Detail Archives    Discuss This Issue    Subscribe to The Detail Fingerprint News Archive       Search Past Details

G o o d   M o r n i n g !
via THE WEEKLY DETAIL
 
Monday, February 20, 2006

 
The purpose of the Detail is to help keep you informed of the current state of affairs in the latent print community, to provide an avenue to circulate original fingerprint-related articles, and to announce important events as they happen in our field.
_________________________________________
__________________________________________
Breaking NEWz you can UzE...
compiled by Jon Stimac

McKie Prosecution Move Defended BBC NEWS, UK - Feb 9, 2006 ...there are calls for an inquiry into the case and the operation of the Scottish Fingerprint Service...

Fingerprint Evidence Unsettles Dina Defence IOL, SO. AFRICA - Feb 17, 2006 ...fingerprint evidence linked accused baby-killer to two of her co-accused...

Body Identified After 27 Years   DES MOINES REGISTER, IA - Feb 17, 2006 ...investigators were able to retrieve only two fingerprints from her body...

Fingerprint Evidence Points to Robbery Suspect   HILLSBORO ARGUS, OR - Feb 14, 2006 ...using fingerprint evidence, detectives were able to connect an alleged serial robber to two burglaries...

__________________________________________
Recent CLPEX Posting Activity
Last Week's Board topics containing new posts
Moderated by Steve Everist

A Matter of Serious Concern
Iain McKie 26 19 Feb 2006 12:10 pm

The honest Marion Ross mistake
redlion62 144 19 Feb 2006 01:11 am

Pilot study on simultaneous (cluster) impressions
John P. Black 823 16 Feb 2006 07:32 pm

Human Identification e-Symposium 2006
Deuby 157 16 Feb 2006 03:13 pm

Forensic Scientist 2 vacancy in MN
janelle loye 159 15 Feb 2006 04:01 pm

mckie
flying monkey 983 15 Feb 2006 12:47 am

Any problems using DFO on an item previously Ninhydrin'd?
Steve Everist 241 14 Feb 2006 11:03 pm

How much training
Mike 359 14 Feb 2006 08:37 pm

(http://clpex.com/phpBB/viewforum.php?f=2)

UPDATES ON CLPEX.com
 
Updated the Training page to include Ivan Futrell's new Fingerprint Comparison and Identification Course.  Four classes have been scheduled in Texas, California, and Missouri.  Check out his course page for details.

_________________________________________

Last week

we looked at a report on the McKie settlement and saw indications of additional controversy as events continue to unfold.

Several weeks ago we looked at a critical view on the accuracy of latent print examination.

This week

we look at a study that addresses many of the concerns in that critical viewpoint in a positive way for our community.

_________________________________________
A Report of Latent Print Examiner Accuracy During Comparison Training Exercises

Journal of Forensic Identification, 2006, 56 (1), 55-93.

by

Kasey Wertheim [1]
Glenn Langenburg [2]
Andre Moenssens [3]

Abstract: During comparison training exercises, data from 108 participants were collected. For each participant, the following were recorded: the number of comparisons performed, the number of correct individualizations made, the number of erroneous individualizations made, the number of clerical errors made, and the assessments of the latent prints regarding the quantity and quality of information present in the latent prints in the exercises. Additional information regarding the training and experience of the participant was also gathered in such a manner that preserved the anonymity of the participant.

Because the training courses were open to participants of any skill level, including participants with no training and experience, the authors separated the data of participants with more than one year of experience from the data of participants with one year of experience or less. The 92 participants with more than one year of experience made 5861 individualizations (identifications) at the highest level of confidence. Fifty-eight hundred of these individualizations were correct and 61 of these individualizations were one of two types of error: 59 were clerical in nature and 2 were erroneous individualizations. This resulted in an erroneous individualization rate of 0.034% and a clerical error rate of 1.01% for the participants with more than one year of experience during these training exercises.

A follow-up experiment was performed involving verification of the errors reported by previous participants. Sixteen participants with more than one year of experience acted as verifiers to previous participants’ results. Each verifier was given a packet to verify containing the results of eight correct individualizations and two errors. These 16 independent reviewers did not verify any of the errors given to them in the verification packet exercises.

Legal Background

Prior to the Daubert decision, the standard for the admission of novel scientific evidence was the one first articulated in Frye v United States [1]. In the Frye case, the court stated:

***
Just when a scientific principle or discovery crosses the line between the experimental and demonstrable stages is difficult to define. Somewhere in this twilight zone the evidential force of the principle must be recognized, and while courts will go a long way in admitting expert testimony deduced from a well-recognized scientific principle or discovery, the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which it belongs. [Emphasis added.]
***

Thus was born the “general acceptance” test for the admission of novel scientific evidence. Over the years, the new rule was applied to a wide variety of forensic techniques in local and state as well as in federal courts. Even after the Federal Rules of Evidence were promulgated in 1975 [2], the Frye principle of general acceptance remained the law on novel expert evidence in criminal cases. Most courts adopted the principle, even those which by then had formulated evidence codes based on the federal rules of evidence. The Frye concept was also followed in the majority of federal circuits, though it was modified here and there.

The United States Supreme Court changed that concept when the Daubert plaintiffs appealed a decision of the 9th Circuit Court of Appeals, which had nonsuited them because the testimony of the experts who sought to testify on behalf of the plaintiffs was deemed inadmissible; their proposed testimony was held not to meet the Frye test! The case went to the Supreme Court on the issue of whether the passage of the Federal Rules of Evidence, and particularly Federal Rule 702 which governs the admissibility of expert testimony, had done away with the Frye rule. At the time of the Daubert appeal, Rule 702 stated:

***
If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise. [3]
***

The federal principle that a proffered expert’s opinion testimony had to assist the trier of fact in order to be admissible under Rule 702 – a concept almost exclusively applied in civil cases – was generally believed to favor the admissibility of all kinds of expert testimony. The threshold was said to be very low. In that regard, Rule 702 was said to represent a much more liberal standard of admissibility than the reputedly conservative general acceptance requirement. Indeed, the liberal attitudes toward admissibility of expert opinions generated somewhat of a hue and cry that “junk science” was flooding the courtrooms. In criminal cases, on the other hand, most federal cases had continued to apply the Frye test even after the passage of Rule 702.

The Supreme Court, when it decided Daubert in 1993, agreed with the plaintiffs’ argument that the passage of the federal rules of evidence had superseded the Frye principle as a rule of admissibility. In order to guide the United States District Courts, whom it made the gatekeepers charged with keeping junk science out of the courtroom, the Supreme Court suggested that trial judges facing a decision of whether an expert’s opinion would be admitted examine the following factors in considering the challenged expert testimony’s reliability.

The factors the Court mentioned were (1) whether the type of evidence can be and has been tested by a scientific methodology, (2) whether the underlying theory or technique has been subjected to peer review and has been published in the professional literature, (3) how reliable are the results obtained in terms of a potential error rate, and finally (4) that general acceptance can yet have a bearing on the inquiry [4].

The Daubert Court was at pains to suggest that the above list of factors was intended to facilitate a “flexible” inquiry into the reliability evaluation and that not all of these factors were absolute requirements. It also suggested that Rule 702 offered a more flexible approach to admitting expert opinions than the more stringent principle of the Frye case. In a later case (Kumho Tire v Carmichael) [5] that also answered the question whether the same Daubert factors to determine reliability would also apply to nonscientific expert testimony in the affirmative, the Court again stressed the flexible nature of the factors to be applied and recognized that not all of these might apply in a given case. Indeed, the Court recognized that for some disciplines, different factors might be more appropriate.

Friction ridge impression evidence – historically referred to as fingerprint evidence – has by now a record of nearly one hundred years of court acceptance as a reliable means by which a person’s identity can be established. Despite this success record, it has not been immune from recent attacks based on Daubert. And when such a Daubert challenge is made, one of the factors on which litigants focus is the error rate.

The Error Rate Factor

Litigants seeking to attack the admissibility of expert opinion testimony that is based on an individualization of a crime scene latent print have argued in some cases that no error rate has ever been established for friction ridge impression evidence.

Little or nothing can be found in the friction ridge impression literature on what the Daubert court suggested trial judges should do when considering “the known or potential rate of error” of the science of friction skin individuality [4]. When lower courts have sought to apply Daubert’s concept of error rates to friction ridge impression individualizations, they have divided this concept into two parts: methodological error and practitioner error.

In the most authoritative appellate decision, which is also the most recent one, United States v Byron Mitchell [6], the court seemed to accept the premise that the methodological error rate, while impossible to calculate exactly, might be close to zero. But the Mitchell court was initially bothered by the lack of definition in what constitutes an error when dealing with fingerprint identifications. It saw at least two different aspects to error: false positives (incorrect affirmative identifications) and false negatives (incorrect findings of dissimilarity). For the purpose of cases wherein fingerprints were used to tie defendants to crimes or crime scenes, the court said that only false positives ought to be considered. Seen in that light, it found that the error rate, “though not precisely quantified”, was indeed zero or close to zero [6].

With regard to practitioner error, on the other hand, this issue was seen as one relating to the competence of the expert witness and therefore not involving friction ridge comparisons as a discipline. Practitioner error, then, falls outside the purview of a Daubert inquiry. Despite this recognition, the Mitchell court suggested that “prosecutors would be well-advised to elicit testimony about their experts’ personal proficiency, rather than relying on the discipline’s good general reputation among lay jurors.” [6] The court remarked that if prosecutors do not follow that advice, cross-examiners are likely to seek to expose incompetent examiners on cross-examination by inquiring about that very issue.

With no research regarding practitioner error rates in the scientific literature, litigants attacking fingerprint evidence and critics of the profession have filled this void using inappropriate measures of examiner reliability. Proficiency tests administered by Collaborative Testing Services (CTS), failure rates of the International Association for Identification (IAI) Latent Print Examiner Certification Examination, and high-profile erroneous individualizations have all been cited as measures of practitioner error rates. *

* The 1995 CTS latent print proficiency test is a common citation for critics [6, 7]. This particular test had the highest error rate for participants (22%) of any CTS latent print test [8]. Contributing to this substantially high error rate was one impression in blood. This impression represented the most similar area of a very close pattern shared by an identical twin brother. The known exemplars of the donor of this impression were not provided, but the known exemplars of the twin brother who did not create the impression were provided.

The authors believe that CTS latent print proficiency tests can meet minimum external proficiency testing standards when used appropriately by agencies to annually monitor the performance of latent print examiners. However, the authors do not support the use of the results of these CTS latent print proficiency tests to estimate practitioner error rate for several reasons:

1) Errors in CTS tests are reported as the number of responses with results that differ from the “manufacturer’s expected results”. There is no distinction in the results between the types of errors (e.g., clerical errors, erroneous individualizations and exclusions, and missed individualizations and exclusions) that were committed by the participants. (Examples from one CTS Latent Fingerprint Examination are shown in Appendix A. [of the article in the JFI])

2) Individuals who are not trained to competency may participate in the CTS latent print proficiency tests, and no distinction is made between the results of these individuals and the results of those who are trained to competency.

3) Participants in the CTS tests are not limited to participants in North America and do include participants from European countries, where a minimum number of minutiae (“point threshold”) is maintained as operational procedure or necessary as a legal standard for courtroom admissibility. Thus, such participants may not declare a match on the test, and this will be scored by CTS as “not identified”. “Not identified” may be construed by some critics as an error [7, 9].

In 1996, the National Research Council (NRC) released its second report on DNA, entitled The Evaluation of Forensic DNA Evidence [10]. The NRC addressed DNA error rates within the report. It stated that proficiency testing and audits are both essential components of quality assurance programs; however, neither is designed to measure error rates. (For further information regarding the NRC report, please see the discussion in the Limitations section.)

A second measure of examiner accuracy that has been cited is the Latent Print Examiner Certification Examination of the International Association for Identification [11]. The authors firmly believe that this is a highly inappropriate measure of accuracy. Approximately half (48%) of the examiners who meet the minimum requirements to take the certification examination fail the test [12]. What is not distinguished is the reason for the failure rate. Any of the following events would result in failing the four-part Latent Print Examiner Certification Examination [13]:

1) Receiving a less-than-passing percentage on any of the four parts of the examination: pattern classification (90%); general knowledge, history, and processing (85%); comparisons (80%); courtroom testimony (pass/fail)

2) A single erroneous individualization made in the comparison portion of the exam

3) A single clerical error in the comparison portion of the exam (those administering the examination cannot distinguish between clerical errors or erroneous individualizations)

4) Failure to complete the first three portions of the examination within the required time limit (6 ½ hours)*

* Modified examination requirements were established by the IAI Board of Directors during the 2005 conference [14].

A third inappropriate measure of examiner accuracy is the citation of anecdotal or high-profile erroneous individualizations [15-18]. The existence of such cases merely confirms that erroneous individualizations can and do occur. But without knowing the number of correct individualizations that have been made, these cases of erroneous individualization are simply anecdotal, misleading, and inappropriate for measurement of examiner accuracy.

This study does attempt to compensate for the limitations of the aforementioned inappropriate measures of examiner accuracy. Because this is experimental research, caution must also be used when examining the data presented here to draw conclusions regarding examiner accuracy in actual case work. The limitations of this study are discussed later in this report. Additional studies by the authors are planned to address some of the limitations of this study.

Methods and Materials

Data utilized in this study originated from examiners during latent print comparison training courses. During the courses, participants were given comparison exercise packets. Each packet contained ten latent prints and eight sets of inked exemplars. The difficulty of the packets varied in terms of quality and quantity of friction ridge detail present in the latent prints, the lack of focal points to aid the examiner when searching, the source area of the latent prints (e.g., palm prints, sole prints, etc.), and so forth. The packets ranged in levels of difficulty from 1 (easiest) to 16 (most difficult) and consisted of a pool of approximately 4600 different latent prints. The rating of difficulty for these packets was predetermined by the course instructors (three certified examiners with approximately 45 years of combined experience). This decision was subjective and based primarily on the complexity of the latent prints in the packets and the challenge of the individualizations [e.g., difficulty to locate; degree of distortion; lack of focal points (core, delta, scars, etc.); lack of “helpful searching clues”; and quantity of available minutiae].

The difficulty of the first packet that was received by each participant was based upon the participant’s declared training and experience. The difficulty of subsequent packets that were given to the participant was based on the participant’s performance with the first packet, as monitored by the course instructor.

Because these were training courses whose aim was to improve the comparison ability of the participants, the students were encouraged to challenge and improve their skills by tackling increasingly difficult exercises. Additionally, all of the exercise packets contained latent prints that had been deemed “sufficient for individualization” by course instructors, and all latent prints had been individualized to the exemplars provided (i.e., there were no nonmatches).

When the participant was given an exercise packet, the participant was asked to analyze the ten latent prints in the packet before searching and comparing to the exemplars. He or she was asked to rate each latent print in seven* categories: quantity of details, quality of details, presence of focal points, level of contrast, amount of lateral distortion, amount of deposition pressure, and level of background interference.

* The participants from one of the training classes were asked to rate the latent prints in just six categories (lateral distortion was added to subsequent classes). Furthermore, the rating scheme for this training class was a scale of 1 through 3 (1 the lowest in all categories). The rating scheme in the subsequent training classes was a scale of 1 through 5 (1 the lowest in all categories). The decision to include these data nonetheless was made primarily because these data included an erroneous individualization, and the authors did not wish to discard the data merely on the basis of a slightly expanded analysis of the latent prints.

After the analysis was performed, the participant performed comparisons to the ten sets of exemplars (standard inked fingerprints, palm prints, or sole and toe prints). Participants performed comparisons using standard magnification devices for latent print examiners (typically 4 to 6 X magnification) and ridge counters. If the participant effected an individualization, he or she was asked to record the corresponding name from the exemplar and source (e.g., finger number, right or left palm, etc.). The participant was also asked to record the time it took to complete the procedure (including time for analysis of the latent print). Lastly, the participant was asked to record the level of confidence for the individualization.

Recording the level of confidence of the individualization provided a mechanism that served two purposes:

1) Participants could push their comparison skills in a training environment and attempt exercise packets that may have been beyond their skill level.

2) The authors could differentiate between individualizations made at the highest confidence level and individualizations that the participant may not have felt entirely comfortable making and would not have reported in actual casework.

When the participant completed (or attempted to the best of his or her ability to complete) an exercise packet, a new exercise packet was given to the participant, and the results of the completed exercise packet were examined by the course instructor.

When the course instructor examined the answer sheets of participants, results that were not in agreement with the known answers for the exercise packet were noted. On the basis of past experience and common trends, and in some instances, a discussion with the participant, the instructor made a determination as to whether the error appeared to be an erroneous individualization or a clerical error. If it could not be clearly determined, then the error was determined to be “indeterminate”.

Participants were also asked to provide additional information regarding their training and experience. Any information that revealed the identity of a participant was removed from all
worksheets.

A unique alpha-numeric identifier was associated with each participant’s data, thereby rendering the source of the data anonymous. All data were then pooled together and assessed.

Results and Discussion

Demographics

Data were collected from 108 participants. The mean number of years of experience for the participants was 7.9 years (range = 0 to 30+ years, n = 107*) (Figure 1[in the JFI]).

* One participant did not report this information.

Of the 108 participants, 16 possessed one year of experience or less. Because the training courses were open to participants of any skill level, including participants with no training and experience, the authors arbitrarily selected to separate the participants with more than one year of experience. (Rather than separating participants on the basis of experience, the
more appropriate device to separate participants is to determine whether the participant is trained to competency and performing unsupervised casework. This modification will be added to future studies.)

Summary of Results

Table 1 [in the JFI article] shows the total number of correct individualizations and errors that were made by the participants. The data in the table are separated into three categories: data from participants with one year of experience or less, data from participants with more than one year of experience, and combined data from both groups. A comparison of the two groups in Table 1 shows a higher percentage of errors were committed by participants with one year of experience or less. Thirty-seven errors were committed by 16 less-experienced participants. This equates to 2.3 errors per inexperienced participant. (It must be noted, as will be shown in Figure 3, that some participants did not make any errors at all). In contrast, 81 errors were committed by 92 participants with more than a year of experience. This equates to 0.88 errors per participant with more than a year of experience. Combining these data, the average errors per participant in the study was 1.1. Thus, the inclusion of data from inexperienced individuals penalized the more experienced individuals.

A further examination of the 37 errors made by participants with one year of experience or less shows that 21 of the 37 errors (57%) were erroneous individualizations—four of which were made at the highest level of confidence. This is a sharp contrast to the participants with more than a year of experience. The more experienced group committed 15 erroneous individualizations—two of which were made at the highest level of confidence—out of 81 total errors (19%). These data support the proposition that CTS results, which include data from relatively inexperienced examiners or trainees, may have inflated error rates and should not be applied to latent print examiners who are trained to competency.

Confidence Rating

In Table 2, the results for participants with more than one year of experience are separated into categories of confidence. Participants were instructed to use a designation of confidence when reporting every individualization. In 93 instances, a participant neglected to report a confidence. The scale of confidence (1 through 3) is explained below:

3 = Highest level of confidence. The participant recorded this level of confidence if the participant would report this individualization in casework.

2 = Moderate level of confidence. The participant recorded this level of confidence if the individualization was beyond his or her ability and comfort level. Scenarios used to describe the appropriate use of this level of confidence included “not absolutely certain about the individualization”, “you would consult another colleague before reporting”, or “you would spend more time before reporting”.

1 = Lowest level of confidence. The individualization is far beyond the participant’s ability and comfort level. Scenarios used to describe the appropriate use of this level of confidence included “a strong guess” or “indications of the source of the latent print”.

The above confidence rating scale allowed participants to push their comparison skills beyond their comfort and skill level in a training environment. Participants were encouraged to complete increasingly difficult exercise packets that exceeded their skill level.

On the basis of this designation of confidence, the data of greatest interest are the individualizations made at a confidence rating of 3. Of 5861 individualizations made by examiners with more than a year of experience at a confidence rating of 3, two were deemed erroneous individualizations and 59 were deemed clerical errors. This equated to an erroneous individualization rate of 0.034% and a clerical error rate of 1.01%.

Error Classification

The determination of the error type (erroneous individualization, clerical, or indeterminate) was a decision made on-site by the course instructor. When determining clerical errors, obvious trends were observed to support this decision. If the instructor could not determine the intent of the student, the error was classified as indeterminate.

Clerical errors in this training course usually represented the incorrect recording of the source of the latent print. For example, the source of the latent print was recorded as the left ring finger (#9 finger) when the participant meant the right ring finger (#4 finger). This clerical error will be referred to as transposition transcription error. A second type of clerical error was for the participant to record the correct finger number but incorrectly record the name on the fingerprint card. Both of these types of errors are easily identified by the instructor by merely examining the latent print and recorded exemplar. For instance, if the latent print is a left-slant loop pattern and the exemplar bears a right-slant loop pattern, it is highly unlikely that a participant with even minimal training and experience would effect such an individualization. See Figures 2A and 2B [in the JFI article] for actual examples.

Table 3 [in the JFI article] provides further analysis of the clerical errors.

Approximately twice as many clerical errors were made on latent prints originating from a left hand. However, the number of latent prints originating from right hands in the exercise packets was approximately equal to the number of latent prints originating from left hands in the exercise packets. Participants were more likely to err when recording the source of a lefthanded latent print. Subconscious biases or expectations may contribute to this effect.

There were 15 clerical errors at the highest level of confidence that were not transposition transcriptions, but were identified to the correct individual. Of these, 10 were the next sequential finger (e.g., #7 left index finger was recorded, but the source was #8 left middle finger). This could possibly be from an error when translating the plain impressions at the bottom of the fingerprint card. It would also be of interest to see whether those participants with significant experience or some tenprint experience would be less likely to commit such clerical errors.

Lastly, there were 2 clerical errors at the highest level of confidence in which the participant recorded the correct finger or palm but recorded the incorrect name from the exemplar. On the basis of previous trends noted above, it is highly unlikely that the participant was effecting the individualization to the pattern type or print present in the exemplar. This type of error, though deemed clerical, could have a serious impact in the case, as it incorrectly associates an individual with the case who otherwise could have been excluded. Although clerical errors are often not deemed as serious as erroneous individualization, it is important to recognize that the consequences of an undiscovered clerical error could potentially be quite serious. The obvious difference is that a clerical error will, under most circumstances, be readily apparent when it is reexamined.

Table 4 shows the percentage of each type of error for the total individualizations attempted at each level of confidence rating. An examination of Table 4 shows that clerical error is independent of confidence, although the number of erroneous individualizations increased by two orders of magnitude as confidence decreased.

Multiple Errors

Some participants made multiple errors. The total number of errors made by each participant with more than one year of experience ranged from 0 to 5. Figure 3 shows the number of participants and the distribution of the total number of errors committed by each participant. Over half (48 out of 92) of the participants did not make a single error. Additionally, each column in Figure 3 is separated into the relative percentages of total errors committed attributed to erroneous individualizations and clerical errors. For example, one column in Figure 3 shows that eight participants committed a total of three errors each and 90% of the errors in this column were clerical errors. Figure 3 includes all errors made at all levels of confidence.

Figure 3 illustrates that a decreasingly smaller fraction of the participants in the course were responsible for multiple errors. As the number of multiple clerical errors increased for an individual, so did the number of erroneous individualizations (for all levels of confidence). This trend could be an indicator of an examiner’s sloppy work habit, an absence of double checking work product, or simply rushing through the exercises, keeping in mind that this was not a case-work environment. These data support the argument that the courts should look to each expert’s practitioner error rate rather than the wholesale exclusion of fingerprint evidence in the courtroom or as commentary on an industry-wide error rate.

Anatomy of Errors

Of more interest and value to the latent print examiner community than the reporting of occurrences of errors is why the error occurred. Some insight can be provided by an examination of the two erroneous individualizations reported at the highest level of confidence by participants with more than one year of experience.

Erroneous Individualization #1

The participant (Participant No. 125A) who effected this erroneous individualization reported the following information with respect to training and experience:

2 years of experience
30% of duties are analyzing and comparing latent prints
0 times testifying in court to latent print evidence
Seventy-nine (79) total individualizations were attempted by this individual during the course:
75 at confidence = 3
3 at confidence = 2
1 confidence not reported
1 erroneous individualization was made (confidence = 3)
3 clerical errors were made (confidence = 3)

The latent print that was erroneously individualized is shown in Figure 4[in the JFI article]. The latent print is from an exercise package of difficulty rated at 5 (on a predetermined scale ranging from 1 to 16). This participant worked with eight exercise packets ranging in difficulty from 1 to 7. The participant reported having only two years of experience. An important question arises when analyzing the background of the participant who made this error: Was the latent print (PP5-079) beyond the ability and experience of this particular examiner? The combination of a latent print beyond the ability level of the participant and relative inexperience may have contributed to the erroneous individualization. This trend will be examined in future studies.

Erroneous Individualization #2

The participant (Participant 2100B) who effected this erroneous individualization reported the following information with respect to training and experience:

0 years of training (the question “Years of fingerprint training” had been added to the background survey when this participant took the course)
6.5 years of experience
% of duties are analyzing and comparing latent prints (not answered)
12 times testifying in court to latent print evidence
The average rating for all 79 latent prints examined by this participant was 2.30 +/- 0.57 (SD). The average rating for latent PP5-079 was 1.50. This latent print was 1.3 standard deviations below the average difficulty attempted for this participant.
Sixty (60) total individualizations were attempted by this individual during the course:
60 at confidence = 3
No other confidence level was reported by this individual
1 erroneous individualization was made (confidence = 3)
2 clerical errors were made (confidence = 3)

The latent print that was erroneously individualized is shown in Figure 5 [in the JFI article]. The latent print is from a package of difficulty rated at 8. This participant worked with seven packets of latent prints ranging in difficulty from 5 to 12. Was the latent print (F28-127) beyond the ability of this particular examiner? The participant rated the latent print F28-127 as shown in Table 6. The average rating for all 60 latent prints examined by this participant was 3.75 +/- 0.88 (SD). The average rating for latent F28-127 was 4.00. This latent print was within the average range of difficulty for this examiner. It is also important to note that this examiner incorrectly individualized an exemplar that is very similar in appearance to the correct exemplar for this comparison (i.e., a close nonmatch).

Comparison of the Two Erroneous Individualizations Clearly, there is a greater cause for concern with Erroneous Individualization #2 than with Erroneous Individualization #1. The participant who made Erroneous Individualization #2 was experienced and has testified in court. The latent print was not beyond the average difficulty for this examiner in this course. The participant who made Erroneous Individualization #1 was relatively inexperienced, has not testified in court as an expert, and the print was above the average difficulty for this examiner in this course. It is also possible that the less experienced of these two examiners was still in training and was not performing casework at the time. The demographic questions that were initially asked of participants in this study did not adequately determine whether an individual was still in training or performing unsupervised casework. As addressed previously, this is a limitation of the study and the appropriate questions will be asked in future studies.

More information is needed regarding work habits, training, and examiner performance before any meaningful conclusions can be drawn. This is an important area for future study, because the answers to these questions could provide additional guidelines for training, application of methodology, and standards for expert qualifications.

Limitations

There are a number of limitations that must be addressed. Some of these limitations will be minimized or eliminated in future studies proposed by the authors. Each of these limitations is discussed below.

The known limitations of this study were:

• Training environment versus casework
• Limited equipment and facilities
• Pushing difficulty during a training environment
• No nonmatches
• No verification
• Population of participants may not be wholly representative
• In limited instances, classification of error type was subjective
• Background information of participants was not known

Training Environment Versus Casework

An important limitation of this study, and one explanation for why the erroneous individualizations occurred, is that all of these comparisons were performed in a training environment. It is possible that a participant may place less seriousness or value on training product than on casework product. Some participants may have placed undue emphasis on competing with other students in the training courses to make more individualizations. Competitive behavior would not be expected in normal casework.

On the other hand, casework examinations may be affected by a psychological mindset not present in a training environment. Context bias* was recently shown to affect four latent print examiners during one study [19]. Context bias and the pressure of high-profile cases were not present in our study design.

* Context bias or “observer effects” is the phenomenon where context information or expectations of the observer can influence the observer’s conclusions, judgment, perceptions, and decision-making processes [20].

Limited Equipment and Facilities

Some examiners may not have performed as they typically would have had they been in their normal environments. In the training environment, examiners were limited to the tables, chairs, and lighting devices present at the facilities, and the examination equipment consisted of only magnifiers and ridge counters. Digital enhancement software, scanners, and computers were not available to the student. The training course was held during working hours (0800 to 1700), and participants who normally perform comparisons during later hours or on shift work may have had difficulty adjusting to the change.

Additionally, because this was a training course with specific dates and times for the course, daily factors that can affect performance could have an effect on training product that may not necessarily be present in casework. There is generally no set time limit with casework, and the length of time an examiner spends on a case is flexible and discretionary. For example, if an examiner is ill, casework can be postponed until the examiner recovers. Some examiners may be accustomed to minimal comparison activity in one day, unlike the intensive exercise periods required in a training environment. This flexibility was not present in the training environment where the participant was required to be present and to work during the hours of the course. In future studies, additional questions can be added to the survey to identify some of these daily variables that could affect comparison results.

Pushing Difficulty During a Training Environment

Participants were encouraged to push their comparison abilities by attempting increasingly more difficult exercises as they progressed through the course. If a participant expressed to the instructor that he or she had progressed “too far”, then exercises for the participant’s appropriate skill level were administered. The participant always had the option of reporting an individualization with lower confidence (e.g., less than 3); however, it is not known by the researchers whether this option was always utilized appropriately by the participant.

No Nonmatches

Another limitation of this study was the absence of nonmatches. Every latent print had a match in the packets and the participants were aware of this. This design was due to the nature of the course intent, because participants were taking the course to become more efficient at finding and making individualizations. This important issue will be addressed in future studies with the inclusion of exercise packets with nonmatches.

A participant’s knowledge of the absence of nonmatches may have affected the comparison results. One possible result was an increase in the number of erroneous individualizations, because participants were aware that there must be a match in the exemplars. The participant may have attempted to “force” the individualization and found one that was quite similar, yet still a nonmatch. This effect may explain the sharp increase in erroneous individualizations at lower levels of confidence (confidence less than 3) in Table 4, but, notably, the rate of clerical errors did not change. Conversely, knowledge that nonmatches did not exist may have encouraged the examiner to continue searching for an individualization long after that participant may have given up (and potentially erroneously excluded an individual as having made an impression) in actual casework.

No Verification

Individualizations were not verified by other examiners in this experiment. In casework, an examiner following the ACE-V methodology will have all individualizations verified by another examiner who is trained to competency. This did not occur in this course. With a second examiner performing an independent examination of the declared individualization, it seems likely that the number of errors reported (clerical and erroneous individualizations) will be reduced. A follow-up experiment was performed to test this very notion, and the results of that experiment are included following the limitations listed here.

Population of Participants May Not Be Wholly Representative

Because this is a training course to increase the efficiency of the examiner, there may be several subsets of the latent print examiner population that are not represented by this sample of examiners. In particular, examiners certified by the International Association for Identification are unlikely to take this course, because this course is promoted as a helpful tool to pass the certification examination. It is also possible that examiners who need improvements to their skills or training may be more likely than highly skilled examiners to attend this course. Therefore, highly skilled examiners may be inadvertently excluded from this study. On the other hand, examiners working for agencies that minimize or ignore training needs may not receive the opportunity to attend this course. Therefore, poorly skilled examiners may be inadvertently excluded from this study. Both of these scenarios could lead to a sample population that may not be entirely representative of the latent print community at large.

In Limited Instances, Classification of Error Type Was Subjective

Clearly, the choice to classify an error as erroneous individualization, clerical, or indeterminate was an important decision. As previously addressed, specific trends were observed in many clerical errors (transposition transcription, sequential fingers, etc.). In some instances, student behavior was a strong indicator of evidence (typically students who effected an erroneous individualization in the course often appeared irritated, upset with their performance, and were immediately very conservative and less confident in their performance). Beyond identifying obvious trends, ultimately, the decision to classify an error is a subjective decision by the course instructor that is based on the observed trends, student behavior, student’s performance, and honesty of the student. Though the authors are confident in our classification of clerical versus erroneous individualization errors, the worst case scenario is that two errors, which were deemed clerical in nature, erroneously associated another individual with “the evidence”. Thus, of the 61 total errors reported at the highest level of confidence by participants with more than one year of experience, 57 of these errors associated the correct individual to the evidence (but listed the wrong finger, palm, or foot), and 4 of these errors associated the incorrect individual to the evidence. However, our contention is that the clerical errors are likely to be easily spotted during verification by a second examiner.

Background Information of Participants Was Not Known

Initially when the data were gathered, participants in the course were not informed that the data may be used in this or future studies. Therefore, to preserve anonymity, minimal
questions were asked regarding the individual’s background. The data were then pooled for calculations. Thus, in an effort to preserve the source of the data, potentially critical conclusions regarding training and experience were sacrificed.

In spite of the anonymity, other colleagues raised concerns that the students were unaware of how the data might be used. The authors wish to assure the readers that these concerns were taken very seriously. In an effort to address these concerns, an attempt was made to contact all past students whose data may or may not have been included in the study. Because of the anonymity of the alpha-numeric identifier given to each student’s data, even the authors can no longer determine the source of the data. Therefore, mass mailings of notification forms were distributed to potential contributors. Included in the notification forms was an advance draft of this article, in an effort to show how the data were utilized.

Future studies by the authors will include notification and consent forms prior to data collection. It is not known how prior knowledge of testing before participation may affect future results. Also, future studies will include questions that will address training, experience, and background, while still preserving participant anonymity.

Final Caution

It is difficult to quantify an error rate for the human expert. It is a moving target. The error rate for a particularly easy latent print could be significantly different than the error rate for a very difficult, highly distorted latent print. The problem is further compounded by the fact that the error rate will be directly tied to the ability of the examiner.

When examining the errors and assessing the reliability of examiners in this study, it is imperative to note that these are experimental data. This is one experiment, under a specific set of conditions, for a limited sample of experts looking at a closed set of latent prints.

The data in this study could be of value if the population of examiners and the population of latent prints in the course begin to approach an accurate sampling of examiners and latent prints in case work. With more than 4600 different latent prints of varying quantity and quality of ridge detail in the exercises, it is reasonable to assume that the prints in this study would contain an adequate sampling of latent prints encountered in actual casework. However, as previously addressed, because this is a training course, segments of the latent print expert community may continue to be excluded (e.g., certified examiners, experts from underfunded agencies). This is an important limitation of the study. A training environment can provide massive amounts of data and comparisons that are essential for this type of experiment; however, one drawback is that the population of participants will most likely be limited.

Another caution is how one should use these data. The data in this study are descriptive, not predictive. The errors that were identified were specific to the participants, latent prints, and conditions of the study. These data should therefore not be used as a predictor of error or an estimate of reliability for an examiner on the witness stand. At best, attorneys ignorant of science or at worst, unscrupulous attorneys may attempt to apply the data from this study improperly. Questions of the witness, such as, “Isn’t it true that there is a .034% chance that the identification before the court is erroneous or at least a 1% chance that it was a clerical error and therefore not my client?” should be summarily dismissed and would be a gross misuse of these data. The expert witness should be prepared to discuss the limitations of this study and error rates in general.

Historically, the concerns of error rates in forensic science have also been addressed by other disciplines. In the 1996 Report from the National Research Council for the forensic use of DNA, the NRC committee stated the following points with respect to error rates and forensic DNA examinations [10]:

• At issue is not the general error rate for a laboratory or laboratories over time but rather whether the laboratory doing the testing in this particular case made a critical error.

• To accurately estimate error rates from proficiency tests, it would require laboratories to undergo an unrealistically large number of proficiency trials. In effect, laboratories would be performing more proficiencies than actual case work.

• The pooling of proficiency test results across laboratories has been suggested as means of estimating an “industry-wide” error rate. This would penalize the better laboratories; multiple errors on a single test by one laboratory could substantially affect the overall estimated false match error rate. Initial studies have shown that the preponderance of errors originated in a small population of laboratories.

• Using descriptor error rates from a previous set of tests to predict or estimate error rates in future tests (for false matches) is almost certain to be incorrect. When errors are discovered, they are investigated thoroughly so that corrections can be made. A laboratory is not likely to make the same error again, so the error probability is correspondingly reduced.

The NRC further summarized by stating:

…for all those reasons, we believe that a calculation that combines error rates with match probabilities is inappropriate. The risk of error is properly considered case by case, taking into account the record of the laboratory performing the tests, the extent of redundancy, and the overall quality of the results. However, there is no need to debate differing estimates of false-match error rates when the question of a possible false match can be put to direct test… [10].

Their final recommendation is that the best insurance against false incrimination is the opportunity for retesting and minimization of errors through proper quality programs as recommended by the guidelines of appropriate Scientific Working Groups and accrediting bodies (e.g., ASCLD/LAB).

Follow-up Experiment – Verification

A significant limitation of the initial experiment was the lack of verification. In casework, an examiner who is following the ACE-V methodology will have all individualizations verified by another examiner who is trained to competency. This did not occur in this course. With a second examiner performing an independent examination of the declared individualization, it seemed likely that the number of errors reported (clerical and erroneous individualizations) would be reduced.

As a follow-up experiment, verification packets were prepared for a new group of 25 participants. Two verification packets were created. The first packet (marked PP) contained the erroneous individualization #1 (PP5-079) and a second error (an erroneous individualization that was previously reported by a participant at confidence level of 1). The second packet (marked F2) contained the erroneous individualization #2 (F28-127) and a clerical error (a transposition transcription error). Both packets also contained eight correct individualizations reported by previous participants.

In this experiment, participants were given a worksheet previously completed by students, but with no indications of the errors that were present (an example is provided in Figure 6[in the JFI article]). Participants were asked to indicate whether they “AGREE” or “DISAGREE” with the conclusions of the previous participant. Participant were given the option to make comments or notes regarding their verification conclusions. Some participants chose to make notations; some participants did not.

Table 7 shows the results of the verification follow-up experiment. These data are reported similarly to the initial experiment, dividing the data into three groups: an inexperienced group, an experienced group, and combined data for both groups.

Of the 50 possible errors (25 participants, each receiving two errors in his or her packet), 49 errors were detected by the verifiers. Only one error (the erroneous individualization F28-127) was not detected by a verifier, and this verifier was a trainee, had less than one year of experience, and was not performing unsupervised case work.*

* A modified questionnaire and the addition of a consent form were introduced in the follow-up experiment. From the modified questionnaire, it was determined exactly how much latent print comparison experience was possessed by a participant and whether a participant was trained to competency and reporting out unsupervised casework results. This information was not previously available in the initial experiment. For proper comparison of the data between the two experiments, participants were still separated into two groups: those participants with one year or less experience and those with more than one year of experience. Only two participants in the followup experiment possessed more than one year of experience, but were not declared “trained to competency” by their agency, nor were they generating their own casework reports.

Table 7 only displays the results of the errors that were present in the verification packets. Interestingly, not all of the correct individualizations were agreed upon by verifiers. Of the eight verifiers with more than one year of experience who examined verification packet #1 (PP), four verifiers wrote “DISAGREE” for at least one of the correct individualizations. Of the eight verifiers with more than one year of experience who examined verification packet #2 (F2), five verifiers wrote “DISAGREE” for at least one of the correct individualizations.

Verifiers were given several reasons why they could disagree with a previous participant’s conclusion. Examples included, but were not limited to: an error was committed, insufficient agreement to call, inconclusive, would need to spend more time examining with different equipment in the laboratory, beyond my expertise, and so forth. If students possessed any of these concerns, they were instructed to state “disagree”. They were to state “agree” only if they were certain of the conclusion as it was stated on the worksheet, would sign off on it, and would send it out the door, willing to testify to their verification in court.

The results suggest that a more conservative attitude was taken by the verifier. Not only did the experienced verifiers catch all the errors, but they were also hesitant to agree on some of the correct individualizations. One possible explanation is that some of the individualizations (although correct) were above the level of difficulty at which the examiner was comfortable. Unlike the standard exercises in this course, the verification packets were fixed at specific levels of difficulty [levels 5 and 8 (rated out of 16) for packet PP and packet F2, respectively].

Another explanation for the unverified correct individualizations is that verifiers assume a more critical and conservative opinion knowing that they are in a position of quality control. By inference, this may suggest that the original examiner in case work may be less conservative, knowing that someone else is going to review his or her results. Further study of this dynamic within the ACE-V methodology is recommended by the authors.

It should also be noted that the verification worksheet had the results listed by the previous participant. This is similar to the procedures used by many U.S. agencies when performing verification (i.e., not blind verification). Some critics and researchers are calling for blind verification of all results [20-23], and, in fact, one researcher stated that verification is a misnomer; that verification with respect to the ACE-V methodology is “ratification, not verification”[24]. The fact that not all examiners agreed with all of the correct individualizations placed before them, and all examiners with more than one year of experience disagreed with all of the individualizations in error, suggests that the verification process is not “ratification”.

Furthermore, the results of the follow-up verification experiment draw attention to two issues. The first issue is that because one of the errors (and notably, the close nonmatch), Erroneous Individualization #2, was not caught by a trainee, this instance illustrates a potential danger when inexperienced individuals are asked to verify difficult individualizations above their skill, ability, or experience level. The second issue is that in our study, all of the experienced examiners did not verify any errors, even though they were told the results of the initial examiner. The verifying examiners in our study, however, were not given any context information about the initial examiner (e.g., identity, experience, etc.); they were simply given the conclusions of the initial examiner. These results are in contrast to the results of the Dror study [19], in which four out of five examiners were influenced by the context information provided and did provide an erroneous result. Clearly, there is a need to further investigate the potential for bias effects in latent print examinations, and this potential for bias must be checked and balanced with an appropriate blind verification scheme (perhaps under limited circumstances as the Stacey Report [15, 25] suggests).

Finally, it should be noted that the design of this follow-up experiment was actually biased against the participants catching all of the errors. The participants, like all participants of the course, are told at the beginning of the course that they will be only making individualizations in their exercise packets and that there are no nonmatches. The participants spent several days prior to receiving the verification packets making dozens of individualizations from previous exercise packets. Toward the end of the course, the participants were given the verification packets and they were not told that these packets would contain errors. They were simply asked whether they agreed or disagreed with the previous participant’s conclusions. Additionally, in actual casework, it is common practice for an agency to have only one examiner, perhaps two in some agencies, verify an initial examiner’s conclusions. In this follow-up study, each error was subjected to eight separate, independent verifications by examiners with more than one year of experience. In all of these instances, the errors were not verified by the examiner.

Finally, the verification worksheets had been altered by the authors. Initially, some of the conclusions of the previous experts were made at lower levels of confidence. All the results on the verification worksheets in the experiment were altered to a 3 (the highest level of confidence). Therefore, participants were given worksheets that not only presented errors marked as individualizations, but also individualizations made by other experts at the highest level of confidence. In spite of these biases, participants with more than one year of experience still did not verify any error.

Future Research and Direction

Future studies will address some of the limitations previously discussed. In particular, nonmatches in packets, proper background and demographic data (while still preserving anonymity), and continued inclusion of verification will be valuable components. Additionally, collecting data from participants from countries with national standards for training and registration, numeric threshold standards (a minimum number of points), or different methodologies for comparison will be a goal of future studies. These data could then be compared to data from U.S. examiners to determine whether there are significant differences. Such comparisons could support or identify weaknesses in current standards and practices.

Conclusions

The following conclusions were drawn from the data in this experiment:

• 5861 individualizations were attempted at the highest level of confidence by 92 participants with more than one year of experience.

• 5800 correct individualizations were made at the highest level of confidence by participants with more than one year of experience.

• Two erroneous individualizations were made at the highest level of confidence by two participants with more than one year of experience.

• Fifty-nine clerical errors were made at the highest level of confidence by participants with more than one year of experience.

• The most common type of clerical error was a transposition transcription error (74% of the clerical errors).

• Errors occurred twice as often on left-handed latent prints than errors from right-handed latent prints.

• More than half the examiners made no errors (clerical or erroneous individualizations); however, some examiners made multiple errors. As the number of errors increased for a single examiner, so did the number of that examiner’s erroneous individualizations (these data included all errors at all levels of confidence 1 through 3).

• Participants with one year of experience or less made a significantly higher ratio of errors (specifically erroneous individualizations) when compared to participants with more than one year of experience.

• When the confidence reported for a participant decreased, the number of erroneous individualizations increased by two orders of magnitude. The number of clerical errors, however, remained relatively constant. Clerical errors are independent of latent print difficulty or examiner ability.

• A verification follow-up experiment was conducted. Sixteen participants with more than one year of experience did not verify two errors given to each of them amongst eight correct individualizations. Only one erroneous individualization was not caught by a verifier in the follow-up experiment. This verifier was a trainee with less than one year of experience.

The previous conclusions and rates of error were meant to determine the reliability of examiners under a specific set of conditions. These statistics are not intended to be a predictor of reliability for other examiners in case work; rather, they are merely a hint of a nebulous, wispy, industry-wide error rate. Most importantly, the results show that errors can and do happen. Latent print examinations are currently performed by human examiners, and human examiners are fallible. Therefore, the examinations are only as valid as the examiner performing the tests and the conditions under which the tests were conducted. Only through rigorous use of ACE-V, sufficient training, use of precautions to limit bias and potentially other influential factors, and appropriate safeguards of quality assurance programs will errors be reduced and the chances of falsely incriminating an individual based on an erroneous individualization be minimized.

Acknowledgments

The authors would like to thank the following people for their various contributions:

• BCA student workers Kristin Tebow and Trisha A. Evans for their tireless data entry.

• SWGFAST members for their comments and concerns, particularly Alice Maceo, Debbie Benningfield, Bridget Lewis, Mike Campbell, Lyla Thompson, Stephen Meagher, Alan McRoberts, Pat Wertheim, David Grieve, and Christophe Champod for their valuable contributions.

• Drs. Ralph and Lyn Haber for their suggestion of a trilevel indicator of confidence. This was an important tool for the project and an improvement upon an earlier design. Their comments and guidance with respect to the study design were extremely valuable.

• Lastly, and most importantly, the students of these courses for their contribution to the study and the profession. Without their continued support and participation, future studies would not be possible.

Questions regarding this article or these data should be directed to:

Glenn Langenburg
Minnesota Bureau of Criminal Apprehension
1430 Maryland Avenue East
St. Paul, MN 55106
(651) 793-2967
Glenn.Langenburg@state.mn.us


References

1. Frye v US, 293 F. 1013, 1014 (D.C.Cir. 1923).

2. Federal Rules of Evidence; House of Representatives - The Committee on the Judiciary, US Government Printing Office: Washington, DC, 1975.

3. Federal Rules of Evidence; House of Representatives - The Committee on the Judiciary, US Government Printing Office: Washington, DC, amended April 2000.

4. Daubert v Merrell Dow Pharmaceuticals, Inc. 509 U.S. 579, at 593-595 (1993).

5. Kumho Tire v Carmichael, 526 U.S. 137 (1999).

6. US v Byron Mitchell, 365 F.3d 217 (3rd Cir. 2004).

7. US v Plaza, et. al. 188 F. Supp. 2d, Daubert Hearing, 2002.

8. Grieve, D. Possession of Truth. J. For. Ident. 1996, 46 (5), 521-528.

9. Smith, K., Member Latent Print Certification Board, Dulles, VA. Personal communication, 2005.

10. The Evaluation of Forensic DNA Evidence; National Research Council, National Academy Press: Washington, DC, 1996, 75-88.

11. CBS News “Fingerprints: Infallible Evidence?” June 4, 2004. www.cbsnews.com. (accessed January 13, 2005).

12. Thompson, L. Secretary, Latent Print Certification Board, Shawnee Mission, KS. Personal communication, 2004.

13. International Association for Identification. Certified Latent Print Examiner Requirements, Section 3F (2005), www.theiai.org, (accessed January 13, 2005).

14. Message from the Latent Print Certification Board. J. For. Ident. 2005, 55 (6), 858.

15. Stacey, R. A Report on the Erroneous Finger print Individualization in the Madrid Train Bombing Case. J. For. Ident. 2004, 54 (6), 706-718.

16 Shirley McKie.com, Shirley McKie and David Asbury Official Reports, www.shirleymckie.com, (accessed January 13, 2005).

17. The Innocence Project. Case #141-Stephan Cowans, (2004), www.innocenceproject.org, (accessed January 13, 2005).

18. Cole, S. More Than Zero: Accounting for Error in Latent Print Identification. J. Crim. Law & Crim. 2005, 95 (3), 985-1078.

19. Dror, I. E.; Charlton, D.; Péron, A. E. Contextual Information Renders Experts Vulnerable to Making Erroneous Identifications. For. Sci. Int., in press.

20. Haber, R., Haber, L. Mindset in the Latent Print Comparison Process. Presented at IAI Educational Conference, Dallas, TX, August 9, 2005.

21. Saks, M. J.; Risinger, D. M.; Rosenthal, R.; Thompson W. C. Context effects in forensic science: A review and application of the science of science to crime laboratory practice in the United States. Science & Justice 2003, 43 (2), 77-90.

22. Saks, M.; Koehler, J. The Coming Paradigm Shift in Forensic Identification Science. Science August 2005, 309 (5736), 892-895.

23. Steele, L. The Defense Challenge to Fingerprints. Crim. Law Bulletin 2004, 40 (3), 213-240.

24. US v Plaza et al. 188 F. Supp. 2d, Daubert Hearing. Testimony of Ralph Haber.

25. Stacey, R. A Report on the Erroneous Finger print Individualization in the Madrid Train Bombing Case. Presented at IAI Educational Conference, Dallas, TX, August 9, 2005.

Appendix A

An analysis of 13 respondents’ errors from the 2002 CTS Latent Prints Examination (Test 02-516).

Three hundred and three (303) responding participants attempted to identify 10 latent prints (Item 5) to four sets of known exemplars (Items 1 through 4). In this examination, there were a total of 15 errors (reported by 13 participants). This equates to 4.3% of the participants making 1 or more errors. It should also be noted that Errors 2, 4, 7, and 11 all contain responses that are not possible, and these errors are most likely clerical in nature.


______________________________________________________________________
Feel free to pass The Detail along to other examiners.  This is a free newsletter FOR latent print examiners, BY latent print examiners. There are no copyrights on The Detail, and the website is open for all to visit.

If you have not yet signed up to receive the Weekly Detail in YOUR e-mail inbox, go ahead and join the list now so you don't miss out!  (To join this free e-mail newsletter, enter your name and e-mail address on the following page: http://www.clpex.com/Subscribe.htm  You will be sent a Confirmation e-mail... just click on the link in that e-mail, or paste it into an Internet Explorer address bar, and you are signed up!)  If you have problems receiving the Detail from a work e-mail address, there have been past issues with department e-mail filters considering the Detail as potential unsolicited e-mail.  Try subscribing from a home e-mail address or contact your IT department to allow e-mails from Topica.  Members may unsubscribe at any time.  If you have difficulties with the sign-up process or have been inadvertently removed from the list, e-mail me personally at kaseywertheim@aol.com and I will try to work things out.

Until next Monday morning, don't work too hard or too little.

Have a GREAT week!