Scientific Testimony: An Online Journal


The Need for "Blind" Procedures in Forensic Science

An Internet Discussion


Listed below are comments regarding the need for "blind" procedures in forensic testing that were made in early June 1998 on the internet discussion list, Forens-L. Readers are invited to add their own comments to the discussion. Send comments by e-mail to openforum@scientific.org.

Forens-L is an active list for people interested in forensic science. To subscribe to this list, send an email to mailserv@acc.fau.edu. In the body of the message type "subscribe Forens-L [your name]." (Without the quotation marks.)

The discussion was initiated when Brent Turvey posted an article about a psychological study. The study, by Gary Wells of Iowa State University, showed that feedback given to eyewitnesses about whether their identifications were accurate or not greatly affected their confidence. [Read the article about Well's research].


List Members:

The advice given regarding preventing the phenomenon observed by the researchers is the same as that given to unsophisticated law enforcement looking for quick answers who employ the use of psychics; don't let anyone who knows anything about the case near them to prevent information leakage and non-verbal reinforcement.

Brent

Brent E. Turvey, M.S. Forensic Science
Criminal Profiler
bturvey@corpus-delicti.com
http://www.corpus-delicti.com

"No one acts without motivation."
- Vernon Geberth, Practical Homicide Investigation, 3rd Ed.


Brent,

When Joe and I teach our Crime Scene Investigation course (www.NCIT.com) we show a series of evidence "matches", broken knives, paint comparisons, bullet and casing matches, etc. We have one that no one should ever miss - a simple physical match. However, we have found that 90% miss the identification when we:

1) Tell them that it's unfair-too difficult-they aren't experienced enough to make the comparison or

2) tell that it's too easy, no one should miss this comparison.

Feedback can work both ways - mind games can throw off even experienced investigators - many of whom have had considerable interview/interrogation schools and experience.

Jerry


Jerry,

Thanks for sharing that experience. Very interesting.

So how do you recommend that those who perform those types of physical matches avoid that specific pitfall? Is it as straightforward as maintaining a feedback/ information vacuum when performing certain kinds of analysis?

Brent


To the list:

Psychologists have been demonstrating for years that people's judgments about things they examine can be influenced by extraneous information that colors their expectations about what they are likely to see. So it should not be surprising that forensic scientists' evaluations of evidence can be influenced by what they are told about the case. The "bias" induced by such information is usually confirmatory: we look for and tend to find evidence confirming our expectations (sometimes even when it isn't there).

Brent Turvey wrote:

So how do you recommend that those who perform those types of physical matches avoid that specific pitfall? Is it as straightforward as maintaining a feedback/ information vacuum when performing certain kinds of analysis?

Answer: Yes, of course. Scientists in almost every other field use "blind" procedures whenever they need to interpret potentially ambiguous data. They know that "blind" procedures are important in order to avoid being misled by "examiner bias" - i.e., the tendency to mistakenly see what one wants or expects to see in ambiguous data.

In forensic science, by contrast, analysts almost never use blind procedures. Can anyone tell me why that is? I have heard some people argue that the failure of forensic scientists to use "blind" procedures when interpreting test results shows that forensic science procedures aren't really scientific. Is that a valid point?

William C. Thompson


William,

Thanks for the thoughtful response.

My answer would be that for certain type of examinations, which are meant to be more objective, the need for blind testing is absolute.

However there are also subjective analysis in the forensic sciences as well. One good example is the forensic pathologist's need for complete victim history in interpreting something about the nature of the interaction between the victim and the environment in such a a manner as to cause their death. If they want to interpret something as an autoerotic death, for example, they may look for a pattern of that behavior previously, dated pornography, reasonable expectation of privacy, etc... all of which may form a convergence of evidence suggesting autoerotic behavior caused their death, not suicide, leading to a manner of death of accidental. Very subjective process sometimes.

In my own work, I try to maintain objectivity in profiling cases by not getting to know any specific suspect information and by not getting to know and defendant information when hired by an attorney client (no interviews with suspects or defendants). That way I stick to interpreting the physical evidence at the crime scene and avoid the pitfall, as much a possible, of shaping the profile to fit a specific person.

Ironically, I was recently chastised by a prosecutor and a judge for not including suspect information in my analysis of the crime scene. If I had, they argued, it would have been more clear to me what type of offender actually was in operation at the scene. I tried to explain that it was improper to mold a profile to fit a specific person, but they viewed my need for blinded conditions as a fundamental flaw. I again, tried to explain that interviewing or including defendant information in a profile would have invalidated my analysis of the crime scene. Again, this was looked at as an oversight in common sense on my part.

Regardless, I have no intention of changing the way I make my analysis in defense cases.

Anyone else experience this sort of ignorant punishment from the court regarding the exclusion of subjective information from an analysis of any kind?

Brent


William C. Thompson wrote:

In forensic science, by contrast, analysts almost never use blind procedures. Can anyone tell me why that is? I have heard some people argue that the failure of forensic scientists to use "blind" procedures when interpreting test results shows that forensic science procedures aren't really scientific. Is that a valid point?

It's a semantic point — a bit like asking whether or not "Computer Science" is a science. In some definitions of science it is not. It's an engineering discipline.

It also depends on what you mean by "blind." Consider any of the image interpretation disciplines, such as radiology and surgical pathology in medicine or surveillance photograph interpretation in intelligence. If one defines a "blind" study as one in which no information were made available other than the image, then blind studies constitute malpractice. Any pathologist who, for instance, interprets a biopsy without getting the full patient history, impressions from the surgeon, etc. is asking for trouble. These are all important data which are part of the interpretation.

Thus, for any particular "science" within the forensic sciences, I would suggest that different levels of "blindness" are appropriate. If the data should be extraneous, but is taken into consideration, that's bad.

The interesting question may be not whether or not a person should interpret data blindly, but what constitutes extraneous information for different specialties.

billo


Dear Professor Thompson,

In DNA we now have partially automated genotyping using software. Experimental computer based expert systems are being formed to infer genotype from the genoscanner information. Comparison of post assignment samples is predominantly objective. I am comfortable with this level of objectivity in modern DNA evidence.

Passing to something else, the comparison of fired bullets involves placing the teat fired bullets on one side of the macroscope and the evidence one on the other. Significant correllation is sought. I struggle to see how this type of examination could be done blind.

This latter class would include (at least) firearms, toolmarks, paint and fibres. The former might include glass where we can measure the refractive index of the recovered glass before observing the control.

Einstein points out subjectivity in Physics. Persons were comparing the concept of a trajectory for sub-atomic particles with Heinenberg's uncertainty principle. How could a sub-atomic particle have a trajectory at all if its position and time could not be jointly determined. Einstein pointed out that a sub-atomic particle does not have a trajectory, we infer one from a series of bubbles in a bubble chamber. The subjectivity of the scientists "observing" the trajectory was exposed. Do you infer ffrom this that Physics is not a science or if you prefer that sub-atomic particles do not have a trajectory? Your choice?

You have been quite loose over the years with your use of the term bias. I think it must be about time you were a little more specific as to what you meant?

------------------------------

John Buckleton
Program in Statistical Genetics
Dept. of Statistics
North Carolina State Univ.
Raleigh, NC 27695-8203
buckle@statgen.ncsu.edu


billo wrote:

The interesting question may be not whether or not a person should interpret data blindly, but what constitutes extraneous information for different specialties.

Once the right question is asked, the data necessary for answering the question becomes fairly evident. It is asking the right question that is the trick - and that is what distinguishes scientists (the people who ask the questions) from the technicians (the people who gather the data).

Pete Barnett

Peter D. Barnett
Forensic Science Associates
Richmond CA


William C. Thompson wrote:

In forensic science, by contrast, analysts almost never use blind procedures. Can anyone tell me why that is?

A thought-provoking question. Examiner bias can promote injustice. But so can blinders.

I have in mind a DNA rape case. Evidence and reference samples were delivered to the lab. The analyst is un-blinded by the knowledge that the case is rape, that the suspect denies intercourse, etc, but isn't that necessary? The samples are precious, and so are time and tests.

The analyst extracted DNA through various means and ran some gels. Should these perhaps be handed to a second, blinded, person for analysis and interpretation? In a few cases that might be wise, but mostly it would at best avail nothing. After looking at several loci of data, the blinded analyst begins to form an hypothesis, inevitably tries to interpret the rest of the data in light of that hypothesis, and there's your bias anyway.

On the other hand, in many cases the interpretation of one locus -- suppose for example it indicates multiple contributors -- will necessarily and properly influence the interpretation of another. So it would be both impossible in practice and wrong in principle to analyze the gels blindly one locus at a time.

As luck would have it, the autorads were duly analyzed in a way such that several uncomfortable and inconsistent facts were swept under the table. I stipulate that examiner bias has occurred, but it didn't occur because of errors in blinding. It occurred because of human nature, namely the nature to expect the obvious and not to reckon on the unfamiliar, the incongruous, or the vexing. That's not even a human failing -- it just may have been the wrong response this time. The remedy, if there is a remedy, I believe is more likely to be training, diligence, and especially thoughtfulness -- than to be procedural regulations such as blindness.

To emphasize this last point, I ask you to consider what happened next. A spreadsheet containing the DNA types as interpreted by the analyst was shipped off to a statistician. The statistician, thus ignorant of the analysis steps and decisions that underlay the spreadsheet, duly produced blinded computations based on the hypothetical of the spreadsheet. Is that not a wonderful instance of blinding?

Perhaps not so wonderful. On the one hand, an opportunity to review the perhaps dodgy analytical decisions was thereby missed. And on the other, the statistician hadn't even a suspicion such as would motivate a boldly stated disclaimer:

Warning! The computations in this report are based on the spreadsheet, which may or may not reflect the real data of the case!

So the statistician's report -- which after all rests much more strongly on the reputation of its author than on internal logic which few among its audience can understand -- stands as a monument against some poor sucker against whom the real evidence is weak or flawed.

A long story, but perhaps a typical one. If I may draw a moral from my own tale, I see in it parallels with bureaucracy. No single step is wrong, but through a process of obfuscation, inefficiency, and compartmentalization, a wrong result is achieved. The moral that I am tempted to make, therefore, is that blind procedures in forensic analysis are not a panacea, not even a virtue, but a danger and an evil. To depend instead on the perspicacity and objectivity of the analyst may be a poor hope, but I don't see a preferable alternative.

Charles Brenner


buckle@brooks.statgen.ncsu.edu wrote:

Passing to something else, the comparison of fired bullets involves placing the teat fired bullets on one side of the macroscope and the evidence one on the other. Significant correllation is sought. I struggle to see how this type of examination could be done blind.

Is this as painful as it sounds?


William C. Thompson wrote:

In forensic science, by contrast, analysts almost never use blind procedures. Can anyone tell me why that is? I have heard some people argue that the failure of forensic scientists to use "blind" procedures when interpreting test results shows that forensic science procedures aren't really scientific. Is that a valid point?

I would like to second John Buckleton's reply to this statement that this is definitely not true for DNA analysis.

"In DNA we now have partially automated genotyping using software."

Even earlier manual DNA typing procedures in our lab required a second independent analyst to confirm results before any result was acceptable. Added to this all result typing is done in duplicate from source where possible. The only exceptions are minute samples and vaginal smears - even these include process duplication after initial extraction.

"Experimental computer based expert systems are being formed to infer genotype from the genoscanner information."

These expert systems are no longer experimental. We have implemented our own STRgazer expert system since November 1997. This continues the double result assignment requirement by requiring that the analyst assigned result be the same as that assigned by the STRgazer expert system. Without this consensus, results cannot be acccepted and samples must be re-typed from start.

If anybody is interested in obtaining an evaluation version of the STRlab DNA sample administration system, which includes the STRgazer expert system - biored@iafrica.com - attention Stewart Allen

DNA results are objective!!!!!!

Stewart Allen
Senior Forensic Analyst
SAPS, FSL
biored@iafrica.com


To the list:

I appreciate the many responses to my question about why forensic scientists fail to use "blind" procedures when interpreting test results. Here are some reactions to the answers that I received.

1. Blind scoring is impossible. One answer suggested by several writers is that blind scoring just isn't practical. John Buckleton, for example, says that he struggles to see how examinations of firearms, toolmarks, fibers, paint, etc. could be done blind. After all, he says, the analyst needs to know what he is looking at in the microscope. This answer misconstrues what it means to be "blind." No one is suggesting that the analyst be totally ignorant of what he or she is looking at. The suggestion is merely that the analyst be blind (i.e., uninformed) about facts that should not be considered when making the evaluation.

I once heard a forensic DNA analyst defend the scoring of an ambiguous autorad in a rape case by saying "that must be his [the defendant's] band, they found the victim's purse in his apartment." (or words to that effect). IMHO it is clearly improper for a DNA analyst to rely on evidence about the victim's purse to help resolve ambiguities that arise when scoring autorads. (Does anyone disagree with me on this point?) Yet current procedures at most laboratories allow the analyst to be exposed to extraneous information of this sort when scoring autorads, making microscopic comparisons, and interpreting other analytical tests on physical evidence. In my view, current procedures invite "examiner bias."

The term bias, as used by psychologists, and as I use it here, does not connote any maliciousness or intent to distort. One can be biased without being aware of it. Indeed, the phenomenon of "examiner bias" is dangerous precisely because we are not aware of the extent to which our perceptions may be colored by extraneous information. [I hope this answers your question about defintions, John].

Surely it is possible to make analysts blind (at least temporarily) to most information that they should not consider. Some of the necessary elements of blindness could be achieved by a separation of function within the forensic laboratory. One person could receive the evidence samples, communicate with the authorities, determine what items should be tested, with which procedures, and in what order. This first person could and should know all about the case. A second person could carry out the tests, make subjective comparisons of samples, and interpret the test results.

Under this procedure, the second person could be blind (uninformed) about many extraneous aspects of the case, such as the fact that the victim's purse was found in the apartment of suspect A, that suspect A has a criminal record, or that the victim identified suspect A in a line up. In some cases, additional steps could be taken to prevent bias arising from information internal to the test. If I ran a forensic DNA lab, for example, I would make forensic DNA analysts determine what alleles are present in evidentiary samples (particularly mixed samples) before they know the genotypes of the reference samples. This procedure would avoid the kind of bias Charles Brenner mentioned that might arise from the tendency to "see" the suspect's pattern in an ambiguous evidentiary profile.

Once the analytical tests are interpreted (and the results recorded in some permanent fashion) the results could be communicated back to the first person, who knows the big picture. The first person could think about what the test results mean in the context of the whole case, what additional testing might be necessary, and so on. The question in my mind is not whether blind procedures of this type are possible, but why forensic scientists have failed to adopt them.

2. Blind scoring is "dangerous" and "evil." Charles Brenner suggests that blind scoring may do more harm than good. As I understand it, he thinks that the quest for "blindness" may prevent forensic scientists from having the broad view of a case needed to reach sensible conclusions. I think this argument rests on the faulty assumption that everyone in the lab who deals with a case must be blind. As I have just explained, it is possible to "blind" the analysts who read autorads or make microscopic comparisons without blinding the people who communicated with detectives, decide what comparisons are necessary, and consider what the results of the analytical tests might mean in a particular case. I fail to see any danger or evil in such a division of responsibility. Am I missing something, Charles?

3. Physicians don't use blind procedures, why should we? Several people argued that it would be malpractice for a physician to interpret laboratory test results without knowledge of the patient's clinical history and therefore, by analogy, that it would be inappropriate for a forensic scientist to interpret the results of analytical tests on physical evidence without knowing the history of the case.

This analogy is inapt and misleading. Physicians are often asked to make determinations on some ultimate issue, such as the diagnosis of a disease. Because the physician makes the ultimate determination, the physician can and should consider all relevant information. Forensic scientists, by contrast, usually are not asked to determine an ultimate issue (e.g., whether the defendant is guilty or not) but merely to provide factual information that will help someone else (e.g., a judge or jury) determine the ultimate issue.

Because forensic scientists do not decide ultimate issues, they do not need to consider all of the information in the case. Indeed, it is wrong for them to do so because the information they provide to the ultimate decision maker is supposed to be independent of other facts in the case. Failure to use blind procedures can easily lead to what Bayesians call double-counting of evidence, which is unfair to defendants.

A final note on physicians is that they are not immune to "examiner bias." A number of studies have found that physicians' failure to used blind procedures for interpreting laboratory tests is associated with high rates of false positive errors. For example, one study found a "clinical information bias" in the interpretation of echocardiograms. Physicians who knew the patients clinical history made a lot more errors when interpreting the echocardiograms than those who did not. The errors were false positives resulting from a tendency to mistakenly interpret the test results in a manner consistent with the clinical history (Tape & Panzer, Echocardiography, Endocarditis, and Clinical Information Bias, J.Gen. Intern. Med. 1, 300-304, 1986). The authors say "Clinical information may also bias other imaging tests, including radionuclide scans, computed tomography, noncardiac ultrasound, magnetic resonance imaging, and conventional radiography." So it is by no means clear that the practices of physicians in this area should serve as a model for forensic science.

4. We already use blind procedures. A final response is that blind procedures are not really necessary because forensic tests are objective. John Buckleton and Stuart Allen proudly announce that "DNA results are objective!!!!!!" (exclamation points in original). Very good, if true. But it appears that they are talking only about STR tests and only about that subset of STR tests that is scored by computer using expert systems. The interpretation of PM/DQ-alpha tests certainly requires subjective judgment, and subjective judgment sometimes plays a crucial role in the interpretation of RFLP test results as well (see W. Thompson, Subjective interpretation, laboratory error and the value of forensic DNA evidence: three case studies, Genetica 96: 153-168, 1995). Beyond DNA testing, there are a great many analytical tests on physical evidence that require subjective judgment. Why aren't these procedures done blind?

Returning to DNA for a moment, I must say that I have doubts about whether the use of "partially automated" genotyping software truly makes the scoring of DNA tests objective. The programs of this type that I have seen allow the operator to override the machine's scoring determinations at will on the basis of subjective judgment (e.g., to eliminate what the analyst deems to be "stutter bands"). As a result, conclusions that purport to be objective can, on close examination, turn out to be highly subjective (and wrong). To make matters worse, some labs fail to document operator overrides, thereby hiding the fact that subjective evaluation played a role in the outcome of the test.

My doubts arise partly from my experience litigating an RFLP case two years ago in which the laboratory analyst claimed that the results were scored "objectively" because a BioImage machine was used to perform band sizings. Because I doubted the claim of objectivity, I obtained a court order requiring the laboratory to re-score the autorads while I and a defense expert looked on. This re-scoring failed to replicate the results originally reported. It appeared that in order to produce the original "objective" results the analyst had used operator overrides to delete a large number of "bands" that did not match my client from the vaginal lane, leaving only bands that matched my client. How did he know which bands to delete? He said he could "tell by looking" which bands were true bands and which were not. So much for objectivity!!!

To make matters worse, the re-scoring produced different band sizings than were initially reported. The new sizings appeared to exclude rather than include my client as a source of some of the vaginal bands. Anyone who is interested in this case can read a detailed account (and see the autorads) in the online journal Scientific Testimony. [Editor's Note: See the discussion of People v. Marshall in CASE IN POINT]

To summarize an overly long message (I do get carried away when I start typing), I have yet to see a convincing reason offered for the widespread failure of forensic scientists to use blind procedures for interpreting the results of analytical tests on physical evidence. I believe that the failure to use blind procedures when making subjective evaluations of potentially ambiguous test results is a poor scientific practice that invites error. Given the importance of these test results in legal proceedings, I think that more rigorous procedures are called for. If forensic scientists do not take steps on their own to improve their analytical rigor, contraints may eventually be imposed from outside the field.

One final thought. In my opinion many of the scientific atrocities documented by John Kelly and Phillip Wearne in their book Tainting Evidence could have been avoided had the FBI laboratory followed the "blind" procedures that I have suggested. Those members of this list who are interested in reading books rather than burning them would be well advised to read Kelly and Wearne.

William C. Thompson


To the list:

I appreciate the statement by Prof. Thompson for he clarified many of the issues concerning examiner bias. Furthermore his publication of "Examiner Bias in Forensic RFLP Analysis" and "People v. Marshall: The Legal Story" (more fascinating reading than fiction) (www.scientific.org); are educational and illustrative of problems found in the interpretation and reporting of conclusions for use in court. Certainly bias plays a strong role in divergent opinions in courtrooms. I suspect, nay I am sure, that bias based upon who pays the piper plays a greater adverse effect upon truth than subtleties of psychologically associated subconscious bias. As for the forensic pathologist, certainly the autopsy must be interpreted in the light of the circumstances. Not who did it but what happened. Leave the who to the police. Anyway, more case examples such as People v. Marshall: The Legal Story should be published. Nothing like letting the rest of us learn, as Paul Harvey states, "The other side of the story." Thanks to Bill Thompson.

Joseph H. Davis, M.D.


What a pleasure to read the recent exchanges on forens-L:

I wanted to add a short personal comment on the question of the blindness of the expert advocated by Prof. Thompson. Let us consider the opportunity of blindness at two different stages:

(1) at the analysis stage

(2) at the interpretation stage

(1) It is the wish of every scientist to avoid all biais at this stage. This temptation to suppress the subjectivity during the analysis of the data (what concordances and discordances do we have between a questionned and a known sample?) is very strongly anchored in the heart of forensic scientists. It can be possible for some type of evidence (DNA, drug testing, fire residue analysis, etc.) because information can be gathered directly from the traces (without the control): these evidence are standing alone, we know what you are looking for. For other types of evidence, when the proof mostly derives from a visual comparison between knowns and unknowns (firearms, fingerprints, shoeprints, document examinations, etc.) it is very difficult (or impossible, perhaps counterproductive) to avoid the so-called biais provoked by the presence of the control material in examination. But, this quest for objectivity fails to recognize this subjective nature of science. This quest for objectivity in the analytical or visual determination is an illusion. Every observation is conditioned upon prior knowledge, then every observation is subjective by nature. The pure blindness does not exist, so we had better to improve the scientist to see relevant features (by giving him adequate prior knowledge and procedures for inference) than to improve his blindness. Bertillon a century ago was saying (roughly) concerning crime scene investigation: "we see what we have in mind and we have in mind only what we are thinking about". Nobody would call (I hope) for a total blind approach of the crime scene for the forensic scientist.

(2) To interpret the data, the scientist need to know the questions (as Peter Barnett indicated). Consequently, the total ignorance of the context is impossible. What the scientist has to do is to try as much a possible to keep the questions as the hypothesis under which the evidence will be assessed and not the hypothesis to demonstrate. In other words, the scientist must give the value of the evidence given the hypotheses and not the inverse. Blindness is not the good tool to avoid transposing the conditional. Education is better.

I hope to read more on the topic

Regards

Christophe

_______________________________________

Christophe Champod, Ph.D.
christophe.champod@ipsc.unil.ch
Professeur assistant
Institut de police scientifique et de criminologie
[School of Forensic Science]
University of Lausanne
BCH
CH-1015 Lausanne-Dorigny (Switzerland)
tel. (+4121) 692 46 29
fax. (+4121) 692 46 05


I was most interested in Christophe Champod's quote of Bertillon-"We see what we have in mind..and have in mind only what we are thinking..". Bertillon went on to prove his own point by refusing to accept the validity of finger-print identification, and by declining to consider exculpatory evidence in the Dreyfus case.Maybe there is a moral here.

EJ Wagner


EJ Wagner wrote:

I was most interested in Christophe Champod's quote of Bertillon - "We see what we have in mind..and have in mind only what we are thinking..". Bertillon went on to prove his own point by refusing to accept the validity of finger-print identification

And now, contrary to Bertillon's initial skepticism, we all believe fingerprint indentification to be absolute. All you need is 6, or is it 12, (or 16, or does the number really matter) points of identification, or are they points of similarity, or minutae, or corresponding minutae. And, of course, no dissimilarities, or, at least, unexplainable dissimilarities (the ability to explain dissimilarities is, I guess, a function of experience). And if you don't like minutae, you can always use ridgeology.

We all hope the DNA identification techniques will, in a few years, progress to such a high level of certainty.

In the meantime, and while waiting for the wrinkles in all of this DNA stuff to be worked out, I still like microscopy.

Pete Barnett

Peter D. Barnett
Forensic Science Associates
Richmond CA


William C Thompson gives us his final word. Actually four pages on my screen. I appreciate this brief summary. Some balance had come into his argument. But I remind the list that Professor ThoMpson's two opening tenents were:

subjective = bias AND subjective = non-scientific.

There is much that could be said about the detail of his reply, in fact there is much reply. However it is good to see him abandon so charitably his (wrong) opening hypotheses.

-----------------------------

John Buckleton
buckle@statgen.ncsu.edu


John Buckleton writes:

I remind the list that Professor Thonpson's two opening tenents were: subjective = bias AND subjective = non-scientific.

John,

I never set forth any "opening tenents" and if I had they would not have been the ones you attribute to me. Your misreading of my position illustrates the very point that we are discussing. People tend to interpret evidence in a manner consistent with their expectations. Perhaps you expected me to say something unreasonable? Had you been "blind" to the identity of the author I suspect you would have given my words a more charitable (and accurate) interpretation.

It is not without reason that academic journals use blind review of manuscripts. Indeed, blind procedures are commonly used in academic science whenever important determinations rest on subjective judgments. In my initial posting I asked why blind procedures are not used more commonly in forensic science. I STILL haven't heard any good answers to this question.

I did not equate subjective judgment per se with bias or with the non-scientific. What I suggested was that the failure to use blind procedures when making subjective judgments is conducive to bias and is therefore a bad scientific practice. As far as I can tell, most people agree with this suggestion. What we should now discuss is how best to implement blind procedures in forensic labs. I have given you my thoughts. What are yours?

William C. Thompson


Bill Thompson wrote:

Charles Brenner suggests that blind scoring may do more harm than good.

A tough and subtle thesis to defend! What I actually said was:

"In a few cases that might be wise, but mostly it would at best avail nothing."

in the context of blindly scoring autorads in mixed stain situations. You may have noted my concern about "more harm than good" when I discussed the more general idea of bureaucratizing and compartmentalizing the lab. Perhaps you then juxtaposed two parts of my message.

The point is not crucial, but were it not for the good nature for which I am justly famous I might wonder if you are the same person who lately chided Buckleton for misreading a message.

As regards blind scoring I have, on occasion, taken your position against the "interactive" (i.e. non-blind) way a multilocus pattern was scored.

Those members of this list who are interested in reading books rather than burning them

Do the book-burners know who they are? I don't understand this comment.

I fail to see any danger or evil in such a division of responsibility. Am I missing something, Charles?

Yes, quite a lot.

I gave an example already, about a statistical analysis. Was the point not clear?

Some of the contributors to this subject—like Turvey, Davis, Barnett, Champod, and Oliver — present points of view that come across, at least to me, as the persuasive and unassailable kind of common sense that appears to be born from intelligent reflection on long personal experience.

What their varied responses say to me, and what my own story I thought illustrated, is that the issue is complex and fraught with subtlety.

You presented your idea for discussion a few days ago, stimulated an excellent discussion, and now apparently unmoved by anything anyone has said propose that no one has disabused you and that we therefore move ahead to the next step. That bespeaks an agenda more than a discussion.

Charles Brenner


Charles Brenner writes:

You presented your idea for discussion a few days ago, stimulated an excellent discussion, and now apparently unmoved by anything anyone has said propose that no one has disabused you and that we therefore move ahead to the next step. That bespeaks an agenda more than a discussion.

Charles,

I am happy to continue the discussion. My proposal to move to "the next step" was just an effort to introduce a new topic. I did not mean to cut off discussion of the old one. If my previous comment gave this impression, it was simply a poor choice of words.

Let me begin by explaining further why I am unmoved (as yet) by various arguments that have been offered in defense of non-blind procedures.

You argued that blind procedures "would avail nothing" because "[a]fter looking at several loci of data, the blinded analyst begins to form an hypothesis, inevitably tries to interpret the rest of the data in light of that hypothesis, and there's your bias anyway." But in the situation you posit the analyst is not entirely blind to potentially biasing information. The bias you describe could be eliminated if the laboratory typed each sample separately in the manner described by Stuart Allen. In that situation the analyst has no basis for forming a hypothesis so the potential bias is eliminated. Making analysts type the most problematic samples (such as mixtures) before knowing the types of suspects would also be a good approach. I accept your point that half-baked efforts to "blind" analysts may not be entirely effective. But the fact that some procedures for making analysts "blind" will not work perfectly is hardly a persuasive argument against blind procedures in general, particularly when effective procedures do exist.

Christophe Champod makes the "post-modernist" argument that blind procedures are pointless because perfect objectivity is an illusion:

But, this quest for objectivity fails to recognize this subjective nature of science. This quest for objectivity in the analytical or visual determination is an illusion. Every observation is conditioned upon prior knowledge, then every observation is subjective by nature. The pure blindness does not exist, so we had better to improve the scientist to see relevant features (by giving him adequate prior knowledge and procedures for inference) than to improve his blindness.

From a psychological perspective it is true that every observation is conditioned upon (and affected by) prior knowledge, as Champod suggests. But that is the very reason that blind procedures are appropriate. We can all agree (I hope) that some types of prior knowledge should influence an analyst's judgments and other types of prior knowledge should not. Knowledge that the victim's purse was found in the suspect's apartment, for example, falls in the latter category. So how do we deal with the fact that analysts may come to know certain facts that should not influence their judgment? Champod argues that the best way is to educate the analyst to recognize which types of information should and should not influence the judgment. I believe that this approach is psychologically naive. People are not very good at monitoring and controling their own thought processes and often lack insight into the factors that influence their own judgments. A great deal of research has been done on "intentional forgetting" (people's ability to disregard known facts when making judgments). The general finding is that people are not very good at it and that training and instruction does little to help. Thus, the analyst who hears about the purse in the apartment is likely to be influenced by that information even if he knows that he should not be. I am not against educating analysts about what should and should not be considered when analyzing evidence (in fact, my impression is that instruction of that type is desperately needed). But I am not convinced that instruction of that type will substitute for the use of appropriately blind analytic procedures. If Champod were right, there would be no need to use blind procedures in academic science. The fact that blind procedures are widely used in academic science is testiment to the fact that most scientists reject his position.

As far as I can tell, I have now (in this posting or previous postings) addressed every argument that has been advanced against blind procedures. If anyone thinks I have missed important points, or that my arguments are unpersuasive, I hope they will say why. I find this discussion interesting on many levels and hope that it continues.

William C. Thompson

P.S. My comment about burning books rather than reading them was a reference to Ted Mozer's posting on May 16th about burning the Kelly and Wearne book. You probably forgot it, but it certainly stuck in my mind.


Bill

The "blind" situation that you report is exemplified by the incoming letter in the Boots and Proctor case that came up recently on this list. The incoming letter states that:

"As per our phone conversation of March 6, 1986 I am submitting the partially burned flakes of double base powder out of our Oliver homicide.

This is a murder case that took place in June 1983. The killer or killers entered a local 7-11 store in the late evening hours and forced the young male clerk into the back room (cooler) and broke a full 10 ounce bottle of Orange Crush over his head and then shot him in the head three times with a .22 caliber weapon (probably a Hi-Standard revolver.) Due to some interagency problems the case to date has not been prosecuted, but will be soon.

Going through the trace evidence, some of which had been analyzed by SEM-EDAX, I found a partially burned double base powder flake on one of the planchets. The flake was originally found on the trousers of one of our suspects. We want, if possible for you or Ed to compare this flake (B) to some partially burned flakes (A) found on the body of our victim. The only difference between the treatment of the flakes is that flake B has been carbon coated to prepare it for SEM work.

Exhibits

Both A and B are sandwiched between the glass slides and clearly circled and labeled. (I have tried to get them to move by tapping the slide but they appear to be stationary.)

(Sample A) Several partially burned flakes of double base powder from the victim.

(Sample B) One piece of partially burned flake of double base powder from the trousers of a suspect.

Request

If possible, please compare A to B.

Time is of the essence now because of a lawsuit one of the suspects is bringing against the police department for false arrest.

I would appreciate any help you can give. Thank you very much."

For those of you who missed the thread about Boots and Proctor, those two fellows spent eight years in prison for that homicide before the individual who actually committed the murder admitted and was determined to have actually committed it. Boots and Proctor were freed and sued again and won $2 million for their troubles.

Fred Whitehurst


An interesting case I wanted to share-related to facts and conclusions. At a murder scene, a pair of bloody, relatively new gym shoes, size 45, along with matching bloody shoe impression's were found. The victim's shoes were size 42. Everyone at the scene said that the murderer must be a large person, or at least someone with a large foot. The murderer was eventually apprehended. I saw the video confession and noted that the suspect was not particularly of large dimensions. It turned out that his shoe size was 43.He was questioned regarding the size 45 shoes that he had left at the crime scene. He explained that when he went to buy shoes, he had seen this pair that he really liked but the store had them only in size 45, so he bought them anyway.When his shoes were bloodied at the crime scene, he left them there and left the scene wearing a pair of the victim's size 2 shoes.


As long as we relating interesting cases about shoe sizes; I had a case once in which a man killed a rural mailman who delivered during the day. He wore his wife's shoes, 2 sizes (US sizes) too small when he did it. The mail box was in a sandy area. Since his wife was working in town at the time of the murder, he figured she would not be suspected and since only her shoetracks were there, neither would he.

You also have to be careful about drawing conclusions relating height to shoe size even when the individual is wearing the correct size shoe. When I was in college, all males had to participate in ROTC. Trying to get them to line up in a straight row was a challenge for some of them so I took it in steps. One, line up their feet and then two, get them to stand up straight. My platoon were all six feet tall +/- inch. One man's toes stuck out in front of the rest by almost two inches so I walked behind him to get him to step back but his heels were in line. When his mama told him to keep his feet on the ground he had plenty of feet to do it with.


billo wrote:

"If the data should be extraneous but taken into consideration, that's bad."

He is correct. Psychologists who have studied medical diagnostic thinking processes, and that includes forensic pathology opinions, have noted that the most common error in medical diagnosis is over reliance upon non-contributory data (Elstein, Shulman and Sprafka, Medical Problem Solving:An Analysis of Clinical Reasoning, Harvard University Press, Cambridge, MA, 1978). The forensic pathologist who interprets the autopsy findings absent correlation with circumstances is a danger to the community.

Joseph H. Davis, M.D.


If by "forensic pathologist" you mean a physician who runs an autopsy on a body, notes the input from the authorities and proceeds to determine the cause and manner of death, I do not agree.

When a pathologist can determine the certain cause of death based strictly on what she sees on the table and in the associated laboratory analyses, there is no problem. But if all she knows comes from the table, the lab, and the authorities statements, it is, in my opinion, unacceptable for her to declare the manner of death or even an overly suggestive cause. Her opinion in this matter is based in part on hearsay.

The M.D. or O.D. who does not become involved with all the evidentiary aspects of a case is no more qualified to render an unequivocal opinion of homicide than is any other scientist or technician who has not seen all the evidence in a case. Unfortunately, the state seems to pine for the closure of "Manner of death: homicide" — even though the autopsy report is usually written before the police investigation gets into full swing and long before there is any input from the yet-to-be-named defendant.

I suggest that if there simply has to be some comment from the pathologist about how the knife came to be in the victim's back, it be limited to "PRESUMED Manner of death." This procedure would mitigate the effect of the circular reasoning associated with the DA telling the doctor how it happened and then relying on her opinion and the prestige of the M.D. title to sway the jury.

Jerry


Jerry,

When you go to the doctor you give a history. The diagnosis is not based upon a physical examination plus laboratory tests alone. Why? Because laboratory test results and some physical examination findings may be representative of multiple causes.

Example: An airline pilot developed cardiac rhythm disturbances. His job may be lost over this. Is it truly organic heart disease in which case he is a danger to the public who are his passengers and therefore he must lose his job? Physical examination plus laboratory tests can not guarantee absence of organic heart disease. A history is essential in the diagnostic process.

Circumstances: He did not drink coffee but drank numerous Coca-Colas during the day. When he was told to stop drinking the beverage, the heart rhythm problems ceased. Why? Caffeine is the stimulant in Coca-Cola. (True story).

Pathologist: " I found myocardial scarring plus a narrowed coronary artery. Therefore the cause of death is heart disease."

Wrong! In the first place, that heart disease found at autopsy was there last week before he died. Yes, it may be the clue that the terminal episode was ventricular fibrillation due to heart disease but was it?

Answer: It was the defective electric lawn edger he was using when he died. Over half of 110 volt appliance electrocutions leave no burn marks or any other clues on or in the body as to what really happened. Only a very few deaths are solely autopsy dependent in order to make a proper diagnosis.

Question for all who utilize autopsy derived information: Did he die of the autopsy finding or only with the autopsy finding?

Answer: Only the circumstatntial investigation shall determine the truth.

Manner of death? The death certificate must be signed and in a timely manner. It is an instrument for Health Department vital statistics purposes. Almost all death certificates are signed by physicians.

Manner is going to be one of four: Natural, Accident, Suicide or Homicide (Undetermined means only means that you have not yet figured out which of the four it should be). The entered manner is "presumptive" which is suitable for vital statistics purposes. The medical examiner relies upon police and other information sources for this purpose. In Florida the certificate form is pre-printed with "Probable" after which the physician certifier places the manner. The death certificate is not the determiner of what develops in court. The court uses the original facts plus more to make its determinations. Physicians testify in court as to their opinions whch are based upon their knowledgeof the patient, knowledge which is derived from records and information created by others (an exception to the hearsay exclusion rule--otherwise no more testimony by physicians in court!).

I could continue but that became a four hour presentation at our thrice yearly weeklong course on police/medical investigation of death. Space at this site does not permit.

Joseph H. Davis, M.D.


List Members:

To add just a little more fuel to the discussion, I was directed to an article in the Atlanta Constitution this morning that is extremely relevant to our conversation regarding witness suspect "Positive" ID's.

There is a school of thought within law enforcement and prosecutorial bodies (how pervasive is anyone's guess as we can each only speak to our experiences on the subject) that once a known criminal has been arrested for a crime, it harms everyone to let those charges go away. Because, after all, even if they are not guilty of the crime at hand, they are guilty of something (how often do we hear that?). So kudos to the federal prosecutors for being willing to drop the charges.

*****************************************************

June 5, 1998

Suspect's jail time saves him

Steve Brown was freed from a jail term for robbery because his lawyer, Natasha Silas, discovered that he was already in jail when he was accused of robbing two convenience stores.

By Bill Rankin, The Atlanta Journal-Constitution

It appeared certain that Steve Brown was headed for prison. The only question was for how many decades.

Two clerks picked his face out of a 12-photo lineup provided by Atlanta police, who arrested Brown for two armed robberies of a Chevron Food Mart near the Lindbergh MARTA station. If convicted at trial, scheduled to begin June 15, Brown faced 30 to 60 years in prison without parole.

But on Wednesday, federal prosecutors dismissed the charges after being told Brown could not have committed the Oct. 1 and Oct. 19 armed robberies because jail records show he was in the DeKalb County Jail during the first robbery and in the Atlanta Jail during the second.

"I feel great," said Brown, 32, wide-eyed and beaming, but still a bit unsteady on his feet with his newfound freedom. "It's by the grace of God it turned out this way."

Since his Nov. 8 arrest, Brown has been in custody. Federal defender Natasha Perdew Silas said that from the moment she first met him, Brown insisted he did not commit the crimes.

Silas discovered that while Brown had more than a dozen minor shoplifting convictions, he never used a weapon. But she knew two victims identified him, and a surveillance videotape showed a robber who resembled Brown.

"I told him I was scared for him," Silas said. "The evidence looked very strong."

One of the Chevron clerks still wants to testify. "He's the guy and I know it," Paul Rovick said Thursday. "It was a positive identification."

Rovick has been robbed at gunpoint five times and wanted vindication. "I think he slipped out of jail somehow," he said. "I'm really upset about this."

But Silas obtained mug shots taken when Brown entered the jails and logs showing when he was released. Silas questioned the methods used with the photo lineup. Of the 12 photos, she said, only Brown's had a distinctive receding hairline like that described of the actual robber.

"Eyewitness identifications are often unreliable," Emory University research psychologist Steven Cole said. "First, there is the normal fallibility of perception and memory and, second, there is the mind's susceptibility to suggestive influences."

A 1997 study by the National Science Foundation found that more than 75,000 people become U.S. crime suspects each year based on identifications from lineups and photo spreads. But in 24 of a sample of 28 cases involving suspects released from prison based on DNA evidence proving their innocence, eyewitnesses made false identifications from photos and lineups, the NSF study found.

"I want people in Atlanta to know that I could have been convicted for something I didn't do," Brown said. "There are people in jail right now serving time for crimes they didn't commit."

Remarkably, Brown, a Chicago native who has lived in Atlanta more than 10 years, was not upset for being incarcerated almost seven months on charges that were dismissed. "I've been living in the fast lane . . . doing wrong," he said. "I think God was sending me a message that he wanted me to stop this foolishness."

Copyright 1998 Cox Interactive Media

*****************************************************

Brent E. Turvey, M.S. Forensic Science
bturvey@corpus-delicti.com


William C. Thompson wrote:

4. We already use blind procedures. A final response is that blind procedures are not really necessary because forensic tests are objective. John Buckleton and Stuart Allen proudly announce that "DNA results are objective!!!!!!" [exclamation points in original]. Very good, if true. But it appears that they are talking only about STR tests and only about that subset of STR tests that is scored by computer using expert systems. The interpretation of PM/DQ-alpha tests certainly requires subjective judgment, and subjective judgment sometimes plays a crucial role in the interpretation of RFLP test results as well.

You clearly were blind when you read my reply. The basis of current computerised systems stems from the way that DNA results should be read for manual procedures. That is each result read independently and verified by a second independent analyst. Therefore if HLA/PM and RFLP results are read in this way (the standard approach always followed by our LAB) then these results are objective. I do agree that incorrect procedure was followed in the examples you note, but please do not generalise the forensic community based on a few bad eggs.

Returning to DNA for a moment, I must say that I have doubts about whether the use of "partially automated" genotyping software truly makes the scoring of DNA tests objective. The programs of this type that I have seen allow the operator to override the machine's scoring determinations at will on the basis of subjective judgment etc...............

Any system which allows analyst flexibility will allow "manipulation". The only way to prevent biased actions is as I have already stated:

"The basis of current computerised systems stems from the way that DNA results should be read for manual procedures. That is each result read independently and verified by a second independent analyst."

The STRgazer expert system which we employ requires that two independent (blind) result assignments must concur in order for a result to be accepted. Additionally all modifications are automatically logged.

You are right in stating that blind procedures are necessary to ensure objectivity - but please don't generalise!

Stewart Allen


Stewart Allen writes:

Any system which allows analyst flexibility will allow "manipulation". The only way to prevent biased actions is as I have already stated: <snip>

That is each result read independently and verified by a second independent analyst."
The STRgazer expert system which we employ requires that two independent (blind) result assignments must concur in order for a result to be accepted. Additionally all modifications are automatically logged.

You are right in stating that blind procedures are necessary to ensure objectivity - but please don't generalise!

Thank you for supporting my argument for the necessity of blind procedures to assure objectivity. The procedures that you describe for scoring STR results do sound rigorous and appropriate. They are exactly the kind of procedures I had in mind when urging that forensic laboratories interpret test results in a "blind" manner. Your comments show that it is quite possible for forensic laboratories to adopt "blind" interpretive procedures IF THEY CHOOSE TO DO SO. The question I am raising is why many forensic laboratories have chosen NOT to adopt blind procedures when, as you say, blind procedures are "necessary to achieve objectivity."

I do agree that incorrect procedure was followed in the examples you note, but please do not generalise the forensic community based on a few bad eggs.

You are mistaken if you think that blind procedures like those that you have adopted in your laboratory are widely used in forensic science. I believe that the majority of the forensic DNA laboratories in the U.S. make little or no effort to "blind" the analysts who interpret test strips or read autorads. I have examined many cases from major laboratories in which it was clear that the analyst who made critical interpretive judgments (e.g., the scoring of messy results from mixed evidentiary samples) not only knew the genotypes of the suspects but also knew many extraneous facts about the case. Indeed, I wonder whether there are ANY forensic DNA laboratories (other than yours) that have adopted the kind of blind procedures that you describe. If so, I would like to hear about it. I also believe that blind procedures are rarely used for trace evidence analysis. I hope that people will correct me if I am wrong, but my impression is that in the U.S. the "bad eggs" (as you use that term) are in the majority and that "good eggs" like your laboratory are few and far between. Again, if I am wrong on this point I hope that people will tell me.

I think it will help the discussion of blind procedures if the "good eggs" identify themselves and describe what steps they take to "blind" their analysts. I hope that the "bad eggs" out there (you now know who you are) will at least give some thought to whether more rigorous procedures might be in order.

William C. Thompson


To those outside of the USA I apologize for taking up the bandwidth, but the following is intended to those who continually mock the USA legal system with their negative attitude.

I cannot speak for others on this list nor will I try, but I'm a little irritated. EVERYTIME I download my mail from this list I get a whole lot of gripping about how bad things are, Brent and others seem to enjoy finding those little tid bits of information about "bad science", "bad analysts", "bad forensic science" and shame on the FBI. I have a few questions to these folks, "Is there another legal system in the world that is better then ours, as a whole?" Sure we have some problems. But I can think of a few countries that you have no freedom, no justice, and no "rights". Bad Science? How about none at all!

I have no complaint about finding a problem in our justice system and identifying the cause & solution. But day after day after day to have nothing to say but shame on them....shame on you! How many times has there been "Good Science" that has helped the prosecution or defensive in arriving at the truth? I'm sure more times then what is identified on this list.

I joined this list to exchange ideas, learn new information and to have open discussions with my peers. Yet lately all I'm seeing is this negativism about our legal system. I recently asked questions about DNA. The answers I received I thought were from 3rd graders, except for their signatures. Some of you may think the replies you sent were humorous, I didn't....... as others didn't, who e-mailed my privately about the "morons". We have lost some input from many of the members of this list, some have left while others are just there in the shadows, waiting and hoping the list reverses itself.

Please, lets get back to the healthy discussions about science, physical evidence, procedures, techniques and methodology. I really enjoyed the "old" forens-L list, this one has been taken over by negativism and I for one have had enough. Yes I know what a delete button is and I use it too often!

My two cents worth. IMHO

Hayden B. Baldwin
Forensic Enterprises, Inc.
http://www.feinc.net
hbaldwin@interaccess.com


In our DNA laboratory, we independently score the DNA typing strips/ gels with a numbering system. Another analyst co-reads the results and if there are any disagreements (major/minor source) a third individual is present to break the tie, or to discuss the possibility of calling the result inconclusive.

In all cases, we run the evidentiary profiles prior to running the controls for the suspect (s) or victim (s) , (unless evidence comes to the surface after the controls are run), interpret the results of the evidentiary exhibits and then run and call the reference samples. Once the scoring is done the puzzle is pieced together. Interpretation of mixtures of DNA can be difficult depending on the source of the sample. This is where careful forensic interpretation in the context of the case is necessary. For example in a sexual assault situation, you may have a mixture in the sperm fraction that is indicative of more than one individual. The victim for certain may be a source since the evidence may have been collected from vaginal swabbings. Therefore this victim's profile must be interpreted in the mixture context. Secondly, it may be informative to establish if the victim had any reason to have multiple semen sources at the time of the incident, (i.e. prior conceptual intercourse in close temporal relation to the attack etc...) In the case of lack of consensual evidence it may be proper to interpret the mixture as having arrived from a single assumed semen source (from the attack) that is mixed with the female's DNA. I think the initial results in a case like this would normally be run with the epithelial fraction DNA from the vaginal source and this should type as victim, (although some times it may not for good reason). Also it is important to interpret the profile with as much information as possible. Both test strips and gels/ pherograms should be used in concert to arrive at the conclusions. I am sure that Stewart has run into mixtures in his work, and I know he has to interpret these results objectively ( the scoring) however whether individuals are included or excluded as potential sources of DNA to the mixture must include some case facts. I am of the opinion that if any results are to be drawn from a mixture that one should state up front the assumptions that they are interpreting under, ( i.e an assumed single semen source), this allows both the prosecution and defence to challenge/defend those assumptions and by extension the DNA results.

Greggory S. LaBerge
Denver Police Department Crime Laboratory Bureau
University of Colorado Health Sciences Center, Department of Preventive Medicine and Biometrics
glaberge@msn.com


Greggory LaBerge wrote:

I am sure that Stewart has run into mixtures in his work, and I know he has to interpret these results objectively ( the scoring) however whether individuals are included or excluded as potential sources of DNA to the mixture must include some case facts.

Of course - A DNA analyst who has not experienced mixture results is definitely not a DNA analyst. The objectivity issue in this regard has also been resolved by the STRlab system. The DNA profiles search engine, STRquest will automatically indicate all profiles included within the mixture profile. This is done without regard to case information and may even include profiles from totally independent cases (which gives an idea of the likelihood of adventitious mixture "hits"). This is the similar to manual comparison - if all suspect alleles are present in the mixture, then this person cannot be excluded as a possible contributor (unfortunately the manual method does not tell you how many other profiles on your database are included by chance within the mixture - STRquest does!).

Partial profiles on either side, mixture or control will not influence the STRquest search algorithm. An exception where this objective procedure may be considered as "flawed" is when single alleles are lost due to complex matrix effects of mixtures. This is not a problem if a person is completely objective:

If an allele is lost in a mixture which is required to include a suspect (the allele exists in the suspect but is missing in the mixture, but all other alleles are present), then according to protocol the suspect must be excluded. It may be possible to postulate a reason for the one allele omission or state some statistical probability of such a one allele omission, but this may only be sensible when comparing two non-mixture profiles, as you would have to state the probability of one allele omissions on all permutations included in the mixture - not to mention the biased view that the suspect's extra allele is the only one to consider.

Therefore even mixture results can be scored blind and even interpreted without looking at casework information.

If rigorous standard procedures are not followed for DNA analysis then we destroy the whole reason for advances in DNA analysis - undisputed human identification - If DNA is a human identifier, then why do you as forensic analyst have to look elsewhere to "get a match"?. Other circumstances may influence these results but this is for the courts to resolve.

I state again : DNA results are objective!!!!

Stewart Allen
Senior Forensic Analyst
South African Police Service - Forensic Science Laboratory
biored@iafrica.com

 


Billo wrote:

>Consider any of the image interpretation disciplines, such as radiology and surgical pathology in medicine or surveillance photograph interpretation in intelligence. If one defines a "blind" study as one in which no information were made available other than the image, then blind studies constitute malpractice. Any pathologist who, for instance, interprets a biopsy without getting the full patient history, impressions from the surgeon, etc. is asking for trouble. These are all important data which are part of the interpretation.<

Would it be reasonable for the image interpreter to have a "first look" blind, record his/her impressions, and then consider outside information? This might "waste" a certain amount of time, but the scientific benefit seems clear. In some cases the time factor would be more weighty, and in other cases it may be difficult or impossible to perceive the relevant image without the focus provided by outside information.

Submitted 2/4/00

 


 

 

 

 


CLOSE