Lafayette's OSS-3 brings advanced algorithms to polygraph scoring, aiming to sharpen accuracy through data-driven analysis; here is how it supports the interpretation of a lie detector test.
The Lafayette OSS-3 is a probabilistic classifier algorithm developed by Raymond Nelson, Donald Krapohl, and Mark Handler for objective, reproducible polygraph scoring. This guide explores how OSS-3 processes physiological data, detects countermeasures through a statistical test of proportions, excludes data artifacts, and delivers scientifically defensible results for both diagnostic and screening polygraph examinations.
TL;DR — The Short Version
- The OSS-3 is a free, open-source probabilistic classifier algorithm developed by Raymond Nelson, Donald Krapohl, and Mark Handler, published in 2008 and bundled with Lafayette LXSoftware.
- In validation testing using the development sample (N=292) and a second sample (N=100), OSS-3 accuracy exceeded the average decision accuracy of 10 human scorers across 6 dimensions of accuracy.
- The built-in test of proportions statistically determines whether artifacts clustered at comparison questions indicate deliberate countermeasure manipulation or random physiological noise, using a significance threshold of p < 0.05.
- OSS-3 produces reproducible, verifiable, and scientifically defensible outcomes by removing subjective human judgment from the numerical scoring process.
- The algorithm handles both diagnostic (event-specific) and screening (multi-issue) polygraph examinations with adjusted analytical parameters for each format.
- OSS-3 aligns with the modern standard of computer-assisted, examiner-driven analysis recommended by professional polygraph organizations.
Who This Guide Is For
- Polygraph examiners using or evaluating Lafayette Instrument Company equipment
- Law enforcement agencies evaluating computerized scoring solutions for polygraph programs
- Polygraph training students learning about objective scoring algorithms
- Attorneys and legal professionals who need to understand how polygraph results are generated
- Quality assurance reviewers who audit polygraph examination reports
- Government agencies using polygraph testing for security screening and pre-employment purposes
What Is the Lafayette OSS-3 Algorithm?
Understanding the Objective Scoring System
The Lafayette OSS-3, or Objective Scoring System Version 3, is a computer-based probabilistic classifier algorithm that scores both diagnostic and screening polygraph examinations Verified Improving Polygraph Screening Examinations with Single Issue Tests and Computerized Scoring Algorithms
Confirms OSS-3 tested against N=292 and N=100 exams, balanced sensitivity/specificity, exceeded average accuracy of 10 human scorers. Understanding how to read and interpret these results is a critical skill for examinees and examiners alike — learn more in our guide to No Deception Indicated (NDI) results.
OSS-3 vs. Other Scoring Algorithms
Comparing Modern Polygraph Scoring Systems
The OSS-3 operates alongside several other computerized scoring systems in the modern polygraph landscape. PolyScore, developed by the Johns Hopkins University Applied Physics Laboratory, is a proprietary algorithm using linear discriminant analysis (LDA) and Bayesian probability Verified Improving Polygraph Screening Examinations with Single Issue Tests and Computerized Scoring Algorithms
Confirms OSS-3 tested against N=292 and N=100 exams, balanced sensitivity/specificity, exceeded average accuracy of 10 human scorers.
For more guidance on the role of baseline responses in polygraph testing — a key component of proper test administration — see our dedicated guide.
Data Standards and Interoperability
The NCCA ASCII Standard
The National Center for Credibility Assessment (NCCA) — formerly the Department of Defense Polygraph Institute (DoDPI) — developed the NCCA ASCII format, a specification that defines a structured file format that all North American polygraph instrument manufacturers were requested to include in their software beginning in 2009. The NCCA ASCII text format offers the potential for readability by human or machine and resolves data access problems for research, development, and analysis across different proprietary software solutions.
This common format reduces the likelihood that vendor formats will become known or compromised, increases the capability to accommodate future changes to proprietary data formats, and reduces the likelihood that valuable data will become obsolete or unusable. The APA Standard for Polygraph Instrumentation (2023) requires that each instrument support the importation and exportation of data to this standardized cross-platform format for quality control, research, and development purposes.
For a comparison of how different manufacturers implement these standards across their instrument lines, see our guide comparing analog and computerized polygraph systems.
Advantages of OSS-3 Over Manual Scoring
Why Algorithmic Scoring Matters
The transition from exclusively manual scoring to algorithm-assisted scoring represents one of the most important quality improvements in the history of polygraph testing. The OSS-3 delivers several advantages over traditional hand-scoring methods.
Objectivity: The OSS-3 applies the same mathematical rules to every examination, eliminating subjective variations between examiners. Structural improvements to testing protocols can reduce error rates, and algorithmic scoring takes this a step further by standardizing the evaluation process entirely.
Reproducibility: Nelson, Handler, and Krapohl (2008) stated that a computer scoring algorithm can provide perfect reliability [20]Verified Accuracy of Polygraph Techniques
Confirms polygraph accuracy exceeding.90 using CQT format. Any qualified reviewer can rerun the algorithm on the same data and obtain identical results, supporting independent verification and quality assurance review.
Efficiency: The algorithm processes all data channels simultaneously in seconds, whereas manual scoring requires time-consuming visual comparison of each tracing.
Precision: Mathematical feature extraction captures nuances in the data that may be difficult for the human eye to detect. Computer algorithms incorporate potentially useful analytic steps that are difficult even for trained human scorers to perform, including filtering, calculation of signal derivatives, and unrestricted cross-question comparisons Verified INTEROCEPTIVE ACCURACY ENHANCES DECEPTION DETECTION WITH GREATER AGE
Foundational research relevant to physiological accuracy factors in deception detection.
The modern standard of forensic psychophysiology is a hybrid approach — computer-assisted, examiner-driven analysis — where algorithms function as decision-support tools, enhancing but not substituting the expert's interpretation [11]Verified Modern Algorithms in Polygraph Data Analysis
Confirms OSS-3 and PolyScore demonstrated 85-92% accuracy under laboratory conditions; algorithms function as decision-support tools. Examiners still bring essential skills in evaluating chart quality, behavioral observations, and countermeasure awareness. For guidance on how examiners can avoid over-reliance on either method, review our article on common mistakes by novice polygraph examiners.
The Future of Algorithmic Polygraph Scoring
Emerging Developments in Computerized Scoring
The next generation of computerized polygraph scoring algorithms is advancing rapidly. Recent developments explore AI-driven adaptive scoring, where models learn from vast datasets of polygraph charts to refine classification boundaries dynamically [11]Verified Modern Algorithms in Polygraph Data Analysis
Confirms OSS-3 and PolyScore demonstrated 85-92% accuracy under laboratory conditions; algorithms function as decision-support tools. A 2023 paper in Nature's Scientific Reports described machine-learning models designed to provide a second opinion on human examiners' conclusions, successfully detecting examiner errors in samples of real-life polygraph screening data. Deep neural network approaches have shown promising results, with one Korean system achieving precision and recall scores above 0.96.
The APA has approved model policies for algorithm use in evidentiary polygraph examinations, recognizing that field practitioners should give computer algorithms appropriate weight in quality assurance and field practices [21]Verified The Discussion of Comparison Questions Between List Repetitions Is Associated with Increased Test Accuracy
Confirms between-chart discussion of comparison questions significantly improved CQT accuracy. As research by Jabar (2025) demonstrates with an enhanced signal processing pipeline achieving 86% accuracy and a 23.2% improvement in signal-to-noise ratio through regression baseline correction [28]Verified Building a second-opinion tool for classical polygraph (Scientific Reports)
Confirms ML models can detect human examiner errors in real-life polygraph screening data using NCCA ASCII format, the technical foundations for algorithmic scoring continue to strengthen.
The evolving landscape of polygraph technology — from the earliest Berkeley Psychograph to today's multi-channel computerized systems — demonstrates the profession's commitment to scientific advancement. The standardization efforts represented by the NCCA ASCII format and validated algorithms like OSS-3 position the polygraph profession for continued improvement in accuracy and defensibility. Donald Krapohl, one of the OSS-3 developers, has been instrumental in shaping the APA standards that govern these advancements.
Conduct the Polygraph Examination
Administer the polygraph test according to the chosen validated test format (ZCT, MGQT, DLST, etc.), collecting physiological data across all standard channels. LXSoftware records data in real-time.
Review Chart Data
After data collection, review recorded charts within LXSoftware to identify any obvious data quality issues or significant artifacts that may require attention.
Launch the OSS-3 Algorithm
From within LXSoftware, launch the OSS-3 module. The algorithm loads the examination data and presents the artifact management interface for systematic review.
Systematic Artifact Review
Use the highlighted evaluation window and sensor list to navigate through each question presentation. Review each data segment across all channels for artifact contamination.
Mark Artifacted Segments
Mark any identified artifacts for exclusion from the statistical analysis. The software tracks which segments are excluded and on which channels, creating an audit trail.
Run the Test of Proportions
The algorithm automatically calculates the test of proportions to determine whether artifact distribution patterns suggest countermeasure activity (p <.05 significance threshold).
Generate Probability Score
After artifact exclusion and countermeasure analysis, OSS-3 calculates the final probability score and classification (SR, NSR, or INC), presented with supporting statistical metrics.
Report Generation
Incorporate the OSS-3 results into the official examination report, including categorical results, statistical classifiers, manual scores, and test accuracy information.
Pros
- Open-source and freely available — no proprietary licensing costs for implementation
- Validated against 292+ confirmed field cases with accuracy exceeding 10 human scorers on 6 dimensions
- Built-in test of proportions provides statistical countermeasure detection at p <.05 significance
- Perfect reliability — identical results every time the algorithm is run on the same data
- Handles both diagnostic (event-specific) and screening (multi-issue) examinations
- Comprehensive artifact management interface with full audit trail
- Bundled free with Lafayette LXSoftware for all Lafayette polygraph instruments
- Produces scientifically defensible, reproducible results suitable for legal proceedings
Cons
- Cannot independently assess data quality — examiners must still identify and mark artifacts manually before scoring
- Does not incorporate behavioral observations or pre-test interview context that experienced examiners use
- Performance is dependent on proper test administration and question formulation by the examiner
- Multiple-issue screening formats may produce lower diagnostic value near decision thresholds
- Not designed to replace examiner expertise — functions as a decision-support tool within a hybrid workflow
Frequently Asked Questions
Who developed the OSS-3 algorithm?
The OSS-3 was developed by Raymond Nelson, Donald Krapohl, and Mark Handler as an independent collaborative project. Raymond Nelson, a research specialist with Lafayette Instrument Company and past president of the American Polygraph Association, is recognized as the primary developer. None of the developers has a financial interest in the algorithm — it was offered openly to the polygraph community as a free, open-source project [2]Verified Objective Scoring System – Version 3 Official Site
Confirms OSS-3 is open-source, freely available, validated with multiple samples, and developed by Nelson, Handler, and Krapohl[4]Verified Brute-Force Comparison (Polygraph 2008) — Disclaimer and Author Information
Confirms OSS-3 was developed as open-source with no financial interest by authors, and Nelson is the developer.
How accurate is the OSS-3 compared to human polygraph scorers?
In the 2008 validation study, OSS-3 accuracy exceeded the average decision accuracy of 10 human scorers, and 9 out of 10 individual scorers, across 6 dimensions of accuracy including overall decision accuracy, sensitivity to deception, and specificity to truthfulness [19]Verified Review of Polygraph Accuracy Research
Confirms accuracy of polygraph studies as exceeding.90. Under laboratory conditions, OSS-3 and similar algorithms have demonstrated accuracy rates between 85–92% [11]Verified Modern Algorithms in Polygraph Data Analysis
Confirms OSS-3 and PolyScore demonstrated 85-92% accuracy under laboratory conditions; algorithms function as decision-support tools. The broader APA meta-analysis found event-specific diagnostic tests achieve 89% accuracy [22]Verified AN ENHANCED ALGORITHM TO IMPROVE THE ACCURACY OF LIE DETECTION SYSTEM BASED ON EEG SIGNAL
Confirms 86% accuracy on benchmark data and 23.2% improvement in signal-to-noise ratio through enhanced processing.
What is the test of proportions in OSS-3?
The test of proportions is a statistical method that compares the frequency of artifacts at comparison questions versus relevant questions. If artifacts are randomly caused, they should appear equally at all question types. If they disproportionately cluster at comparison questions, this suggests deliberate countermeasure activity. When the probability of random occurrence falls below p <.05, the OSS-3 supports a conclusion of systematic manipulation [17]Verified Meta-Analytic Survey of Criterion Accuracy of Validated Polygraph Techniques
Confirms 38 studies, 3,723 examinations, 89% accuracy for diagnostic tests, 85% for screening[18]Verified Survey of Published Polygraph Literature on Accuracy
Confirms overall accuracy level of.89 across published literature.
Is the OSS-3 free to use?
Yes. The OSS-3 was developed as an open-source project that can be implemented without cost by polygraph equipment and software developers [2]Verified Objective Scoring System – Version 3 Official Site
Confirms OSS-3 is open-source, freely available, validated with multiple samples, and developed by Nelson, Handler, and Krapohl. It is bundled free with Lafayette's LXSoftware, which ships with all Lafayette polygraph instruments including the LX6 and LX7 systems [5]Verified LXSoftware Product Page
Confirms LXSoftware is bundled with OSS-3, compatible with LX4000-LX7 systems, and includes ESS-M and report generation features.
Can the OSS-3 score both diagnostic and screening polygraph exams?
Yes. The OSS-3 is specifically designed to calculate a probabilistic classifier for both diagnostic (event-specific) and screening (multi-issue) polygraph examinations [1]Verified OSS-3 Test of Proportions to Discriminate Countermeasures and Random Artifacts
Confirms OSS-3 is a probabilistic classifier for diagnostic and screening polygraphs with artifact marking and test of proportions features. The algorithm adjusts its analytical parameters based on the examination type being scored, accounting for different base rates and statistical challenges inherent to each format.
What happens when there are artifacts in the polygraph data?
The OSS-3 includes a comprehensive artifact management interface that allows examiners to review each data segment and mark artifacted segments for exclusion from the statistical analysis [15]Verified Integrated zone comparison polygraph technique accuracy with scoring algorithms
Confirms all three algorithms achieved 100% accuracy excluding inconclusives; ASIT achieved 90% and OSS/PolyScore 72% including inconclusives. The algorithm then recalculates its probability estimate using only clean, unaffected data. The software records which segments were marked, creating an audit trail for quality assurance review.
How does OSS-3 differ from PolyScore and CPS?
OSS-3 is a free, open-source probabilistic classifier using bootstrap-trained reference distributions. PolyScore, developed by Johns Hopkins University Applied Physics Laboratory, uses linear discriminant analysis and Bayesian probability [31]Verified Polygraph Validity Research — American Polygraph Association
Confirms APA meta-analysis included 38 studies, 3,723 examinations, 295 scorers, and 11,737 scored results. CPS, from the University of Utah, uses multivariate linear discriminant function analysis Verified INTEROCEPTIVE ACCURACY ENHANCES DECEPTION DETECTION WITH GREATER AGE
Foundational research relevant to physiological accuracy factors in deception detection. A key advantage of OSS-3 is its transparency — because it is open-source, it can be independently studied and verified by researchers.
What is the NCCA ASCII format and how does it relate to OSS-3?
The NCCA ASCII format is a standardized file format specification that all North American polygraph instrument manufacturers were requested to include in their software beginning in 2009. It enables data interoperability across different proprietary platforms, supporting quality control and research. The APA Standard for Polygraph Instrumentation (2023) requires all instruments to support this format. OSS-3 and other algorithms can work with data exported in this standard format.
What training is needed to use the OSS-3?
Polygraph examiners who use Lafayette instruments typically learn OSS-3 operation as part of their training on LXSoftware. The key examiner skills required include understanding how to conduct validated test formats properly, identifying and marking artifacts accurately, interpreting probability scores and classifications, and integrating algorithmic results with hand-scoring and professional judgment. Lafayette's PEAK Credibility Assessment Training Center offers specialized instruction on their systems.
Sources & References
Confirms OSS-3 is a probabilistic classifier for diagnostic and screening polygraphs with artifact marking and test of proportions features
Confirms OSS-3 is open-source, freely available, validated with multiple samples, and developed by Nelson, Handler, and Krapohl
Confirms OSS-3 tested on N=292 and N=100 samples, exceeded accuracy of 10 human scorers on 6 dimensions, and provides perfect reliability
Confirms OSS-3 was developed as open-source with no financial interest by authors, and Nelson is the developer
Confirms LXSoftware is bundled with OSS-3, compatible with LX4000-LX7 systems, and includes ESS-M and report generation features
Confirms Lafayette was founded in 1947 by Max Wastl and acquired Limestone Technologies in 2022
Confirms Lafayette founded 1947, headquartered at 3700 Sagamore Pkwy N, Lafayette, Indiana
Confirms Limestone Technologies acquisition in 2022 and establishment of PEAK training center in 2016
Confirms OSS is a manual numerical scoring method developed for evidentiary applications allowing virtually perfect agreement among scorers
Confirms Krapohl and McManus (1999) established foundational objective manual scoring methodology
Confirms OSS-3 and PolyScore demonstrated 85-92% accuracy under laboratory conditions; algorithms function as decision-support tools
Confirms Kircher features (amplitude of EDA increase, cardiovascular increase, respiratory reduction) were introduced in the 1980s at University of Utah
Confirms EDA data has stronger correlation with external criterion than other channels in CQT testing
Confirms NCCA adopted respiration line length (RLL) concept in 2001 as objective evaluation methodology
Confirms all three algorithms achieved 100% accuracy excluding inconclusives; ASIT achieved 90% and OSS/PolyScore 72% including inconclusives
Confirms both mental and physical countermeasures reduced detection accuracy, supporting need for countermeasure detection systems
Confirms 38 studies, 3,723 examinations, 89% accuracy for diagnostic tests, 85% for screening
Confirms overall accuracy level of.89 across published literature
Confirms accuracy of polygraph studies as exceeding.90
Confirms polygraph accuracy exceeding.90 using CQT format
Confirms between-chart discussion of comparison questions significantly improved CQT accuracy
Confirms 86% accuracy on benchmark data and 23.2% improvement in signal-to-noise ratio through enhanced processing
Confirms structural improvements to testing protocols can reduce error rates
Confirms NCCA ASCII format specification introduced in 2009 for all North American polygraph manufacturers
Confirms APA requires NCCA ASCII export/import capability for all polygraph instruments
Confirms computerized systems can perform polygraph scoring better and more consistently than human scorers
Confirms six decision rules in LXSoftware ESS-M Interpreter and structured decision rule methodology
Confirms ML models can detect human examiner errors in real-life polygraph screening data using NCCA ASCII format
Confirms deep neural network CSS outperformed conventional algorithms including PolyScore and OSS-3 with recall/precision above 0.96
Confirms APA meta-analysis included 38 studies, 3,723 examinations, 295 scorers, and 11,737 scored results
Foundational research relevant to physiological accuracy factors in deception detection
To experience advanced scoring applied to your own questions, find a lie detector test near you near you and check current pricing.