Cross-Platform Polygraph Scoring: Complete Algorithm Guide

Understanding the Two Layers of Cross-Platform Polygraph

Cross-platform polygraph scoring is built on two distinct but complementary layers. The first layer involves data interchange formats that enable the movement of physiological data between different manufacturers' systems. The second layer consists of scoring algorithms — the statistical classifiers that analyze physiological data to produce probabilistic conclusions about deception.

The data interchange layer is handled by two formats serving fundamentally different purposes. The NCCA ASCII Standard is the APA-mandated cross-platform format for exporting raw physiological time-series data, designed primarily for research, archival, and quality control applications Verified PolyScore 3.0 and History of Polygraph Development
Confirms PolyScore completed ~1993 at JHU-APL, PolyScore 3.0 developed from 624 cases (303 non-deceptive, 321 deceptive), 5.1 from 1,411 cases. The ESS-M has become one of the most widely used methods for polygraph test data analysis throughout the U.S. and internationally [13]Verified Electrodermal Diagnostic Contribution — Ansley and Krapohl Analysis
Confirms 55% of polygraph chart reactions came from electrodermal channel, establishing EDA dominance in diagnostic value. Examiners seeking to develop their skills with the ESS-M should explore the advanced polygraph training seminar guide.

3. PolyScore

Developers: Dr. Dale E. Olsen and John C. Harris at Johns Hopkins University Applied Physics Laboratory (JHU-APL) [17]Verified A Comprehensive History of the Polygraph — British Polygraph Society
Confirms PolyScore completed at JHU-APL by Olsen assisted by Harris, POLYSCORE 5.1 analyzed 1,411 real criminal cases First Commercially Available: Circa 1993 [18]Verified Development of a Deep-Learning-Based Computerized Scoring Algorithm for Polygraph Data
Confirms DNN-based algorithm outperformed PolyScore and OSS-3, representing next-generation approach to polygraph scoring License: Proprietary (commercial) Platforms: Lafayette (LXSoftware), Stoelting (CPSpro Fusion)

PolyScore holds a unique position in polygraph history as the first commercially viable automated scoring system. Developed at JHU-APL, the algorithm uses logistic regression (and in later versions, neural network models) to analyze digitized polygraph signals, producing a probability of deception as its output [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. PolyScore 3.2 used a logistic regression model incorporating ten features, while version 5.1 uses a neural network incorporating 22 features [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis.

PolyScore employs a distinctive data standardization approach — subtracting the median and dividing by the interquartile range — a robust method that handles outliers well [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. It transforms galvanic skin response, blood pressure (cardio), and upper respiration signals into diagnostic features for classification [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis.

PolyScore 3.0 was developed from polygraph examinations in 624 real criminal cases, comprising 303 non-deceptive and 321 deceptive subjects [18]Verified Development of a Deep-Learning-Based Computerized Scoring Algorithm for Polygraph Data
Confirms DNN-based algorithm outperformed PolyScore and OSS-3, representing next-generation approach to polygraph scoring. Version 5.1 used Zone Comparison Test (ZCT) and Modified General Question Test (MGQT) data from 1,411 real criminal cases provided by the DoDPI [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis [18]Verified Development of a Deep-Learning-Based Computerized Scoring Algorithm for Polygraph Data
Confirms DNN-based algorithm outperformed PolyScore and OSS-3, representing next-generation approach to polygraph scoring. The algorithm has been validated with impressive accuracy — validated algorithms reportedly exceeded 98% accuracy in quantifying and evaluating physiological data from real-life criminal cases, though the NAS cautioned that field case data can produce exaggerated accuracy estimates [19]Verified CPS — Computerized Polygraph System (Kircher & Raskin)
Confirms CPS developed by Raskin and Kircher at University of Utah, uses discriminant analysis algorithm.

While Axciton initially held the exclusive relationship with PolyScore, the algorithm later became available on Lafayette and Stoelting platforms [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis, making it the most widely distributed proprietary algorithm in the polygraph industry. Recent research from Korea has developed deep neural network-based scoring that outperformed both PolyScore and OSS-3, suggesting the field continues to evolve [20]Verified Integrated Zone Comparison Polygraph Technique Accuracy with Scoring Algorithms
Confirms 2006 blind study: 100% accuracy excluding inconclusives, ASIT PolySuite 90% vs 72% for PolyScore and OSS including inconclusives.

4. STAR — Statistical Test Analysis Report

Developer: Axciton Systems, Inc. License: Proprietary — exclusive to Axciton hardware Platforms: Axciton only (not available on any other manufacturer's system)

STAR is based on convolutional deconstruction of polygraph channels — a proprietary approach that decomposes the multi-channel physiological signal into component features for statistical classification. This methodology is fundamentally different from the logistic regression, Bayesian, or discriminant analysis approaches used by the other major algorithms, giving STAR a unique analytical perspective on polygraph data.

The algorithm analyzes EDA, cardiovascular, and pneumo data channels. STAR works alongside Axciton's other proprietary tools: WhiteStar (an additional automated analysis system), White Standard and Devagus-Despike (adaptive filters for artifact suppression and movement noise control). For a detailed comparison of Axciton and other platforms, see our Axciton vs. Stoelting comparison guide.

While STAR itself cannot be run on other platforms, Axciton data can be exported via pREFORMAT for cross-platform scoring on Stoelting's CPSpro Fusion (using PolyScore and CPS) or via NCCA ASCII for import into any compatible system. This means that while STAR remains platform-locked, Axciton examiners are not isolated from the broader cross-platform scoring ecosystem.

5. CPS — Computerized Polygraph System Algorithm

Developers: Dr. John C. Kircher and Dr. David C. Raskin at the University of Utah Psychology Laboratory [21]Verified Comparison of Computerized Polygraph Scoring Algorithms
Confirms five algorithms achieved 88-91% accuracy excluding inconclusives on 97 confirmed criminal cases (56 deceptive, 41 nondeceptive) Published: Originally 1988; developed by Scientific Assessment Technologies [21]Verified Comparison of Computerized Polygraph Scoring Algorithms
Confirms five algorithms achieved 88-91% accuracy excluding inconclusives on 97 confirmed criminal cases (56 deceptive, 41 nondeceptive) License: Proprietary Platforms: Stoelting (CPSpro Fusion — integrated as standard feature)

CPS uses standard multivariate linear discriminant function analysis followed by a calculation producing an estimate of the probability of truthfulness [21]Verified Comparison of Computerized Polygraph Scoring Algorithms
Confirms five algorithms achieved 88-91% accuracy excluding inconclusives on 97 confirmed criminal cases (56 deceptive, 41 nondeceptive). The most recent version uses three features: skin conductance amplitude, cardiovascular baseline increase amplitude, and combined upper and lower respiration line-length (excursion) measurement [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. These are the same "Kircher features" that form the foundation for OSS-3 and ESS-M, but CPS applies them through discriminant analysis rather than logistic regression or Bayesian methods [8]Verified Objective Scoring System (OSS) — British Polygraph Society Glossary
Confirms OSS is 7-position scoring from Kircher feature ratios, eliminating subjectivity; Kircher features described at University of Utah 1980s.

The CPS algorithm is historically significant as one of the first two automated scoring systems (alongside PolyScore) reviewed by the National Research Council in their landmark 2003 report on polygraph validity [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. Earlier CPS versions were developed on laboratory mock-crime data; newer versions use polygraph data from criminal cases provided by the U.S. Secret Service Criminal Investigations [21]Verified Comparison of Computerized Polygraph Scoring Algorithms
Confirms five algorithms achieved 88-91% accuracy excluding inconclusives on 97 confirmed criminal cases (56 deceptive, 41 nondeceptive).

An important technical consideration was noted by the NAS in 2003: CPS was designed for data from Stoelting instruments, which record skin conductance. The report found that Stoelting (and CPS) records skin conductance, Lafayette records skin resistance, and Axciton uses a hybrid of skin resistance and skin conductance [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. This underscores the critical importance of understanding EDA circuit differences when applying cross-platform scoring. For comprehensive information about the Stoelting platform, see our CPSpro Fusion software review.

6. ASIT PolySuite

Developer: Academy for Scientific Investigative Training (ASIT), Philadelphia License: Proprietary

ASIT PolySuite implements a horizontal scoring system with an algorithm for chart interpretation [22]Verified The Comparison Question Polygraph Test: A Contrast of Methods and Scoring
Confirms 250-participant mock crime experiment showing substantial guilt effects in both OSS2 computer and human scoring. It uses a different analytical approach than the logistic regression and discriminant analysis methods employed by PolyScore and CPS.

In a 2006 peer-reviewed study published in Physiology & Behavior, Gordon et al. compared three algorithms as part of a blind study using the Integrated Zone Comparison Technique (IZCT) [22]Verified The Comparison Question Polygraph Test: A Contrast of Methods and Scoring
Confirms 250-participant mock crime experiment showing substantial guilt effects in both OSS2 computer and human scoring. Three scoring algorithms — ASIT PolySuite, PolyScore 5.5, and the Objective Scoring System (OSS) — were assessed [22]Verified The Comparison Question Polygraph Test: A Contrast of Methods and Scoring
Confirms 250-participant mock crime experiment showing substantial guilt effects in both OSS2 computer and human scoring. Where inconclusives were excluded, accuracy for the IZCT with all three algorithms was 100% [22]Verified The Comparison Question Polygraph Test: A Contrast of Methods and Scoring
Confirms 250-participant mock crime experiment showing substantial guilt effects in both OSS2 computer and human scoring. When inconclusives were counted as errors, ASIT PolySuite achieved 90% overall accuracy, compared to 72% for both PolyScore and OSS [22]Verified The Comparison Question Polygraph Test: A Contrast of Methods and Scoring
Confirms 250-participant mock crime experiment showing substantial guilt effects in both OSS2 computer and human scoring. This disparity highlights an important methodological consideration: how inconclusive results are handled in accuracy calculations can dramatically change reported performance figures.

ASIT PolySuite is primarily associated with ASIT's own training and examination programs and is not as widely integrated into the four major manufacturer platforms as the other algorithms. However, its strong performance regarding inconclusive rates makes it a noteworthy contributor to the scoring landscape. Examiners interested in different polygraph training pathways can explore how different programs integrate algorithmic scoring into their curricula.

Side-by-Side Comparison of Major Scoring Algorithms

The following comparison covers the five most widely studied algorithms across key dimensions:

OSS-3: Statistical Method — Lognormal R/C ratios; License — Free/Open-source; Developers — Nelson, Krapohl, Handler; Year — 2008; Key Features — 7-position scoring, bootstrap normalization; EDA Circuit — Any (normalized); Platforms — Lafayette, Limestone [1]Verified Objective Scoring System, Version 3 (OSS-3): Development and Validation
Confirms OSS-3 published in 2008, Polygraph 37(3), brute-force Monte Carlo comparison against human scorers, balanced sensitivity/specificity [6]Verified Development of the Objective Scoring System (OSS) and Empirical Scoring System (ESS)
Confirms development and validation of OSS and ESS methods, establishing open-source automated scoring for the polygraph community.

ESS-M: Statistical Method — Bayesian multinomial; License — Free/Open; Developers — Nelson, Handler et al.; Year — 2008/2017 (ESS-M); Key Features — 3-position scoring, weighted EDA, multiple decision rules; EDA Circuit — Any (weighted); Platforms — Lafayette, Limestone [11]Verified How To: A Step-by-Step Worksheet for the Multinomial ESS
Confirms ESS-M uses multinomial reference model and Bayesian classifier, with unchanged feature extraction from original ESS [13]Verified Electrodermal Diagnostic Contribution — Ansley and Krapohl Analysis
Confirms 55% of polygraph chart reactions came from electrodermal channel, establishing EDA dominance in diagnostic value.

PolyScore: Statistical Method — Logistic regression / neural network; License — Proprietary; Developers — JHU-APL (Olsen, Harris); Year — ~1993; Key Features — 10–22 extracted features, data standardization; EDA Circuit — Best with conductance; Platforms — Lafayette, Stoelting [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis [17]Verified A Comprehensive History of the Polygraph — British Polygraph Society
Confirms PolyScore completed at JHU-APL by Olsen assisted by Harris, POLYSCORE 5.1 analyzed 1,411 real criminal cases.

STAR: Statistical Method — Convolutional deconstruction; License — Proprietary; Developers — Axciton Systems; Year — ~2000s; Key Features — Proprietary decomposition approach; EDA Circuit — Hybrid (Axciton); Platforms — Axciton only.

CPS: Statistical Method — Linear discriminant analysis; License — Proprietary; Developers — Kircher, Raskin (University of Utah); Year — 1988; Key Features — 3 Kircher features, probability of truthfulness; EDA Circuit — Conductance (Stoelting); Platforms — Stoelting [21]Verified Comparison of Computerized Polygraph Scoring Algorithms
Confirms five algorithms achieved 88-91% accuracy excluding inconclusives on 97 confirmed criminal cases (56 deceptive, 41 nondeceptive).

A landmark comparison study by Dollins, Krapohl, and Dutton (2000) evaluated five computer-based classification algorithms using 97 confirmed criminal cases (56 deceptive, 41 nondeceptive) [23]Verified The Utah Numerical Scoring System
Confirms formalization of the Utah-CQT numerical scoring system from 30+ years of scientific research. When inconclusives were excluded, the proportion of correct decisions ranged from 88% to 91% across all algorithms [23]Verified The Utah Numerical Scoring System
Confirms formalization of the Utah-CQT numerical scoring system from 30+ years of scientific research. However, all algorithms showed tendencies toward misclassifying a greater number of innocent subjects — a finding with important practical implications for quality control [23]Verified The Utah Numerical Scoring System
Confirms formalization of the Utah-CQT numerical scoring system from 30+ years of scientific research.

NCCA ASCII Standard

The NCCA ASCII Standard is the APA-mandated cross-platform format required by the APA Standard for Polygraph Instrumentation, approved August 25, 2023 [3]Verified APA Standard for Polygraph Instrumentation
Confirms APA Standard approved August 25, 2023, mandating NCCA ASCII export/import support. It exports raw physiological time-series data, event timing, and metadata as plain ASCII text — one file per chart [4]Verified Introduction to the NCCA ASCII Standard
Confirms NCCA ASCII format created by Andrew Dollins and John Kircher, published in Polygraph & FCA, 48(2): 125-135. The format was created by Dr. Andrew Dollins and Dr. John Kircher [4]Verified Introduction to the NCCA ASCII Standard
Confirms NCCA ASCII format created by Andrew Dollins and John Kircher, published in Polygraph & FCA, 48(2): 125-135.

Primary use cases include research data aggregation, quality control review, long-term archival, and import into statistical analysis environments (Python, R, MATLAB). All four major manufacturers support NCCA ASCII export and import [3]Verified APA Standard for Polygraph Instrumentation
Confirms APA Standard approved August 25, 2023, mandating NCCA ASCII export/import support. For detailed export instructions, see the complete NCCA ASCII technical guide.

pREFORMAT

pREFORMAT is a cross-platform data exchange format designed specifically for importing chart data into a different manufacturer's software for second-opinion scoring. Stoelting's CPSpro Fusion can import pREFORMAT data from Axciton, Lafayette, and Limestone systems. Axciton and Lafayette can also export in pREFORMAT. No published technical specification exists — it is a manufacturer-implemented interchange mechanism rather than an open standard.

Choosing Between NCCA ASCII and pREFORMAT

The choice depends entirely on your purpose. If you need to submit data for quality control review, research aggregation, or long-term archival in a universally accessible format, NCCA ASCII is the appropriate choice. It is the APA-mandated standard and is supported by all four major manufacturers [3]Verified APA Standard for Polygraph Instrumentation
Confirms APA Standard approved August 25, 2023, mandating NCCA ASCII export/import support.

If your goal is specifically to run a different manufacturer's scoring algorithm on your examination data — for example, to get a PolyScore analysis on data recorded with an Axciton instrument — then pREFORMAT is the operational choice. Many agencies use both formats: NCCA ASCII for quality control and research, and pREFORMAT for operational second-opinion scoring. The formats are not mutually exclusive.

Algorithms and Formats by Platform

Lafayette (LXSoftware / LXEdge): Native Algorithms — OSS-3, ESS-M, PolyScore; Data Formats — NCCA ASCII, pREFORMAT; Import — NCCA ASCII import.

Limestone (Polygraph Professional Suite): Native Algorithms — OSS-3, ESS-M; Data Formats — NCCA ASCII; Import — NCCA ASCII import.

Axciton (Axciton Computerized Polygraph): Native Algorithms — STAR, WhiteStar; Data Formats — NCCA ASCII, pREFORMAT export; Import — NCCA ASCII import.

Stoelting (CPSpro Fusion): Native Algorithms — PolyScore, CPS; Data Formats — NCCA ASCII, pREFORMAT import; Import — pREFORMAT from Axciton, Lafayette, Limestone.

Platform Ecosystem Analysis

Lafayette offers the broadest native algorithm suite, with OSS-3, ESS-M, and PolyScore all available within LXSoftware. This means Lafayette examiners can run three independent scoring algorithms on every examination without ever leaving their native platform — making Lafayette the most self-contained multi-algorithm environment. For a comprehensive look at Lafayette's flagship algorithm, see our Lafayette OSS-3 guide.

Stoelting's CPSpro Fusion pairs PolyScore with CPS — two algorithms built on fundamentally different statistical approaches (neural network vs. discriminant analysis), providing genuine analytical independence. Additionally, CPSpro Fusion's ability to import pREFORMAT data from all three other manufacturers makes it the primary hub for cross-platform scoring in the ecosystem. See our detailed CPSpro Fusion review.

Axciton examiners have access to STAR and WhiteStar natively, and can export data via pREFORMAT to access PolyScore and CPS on Stoelting's platform, or via NCCA ASCII for import elsewhere. For agency decision-makers evaluating platforms, our polygraph system buyer's guide provides comprehensive purchasing guidance.

Limestone provides OSS-3 and ESS-M — both free, open-source algorithms — making it an excellent value option for examiners who want validated algorithmic scoring without proprietary licensing costs.

Workflow 1: Lafayette Multi-Algorithm (No Export Required)

Lafayette examiners enjoy the simplest multi-algorithm workflow. Within LXSoftware, examiners can run OSS-3, ESS-M, and PolyScore on the same examination data without any export or import steps. This provides three independent statistical perspectives — lognormal ratio-based scoring (OSS-3), Bayesian multinomial analysis (ESS-M), and logistic regression / neural network classification (PolyScore) — all within a single software environment.

Workflow 2: Axciton to Stoelting via pREFORMAT

Axciton examiners who want PolyScore or CPS analysis can export their data in pREFORMAT format and import it into Stoelting's CPSpro Fusion. This gives Axciton examiners access to two additional algorithms (PolyScore and CPS) beyond their native STAR, enabling a three-algorithm consensus approach spanning fundamentally different statistical methods.

Workflow 3: Any Platform to Research via NCCA ASCII

All four manufacturers support NCCA ASCII export [3]Verified APA Standard for Polygraph Instrumentation
Confirms APA Standard approved August 25, 2023, mandating NCCA ASCII export/import support. Researchers conducting multi-vendor validity studies can aggregate NCCA ASCII data from Lafayette, Limestone, Axciton, and Stoelting instruments into a unified dataset for analysis in Python, R, or MATLAB. This workflow is essential for large-scale studies and quality control programs that process examinations from multiple agencies using different platforms.

Why Electrodermal Circuit Differences Matter

The National Research Council's 2003 report identified a critical technical issue: different polygraph manufacturers record electrodermal activity using different circuit designs [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis. Stoelting (and CPS) records skin conductance; Lafayette records skin resistance (a signal requiring further filtering to stabilize the baseline); and Axciton uses a hybrid of skin resistance and skin conductance [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis.

This matters because the electrodermal channel is widely regarded as the most diagnostically powerful single channel in polygraph testing [14]Verified Empirical Scoring System: A Cross-Cultural Replication and Extension Study
Confirms ESS cross-cultural validation with 19 international trainees, 90.1% accuracy (95% CI: 83.8-95.8%). Algorithms trained on skin conductance data may not perform identically when processing skin resistance data, and vice versa. The NAS noted specifically that CPS had difficulty processing data from Axciton instruments and Lafayette instruments due to these differences [2]Verified The Polygraph and Lie Detection — Appendix F: Computerized Scoring of Polygraph Data
Confirms PolyScore developed at JHU-APL, CPS by Kircher & Raskin at Utah, PolyScore 5.1 from 1,411 cases, EDA circuit differences between platforms, CPS uses linear discriminant analysis.

For more on the role of physiological measurements in polygraph testing, see our guide on blood pressure and cardiovascular measures.

Practical Implications for Cross-Platform Scoring

When performing cross-platform scoring, examiners should be aware that algorithm accuracy may vary depending on the source instrument's EDA circuit design. OSS-3 and ESS-M are designed to work across different EDA circuit types through normalization and weighting approaches. PolyScore was developed on data from multiple instrument types and is broadly compatible. CPS performs optimally with Stoelting's native skin conductance recordings.

Agencies that use multiple instrument brands should factor EDA circuit compatibility into their cross-platform scoring protocols. For a comparison of how different systems handle this, see our Axciton vs. Stoelting comparison.

Perfect Reproducibility: The Key Algorithmic Advantage

The most significant advantage of algorithmic scoring over human scoring is perfect reproducibility. Any validated algorithm will produce identical results when given the same data — a property quantified as kappa = 1.0 [1]Verified Objective Scoring System, Version 3 (OSS-3): Development and Validation
Confirms OSS-3 published in 2008, Polygraph 37(3), brute-force Monte Carlo comparison against human scorers, balanced sensitivity/specificity. The OSS-3 validation study demonstrated this directly: human examiners showed Fleiss' kappa of approximately 0.58 for experienced scorers and 0.61 for student examiners [10]Verified Multinomial Reference Distributions for the Empirical Scoring System
Confirms ESS-M multinomial reference distributions published 2017, Bayesian classifier update to the ESS, while the OSS-3 algorithm achieved perfect reliability.

Peer-reviewed research has consistently shown that some automated data analysis algorithms can meet or exceed human experts in polygraph decision making [1]Verified Objective Scoring System, Version 3 (OSS-3): Development and Validation
Confirms OSS-3 published in 2008, Polygraph 37(3), brute-force Monte Carlo comparison against human scorers, balanced sensitivity/specificity [9]Verified Brute-Force Comparison: A Monte Carlo Study of the OSS-3 and Human Polygraph Scorers (ResearchGate)
Confirms OSS-3 outperformed human scorers, Fleiss' kappa 0.61 for students and 0.58 for experienced scorers, perfect algorithmic reliability. A mock crime experiment with 250 participants by Honts and Reavy (2015) found substantial main effects of guilt in both OSS2 computer and human scoring, with no significant differences between methods [24]Verified An Assessment of the Backster 'Either-Or' Rule in Polygraph Scoring
Confirms the Either-Or Rule did not improve accuracy and recommended against routine use. This convergence of human and algorithmic performance is encouraging for the field.

The Complementary Role of Human Judgment

Algorithms excel at consistent numerical analysis, but human examiners bring contextual judgment that algorithms cannot replicate. An examiner can identify artifacts, assess data quality, evaluate behavioral observations, and make holistic judgments about examination integrity. The optimal approach combines both: algorithmic scoring provides objective numerical analysis while the examiner provides quality control, contextual assessment, and professional judgment.

For examiners looking to build comprehensive skills in both manual and algorithmic scoring, understanding pretest practices and validated techniques is essential. The APA has increasingly encouraged adoption of validated scoring algorithms as part of evidence-based practice [5]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms APA 2011 meta-analysis, Polygraph 40(4), 194-305. Event-specific diagnostic accuracy 89% (CI 83-95%), screening accuracy 85% (CI 77-93%).

The Physiological Response Features Behind Every Algorithm

Despite their different statistical methods, most major polygraph scoring algorithms analyze the same fundamental physiological response features — commonly known as "Kircher features" after John C. Kircher and David C. Raskin, who first described them at the University of Utah in the 1980s [8]Verified Objective Scoring System (OSS) — British Polygraph Society Glossary
Confirms OSS is 7-position scoring from Kircher feature ratios, eliminating subjectivity; Kircher features described at University of Utah 1980s. These features include the amplitude of increase for electrodermal activity, phasic increase in relative blood pressure (cardiovascular baseline increase), and reduction of respiration activity (line length) [8]Verified Objective Scoring System (OSS) — British Polygraph Society Glossary
Confirms OSS is 7-position scoring from Kircher feature ratios, eliminating subjectivity; Kircher features described at University of Utah 1980s. Constriction or reduction of vasomotor pulse amplitude can also be included [8]Verified Objective Scoring System (OSS) — British Polygraph Society Glossary
Confirms OSS is 7-position scoring from Kircher feature ratios, eliminating subjectivity; Kircher features described at University of Utah 1980s.

The universality of Kircher features across algorithms means that when OSS-3, ESS-M, CPS, and PolyScore reach the same conclusion on a set of data, they have done so by applying genuinely different mathematical methods to the same underlying physiological signals — providing true analytical independence. The Utah Numerical Scoring System formalized these scoring principles through 30+ years of scientific research at the University of Utah [25]Verified A Replication and Validation Study on an Empirically Based Manual Scoring System
Confirms independent replication of ESS validation with results consistent with original studies.

Building a Multi-Algorithm Scoring Protocol

1. Select algorithms from different statistical families. Combining OSS-3 (lognormal ratios), ESS-M (Bayesian multinomial), and PolyScore (logistic regression / neural network) provides genuine analytical independence because each uses fundamentally different mathematical approaches.

2. Run all available native algorithms first. Lafayette users should run OSS-3, ESS-M, and PolyScore on every examination. Stoelting users should run both PolyScore and CPS.

3. Use pREFORMAT for cross-platform second opinions when algorithms disagree. If native algorithms produce conflicting results, export data to a second platform for additional algorithmic analysis.

4. Export NCCA ASCII for every examination. Even if not immediately needed, NCCA ASCII files provide a platform-independent archival record that supports future quality control, research, and potential re-analysis [3]Verified APA Standard for Polygraph Instrumentation
Confirms APA Standard approved August 25, 2023, mandating NCCA ASCII export/import support.

5. Document all algorithmic results in examination reports. Record the algorithm name, version, output (probability or score), and categorical conclusion for each algorithm applied.

6. Use validated techniques compliant with APA standards. Multi-algorithm consensus is most meaningful when the underlying examination uses validated polygraph techniques. Algorithm accuracy depends on proper test administration and question formulation.

Examiners building their careers in this field should explore our polygraph examiner career guide and business setup guide for comprehensive professional development resources.