also: Adaptive Questioning, Base Rate, Bayesian Analysis, Bootstrap Method, Chart, CPS.

Further Reading

For more information on polygraph terminology and the science of lie detection, explore the full Polygraph Glossary with over 250 defined terms. Visit the Polygraph Examiner Hub for professional resources, the research database for published polygraph studies, or learn how a polygraph works if you are preparing for an examination.

Key Takeaways

Cross-Validation is an important concept within the field of Psychophysiological Detection of Deception. Whether you are a polygraph examiner, student at an APA-accredited training programme, legal professional evaluating polygraph evidence, or a private individual preparing for a lie detector test, understanding this term and its relationship to other polygraph concepts will help you navigate the field with greater confidence and knowledge.

The science of polygraph testing has evolved significantly over more than 100 years, from the early work of William Moulton Marston and Leonarde Keeler to today’s digital instruments and validated scoring algorithms. Every term in this glossary represents a piece of that scientific and professional heritage. For a comprehensive overview, visit the Polygraph Examiner Hub or use our examiner selection guide to find a certified examiner near you.