Analyzing the Psychological and Social Contents of Evidence-Experimental Comparison between Guessing, Naturalistic Observation, and Systematic Analysis

Abstract

To improve inferences about psychological and social evidence contained in pictures and texts, a five-step algorithm—Systematic Analysis (SA)—was devised. It combines basic principles of interpretation in forensic science, providing a comprehensive record of signs of evidence. Criminal justice professionals evaluated the usefulness of SA. Effects of applying SA were tested experimentally with 41 subjects, compared to 39 subjects observing naturally (naturalistic observation) and 47 subjects guessing intuitively intuitive guessing group. After being trained in SA, prosecutors and police detectives (N = 217) attributed it a good usefulness for criminal investigation. Subjects (graduate students) using SA found significantly more details about four test cases than those observing naturally (Cohen’s d = 0.58). Subjects who learned SA well abducted significantly better hypotheses than those who observed naturally or who guessed intuitively. Internal validity of SA was a = 0.74. Applying SA improved observation significantly and reduced confirmation bias.

Full text

PAPER
PSYCHIATRY & BEHAVIORAL SCIENCES
Henriette S. Haas,
1
Ph.D.; Maja Pisarzewska Fuerst,
1
M.Sc.; Patrick T€
onz,
1
M.Sc.;
and Jutta Gubser-Ernst,
2
B.Sc.
Analyzing the Psychological and Social
Contents of Evidence—Experimental
Comparison between Guessing, Naturalistic
Observation, and Systematic Analysis*
ABSTRACT: To improve inferences about psychological and social evidence contained in pictures and texts, a five-step algorithm—System-
atic Analysis (SA)—was devised. It combines basic principles of interpretation in forensic science, providing a comprehensive record of signs
of evidence. Criminal justice professionals evaluated the usefulness of SA. Effects of applying SA were tested experimentally with 41 subjects,
compared to 39 subjects observing naturally (naturalistic observation) and 47 subjects guessing intuitively intuitive guessing group. After being
trained in SA, prosecutors and police detectives (N=217) attributed it a good usefulness for criminal investigation. Subjects (graduate students)
using SA found significantly more details about four test cases than those observing naturally (Cohen’sd=0.58). Subjects who learned SA
well abducted significantly better hypotheses than those who observed naturally or who guessed intuitively. Internal validity of SA was
a=0.74. Applying SA improved observation significantly and reduced confirmation bias.
KEYWORDS: forensic science, systematic analysis, naturalistic observation, intuitive guessing, abduction of hypotheses, psychosocial
evidence, confirmation bias, experiment
In most criminal cases, social and psychological facts constitute
an important part of the evidence: Written documents (letters or
drawings), recorded statements, and crime scene photographs offer
clues about the culprits’and witnesses’minds and personalities,
namely memories, motives, plans, cognitions, and emotions as
well as roles, relationships, proceedings, and capacities. Because
only the mind’s output can be observed, we must analyze (prop-
erly recorded) behavior and language and draw inferences based
upon them.
The observation and interpretation of psychological and social
aspects of evidence is an amorphous problem: It has not (yet)
been clearly defined where to begin, how to proceed, and when
the task is accomplished. In the absence of an established proce-
dure to analyze the psychosocial contents of text and pictures,
materials are often appraised by normal reading and then naively
compared to the rest of the evidence. Countless studies in foren-
sics (e.g., 1,2,3) have shown that under these circumstances, the
interpretation of the evidence is prone to be biased. “Confirma-
tion bias is a proclivity to search for or interpret additional infor-
mation to confirm beliefs and to steer clear of information that
may disagree with those prior beliefs [...].”(1).
An analytic approach based on generally accepted scientific
principles should help to avoid considering only salient aspects
of the observandum and falling prey to observer biases. Then
again, some researchers in the decision-making field (4) advo-
cate intuition over analytical thinking. They believe that account-
ing for too much information at a time can cause a cognitive
overload and thus be counterproductive to good abduction.
Behind this controversy is the information-processing predica-
ment (2): “[...] the human brain has two distinct decision-mak-
ing systems. The first is more analytic, rational, controllable and
objective, whereas the second is more experiential and relatively
independent of language, but is much faster.”
Epistemological Principles of Systematic Analysis
To assist criminal investigators in unresolved cases, a method
to improve the interpretation of social and psychological aspects
revealed by the evidence was devised (5). Five rules, each one
an undisputed principle in the scientific community, were orga-
nized into the algorithm of systematic analysis (SA) so as to
observe questioned objects, pictures, and text from a social sci-
ences point of view.
In everyday forensic practice, SA can be applied to anony-
mous letters and threats in order to establish an offender profile
(6–8). Other fields of application are the interpretation of suicide
notes, diaries, and last wills, of 911 calls, as well as of websites
(9), drawings (10), plans, and photographs made by suspects or
1
University of Z€
urich, Z€
urich ZH 8050, Switzerland.
2
University of Basel, 4003 Basel, Switzerland.
*Supported by a research grant by the Foundation “Suzanne and Hans
Bi€
asch for the promotion of applied psychology”(Stiftung Suzanne und Hans
Bi€
asch zur F€
orderung der angewandten Psychologie, c/o Herr Prof. Dr. Ch.
Steinebach, ZHAW Z€
urcher Hochschule f€
ur Angewandte Wissen-schaften,
Departement Angewandte Psychologie, Minervastr. 30, Postfach, CH-8032
Z€
urich.
Received 6 Aug. 2013; and in revised form 26 Jan. 2014; accepted 20
April 2014.
1©2015 American Academy of Forensic Sciences.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and
distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
J Forensic Sci,2015
doi: 10.1111/1556-4029.12703
Available online at: onlinelibrary.wiley.com

witnesses, and crime scene photographs. Furthermore, analyzing
recordings of interviews can help the interrogator to design more
specific questions for follow-up interviews (7). We also recom-
mend the method to describe crucial pieces of evidence about
mens rea, such as letters, threats, or drawings in a murder charge
(10). A context-blind expert can analyze these separately, to
avoid a confirmation bias or the impression of a confirmation
bias in the interpretation of the elements of mens rea.
First Principle: Using Schemata to Improve Perception
Theories of perception (11,12) assume that perception is a
cognitive activity of integrating and comparing stimuli with
mental schemata about the world. They reflect a structural under-
standing of the domain, thus reducing distraction by surface
details and providing substantial information about the object in
question. As a first rule, appropriate schemata of the observan-
dum should be searched to insure that no important piece of
information remain unacknowledged. To detect subtle differences
and similarities, some schemata must be put side by side with
the object of observation; for example, for the analysis of anony-
mous letters, the address on the envelope should be put physi-
cally next to the one in the phone book. The same is necessary
for the comparison of addresses found on letters presumed to
stem from the same author. When analyzing interview protocols,
the questions form the schemata for all of the interviewee’s sub-
sequent answers. Provided that they have never been used in a
preceding question before, all newly introduced words can be
attributed unequivocally to the interviewee’s mind (10).
Some schemata consist of ideal or statistical norms, of stan-
dards and plans; others consist of typologies of deviance (e.g.,
empirical studies about extortion letters). The comparison with a
schema can do more than attract attention to details otherwise
neglected; it may lead to new, unknown, or atypical aspects of
the observandum.
The relevance of schemata to improve problem-solving skills
was confirmed by several empirical studies (13–15).
Summing up Rule I: Compare the observandum with appro-
priate schemata, standards, or norms.
Second Principle: Recording Signs
The next step is to observe and describe all stimuli offered by
the material. The smallest unit of perception is a sign, an entity
with two inseparable faces, its form, and its contents. Peirce (16)
defined three types of signs: indexes, icons, and symbols. An
index is what we normally understand by sign of evidence, “it
refers to the object that it denotes by virtue of being affected by
that object”.Anicon (i.e., normally a picture) is a sign imitating
the object which it stands for. A symbol is a sign referring to the
object that it denotes by virtue of convention, tradition, or law.
Observing the formal aspects of a text involves a description of
its grammatical characteristics, of its choice of words such as
professional languages, regional dialects, and finally elements of
style (e.g., denials, use of passive voice). This part of the analy-
sis can be enhanced using linguistic tools such as the LIWC
software (17) and comparing the results with those from psycho-
linguistic studies (e.g., 18).
Empirical studies about the perception of form versus contents
can be found in the fields of visual perception (19) and linguis-
tics. Studying the recognition of formal syntactical structures,
experiments showed that adults dismiss formal anomalies in
texts more readily than contents anomalies (20).
Summing up Rule II: Describe the form of a sign not only its
presumed meanings.
The nature of the sign has another procedural impact on scien-
tific observation. A sign is to be understood as a coded signal
within a specific social, cultural, and individual context (21).
Consequently, signs should be interpreted within their context,
which leads to the third rule.
Third Principle: Comprehensiveness and Structure
Unfortunately, we often do not know the entire context of a
given case. The set of circumstances must also be reconstructed.
Even if almost all cases involving living beings present complex
open systems, we should strive for comprehensiveness. So refer-
ence classes should be chosen on the basis of all relevant knowl-
edge available prior to the explanandum (22). To get as
comprehensive record of all relevant signs as is possible, observ-
ers must apply the previously found schemata. They determine
the underlying structures of the objects of the inquiry (e.g., pho-
tographs, drawings, letters, statements, web pages).
The third principle has been tested experimentally in the field
of decision-making (23): When people organize their behavior
into subgoals by segmenting their tasks according to a hierarchi-
cal structure, they consistently learn and observe better than
when they do not. Observing each component of a structure is
quite the same as segmenting behavior into subgoals.
When it comes to applying the rules II and III, the distinction
between form and contents is relative within several layers of
classification. For the purpose of analyzing a letter, we can list a
relative relationship between form and contents in five different
structures: (1) materials (e.g., DNA, ink ...), (2) graphics of the
document, (3) aspects of language, (4) elements of the letter
type, and (5) actors and themes in the text (Table 1). Each of
these structures is more formal than its neighbor on the right and
more content oriented than its neighbor on the left. Observing
all those structures of a letter (and not only structure 1) provides
insight into an author’s personality, his motives, skills, roles,
and social environment. Observing actors includes identifying all
grammatical actors (also pronouns and hidden actors in passive
voice verbs or in nouns such as “a phone call”), but also modal
verbs (want, must, may, need, can) and other latent pillars of
meaning. These point toward Freudian slips hidden in a text.
For the analysis of pictures, all icons need to be translated
into text to improve observation in criminal investigation (24).
Summing up Rule III: Observe each component of the
object’s structure as determined by the models.
Fourth Principle: Checking for Incoherence
The inventory of signs resulting from applying rules I to III
might not be coherent. Anomalies, contradictions, and inconsis-
tencies may appear. According to Aristotle’sprinciple of non-
contradiction, any valid hypothesis concerning the evidence
must resolve such inconsistencies (25).
Empirical studies show that human propensity to overlook inco-
herence in pictures and in text is considerable. In an experiment on
visual perception (19), it was found that people are quite insensi-
tive to inconsistencies of illumination direction, even when they
are specifically told to look out for them. Another study addressed
the detection of incoherence in text (26): “the ability to detect
inconsistencies or contradictions is an important aspect of compre-
hension. Indeed, a failure to do so may lead to unwarranted accep-
tance of a set of erroneous arguments and assumptions.”It was
2JOURNAL OF FORENSIC SCIENCES

found that subjects who were instructed about the presence of con-
tradictions doubled the number of identifications of them, but they
also doubled the number of false positives. Testing children’s fail-
ure to notice inconsistencies in text (27) researchers found that
detection failures were related to poor recall of incoherence,
caused by the inability to mentally represent inconsistencies.
These results lead us back to the necessity of supporting memory
by consulting appropriate schemata.
Summing up Rule IV: Detect inconsistencies, anomalies, and
contradictions.
TABLE 1–– Checklist for Systematic Analysis (2014 version): follow the rules no matter whether you expect a promising result or not.
Rule I. Find Schemata and Models for the Object of Observation
Written Documents and Texts Drawings and Photographs
Addresses on Internet and phone book
Business or private letter (if necessary
use culturally determined models)
For other documents, use official standards
Recorded interviews: text of questions figures as
model for subsequent answers
Schemata of deviance: empirical studies about
extortion, threat, and suicide letters, about 911 calls
Drawings: made by children, artists, cavemen, scientific drawings, and drawing instructions
Photographs: Several photographs of a case serve as models for each other
Photographs: compare the conformity of human or animal anatomy
and actions to “actions”visible in the photograph
Schemata of deviance: drawings by mentally ill, drugged, or intoxicated
persons, photographs and drawings by sex offenders
Rule II. Observe Formal Aspects of the Signs
Written Documents and Texts Drawings and Photographs
Documents: observe graphics, that is, format, layout, margins,
edges, fonts/handwriting, representation of numbers, horizontal lines
versus nonaligned text, use of icons, stains
Text: observe language
Punctuation, spelling
Sentence structure (subordinate clauses,
questions, denials, passive voice, etc.)
Unusual words
Style: local dialects, professional language, slang,
malingering a foreign language speaker
Drawings: Two-dimensional versus three-dimensional (perspective),
b&w versus color, secure/insecure hand, text, and figures in a drawing
Drawings: observe proportions
Photographs: light and reflections, dynamic versus static, undisturbed versus
disturbed, alive versus dead
Time if relevant: sequence, time of day, or season
Space if relevant: relative position, entry versus exit, geography, cardinal directions
Rule III. Dissect the Object into Its Structural Components According to the Relevant Schemata
Written Documents and Texts Drawings and Photographs
Letters: observe the elements, that is, return address, inside address,
city and date, salutation (according to purpose and culture), main body
(according to purpose), closing, signature
Texts: List all actors (humans, pronouns, institutions, animals), also hidden
subjects (lack of commitment from speaker). Anonymous writers are often
on this list (not only as “I”or “we”). List relevant themes, list terms used for
weapons, threats, money, vehicles, drugs. Observe modal verbs (who may “want”,
who must “have to do things”?). List actions, cognitions, and emotions
Describe every component of the picture
Specify human actors, objects, and animals
Specify foreground versus background
Describe potential human actions
Describe potential natural influences
Describe potential technological influences
Rule IV. Note Inconsistencies, Anomalies, and Contradictions
Written Documents and Texts Drawings and Photographs
Between one sentence and another
Between the grammatical construction on the
one hand and the meaning of words or sentences on the other hand
Evolution: Observe how the same actor, the same
object, or the same theme changes names along the text (in list of rule III)
Between one picture and the other
Within the components of a picture
Within the drawing style, photographic conditions
Contradicting human anatomy or human actions with the “actions”in the picture
Check for contradictions between text and pictures
Rule V. List What is Missing or Superfluous
Missing or superfluous formal elements or signs see list for rule II
Missing or superfluous components or aspects see list for rule III
VI. Make a hypothesis check with a three-column table (for each relevant hypothesis)
Signs that are contradicting H0 Signs that are inconclusive with respect to H0 Signs that are compatible with H0
HAAS ET AL. .ANALYSIS OF PSYCHOSOCIAL EVIDENCE 3

Fifth Principle: Recording Missing Elements
The unconscious assumption that what you see is all there is con-
stitutes a main source for errors of judgment (28). Scientists should
make missing signs explicit in all inquiries. The comparison of the
evidence with the structures of the relevant schemata facilitates the
awareness of what should or could be there, yet is not. The impor-
tance of missing data is basic to probabilistic reasoning (29,30),
which might be applied to the material in a later step to resolve
questions of proof. Then again, we want to notice apparently super-
fluous details (such as rejected valuables in a burglary or such as
elements of an extortion letter that are not necessary to accomplish
the crime), because they can also reveal important information.
Two studies (31,32) tested undergraduates’ability to identify
sufficient, missing, and irrelevant information in math problems.
Identifying those problems with sufficient information was the task
best solved with a standardized score of M=0.84 (SD =0.22),
while the discovery of missing information and that of irrelevant
information received much lower scores with M=0.35
(SD =0.25) and M=0.24 (SD =0.23), respectively (31). Then,
the perception of both contradictions and missing information
within math problems and brainteasers was also examined (33):
Undergraduates received a booklet containing tasks to solve. Some
of those were in the original version, whereas others had been edi-
ted to contain contradictions or had some information deleted. The
doctored problems were technically impossible to solve except by
generating inferences. Subjects (not knowing this) were encour-
aged to formulate questions after having read a task. The results
showed that the chances that the subjects did indeed ask questions
(because they had spontaneously detected the insolvability of a
task) were low, namely LR =0.5 (SD =0.35) for the missing
information and LR =0.31 (SD =0.32) for the contradictions.
Summing up Rule V: Take note of all missing and superflu-
ous elements or signs.
Applying the Algorithm of the Five Rules in Practice
In many criminal investigations, the first hours and days count
most (34). Under the pressure to solve a major case, it may be
difficult to figure out all potential schemata with their different
structures. Therefore, we created a one-page checklist how to
apply the method on pictures and on text (Table 1).
One of the test cases in the experiment shall illustrate the pro-
cedure. As part of a scientific study (35) in 1959, an adult male
in full physical and mental health drew the sketch of an elephant
and himself next to it (Fig. 1). The riddle to solve was: “Why
would somebody draw like this?”
Rule I: We can use a scientific picture of an elephant as nor-
mative schema, but we will also consider other drawings such as
children’s or cave men’s. In the drawing, we should compare
the human and the animal figures as models to each other. Rule
II, describing formal aspects of the observandum: The ques-
tioned drawing is a black and white pencil sketch, done with an
insecure shaky hand. It lacks any notion of depth or perspective:
There is neither foreground nor background. Nonetheless, pro-
portions between man and elephant and between the animal’s
torso and its legs are correct. Rule III: All components of the
icons need to be described according to the logic of the drawing
(here: anatomy). If this is impossible, each sector of the paper
can be numbered and described. The questioned elephant has
four legs drawn like sticks, a large torso, a tail, and a trunk, but
no head and no feet. The man has a head and a torso, whereas
his limbs are too short, almost missing. His face is empty. Fur-
thermore, the limbs of the elephant connect to its body and the
head of the man connects to his body. Rule IV: The drawing
contradicts anatomy in a grotesque way. Comparing it to cave
men’s drawings shows that they knew the anatomy of animals
quite well and they also drew with a secure hand despite their
primitive utensils. Contradicting the drawing’s general awkward-
ness, the proportions of the figures seem just about right. Rule
V: The most bizarre characteristic is that the elephant has no
head at all (thus no ears, tusks, mouth, eyes) and that the man’s
face lacks eyes, nose, and mouth. A comparison of these features
with children’s drawings, where the head is the most prominent
characteristic of all, reveals an unusual situation.
Abducting Hypotheses
Based on the inventory of signs, some hypotheses spring to
the observers’minds. Abduction, according to Peirce (36), is the
creative process of adopting an explanatory hypothesis that fits
the data in retrospect. It is the third mode of logical inference
besides deduction and induction. Every criminalist knows that
diligent observation does not automatically lead to the abduction
of the correct hypothesis. Most cases being open systems, there
can be several plausible explanations fitting the available facts:
“The identification of pattern in crime investigation may perhaps
be defined simply as the identification of a deterministic
sequence in a series of apparently chance events.”(37).
At this stage of the inquiry, one may need to review the evi-
dence in a re-iterative process, when the obtained results yield new
insights and offer new ideas for consulting yet another model.
In the case of the elephant drawing, the hypothesis of drug
intoxication might have sprung to our mind, so we might want
to search for drawings made under the influence of drugs.
Another explanation could be that the author has never seen an
elephant in his life. Here, we could consider a sculpture made
by a medieval artist who has obviously heard about elephants
but never seen them (38). This sculpture is quite the opposite of
the questioned drawing because it contains all the anatomical
parts of an elephant but in distorted proportions. Other hypothe-
ses are as follows: Was the man blindfolded? Was he drawing
with his subdominant hand? Was he an artist? Was he raised in
seclusion or in the wilderness? Was he under stress? Did he sab-
otage research? and Was he malingering?
Comparing Different Hypotheses, Checking for Validity
Finally, the most likely among different explanations must be
chosen. For this purpose, all signs of evidence can be recorded
FIG. 1–– Drawing of an elephant (35).
4JOURNAL OF FORENSIC SCIENCES

within a three-column table (Table 1): those contradicting a spe-
cific hypothesis, those being inconclusive, and those that fit. The
collection of evidence must now be criticized from all angles. To
avoid a fragmentation of the evidence and the resulting fruitless
discussions or erroneous decisions, we must be transparent about
any decisions to include or exclude signs. Only a comprehensive
table of all evidence can ensure falsifiability of the favored
hypothesis (39). The table also permits to move signs from one
column to another to follow up on the effect of temporary exclu-
sion or inclusion of certain signs on the overall picture. Last but
not least, a three-column validity check provides a record explain-
ing a decision that seems to be the most reasonable in light of the
available evidence—in case it must be defended against media or
political pressure or in case some surprising developments in the
investigation should later occur.
For the elephant drawing, the blindfold hypothesis must be
rejected because it cannot explain why the legs are connected to
the torso. The hypotheses of drawing with the wrong hand, an
upbringing under seclusion, or in the wilderness cannot explain
the missing head and face. The hypotheses of psychopathology
(e.g., autism), neurological disorders (e.g., neglect, face-blind-
ness), or severe visual impairment contradict the initial informa-
tion, the fact that the man was physically and mentally sane at the
time. Only the hypotheses of sabotage, artistry, malingering, and
drug intoxication can be retained as plausible in light of the avail-
able evidence. The solution to the riddle is hard to find, because it
is such an unusual scenario: The author of the drawing had been
congenitally blind when a new type of surgery was performed on
him, which gave him full vision. A team of cognitive psycholo-
gists had visited him in the hospital and had him draw things he
had never seen, before showing them to him (35).
Experts’Opinion about the Algorithm
To determine the usefulness of the five rules for the criminal
investigation, we analyzed anonymous evaluations of eleven 1-
day courses in continuing education given to state prosecutors
and police detectives (N=217 participants since 2008). They
graded the training in the five rules of analysis and hypotheses
abduction on a scale from 1 (very poor) to 6 (very good).
The 43 prosecutors were 58% males and 42% females with an
average age of M=38.5 years (SD =8.6) and a professional
experience in criminal investigation of M=9.7 years
(SD =8.3). They appraised the “usefulness”with a mean grade
of M=5.6 (SD =0.6) and the “intelligibility”with a mean of
M=5.7 (SD =0.5).
Chiefs of Police of eight local police corps had declared the
participation of courses in SA mandatory for all their detectives.
So this sample covers entire cohorts and includes the whole
range of motivation, professional experience, and skill. The 174
detectives (91.3% males and 8.7% females) averaged
M=41.8 years of age (SD =8.1) and had M=16.8 years
(SD =8.9) professional experience in criminal investigation.
They graded the usefulness of the method with a mean of
M=4.7 (SD =0.7), and they valued it as a good critical think-
ing tool with a grade of M=5.0 (SD =0.6).
Experimental Procedure
Participants
Participants of the experiment were graduates regularly
enrolled in classes of forensic psychology and criminology. Dif-
ferent semesters built three quasi-random samples. The “SA”
group was instructed with the five rules and practiced them; it
consisted of n=41 participants (33 females, six males, 2 miss-
ing). They were told that they would learn criminal profiling
during the experiment. Another n=39 students (33 females,
four males, 2 missing) were in the “naturalistic observation”
group (NO). They were told that they would get the opportunity
to “learn by doing casework”in an experiment (40). Those 80
students stayed for all three sessions, completing the whole test
series and filling out the questionnaires, while another 15 stu-
dents dropped out before completing the experiment. The intui-
tive guessing group (IG) consisted of 47 students (32 females,
14 males, 1 missing). They were instructed that the goal was to
compare intuition to careful observation and description in case-
work. They completed the tests cases by pure guessing within
one session. All 127 students were white Europeans and profi-
cient in the language of instruction. The mean age of the SA
group was M=25.9 years (SD =5.6), that of the NO was
M=26.1 years (SD =5.4), and that of the IG was
M=26.8 years (SD =5.6). Age differences were not signifi-
cant. The participants’sex had no significant effect on the out-
come variables.
Experimental Setting
The test cases (Table 2) were selected from several dozen.
They had to have a proven ground truth and offer the opportu-
nity to apply all five rules. We wanted to present a variety of
themes and causes.
TABLE 2–– Test cases presented in the experiment.
Test Case Description
A. Broken window Photographs of a broken window (star-shaped hole) next to the entrance of a suburban home. What happened? The ground truth:
No human mind was involved. A hail storm had hit the house. Clues to detect: No object having caused the hole was visible.
The broken window was barred and provided no opportunity to enter the house, while a plant on the windowsill was is shreds
B. Elephant drawing The elephant drawing (35) was done as part of a scientific study. Why would a mentally and physically healthy man draw like this?
The ground truth: A man recovering from congenital blindness (by surgery), who had never seen an elephant before, had drawn
the picture. Clues to detect: His insecure drawing hand, the animal’s head is missing, while the limbs are connected to the body
and the figures are in adequate proportions
C. Murder scene Photographs of a crime scene (44) and a description of the events leading up to the discovery of a dead female in her apartment
were presented. Who was the main suspect? The ground truth: The murderer was the husband who had staged the crime scene
to appear as a burglary. Clues to detect: The inconsistency of the manipulated crime scene combined with the strange behavior
of the husband when the crime was discovered
D. Citizen’s letter An individual mentioned in a homeland security report on extremism, as being a leading negationist (Holocaust denier), wrote a letter
of protest to the Prime Minister. Was he a negationist or not? The ground truth: The author was a notorious negationist convicted for
organizing conferences and propagating racism. Clues to detect: Numerous and superfluous grammatical negations, a total absence
of terms specifying historical facts (e.g., “genocide”), combined with a reversal of the meaning of the word “negationist”
HAAS ET AL. .ANALYSIS OF PSYCHOSOCIAL EVIDENCE 5

Concerning the final hypothesis resulting from the observa-
tions, we chose two ways to request it on the answering sheets.
For the two criminal cases, the murder scene and the citizen’s
letter, we presented a choice of several hypotheses. For the two
riddles, the broken window and the elephant drawing, subjects
depended entirely on their imagination as we just left open space
for them to state their best hypothesis. Furthermore, we
instructed them to define only one final hypothesis.
To avoid that the experimental effect be flawed by inevitable
differences in the test cases, the participants of each group were
randomly assigned to subgroups. Subgroup 1, representing half
of the class, had to treat cases A and D during the first test,
whereas subgroup 2 had to treat cases B and C at first. During
the second test, the participants received the inverse pair of cases
to test their performance.
During the tests, participants were working on their own in a
quiet classroom, seated with an empty chair between each other.
They were told not to exchange any information about the test
cases during the whole experiment. This examination-like proce-
dure was strictly enforced.
The SA and the NO group both started with the first 1-h test
about two cases to establish the baseline of their capacity to
observe and interpret. They would carefully read the cases, study
the pictures, take notes about their observations, and finally,
state their best hypothesis. The participants in the observation
groups (SA and NO) received lined sheets of paper with enough
space to describe all their observations. The experimental group
in SA then received 5 h of training, spread over the same and
the next afternoon. It included a 90-min presentation explaining
the rules, illustrated by analyzing the criminal manifesto of the
anthrax attack (5,6,41). The students also practiced the system-
atic application of the rules with four exercises (a photograph,
an anonymous threat letter, a drawing, and a newspaper inter-
view with a terrorist suspect). Results of those exercises were
discussed in plenum. For the second test, the SA group received
the checklist (i.e., a previous version). During the same time
span, the NO control group received presentations about differ-
ent topics of forensic psychology not related to any of the rules
of SA. During the third afternoon, each group was asked to take
the second 1-h test on the inverse pair of cases.
The IG received the instruction to write down the first hypothe-
sis springing to their mind immediately upon reading a case. This
was practiced a few times. To ensure that all subjects spend the
same amount of time reading, each case was read aloud slowly
while the students would also follow it on paper. Then, they were
given 60 sec to write down their best hypothesis about the case.
The answer sheets left no space to put down any observations.
After the first test, they changed rooms and did the inverse pair of
test cases.
Operationalization of the Rules
The subjects’use of rules in the SA and NO groups was mea-
sured using the categories specified in the checklist (counting the
number of schemata, formal aspects, components, inconsistencies,
and missings according to the old version of the checklist). The
overall wealth of signs that participants had observed was mea-
sured by taking the sum of observations according to all rules.
To measure the predictive validity of the three different
approaches, we took the quality of the favored hypothesis. All
those hypotheses, which were either plausible or a hit of the
ground truth were considered to be logically valid. Finding a
valid hypothesis involves two steps: first, the abduction of good
ideas, and second, critical thinking whether the favored hypothe-
sis is consistent with all signs of evidence. Those two additional
reasoning modes and the instruction of inserting all signs of evi-
dence into a three-column table were just mentioned in class and
not practiced with the SA group.
Rating the Participants’Answers
All authors participated in the coding of the participants’
answers. To ensure double-blinding, the control variable (first vs.
second test) was covered on the subjects’response sheets. Unfor-
tunately, it was sometimes possible to spot analyses carried out by
trained participants, because they explicitly referred to the rules.
To rate the application of each rule, we established lists for all
answers falling under a given rule. The attribution to rules I and
III was less ambiguous than the one to the rules II, IV, and V.
Certain observations could be rated under either of two of the
latter rules. We coded such answers under one rule only (always
the same for all subjects).
The logical validity of the subjects’favored hypotheses about
a case was determined by the first and second co-author. Beside
hits of the ground truth, we coded those hypotheses as a plausi-
ble that explained all the presented evidence of the case and con-
tradicted none of it. All hypotheses that explained the facts only
partially (e.g., the man has never seen an elephant), which con-
tradicted some of the evidence (e.g., the man was blind) or
which were too general to be refutable (e.g., the man was bad at
drawing), were rated as “not valid”.
To check the inter-rater consensus, the first, third, and forth
co-authors rated the proficiency of the experimental group’s use
of the rules. The correct use of the rules on the test cases was
graded on a 6-point scale (from 1 =“no use”to 6 =“excellent
use”). When the joint application of all rules is considered, the
correlation between different judgments about the cases treated
by the trained subjects was Cronbach’sa=0.75. Differences
before and after learning the method showed no consistent pic-
ture. Obviously, a good agreement can be achieved on the total
absence of guidance by a rule before it had been taught. Inter-
rater disagreements pointed out some weaknesses of the check-
list. Some subjects had found good proceedings not mentioned
in the checklist.
Statistical Analyses
Statistical analyses were performed using SAS
â
9.3 (PROC
MEANS, PROC GLM, PROC ANOVA, PROC CORR, PROC
FREQ, PROC UNIVARIATE SAS Institute Inc., Cary, NC, USA).
Results of the Experimental Testing
Effects of Learning the Rules
Not everybody who has been taught a method can integrate
the new knowledge and apply it correctly, especially if there is
no examination to enforce serious studying. Thus, we examined
the difference between the NO and the SA groups, but we also
considered the “good learners”among the latter (n=21 or
51.2%). Those were defined (ex post) as the subjects who had—
in both tasks of the second test—mentioned at least one schema.
Between the “good learners”and the other 20 experimental sub-
jects of the SA group, there were no significant differences dur-
ing the first test: neither in the wealth of details observed, nor in
the quality of their hypotheses.
6JOURNAL OF FORENSIC SCIENCES

Effect on the Wealth of Observations
We tested the external validity of the five rules by taking the
mean intra-individual differences between the first and the sec-
ond test of the SA and NO groups. Table 3 shows the improve-
ments in the wealth of observations over two case analyses after
the second test.
The NO group showed no significant progress, whereas the
SA group increased the number of relevant observations during
the second test significantly, effect size estimates being medium.
Describing the structural components of an object (rule III)
turned out to be the default mode of naturalistic observation,
whereas the rest of the principles of SA must be taught and
practiced. For the good learners, the effects of learning the five
rules were very strong.
Predictive Validity in the Validity of Hypotheses
Does SA improve the quality of abducted hypotheses com-
pared to NO and to IG? Being a creative task, the generation of
well-fitting hypotheses cannot depend on exhaustive analysis
alone. Creativity as an individual gift might explain an important
part of the variance (42). Yet another (uncontrolled) influence on
the quality of abduction may be general knowledge about the
world and professional experience (4,11,28). In the experiment,
such “general wisdom”concerned knowing how children or
cavemen draw animals or knowing that many people tend to
deceive others by omission rather than by overt lying. Contrary
to analyses carried out in real life, the experimental subjects had
to resort to their memories as models when it came to find sche-
mata for drawings.
Finding out the ground truth was more difficult for the riddles,
where it depended entirely on the subjects’imagination, than for
the criminal cases, where a choice of multiple options was offered
(Table 4). The SA group improved their hypotheses somewhat in
the second test contrary to the NO group, but the effect size esti-
mates were small. However, applying the five rules proficiently,
just like the “good learners”had done, leads indeed to a significant
increase of valid hypotheses. (Not shown in Table 4: For the IG
group, we did the same comparison and found no significant
improvement between the first and second test.)
Looking at absolute values in the quality of the abducted
hypotheses (Table 5) between the three groups, we were quite
surprised that IG on the spot worked just as well as (untrained)
careful reading and describing the evidence (NO) for an hour
per two cases. Thus, NO (following no specific rules) provided
no added value over intuition after reading the case for the first
time. We interpret this result as an expression of the confirma-
tion bias. Maybe participants had formed an instant opinion
about a case and had stuck with it throughout their observations.
In the absence of an objective frame of reference, people tend to
take their own first opinion as the reference and then reason
backwards observing only those details confirming their initial
hunch.
Applying the algorithm of the five rules did increase the qual-
ity of abducted hypotheses; especially the good learners scored
more than 70% valid hypotheses compared to only about 40%
for IG and NO.
Training in SA increased the odds to hit the ground truth by a
LR =1.55 compared to the results of untrained observers (with
a95% CI of [0.93–2.60] and p<0.09). Good learners’results
showed an improvement by a LR =2.24 to find out the truth of
the case (with a 95% CI of [1.16–4.35] and p<0.02) in com-
parison with the results of untrained observers. The training also
improved the abduction of a hypothesis that was at least plausi-
ble (instead of being false, illogical, incomplete, or too vague),
even if it did not correspond to the ground truth. The odds to
find a plausible (yet untrue) hypothesis by applying SA were
increased by a LR =2.25 (95% CI of [1.02–4.94], p<0.04).
For the good learners, it was a LR =3.05 (95% CI of [1.22–
7.60], p<0.01).
Internal Validity of the Five Rules
What about the internal validity of the fives rules, do they
form an algorithm, or are they just a loose bundle of recipes?
TABLE 3–– Intergroup comparison of the individual improvement in the wealth of observations during the second test.
Operationalization
Naturalistic
Observation
n=39
Systematic Analysis
n=41 (Good
Learners n=21)
GLM
(Unbalanced Data)
Effect Size
Measures
MSD 95% CI MSD 95% CI SS F
1,76
(F
1,58
)p<Partial g
2
Cohen’sd
Effect
size r
Rule I
Schemata used
0.03 1.13 0.39, 0.34 0.88
*(1.52)
1.29
*(1.36)
0.47, 1.28
*(0.90, 2.14)
16.3
–*
11.0
–*
0.001
–*
0.124
–*
0.744
–*
0.349
–*
Rule II
Formal aspects
observed
0.33 2.73 1.22, 0.55 1.54
(2.19)
2.43
(2.69)
0.77, 2.30
(0.96, 3.41)
69.9
(86.9)
10.5
(11.8)
0.002
(0.001)
0.119
(0.170)
0.724
(0.926)
0.340
(0.420)
Rule III
Components
observed
0.21 5.41 1.55, 1.96 0.61
(2.38)
5.94
(6.07)
1.27, 2.48
(0.38, 5.14)
3.3
(64.6)
0.1
(2.0)
0.751
(0.160)
0.001
(0.034)
0.071
(0.382)
0.036
(0.188)
Rule IV
Incoherences
observed
0.10 2.77 0.80, 1.00 1.27
(1.29)
2.09
(2.03)
0.61, 1.93
(0.36, 2.21)
27.2
(19.1)
4.6
(3.0)
0.036
(0.091)
0.036
(0.049)
0.475
(0.505)
0.231
(0.245)
Rule V
Missings observed
0.10 1.70 0.65, 0.45 1.24
(1.90)
2.75
(2.91)
0.37, 2.11
(0.58, 3.23)
36.2
(55.0)
6.8
(11.4)
0.011
(0.001)
0.011
(0.165)
0.588
(0.841)
0.282
(0.388)
Total wealth of
observation (all five
rules combined)
0.15 7.02 2.43, 2.12 5.54
(9.29)
11.97
(12.81)
1.76, 9.32
(3.46, 15.12)
647.2
(1216.3)
6.6
(13.7)
0.012
(0.001)
0.078
(0.191)
0.580
(0.921)
0.278
(0.418)
N=80 subjects (observations over two analyzed cases per test).
*Tautological as “good learners”are defined by having used at least one schema for analyzing each of their cases in the second test.
HAAS ET AL. .ANALYSIS OF PSYCHOSOCIAL EVIDENCE 7

The mechanics of the rules (Fig. 2) show that they are corre-
lated in a meaningful way. Finding schemata (rule I) associates
only with the observation of the underlying formal and contents
structures of the observandum, that is, rules II and III combined.
The latter rules do not depend on each other, nor do rules IV
and V. The role of rule III as the default mode of NO is also
reproduced within the mechanics of the algorithm: Contrary to
rule II, there is no significant correlation to the use of schemata.
Then again, a rich record of components according to rule III
provides a better recognition of anomalies and inconsistencies
(rule IV). Finally, there is a link between the observation of for-
mal elements and missings. Internal validity between the first
part of the analysis (rules I to III) and the second part tested
with a Cronbach’sa=0.74 for standardized variables (and
a=0.67 for raw variables). Thus, the five rules are built upon
each other and not redundant.
Interpretation of the Experimental Results
The answering sheets of the participants observing naturally
(NO) made it obvious that looking for psychosocial facts in the
TABLE 4–– Intergroup comparison of the individual improvement in the quality of the abducted hypothesis after the second test.
Naturalistic Observation
n=39
Systematic Analysis
n=41 (Good Learners n=21) GLM (Unbalanced Data) Effect Size Measures
MSD 95% CI MSD 95% CI SS F
1,78
(F
1,58
)p<Partial g
2
Cohen’sd
Effect Size
r
Hits of the
ground truth only
0.15 0.59 0.04, 0.34 0.17
(0.38)
0.70
(0.50)
0.05, 0.39
(0.15, 0.61)
0.01
(0.70)
0.01
(2.26)
0.908
(0.138)
0.000
(0.038)
0.026
(0.418)
0.013
(0.204)
Plausible
hypotheses only
0.06 0.54 0.24, 0.13 0.20
(0.33)
0.58
(0.69)
0.00, 0.40
(0.01, 0.67)
1.16
(1.81)
3.66
(5.71)
0.060
(0.027)
0.051
(0.092)
0.457
(0.632)
0.223
(0.301)
Valid hypotheses
(plausible or hits)
0.08 0.70 0.15, 0.30 0.37
(0.67)
0.83
(0.73)
0.10, 0.63
(0.33, 1.00)
1.69
(4.75)
2.81
(9.35)
0.098
(0.003)
0.035
(0.139)
0.376
(0.822)
0.185
(0.380)
N=80 subjects (means over two cases per test, only one hypothesis was allowed par case), hypotheses of the two subjects who had known the elephant case
were counted as not valid.
TABLE 5–– Quality and specificity of the abducted hypotheses according to the procedure (Systematic analysis [SA] and intuitive guessing group [IG] com-
pared to naturalistic observation [NO]).
Test Case
1 Min Per Case 30 Min Per Case
IG NO SA
Hits
Plausible
Hypothesis
Valid Hypothesis
(Plausible or Hit) Hits
Plausible
Hypothesis
Valid Hypothesis
(Plausible or Hit) Hits
Plausible
Hypothesis
Valid Hypothesis
(Plausible or Hit)
Elephant drawing
(open question)
n=46 cases n =57 cases n =21 cases (all SA)
n=13 cases (good learners)
0.0
ns
13.0
ns
13.0
ns
0.0 14.0 14.0 0.0
ns
0.0
ns
38.1
†
53.9
‡
38.1
†
53.9
‡
Broken window
(open question)
n=47 cases n =61 cases n =19 cases (all SA)
n=8 cases (good learners)
4.3
ns
0.0
ns
4.3
ns
1.6 4.9 6.6 10.5*
25.0
‡
10.5
ns
12.5
ns
21.0*
37.5
‡
Murder scene
(multiple choice)
n=47 cases n =58 cases n =22 cases (all SA)
n=13 cases (good learners)
72.3
ns
2.1
ns
74.5
ns
58.6 3.5 62.1 72.7
ns
92.3
†
4.6
ns
0.0
ns
77.3
ns
92.3
†
Citizen’s letter
(multiple choice)
n=47 cases n =61 cases n =19 cases (all SA)
n=8 cases (good learners)
76.6
ns
2.1
ns
78.7
ns
70.5 6.6 77.1 89.5*
100.0*
5.3
ns
0.0
ns
94.7*
100.0
ns
All four test
cases together
n=187 cases n =237 cases n =81 cases (all SA)
n=42 cases (good learners)
38.5
ns
4.3
ns
42.8
ns
32.9 7.2 40.1 43.2*
52.4
†
14.8
†
19.0
†
58.0
‡
71.4
§
N=127 subjects over 508 test cases.
Significance of the chi-square test compared to naturalistic observation: *p<0.1,
†
p<0.05,
‡
p<0.01,
§
p<0.001.
FIG. 2–– Internal validity of the five rules of systematic analysis, N=82
case analyses carried out by the 41 trained experimental subjects.
8JOURNAL OF FORENSIC SCIENCES

evidence is an ill-defined task. Already during the experiment,
we could directly observe that most untrained subjects were lost;
they did not know where to begin and many had finished the
tasks before the end of the test period (but they were not
allowed to leave or to talk). The trained participants on the other
hand were more focused and many would have preferred more
time to solve the tasks. Then again, the procedure demands a
certain discipline, and so the benefit of applying the five rules
on the quality of the hypotheses manifested itself more clearly in
the “good learners”.
While the SA training raised the likelihood to find the ground
truth, it also raised the likelihood to abduct a plausible yet
untrue hypothesis (instead of hypotheses clearly contradicting
the evidence which were generated by subjects without training).
Yet abductions of plausible but untrue hypotheses constitute
somewhat of a quandary. While it is desirable to open up one’s
mind to as many plausible alternatives when treating unresolved
cases, such hypotheses compete with the unknown ground truth
of the case. The example of the elephant drawing illustrates the
danger that professionals stop looking for alternatives after find-
ing a first plausible explanation (e.g., drug intoxication). Some
plausible but mistaken hypotheses can be quite convincing and
constitute pitfalls for criminal investigation.
During the rating process, we realized that improvements must
be made to the checklist, especially for the analysis of pictures.
According to our results, all five rules build upon each other and
are necessary steps in the process of analysis; none of them can
be dismissed.
Applying this method yielded a more comprehensive picture
of the evidence, which in turn helped to find better explanations
for the facts. But given the fact that this experiment is only the
first of a series to come and was performed by those involved in
creating the method, its results need to be appraised with some
caution. As the first author who conceived the five-step algo-
rithm is also the one who taught it, and who taught “learning by
doing”and “guessing by intuition,”the results of the experiment
might be influenced by an interpersonal expectancy effect (43).
Another concern might be the fact that a perfect blind rating was
not always possible (because subjects had revealed that they
were working according to SA), while at the same time, the rat-
ings of the subjects’questionnaires were not independent from
the creation of the method. Furthermore, it could be criticized
that the criterion for “good learners”had been stated ex post and
not ex ante. Finally, the test cases had to be selected for solv-
ability, thus providing the opportunity to obtain better results
than what is to be expected otherwise.
All these issues of concern are weaknesses encountered typi-
cally when a method undergoes its first scientific testing. So if
the present results raise hopes for applying systematically the
scientific principles of forensic science to psychosocial contents
of the evidence, they need to be reproduced by independent
researchers before drawing definite conclusions. Furthermore,
one could argue that this experiment does not demonstrate the
effect of those five rules specifically. Maybe people pay more
attention to detail when following any well-defined procedure
rather than just observing naturally. Therefore, further experi-
menting is required to test effects of a placebo procedure and
compare it with SA.
Nevertheless, the results of this first experiment fall in tune
with all theories mentioned in the introduction and even resolve
some controversy between them. Intuition—that is pure guessing
(IG)—turned out to be better than its reputation (5) because NO
was prone to be influenced by a confirmatory bias (1–3). Just like
applying one or two of the rules showed improvements in prob-
lem solving (23,26,28,31,32), the SA algorithm as a whole did
also increase the wealth of details observed and the validity of
abductions made. Thus, its application decreased the risk to fall
prey to biased perception, especially when the method was profi-
ciently applied (2,23,28). Table 6 offers a synopsis over advanta-
ges and disadvantages of each one of the three approaches.
Finally, we must mention a general risk encountered in all
methods used to abduct plausible hypotheses. Professionals with-
out training in empirical research often attribute more predictive
validity to the results of an analysis or an offender profile than
they actually have (sometimes even mistaking them for facts).
All reports based on applying methods to reveal psychosocial
contents in criminal cases must therefore begin with some intro-
ductory remarks about the procedures’limitations, namely about
the base rates of true hits and about the occurrence of plausible
yet mistaken hypotheses.
Conclusion
According to this first experiment, the five general scientific
principles united under the algorithm of SA provide a broader
mind-set of what can be perceived about social and psychologi-
cal aspects contained in the evidence and of what must be con-
sidered before an observer can abduct far-reaching hypotheses.
Some other features of the method should yet be examined:
What are potential pitfalls and sources of errors related to the
application of this method and what is the inter-rater reliability
of the five rules over many different cases?
TABLE 6–– Synopsis of advantages and disadvantages of the three procedures (IG, NO, SA).
Procedure
IG NO SA
Procedure is quick Yes No No
Procedure is low cost Yes No No
No need for particular intellectual effort or increased concentration Yes Yes No
No need for special instruction Yes Yes No
Procedure favors problem-solving and critical thinking skills in general No No Yes
Procedure provides a neutral reference frame for the object of observation (unrelated to the case’s context) No No Yes
Procedure leads to a greater wealth of observed details No No Yes
Procedure leads to the abduction of better hypotheses No No Yes
Procedure reduces confirmation bias No No Yes
Procedure provides an overview of all signs of evidence and their interpretation
so that results or decisions can be criticized from all angles (refutability)
No No Yes
Procedure can increase the chance to abduct plausible yet mistaken hypotheses No No Yes
IG, intuitive guessing; NO, naturalistic observation; SA, Systematic analysis.
HAAS ET AL. .ANALYSIS OF PSYCHOSOCIAL EVIDENCE 9

The following points could be interesting for future experi-
ments: As a consequence of the large part of unexplained vari-
ance in the quality of the abducted hypotheses, individual
intelligence differences namely in deductive reasoning and
creativity could be examined and compared to the success of
analyzing test cases with different methods.
In prosecutors’and police detectives’opinion, the five-step
procedure provides a useful tool for criminal investigation, help-
ing them to decide which tracks to pursue and how to bring
together all available signs of evidence in a falsifiable way. Yet
an approval cannot replace actual testing. A similar experiment
with experienced professionals is essential.
Acknowledgments
We want to express our thanks to Ren
e Hirsig, professor of
methodology, to Brigitte Boothe, professor of clinical psychol-
ogy, and to Thomas Hansjakob, JD, chief prosecutor of St. Gall,
for their generous advice and support and to Marianne Pithon
for her close reading of the manuscript.
References
1. Budowle B, Bottrell MC, Bunch SG, Fram R, Harrison D, Meagher S,
et al. A perspective on errors, bias, and interpretation in the forensic sci-
ences and direction for continuing advancement. J Forensic Sci 2009;54
(4):798–809.
2. Dror IE. A novel approach to minimize error in the medical domain:
cognitive neuroscientific insights into training. Med Teach 2011;33:34–8.
3. Dror IE, Charlton D, Peron A. Contextual information renders experts vul-
nerable to making erroneous identifications. Forensic Sci Int 2006;156
(1):74–8.
4. Klein G. The power of intuition. New York, NY: Currency Books,
2003;21,271,284.
5. Haas H. Kriminalistischer Erkenntnisgewinn durch systematisches Beo-
bachten. Kriminalistik 2003;57:93–100.
6. Haas H. Systematic observation as a tool in combating terrorism. In: Pieth
M, Thelesklaf D, Ivory R, editors. Combating the financing of terrorism.
Basle Institute on Governance. New York, NY: Peter Lang, 2009;59–93.
7. Haas H. Verlaufsanalysen von h€
auslicher Gewalt. In: Bundesministeri-
ums f€
ur Justiz, editor. Lebensform Familie—Realit€
at & Rechtsordnung.
€
Osterreichische RichterInnenwoche Laa an der Thaya 26. bis 30. Mai
2008. Schriftenreihe des BJ Band 141, Wien, A: Neuer Wissenschaftli-
cher Verlag, 2009;121–43.
8. Haas H. Analyse anonymer Schreiben unter Ber€
ucksichtigung von Gen-
der-Aspekten. In: Steins G, editor. Handbuch Psychologie und Gesch-
lechterforschung. Wiesbaden, D: VS Verlag f€
ur Sozialwissenschaften,
2010;249–70.
9. Renfer M, Haas H. Systematic analysis in counterterrorism: messages on
an Islamist Internet-forum. Int J Intell CounterIntell 2008;21(2):314–36.
10. Haas H, Sutter L. Methoden der Feststellung von psychischen Tatsachen
im Strafrecht. Aktuelle juristische Praxis (AJP)/Pratique Juridique Actu-
elle (PJA) 2013;11:1589–99.
11. Gregory RL. The Medawar lecture 2001. Knowledge for vision: vision
for knowledge. Philos Trans R Soc Lond B Biol Sci 2005;360:1231–51.
12. Neisser U. Cognition and reality. San Francisco, CA: Freeman, 1976;7.
13. Gick ML, Holyoak KJ. Analogical problem solving. Cogn Psychol
1980;12:306–55.
14. Brown AL, Kane MJ. Preschool children can learn to transfer: learning
to learn and learning from example. Cogn Psychol 1988;20:493–523.
15. Brown AL, Kane MJ, Long C. Analogical transfer in young children:
analogies as tools for communication and exposition. Appl Cogn Psychol
1989;3:275–93.
16. Peirce CS. Elements of logic. In: Hartshorne C, Weiss P, editors. Col-
lected papers of Charles Saunders Peirce, Vol. II. Cambridge, MA: Har-
vard Univ Press, 1978/1931;2.275.
17. Pennebaker JW, Booth RJ, Francis ME. Linguistic inquiry and word
count: LIWC. Austin, TX: 2007; www.liwc.net (accessed December 30,
2013).
18. Argamon S, Koppel M, Pennebaker JW, Schler J. Automatically profiling
the author of an anonymous text. Commun ACM 2009;52(2):119–23.
19. Ostrovsky Y, Cavanagh P, Sinha P. Perceiving illumination inconsisten-
cies in scenes. Perception 2004;34:1301–14.
20. Braze D, Shankweiler DP, Ni W, Palumbo LC. Readers’eye movements
distinguish anomalies of form and content. J Psycholinguist Res 2002;31
(1):25–44.
21. Eco U. Segno. Milano, I: Istituto Editoriale Internazionale, 1973;5.19.
22. Taroni F, Aitken C, Garbolino P, Biedermann A. Bayesian networks and
probabilistic inference in forensic science. Chichester, West Sussex,
U.K.: John Wiley & Sons, Ltd., 2006;25.
23. Hard BM, Lozano SC, Tversky B. Hierarchical encoding of
behavior: translating perception into action. J Exp Psychol 2006;135
(4):588–608.
24. Oevermann U. Die Methode der Fallrekonstruktion in der Grundlagenf-
orschung sowie der klinischen und p€
adagogischen Praxis. In: Kraimer K,
editor. Die Fallrekonstruktion. Sinnverstehen in der sozialwissenschaftli-
chen Forschung. Frankfurt a. M., D: Suhrkamp, 2000;58–156.
25. Aristotle. Metaphysics. In: Barnes J, editor. The complete works of Aris-
totle: the revised Oxford translation, Vol. II. Princeton, NJ: Princeton
University Press, n.d./1984;1552–728;1597.
26. Ellis HC, Ottaway SA, Varner LJ, Becker AS, Moore BA. Emotion,
motivation, and text comprehension: the detection of contradictions in
passages. J Exp Psychol 1997;126(2):131–46.
27. Vosniadou S, Pearson PD, Rogers T. What causes children’s failures to
detect inconsistencies in text? Representation versus comparison difficul-
ties. J Educ Psychol 1988;80(1):27–39.
28. Kahneman D. Thinking, fast and slow. London, U.K.: Peguin Books,
2011;86.
29. Dodge Y. Analysis of experiments with missing data. Chichester/West
Sussex, U.K.: John Wiley & Sons, 1985.
30. Lessler J, Kalsbeek W. Nonsampling error in surveys. New York, NY:
John Wiley & Sons, 1992.
31. Birney DP, Fogarty GJ, Plank A. Assessing schematic knowledge of
introductory probability theory. Instr Sci 2005;33(4):341–66.
32. Low R, Over R. Detection of missing and irrelevant information within
algebraic story problems. Br J Educ Psychol 1989;59:296–305.
33. Graesser AC, McMahen CL. Anomalous information triggers questions
when adults solve problems and comprehend stories. J Educ Psychol
1993;85:136–51.
34. Keppel R, Weis W. Time and distance as solvability factors in murder
cases. J Forensic Sci 1994;39(2):386–400.
35. Gregory RL, Wallace JG. Recovery from early blindness—a case study.
Experimental Psychology Society Monograph 2, 1963;31; www.richard-
gregory.org/papers/recovery_blind/contents.htm (accessed December 30,
2013).
36. Peirce CS. Pragmatism and pragmaticism. In: Hartshorne C, Weiss P,
editors. Collected papers of Charles Saunders Peirce, Vol. V. Cambridge,
MA: Harvard University Press, 1978/1931;5.171.
37. Kind S. The scientific investigation of crime. Manchester, U.K.: Forensic
Science Services, 1987;43.
38. Wikimedia. Sculptures on the Basle M€
unster (CH). http://commons.wikime-
dia.org/wiki/File:Elefant_am_Basler_Muenster.jpg http://www.baslermuen-
ster.ch/bauwerk/rundgang/aussenbau/chor (accessed December 27, 2013).
39. Russell B. The problems of philosophy. Oxford, U.K.: Oxford University
Press, 1912;89.
40. Piserzewska M. Kriminalistisches denken [M.Sc. thesis]. Z€
urich: Univer-
sity of Z€
urich, 2012.
41. Federal Bureau of Investigation (FBI). Headline archives anthrax investi-
gation–closing a chapter. Aug. 6, 2008; www.fbi.gov/page2/august08/
amerithrax080608a.html (accessed December 30, 2013).
42. Carson SH, Peterson JB, Higgins DM. Decreased latent inhibition is
associated with increased creative achievement in high-functioning indi-
viduals. J Pers Soc Psychol 2003;85(3):499–506.
43. Rosenthal R, Rubin DB. Interpersonal expectancy effects: the first 345
studies. Behav Brain Sci 1978;1(3):377–415.
44. Trapp C. Staging des Tatorts und andere Inszenierungen bei T€
otungsde-
likten. Z€
urich: Schulthess Verlag, 2012;13–25.
Additional information and reprint requests:
Henriette Haas, Ph.D.
Adjunct Professor of Forensic Psychology
Psychologisches Institut der Universit€
at Z€
urich
Binzm€
uhlestr. 14/1 Box 1
CH-8050 Z€
urich
Switzerland
E-mail: henriette.haas@psychologie.uzh.ch
10 JOURNAL OF FORENSIC SCIENCES