Introduction to the
Objective Scoring System – version 3

by Raymond Nelson, Mark Handler, and Donald Krapohl
** NEW ** Lafayette Instrument's OSS-3 Report

Home
Poster #1 - Development and validation
Poster #2 - Validation with mixed format field investigation polygraphs
Poster #3 - Additional validation with LEPET and PCSOT screening exams
The OSS-3 Report
Other Materials

The Objective Scoring System, version 3 (OSS-3), was developed as an independent collaborative project, with the goal of providing a  robust, widely applicable, well documented, and mathematically sound computerized polygraph scoring model that can be implemented without cost by polygraph equipment and software developers, and is freely available for further study. The OSS-3 computerized scoring algorithm is based on sound polygraph testing principles derived from existing research, and has demonstrable validity with multiple validation samples, including: the original training sample of confirmed investigative (single issue) polygraphs, a validation sample of similar examinations, and a crossvalidation sample confirmed field investigations conducted with multiple variants of the Zone and MGQT formats. Further crossvalidation efforts examined the validity of the OSS-3 algorithm with three samples of screening examinations including LEPET examinations, PCSOT maintenance and PCSOT disclosure examinations.

Previous versions of the Objective Scoring System (OSS) were initially intended to provide a reliable hand-scoring system. Because of its demonstrated validity, available documentation, and understandable structure, several polygraph equipment manufacturers have provided computerized versions of the previous OSS versions in their software application suites. Presently, only one polygraph equipment manufacturer has limited computer scoring algorithm offerings to proprietary or  unpublished methods. In the sister science of psychology, computerized versions of test measures commonly calculate test results in exactly the same manner that professional practitioners would use to score a test by hand. The polygraph profession has been somewhat unique in its tolerance for computerized scoring algorithms that are sometimes incompletely documented proprietary methods which  include decision models that are not always fully understood by the professionals who use them. In a comparison of computer scoring algorithms, Dollins, Krapohl and Dutton (1999) found that that only the CPS-II proprietary scoring algorithm is supported by a published decision model (Kircher and Raskin, 1988; Kircher and Raskin, 2002), and expressed unanswered concerns about the possibility for biased results among some existing scoring algorithms, and the lack of information about how the existing decision models found some cases to be unclassifiable.

Although previous OSS versions provided a robust and well documented scoring system for there intended usage with single issue examinations composed of three relevant questions and three test charts, the underlying cumulative data structure and normative data for those versions could not be theoretically generalized to the variety of polygraph techniques employed in agency and field settings. Nor could previous OSS versions be theoretically or practically applied to multiple-facet investigations, or mixed-issues screening tests, which make of a substantial portion of all polygraph examinations. Early efforts by Krapohl and Norris (2000) to investigate the effectiveness of the original OSS with MGQT cases were unimpressive.

While OSS-3 was developed from the same foundation of empirical principles as previous OSS versions, including the original Objective Scoring System described by Krapohl and McManus (1999) and Dutton (2000), and the updated Objective Scoring System, version 2 (OSS-2) provided by Krapohl (2002), this new version represents a substantial improvement of the mathematical transformation and aggregation model. OSS-3 performs as well or better than previous OSS versions, and overcomes many of the theoretical and practical limitations of previous versions. While it is possible to complete all OSS-3 calculations and final decisions by hand, or with the aid of a scientific hand calculator, it is anticipated that the mathematical complexity of the present revision will mean that the Objective Scoring System, including this new version, has become a defacto computerized polygraph scoring algorithm.

The OSS-3 algorithm was normed using two bootstrap resamples of 10,000 sets of N=292 confirmed single-issue field examinations from the training sample. Validity was further evaluated using a second smaller sample of confirmed single issue field investigation cases. Further crossvalidation experiments revealed roughly equivalent performance, with no significant differences in classification accuracy, using a sample of mixed format field investigation polygraphs, which included multiple variants of the Modified General Questions Technique and variants of the Zone Comparison Technique.

Unlike the cumulative data model of previous OSS versions, OSS-3 is based on weighted and mathematical averaging, and can be theoretically applied to all comparison question polygraph  techniques consisting of two to four relevant questions. OSS-3 can leverage the advantages of three to five test charts, as described by Senter, Dollins, and Krapohl (2004), and Senter and Dollins (2004). OSS-3 retains as its foundation the three primary physiological measurements described by Kircher and Raskin (1988; 2002), Timm (1982), and Dutton (2000). Also retained from previous versions are the use of R/C ratios, described by Krapohl (1999), to achieve a dimensionless mathematical representiation of differential reactivity. The OSS-3 method does not provide integer point totals analogous to hand scores, but employs empirically based decision rules using probability values (p-values) that will be immediately recognizable to all persons familiar with common inferential statistics. Final probability estimates that are obtained through the use of separate cumulative normal distributions for truthful and deceptive classifications, as described by Barland (1985), using normative data obtained from the second of the two bootstrap trainings. This method offers the important advantage of providing probability estimates that not dependent on sample base-rates, and are more easily interpreted in test testing situations.

Because data are not aggregated in simple cumulative fashion, but through weighted and mathematical mean values and standardized measurements of differential reactivity, polygraph test examination data can be evaluated as an overall score, or through analysis of the individual question spots. The use of advanced two-stage decision rules,  as as described by Senter and Dollins (2003) and Senter (2003), is an important feature of OSS-3 that was not possible with previous data model. While the OSS-3 method can be used with single-issue (overall total) decision rules similar to the decision model of previous OSS versions, single-stage rules were observed to provide sub-optimal results compared with two-stage rules. Similarly, spot-scoring (MGQT) rules can be used with the OSS-3 algorithm, but were also observed to provide biased or suboptimal results compared with two-stage rules. With the addition of a Bonferonni correction to the specified decision alpha, to prevent the mathematical inflation of the specified decision threshold when evaluating multiple spot scores, the OSS-3/Senter algorithm provides modest though statistically significant improvements over previous OSS versions, in the form of reduced inconclusive classifications and increased sensitivity to deception with field investigation cases.

The OSS-3 method was designed from the onset with consideration for the important differences between diagnostic/investigative and screening polygraphs, as described by Krapohl and Stern (2003), and includes optimized decision rules for screening polygraphs that are conducted in the absence of a known allegation or event. The OSS-3/Screening algorithm is designed to provide maximum sensitivity to deception with multiple distinct stimulus targets, while constraining spurious results to minimal levels. The OSS-3/Screening algorithm is capable of significantly reducing inconclusive classification of truthful cases, compared with that which can be achieved by traditional spot-scoring/MGQT rules, through the inclusion of an omnibus algorithm for test data that cannot be classified as deceptive. Crossvalidation experiments indicate the OSS-3/Senter and OSS-3/Screening algorithms are capable of providing roughly equivalent performance in their intended applications, with no significant differences in decision accuracy, inconclusive classifications, or erroneous results.  

For further information regarding the OSS-3 algorithm, please contact the developers at OSS3team@gmail.com.

References

Barland G.H. (1985). A method for estimating the accuracy of individual control question tests. Proceedings of Identa-85, 142-147.

Dutton D.W. (2000). Guide for performing the objective scoring system. Polygraph 29(2), 177-184.

Kircher, J. C., & Raskin, D. C. (1988). Human versus computerized evaluations of polygraph data in a laboratory setting. Journal of Applied Psychology, 73(2), 291-302.

Krapohl, D.J. (1999). Short report: Proposed method for scoring electrodermal responses. Polygraph, 28(1), 82-84.

Krapohl, D.J. (2002). Short report: Update for the objective scoring system. Polygraph, 31(4), 298-302.

Krapohl, D.J., & McManus, B. (1999). An objective method for manually scoring polygraph data. Polygraph, 28(3), 209-222.

Krapohl, D.J., & Norris, W.F. (2000). An Exploratory Study of Traditional and Objective Scoring Systems with MGQT Field Cases. Polygraph, 29(2) 185-194.

Krapohl, D.J., & Stern, B.A. (2003). Principles of multiple-issue polygraph screening: A model for applicant, post conviction offender, and counterintelligence screening. Polygraph, 32(4) 201-210.

Senter, S.M. (2003) Modified general question test decision rule exploration. Polygraph, 32(4), 251-263.

Senter S.M., & Dollins. A.B. (2003). New decision rule development: Exploration of a two-stage approach. (DoDPI01-P-0006). Fort Jackson, SC: Department of Defense Polygraph Institute.

Senter S.M., and Dollins. A.B. (2004). Comparison of question series and decision rule: A replication. Polygraph, 33(4), 223-233.

Senter, S. M., Dollins, A. B., and Krapohl, D. J. (2004). A comparison of polygraph data evaluation conventions used at the University of Utah and the Department of Defense Polygraph Institute. Polygraph, 33(4), 214-222. 

Timm, H. W. (1982). Effect of altered outcome expectancies stemming from placebo and feedback treatments on the validity of the guilty knowledge technique. Journal of Applied Psychology, 67, 391-400.