WK5: DNA EVIDENCE

RESEARCH ARTICLE

Case rethinking: a protocol for reviewing criminal investigations

D. Kim Rossmo*

Center for Geospatial Intelligence and Investigation, School of Criminal Justice, Texas State University, San Marcos, TX, USA

Mistakes in decision-making have been identified as the most common type of error in police investigations. Consequently, wrongful convictions and other types of crimi- nal investigative failure may require a complete case ‘rethinking,’ particularly when new evidence disrupts the existing theory. A rush to judgment resulting in a prema- ture shift from an evidence-based to a suspect-based investigation can produce a number of problematic thinking errors. Faulty assumptions, tunnel vision, groupthink, and other cognitive biases and organizational traps hinder evidentiary interpretation and evaluation. This article outlines a protocol for reviewing evidence and rethinking a wrongful conviction or unsolved crime when the existing investigative theory appears to be incorrect. The protocol involves four stages: (1) evidence; (2) interpre- tation; (3) patterns; and (4) analysis.

Keywords: criminal investigation; police detectives; criminal investigative failures; cold cases; wrongful convictions

Introduction

Police agencies increasingly use sophisticated forensic techniques and powerful computers in their response to major crimes. What can sometimes be forgotten in this emphasis on technology, however, is the need to properly think about the information produced by these and other approaches. Police investigations require rational analysis and logical thinking. When cognitive biases and organizational traps distort the process, wrongful convictions and other types of criminal investigative failure can result.

This article presents a case rethinking protocol that has been successfully used on unsolved crimes and wrongful convictions. The focus is on inquiries of crimes, such as murder, where what happened remains unknown. Reviewing a case often involves the effort to discover new evidence; however, this method is concerned with rethinking existing evidence. Aspects of the protocol may also be helpful in guiding new investiga- tions.

The types and causes of thinking errors that can obstruct a criminal investigation are first discussed. A protocol for the systematic review of a wrongful conviction or unsolved case in which the existing investigative theory appears incorrect is then out- lined. The protocol involves four stages: (1) evidence; (2) interpretation; (3) patterns; and (4) analysis.

*Email: [email protected]

© 2014 Taylor & Francis

Police Practice and Research, 2016 Vol. 17, No. 3, 212–228, http://dx.doi.org/10.1080/15614263.2014.978320

Criminal investigative failures

A criminal investigative failure is defined as an unsolved crime that should have been solved or a crime that was incorrectly ‘solved’ (i.e., a wrongful conviction). Many people tend to believe the criminal justice system is fair and competent, with police, prosecutors, and jurors logically basing their decisions on reliable evidence. Unfortunately, as The Innocence Project and similar groups have shown, innocent people are sometimes prosecuted and convicted. The reality is the system makes mistakes. It is difficult to determine how often such errors happen, but estimates of the frequency of wrongful convictions range from 0.5% (all felonies; Huff, Rattner, & Sagarin, 1996), to 4.1% (death row inmates; Gross, O’Brien, Hu, & Kennedy, 2014), to 5% and higher (murder and sexual assaults; Roman, Walsh, Lachman, & Yahner, 2012).

Eyewitness misidentifications, improper forensic science, and false confessions are the major causes of wrongful convictions (Innocence Project, 2013). However, investiga- tive thinking errors often underlie these direct causes. Mistakes in decision-making have been identified as the most common type of error in police investigations (Irvine & Dunningham, 1993). Problems can arise when police rush to judgment during the early stages of a case and prematurely shift the investigation from evidence-based to suspect-based. If this occurs before all the evidence has been collected and analyzed, there is a significant risk of tunnel vision.

Human beliefs are the product of expectations, desires, and evidence; the more ambiguous the last, the stronger the influence of expectations and desires (Snook, 2000). ‘There can be little doubt that our beliefs influence the processes by which we seek out, store, and interpret relevant information’ (Ross & Anderson, 1982, p. 149). It is therefore important to collect as much evidence as possible before theorizing about the crime and suspects; once that has happened, expectations and desires start influenc- ing the thinking process.

The need for an alternative approach may become evident if DNA reveals a wrong- ful conviction or disrupts the dominant theory, or a crime is still unsolved after a lengthy period. However, rethinking an investigation can be extremely difficult and challenging. Cognitive biases and organizational traps may exist that hinder evidence interpretation and evaluation. As Heuer (1999, p. 125) warns, ‘once information rings a bell, the bell cannot be unrung.’

Thinking errors

When my information changes, I alter my conclusions. What do you do, sir? (John Maynard Keynes)

In an ideal world, we would make the best possible decisions after a careful evalua- tion of all available evidence. In reality, our thinking is frequently impaired by cognitive biases. Within the context of a criminal investigation, such systematic errors in thinking can result in an unsolved crime or a wrongful conviction (Jones, Grieve, & Milne, 2008a; Rossmo, 2009). Tunnel vision and confirmation bias are amongst the most prob- lematic thinking errors that can be made by detectives. Faulty assumptions, probability errors, and groupthink sometimes play supporting roles.

Police Practice and Research: An International Journal 213

The initiating problem is often a rush to judgment – reaching a conclusion before all the evidence has been considered. A premature judgment can lead to tunnel vision and confirmation bias, prevent subsequent evidence from being properly evaluated, and result in a criminal investigative failure. These errors and the aggravating role of groupthink are discussed below.

Premature judgment

For many reasons, detectives may jump to conclusions before an investigation is com- plete. Community fear, media and political pressure, organizational stresses, personal ego, and a compelling desire to apprehend a dangerous offender can all result in a pre- mature arrest. The pressures are exacerbated if the crime is particularly heinous or the victim is a young child.

An overreliance on intuition can lock in a premature judgment. Humans have both intuitive and rational decision-making processes (Kahneman, 2003); while most people are familiar with the former, the latter is often misunderstood. Intuition (sometimes mistakenly referred to as ‘gut instinct’) operates at a below-consciousness level and is automatic, fast, and powerful (Myers, 2002). It is learned slowly and typically involves the use of heuristics. Because of its implicit nature, intuition is difficult to control and is prone to error; it exists because it helps promote survival, not because of its accuracy. Different situations require different types of judgment and when a decision has to be made quickly under conditions of uncertainty, intuition can be useful (Wright, 2013). But when we have reliable data and adequate time, reasoning is the better option; com- plex tasks, such as major crime investigations, require analysis and logic (Jones et al., 2008a).

‘Investigators need not be especially sure that they have the right person to sway their investigation toward an early suspect’ (O’Brien, 2009, p. 328). When that happens, police prematurely move from an evidence-based to a suspect-based investigation (Rossmo, 2009). In the former mode, detectives have not yet determined who the offen- der is; they are still searching for and gathering information and evidence to determine what happened during the crime and who might be a suspect. In the latter mode, detec- tives have decided they know who the guilty person is, and their investigation shifts to preparing for prosecution.

The central problem therefore originates from judgments based on only subsets rather than the totality of evidence. Evidence discovered post-judgment is then less likely to be evaluated in an unbiased manner, and relevant evidence that fails to support the investigative conclusion may not be collected or properly analyzed; in some instances, it may not even be recognized as evidence.

If a flawed forensic test is done early in an investigation, it can have a powerful effect because of the authority given to science (Garrett & Neufeld, 2009). However, forensic science is not always very ‘scientific,’ and some forensic analyses are more subjective than objective. There have been a number of wrongful convictions attrib- uted to bad forensic science (National Research Council, 2009). The uncritical accep- tance of flawed forensic results at the beginning of a case can result in premature judgment and poor investigative decision-making, including the dismissal of important evidence later discovered simply because it conflicts with the forensic findings. Inves- tigations have been sent down the wrongful conviction road by several types of unsound forensic analyses, including time of death estimates (Michael Morton, Texas),

214 D.K. Rossmo

sexual assault examinations (David Camm, Indiana), blood testing (Greg Taylor, North Carolina), bite mark comparisons (Willie Jackson, Louisiana), and arson indicators (Cameron Todd Willingham, Texas).

Faulty assumptions

Assumptions are often necessary in a criminal investigation, especially during its early stages when information tends to be limited. As evidence is collected, however, it is important to reevaluate all investigative assumptions and discard those no longer valid. Unfortunately, this does not always happen. Given enough time, it is even possible for an unsubstantiated assumption to reify and harden into ‘fact.’ If investigators working from a faulty assumption develop tunnel vision, they are unlikely to later detect the underlying error. During the DC Sniper case, the assumption that the killers were driving a white van evolved into a near-certainty during the early days of the investigation (Horwitz & Ruane, 2003). However, this white van – in fact, originally reported as a white box truck – did not exist; the killers were driving a blue Caprice sedan.

Probability errors

A detective may have to estimate the probability of a particular event in order to decide if it is significant to the investigation or merely coincidence. However, humans tend to have a poor understanding of probability, particularly at the intuitive level (Gigerenzer, 2002). As a result, probability mistakes commonly occur; in the context of a criminal investigation, these mistakes can create tunnel vision and sustain confirmation bias. Several cases have been sabotaged by probability errors originating from the flawed estimates of investigators, forensic scientists, medical pathologists, profilers, and prosecutors. Problem areas include violating the laws of probability, ignoring base rates, using vague language (‘common,’ ‘rare,’ ‘risky’), and failing to understand the nature of randomness in large investigations involving many suspects (Rossmo, 2009).

The tragic case of British solicitor Sally Clark is an infamous example of injustice caused by a probability error. Clark was convicted of murdering her two sons who died from sudden infant death syndrome (SIDS). The only evidence against her was the assertion that the occurrence of two SIDS deaths in a single family was highly improbable, an estimate calculated by a pediatrician, Roy Meadow. Unfortunately, he made several errors in his analysis: he ignored genetic effects and incorrectly assumed SIDS cases are independent within families; he committed an ecological fallacy by equating individual-level risk to overall population risk; and he failed to realize that SIDS is common enough on a national level that a recurrence happens to some unfor- tunate family in England about once every 18 months (Hall, 1999). The Royal Statisti- cal Society publically criticized Dr Meadow’s estimate and Clark’s conviction was eventually quashed on appeal, but only after she had spent more than three years in prison.

Police Practice and Research: An International Journal 215

Tunnel vision

Tunnel vision (also called incrementalism) results from a narrow focus on a limited range of alternatives.

Tunnel vision is insidious …. It results in the [police] officer becoming so focussed upon an individual or incident that no other person or incident registers in the officer’s thoughts. Thus, tunnel vision can result in the elimination of other suspects who should be investi- gated. (Cory, 2001, p. 37)

Tunnel vision can be a product of satisficing, or the selection of the first identified alternative that appears ‘good enough’ (Findley & Scott, 2006; Simon, 1956). Arrest- ing the first likely suspect, then closing the investigation off to alternative theories, is a recipe for disaster. Not surprisingly, tunnel vision has been identified as a major cause of wrongful convictions (FPT Heads of Prosecutions Committee Working Group, 2004).

Gould, Carrano, Leo, and Young (2013) link tunnel vision to the broader issue of systemic failure in wrongful convictions: ‘tunnel vision helps explain how one error often leads to additional errors in an erroneous conviction. It contributes and facilitates system breakdown because it dismantles the rigorous testing of evidence that makes the investigative and adversarial processes function effectively’ (p. xxi). They also connect tunnel vision to the concept of escalation of commitment (also known as sunk costs; Brockner, 1992; Coleman, 2010; Staw, 1981) to explain why entrenchment of beliefs occur even when strong contradictory evidence has emerged:

As more resources – money, time, and emotions – are placed into a narrative involving a suspect, the actors involved are less willing or able to process negative feedback that refutes their conclusions. Instead, actors want to devote additional resources in order to recoup their original investment. As a result, evidence that points away from a suspect is ignored or devalued, and latent errors are overlooked. At this point, the police are working to rule in rather than rule out the suspect, and prosecutors have moved from ‘inspection’ mode to ‘selling’ mode.1 Escalation of commitment contributes and facilitates system breakdown because it dismantles the rigorous testing of evidence that makes the adversarial process function effectively. (pp. 86–87)

Confirmation bias

Confirmation bias (also called verification bias) is a type of selective thinking. Once a hypothesis has been formed, our inclination is to confirm rather than refute – we tend to look for supporting information, interpret ambiguous information as consistent with our hypothesis, and minimize any inconsistent evidence (Koehler, 1991). Types of confirma- tion bias include biased search for evidence, biased interpretation of information, and biased memory (selective recall).2

Confirmation bias can cause a detective to focus on evidence confirming the investi- gative theory, while ignoring or refusing to look for contradicting evidence (Stelfox & Pease, 2005). Existing evidence is interpreted in a biased manner; evidence that supports the investigative theory is taken at face value, while contradicting evidence3

is skeptically scrutinized (Ross & Anderson, 1982). Research has shown experienced police investigators consider witnesses less credible if they exonerate a preferred suspect than if they help confirm guilt (Ask & Granhag, 2007). Other manifestations of confirmation bias include the failure to search for evidence that might prove a

216 D.K. Rossmo

suspect’s alibi, not utilizing such evidence if found, and refusing to consider alternative hypotheses.

While the recording and documenting functions integral to a professional criminal investigation minimize the risk of biased memory, there have been some instances of detectives ‘forgetting’ to document important evidence. There have also been several cases where a prosecutor decided exculpatory evidence was not relevant to the case and failed to share it with defense counsel. It is difficult to say whether such actions are the result of rationalizations fed by confirmation bias or the product of outright unethical behavior. In the wrongful conviction of Michael Morton for his wife’s murder, Ken Anderson, the district attorney of Williamson County, Texas, failed to inform Morton’s defense counsel of key physical and witness evidence that eventually identified the real killer (Colloff, 2012). Anderson was eventually sentenced to a jail term after being found in contempt of court and disbarred (he had become a judge).

Confirmation bias can be detected in a criminal investigation by identifying:

� evidence ‘missed’ post-judgment (ignored leads, unexplored lines of inquiry, lost evidence, unanalyzed evidence); and

� biased treatment of post-judgment evidence (disregarded, downplayed, or distorted evidence, illogically considered evidentiary implications).

Confirmation bias can be tested for by considering what would happen if the order of evidential discovery was changed. The conclusions reached by detectives should not depend on the particular sequence in which the evidence was discovered. If altering the evidential order changes the case conclusion, there is likely a problem with the investi- gative logic.

The strength and perseverance of confirmation bias is dramatically illustrated by the convoluted theories put forth by those prosecutors and police who ‘never can admit a mistake.’ Prosecutors are normally thankful for DNA evidence in a criminal case; however, their reaction can be quite different if the DNA establishes the innocence of someone they originally believed guilty. After the bodies of two little girls were found in a remote wooded area in Zion, Illinois, police arrested and charged the father of one of the victims with murder. When sperm recovered from the girl’s vagina, mouth, and anus was tested and found not to match her father’s DNA, the prosecutor dismissed the laboratory results, suggesting the victim contacted the sperm while playing in the woods where couples met to have sex (Martin, 2011). Later, after the DNA was matched to a convicted violent serial rapist who was a friend of the victim’s brother, the prosecutor again argued the sperm was unrelated to the crime. This time, he theorized the rapist masturbated while sleeping over at the house of the victim’s brother. Later, while the victim was watching a movie from the same bed, she touched his sperm and subsequently transferred it to her other body parts (i.e., the inside of her vagina, mouth, and anus).

In another example, from a Canadian case, DNA testing of sperm found on the victim’s clothing matched a convicted rapist from her own neighborhood, not the man who spent 23 years in prison for her murder. Rather than acknowledging the failure, some police officers tried to explain away the DNA by suggesting the convicted man had actually killed her. Then, after he left her body in a back alley, the rapist came along and had sex with her frozen corpse before it was discovered (Boyd & Rossmo, 1994).

Police Practice and Research: An International Journal 217

Ross and Anderson (1982) observed:

it is clear that beliefs can survive potent logical or empirical challenges. They can survive and even be bolstered by evidence that most uncommitted observers would agree logically demands some weakening of such beliefs. They can even survive the total destruction of their original evidential bases. (p. 149)

Stubborn and irrational belief perseverance by the state in the face of significant conflicting new evidence is inconsistent with both justice and reality.

Groupthink

Within organizations with strong subcultures like the police, groupthink can enable and exacerbate confirmation bias. Groupthink is the reluctance to think critically and chal- lenge the dominant theory. It occurs in highly cohesive groups under pressure to make important decisions. The main symptoms include (Janis, 1982):

(1) Power overestimation (a) belief in the group’s invulnerability, resulting in unwarranted optimism and

risk taking; and (b) ignoring the ethical consequences of decisions because of a belief in the

morality of the group’s purpose. (2) Close-mindedness

(a) group rationalizations and the discrediting of warning signs; and (b) negative stereotyping of the group’s opponents (they are regarded as evil,

stupid). (3) Uniformity pressures

(a) conformity pressures (those who disagree with the dominant views or decisions are seen as disloyal);

(b) self-censorship (the withholding of dissenting views and counterarguments); (c) shared illusion of unanimity (silence is perceived as consent, and there is an

unsubstantiated belief everyone agrees with the group’s decision); and (d) self-appointed mindguards (group members shield each other from conflict-

ing information).

Groupthink has several negative outcomes (FPT Heads of Prosecutions Committee Working Group, 2004). Afflicted groups selectively gather information and fail to seek expert opinions. They neglect to critically assess their ideas, examine few alternatives, and fail to develop contingency plans. Groupthink in a major crime investigation perpet- uates confirmation bias and preserves flawed decisions.

Case rethinking protocol

The following protocol has been designed for the systematic review of a case in which the investigative theory appears to be wrong.4 Such a realization may follow the discov- ery of significant new evidence that disrupts the existing theory, but it can also emerge after a case has remained unsolved for an extended period of time. Rethinking an

218 D.K. Rossmo

investigation first requires a detective to unravel the case – wipe the slate clean, unlearn what is believed, and abandon existing theories and suspects.

Crimes are solved by information derived from evidence (Willmer, 1970). The purpose of the protocol is to create a focus on the evidence, its strengths and weaknesses, and the overall evidentiary pattern. Facts must be distinguished from beliefs, certainties from suspicions, and probabilities from possibilities. An investiga- tor must be able to answer the question, ‘How do you know what you think you know?’

The protocol involves four stages: (1) evidence; (2) interpretation; (3) patterns; and (4) analysis. Each stage is explained in detail below, and then briefly illustrated in a text box using a short example based on an evidence item – the written statement of an eye- witness – from a sexual murder case.

Evidence

Man prefers to believe what he prefers to be true. (Francis Bacon)

The first step in rethinking a case is to identify the evidence5 in the investigation and assess its reliability. Evidence is a tangible and recorded fact relevant to the crime. In some situations, the absence of something can be considered evidence. Theories, assumptions, and inferences are not evidence.

It is essential to identify the origin of each item of evidence, whether it is a police report, crime scene photograph, witness statement, or laboratory analysis. Nothing should be taken for granted. Rumors emerge in large investigations that can solidify into ‘facts’ after sufficient retelling; ‘creeping credibility’ can turn a hypothesis into received knowledge over time. If a record cannot be found, then there is no evidence. Identifying the evidence and its supporting documentation establishes the ‘facts of the case’ and the information content of the investigation.

There are only three ways to solve a crime: a witness; a confession; or physical evidence (Klockars & Mastrofski, 1991). Detectives typically have a good understanding of the technical nature of these evidence types, including their benefits and problems. However, they do not always fully understand the probative value of evidence (Robertson & Vignaux, 1995). Evidence has both significance and reliability (Griffith & Tversky, 2004). People tend to place more importance on significant evidence even if its reliability6 (the probability of its truthfulness) is low. All evidence has an error rate – eyewitnesses make misidentifications, suspects give false confessions, and scientific tests produce false positives. The possibility of mistakes and human error always exists. It is therefore necessary to estimate the reliability of an item of evidence in order to determine how much weight it should be given. Source reliability, forensic test error rates, research on evidentiary consistency/trustworthiness, and any other known issues should be considered. While it is usually impossible to assign exact probabilities, investigators can roughly evaluate evidence on an ordinal scale from low to high reliability. Evidence should not be considered more reliable because the investigator wants it to be, and it should not be deemed unreliable simply because it is inconsistent with the prevailing investigative theory.

Police Practice and Research: An International Journal 219

Evidence The evidence item used to illustrate the protocol is a written statement from an

eyewitness in a sexual murder that occurred in Saskatoon, Saskatchewan. The victim was Gail Miller, a nursing assistant who was attacked while walking to the bus stop on her way to work on an icy January morning. She was raped and stabbed; her body was later found lying in the snow in a back alley a block from her home (Boyd & Rossmo, 1994).

The witness, Nichol John, was one of three Regina teenagers, along with David Milgaard and Ron Wilson, who set out on a road trip to buy drugs. They drove to Saskatoon to pick up another friend (who had money), but ended up lost before they found his house and got stuck in the snow. It was very cold, so John stayed in the car while Milgaard and Wilson went for help. John told police she then saw Milg- aard attack a woman on the street.

Written statement from witness Nichol John: After we got to Saskatoon we drove around for about 10 or 15 minutes. Then

we talked to this girl. This was in the area where Sgt. Mackie drove me around. Ron was driving the car at this time. He drove to the curb where Dave spoke to

this girl. Dave was on the outside passenger side of the front seat. Dave opened the door

to talk to this girl as she approached along the sidewalk. Dave asked this girl for directions to either down town or Pleasant Hill. He

offered to give her a ride to where ever she was going. She refused the ride. Dave closed the door and said “The stupid bitch.” We started to drive away and only went about half a block when we got stuck.

We ended up stuck at the entrance to the alley behind the funeral home. Ron and Dave got out and they tried to push the car. They couldn’t get it out. I recall Dave going back in the direction we had spoke to the girl. Ron went the

other way past the funeral home. The next thing I recall is seeing Dave in the alley on the right side of the car.

He had a hold of the same girl we spoke to a minute before. I saw him grab her purse. I saw her grab for her purse again. Dave reached into one of his pockets and pulled out the knife. I don’t know which pocket he got the knife from. The knife was in his right hand. I don’t know if Dave had a hold of this girl or not at this time. All I recall seeing is him stabbing her with the knife.

The next I recall is him taking her around the corner of the alley. I think I ran after that. I think I ran in the direction Ron had gone. I recall running down the street. I don’t recall seeing anyone. The next thing I knew I was sitting in the car again. I don’t know how I got back to the car.

Reliability assessment: uncertain. It is unlikely Nichol John is mistaken in her identification of the killer, as she knows Milgaard and was apparently situated near him when the attack occurred. However, John may be deceitful. She is a 16-year-old street girl who uses drugs. This statement is inconsistent with her first statement to the police. She was under pressure by detectives to talk about the Miller case; the day before she gave this statement she was arrested and spent the night in jail. She continued to travel with Milgaard after the murder. She now claims she cannot remember anything about the crime.

220 D.K. Rossmo

Interpretation

Evidence must be explained and interpreted to give it meaning. Potential inferences from the evidence and its context are identified in the protocol’s second stage. These are reasonable deductions that can be based on the evidence. An inference can be thought of as the extrapolation from the core facts of an item of evidence to its logical, proba- ble, or (reasonably) possible conclusions, not all of which will necessarily be correct. For example, DNA recovered from the vaginal vault of the victim in a child sex murder leads to a strong inference it was left by the killer. However, that inference is not as strong if the victim was a sex trade worker.

A likely inference from an unreliable item of evidence is still not probable; similarly, an unlikely inference from a reliable item of evidence is also improbable. While infer- ences are related to theories, they are distinct; the former is an extension from a single item of evidence, while the latter is built up from the totality of evidence in an investi- gation.

The explanation for an item of evidence – the theory of its cause – is an important inference. While there can be a tendency to focus on the most obvious explanation, full consideration should be given to all feasible options. Possible explanations can be ranked by likelihood; as this assessment might change once an item of evidence is con- sidered in relationship to the other evidence in the case, it is important not to prema- turely eliminate possibilities (existing biases become most treacherous at this point).

Interpretation John’s statement is significant and highly probative. If accurate, it identifies

Milgaard as Miller’s murderer. The following high probability inferences can be made from this item of evidence (lower probability inferences are not listed in this example):

(1) Nichol John’s statement provides an exact location for the stabbing (a loca- tion different from where Miller’s body was found), suggesting there should be blood and perhaps other physical evidence at the site.

(2) The location of the attack indicates the likely route walked by Miller to her bus stop the morning of the murder.

(3) As described, the attack dictates Miller should have stab wounds and a matching pattern of cuts in the clothing she was wearing (winter coat and nursing uniform).

Patterns

The evidence in an investigation is simply a collection of facts. To understand what happened in a crime, potential causes and inferences from the evidence have to be con- nected and understood as a narrative or story. The third stage of the protocol involves looking for patterns that can be derived from the constellation of evidence, and deter- mining connections (influences between items of evidence) and sequences (what flows from what). It is also important at this point to analyze consistencies among inferences as these suggest reliable narratives.

Police Practice and Research: An International Journal 221

In many cases, the pattern will be patchy and incomplete. In such instances, it is important to refrain from ‘filling in the blanks’ beyond what can be supported by the available evidence. In other cases, multiple competing patterns of varying probabilities will emerge. Despite the human tendency to avoid uncertainty, it is necessary to con- sider all viable options; this often requires consideration of contradictory possibilities.

Patterns Part of Nichol John’s statement is consistent with the statement of her friend,

Ron Wilson, who also said he saw Milgaard with a knife. However, John’s description of the location of the attack is not consistent with

the route the victim typically took to the bus stop. Miller usually followed the short- est path, a three-minute trip. On the morning of her murder, she left her home five minutes before the bus was due (the temperature was −42° F), and it is unlikely she would have been at the spot where John said the attack occurred.

John’s statement is also inconsistent with some of the physical evidence. No blood was found near the place where she claimed the attack happened. (It is possi- ble she was mistaken about the location, but no blood was found anywhere other than by Miller’s body.) Furthermore, while the pattern of cuts in Miller’s winter coat matched the pattern of her stab wounds, there were no cuts in her nursing uniform. This meant that, at some point, Miller’s coat was taken off, her uniform was pulled down, and her coat was put back on again – none of which was described by John.

Analysis

Up to this point the focus has been on the case evidence. In the analysis stage, evi- dence patterns are now used to generate and assess investigative theories or hypothe- ses (see Jones et al., 2008a). It is best to start with broad comprehensive categories, drilling down and becoming more specific as evidence permits. Not all the evidence may be helpful in this process. Caution is needed when working with ambiguous evi- dence, which is more vulnerable to expectations, desires, and other biases (Snook, 2000). ‘When the evidence is ambiguous, subtle influences on how an investigator sees it could affect the outcome. A weak case can start to look strong when the inves- tigator overlooks potentially fruitful leads in another direction’ (O’Brien, 2009, p. 329). To avoid this trap, it is important to consider the diagnosticity of an item of evidence.7 Diagnosticity originally referred to the ability of a medical test to identify a patient’s disease. In the context of a criminal investigation, the diagnosticity of an item of evidence is a function of its ability to distinguish between different hypothe- ses, such as a suspect’s guilt or innocence.

When developing case theories, investigators should remember Occam’s razor, the Principle of Parsimony – if more than one explanation for the evidence is possible, it is preferable to start with the simplest (the one with the fewest assumptions). The narrative of the crime should not be any more complicated than necessary to properly explain the evidence. An important principle in the physical sciences, Occam’s razor can also be a helpful investigative concept. While many things are possible in a crime, only some are probable.

222 D.K. Rossmo

In the special case of only two competing theories, the available evidence can be used to determine which is the most likely. The probability of an item of evidence given the first theory is compared to the probability of the evidence given the second theory.8

This process is then repeated for all items of evidence and the results combined. The final product indicates which theory is the more likely based on the totality of the evi- dence (Blair & Rossmo, 2010).

The last step involves determining how particular suspects fit into the various crime theories and identifying any suspect-related information that might link back to the crime, such as a relationship with the victim or prior criminal history. It should be remembered, however, that a suspect’s bad character is not direct evidence of a crime, absent information that provides a connecting link (e.g., a motive).

If it becomes necessary to use assumptions at this point, they should be written down and all the conclusions derived from them so identified. To avoid tunnel vision, the alternatives to an assumption (i.e., competing assumptions) must also be identified and considered.

Analysis There are significant inconsistencies between Nichol John’s statement and other

reliable evidence in the case. Consequently, the reliability of her statement should be reduced to low. John most likely lied in order to get the police to leave her alone. Her statement should not be relied upon for theory generation or suspect prioritiza- tion.

Unfortunately, considerable weight was given to John’s statement during the ori- ginal investigation and trial, even though she continued to claim she could not remember anything. Milgaard was convicted of Miller’s murder and spent over two decades in prison. DNA testing eventually identified the real killer, a man who used the same bus stop as the victim every morning (Commission of Inquiry Into the Wrongful Conviction of David Milgaard, 2005).

Visualization

It is important to be thorough when rethinking a case and the results for each stage of the protocol should be comprehensively recorded. It can be helpful to graphically depict the evidence in a diagram. Figure 1 shows an example of such a diagram based on a tri- ple murder case in Indiana. The wife and two children of David Camm, a former Indi- ana State Police trooper, were shot in the garage of their home. Despite having an alibi, David Camm was tried and convicted of their murders, the case relying heavily on blood spatter evidence. Following a successful appeal, DNA found on a sweatshirt recovered from the crime scene was linked to Charles Boney, a violent felon with multi- ple convictions for attacks against women. His palm print was also found on the side of the victims’ vehicle. Boney gave various conflicting explanations for his presence at the Camm house, the sixth and final version of which had him there to sell David Camm a handgun. The district attorney pursued murder charges against Boney, but he also chose to retry Camm. Figure 1 shows a simplified version of the evidentiary support for each of the three theories of the case: (1) David Camm did the murders alone, the first prosecution theory; (2) David Camm and Charles Boney did the murders together, the

Police Practice and Research: An International Journal 223

second prosecution theory; and (3) Charles Boney did the murders alone, the defense theory. Key evidence items are shown in small boxes situated under the particular the- ory they support (interpretations are not shown); the stronger the reliability of the evi- dence, the heavier the outline of the square. Clusters of consistent evidence are circled. The third theory won out in court – in separate trials, juries found Boney guilty and David Camm not guilty.

Discussion

It is easy for detectives to become overwhelmed by the details in a major crime investi- gation. The signal-to-noise ratio is often low and larger patterns can be missed because of a focus on minutiae. There is a tendency to use cognitive heuristics (‘mental short- cuts’) under such conditions, even though these can lead to cognitive biases and then to errors. One need look no further than conspiracy theorists to find examples of how biases can cause individuals to interpret and distort evidence to support the most amaz- ing conclusions.

The purpose of this protocol is to assist investigators rethink a case when the exist- ing theory appears incorrect. It requires a detective to start over by reexamining the structure of the case, beginning with its foundation – the evidence. Investigative hypoth- eses are generated only after the evidence has been inventoried and assessed, explained and interpreted, and connections and patterns identified. By focusing on and working from the evidence, the investigation becomes grounded in reality instead of being based

Figure 1. Case evidence pattern in the Camm triple murder investigation.

224 D.K. Rossmo

on beliefs, hunches, or biases. This method can help correct thinking errors resulting from a rush to judgment, tunnel vision, confirmation bias, and similar cognitive and organizational traps.

The rethinking approach requires an investigator to abandon the existing theory and adopt a completely open mind. However, this is much easier said than done. Research has shown that awareness of cognitive bias does not make it easier to avoid (Heuer, 1999). Consequently, rethinking a case is more difficult for those detectives with prior involvement in the investigation. The protocol works best when followed by someone with no previous connection to the crime, ideally, an investigator from an outside agency. This is the policy in England; unsolved murders are reviewed at prescribed time periods (e.g., 28 days, 12 months) by a senior investigating officer (SIO) who is not involved in the case. In high profile, complex, or sensitive investigations, it is recom- mended the SIO be from another police force (ACPO, 2006). External peer reviewers, for all the psychological and organizational reasons discussed above, are more apt to notice mistakes and omissions and much more likely to point them out.

Conclusion

The purpose of the case rethinking protocol is to shift the focus of an investigation from the prevailing theory/suspect to the evidence. Existing theories and assumptions can cre- ate cognitive traps that prevent a crime from being solved. There is a tendency to ‘explain away’ inconsistent evidence discovered post-theory, rather than letting it con- tribute to the development and prioritization of investigative hypotheses. This is confir- mation bias.

The protocol does not provide a magic equation for the right answer; it is nothing more than a systematic thinking tool, an analytic framework to help investigators objec- tively rethink a case. It is still up to the detective to solve the crime. However, pilots and surgeons have found the use of checklists to be helpful in minimizing mistakes – and their subsequent tragedies (Gawande, 2011). Criminal investigative failures that result in innocent people being convicted or guilty parties going free are no less tragic. Following this protocol can help reduce the number of such outcomes through a system- atic and evidence-based approach to case rethinking.

Acknowledgements I wish to thank Dr David Stubbins, Central Intelligence Agency (ret.), Deputy Chief Constable Doug LePard, Vancouver Police Department, Detective James Trainum, DC Metropolitan Police Department (ret.), Special Agent Gregg McCrary, Federal Bureau of Investigation (ret.), and the anonymous reviewers for their helpful comments and suggestions.

Notes 1. This is similar to moving from an evidence-based to a suspect-based investigation (Rossmo,

2009), or from information generation to case building (Stelfox & Pease, 2005). 2. A study on bias in criminal investigations found participants who were asked to identify a

suspect early in the process showed a greater tendency to confirm that hypothesis, and later ‘remembered’ the case evidence as being consistent with their particular suspect’s guilt (O’Brien, 2009).

3. Police investigators rarely test their theories by searching for disconfirming evidence even though it is more probative than confirmative evidence (e.g., exclusionary DNA). Once a narrative of the crime has been adopted, police tend to focus only on gathering confirmatory evidence (Stubbins & Stubbins, 2009).

Police Practice and Research: An International Journal 225

4. Templates for reviewing major crimes and evaluating cold cases have been previously proposed (e.g., Adcock & Stein, 2013; Jones, Grieve, & Milne, 2008b). The present protocol differs from these tools in its focus on how to rethink a case.

5. It may not be feasible to reanalyze every item of evidence in an investigation. The goal is to list all the nontrivial items that might have an influence on the theory of the crime and its suspects. Trivial, however, does not mean inconsistent or contradictory.

6. Reliability is used here to refer to the accuracy or ‘truthfulness’ of evidence, consistent with most of the legal literature. However, the correct scientific term is validity.

7. Heuer’s (1999) analysis of competing hypotheses matrix can be a useful tool here. 8. The probability of evidence given a theory is not equal to the probability of the theory given

the evidence; confusing cause given effect with effect given cause is known as the prosecu- tor’s fallacy (Robertson & Vignaux, 1995). Bayes’ theorem provides the mathematical rela- tionship between the two probabilities (Taroni, Aitken, Garbolino, & Biedermann, 2006).

Notes on contributor D. Kim Rossmo is the University Endowed Chair in Criminology and the director of the Center for Geospatial Intelligence and Investigation in the School of Criminal Justice at Texas State University. He has researched and published in the areas of environmental criminology, the geography of crime, and criminal investigations. He is a member of the International Association of Chiefs of Police Advisory Committee for Police Investigative Operations and is a full fellow of the International Criminal Investigative Analysis Fellowship. He has written books on criminal investigative failures and geographic profiling.

References ACPO. (2006). Murder investigation manual. Wyboston, Bedfordshire: National Centre for

Policing Excellence (Association of Chief Police Officers). Adcock, J. M., & Stein, S. L. (2013). Cold case models for evaluating unresolved homicides.

Investigative Sciences Journal, 5, 19–28. Ask, K., & Granhag, P. A. (2007). Motivational bias in criminal investigators’ judgments of

witness reliability. Journal of Applied Social Psychology, 37, 561–591. Blair, J. P., & Rossmo, D. K. (2010). Evidence in context: Bayes’ theorem and investigations.

Police Quarterly, 13, 123–135. Boyd, N., & Rossmo, D. K. (1994, February). David Milgaard, the Supreme Court and

Section 690: A wrongful conviction revisited. Canadian Lawyer, pp. 28–29, 32. Brockner, J. (1992). The escalation of commitment to a failing course of action: Toward

theoretical progress. The Academy of Management Review, 17, 39–63. Coleman, M. D. (2010). Sunk costs and the commitment to medical treatment. Current

Psychology, 29, 121–134. Colloff, P. (2012, December). The innocent man, part two. Texas Monthly. Retrieved December 9,

2013, from http://www.texasmonthly.com/story/innocent-man-part-two Commission of Inquiry Into the Wrongful Conviction of David Milgaard. (2005, March 7). Public

documents. Retrieved December 10, 2013, from http://www.justice.gov.sk.ca/milgaard/pub docs/march7/Nichol%20John/018589.pdf

Cory, P. de C. (2001). The Inquiry Regarding Thomas Sophonow. Winnipeg: Queen’s Printer. Findley, K. A., & Scott, M. S. (2006). The multiple dimensions of tunnel vision in criminal cases.

Wisconsin Law Review, 2, 291–397. FPT Heads of Prosecutions Committee Working Group. (2004). Report on the prevention of

miscarriages of justice. Ottawa: Department of Justice. Garrett, B. L., & Neufeld, P. J. (2009). Invalid forensic science testimony and wrongful convic-

tions. Virginia Law Review, 95, 1–97. Gawande, A. (2011). The checklist manifesto: How to get things right. New York, NY: Metropolitan

Books. Gigerenzer, G. (2002). Calculated risks: How to know when numbers deceive you. New York,

NY: Simon & Schuster.

226 D.K. Rossmo

Gould, J. B., Carrano, J., Leo, R., & Young, J. (2013). Predicting erroneous convictions: A social science approach to miscarriages of justice. Final report submitted to the US Department of Justice (Document No. 241389). Washington, DC.

Griffith, D., & Tversky, A. (2004). The weighing of evidence and the determinants of confidence. In A. Tversky (Ed.), Preference, belief, and similarity: Selected writings (pp. 275–299). Cambridge, MA: MIT Press.

Gross, S. R., O’Brien, B., Hu, C., & Kennedy, E. H. (2014). Rate of false conviction of criminal defendants who are sentenced to death. Proceedings of the National Academy of Sciences, 111, 7230–7235.

Hall, C. (1999, December 31). ‘Statistical error’ in child murder trial. Telegraph. Retrieved July 12, 2008, from www.telegraph.co.uk/htmlContent.jhtml?html=/archive/1999/12/31/nsal31.html

Heuer, Jr., R. J. (1999). Psychology of intelligence analysis. Washington, DC: Center for the Study of Intelligence, Central Intelligence Agency.

Horwitz, S., & Ruane, M. E. (2003). Sniper: Inside the hunt for the killers who terrorized the nation. New York, NY: Random House.

Huff, C. R., Rattner, A., & Sagarin, E. (1996). Convicted but innocent: Wrongful conviction and public policy. Thousand Oaks, CA: Sage.

Innocence Project. (2013). The causes of wrongful conviction. Retrieved December 2, 2013, from http://www.innocenceproject.org/understand/

Irvine, B., & Dunningham, C. (1993). Human factors in the quality control of CID investigations. Royal Commission on Criminal Justice Research, Study No. 21. London: HMSO.

Janis, I. L. (1982). Groupthink: Psychological studies of policy decisions and fiascoes (2nd ed.). Boston, MA: Houghton Mifflin.

Jones, D., Grieve, J., & Milne, B. (2008a). The case to review murder investigations. Policing: A Journal of Policy and Practice, 2, 470–480.

Jones, D., Grieve, J., & Milne, B. (2008b). Reviewing the reviewers: A tool to aid homicide reviews. The Journal of Homicide and Major Incident Investigation, 4, 59–70.

Kahneman, D. (2003). A perspective on judgment and choice: Mapping bounded rationality. American Psychologist, 58, 697–720.

Klockars, C. B., & Mastrofski, S. D. (Eds.). (1991). Thinking about police: Contemporary readings (2nd ed.). New York, NY: McGraw-Hill.

Koehler, D. J. (1991). Explanation, imagination, and confidence in judgment. Psychological Bulletin, 110, 499–519.

Martin, A. (2011, November 27). The prosecution’s case against DNA. The New York Times Magazine, p. MM44.

Myers, D. (2002). Intuition: Its powers and perils. New Haven, CT: Yale University Press. National Research Council. (2009). Strengthening forensic science in the United States: A path

forward. Washington, DC: The National Academies Press. O’Brien, B. (2009). Prime suspect: An examination of factors that aggravate and counteract confir-

mation bias in criminal investigations. Psychology, Public Policy, and Law, 15, 315–334. Robertson, B., & Vignaux, G. A. (1995). Interpreting evidence: Evaluating forensic evidence in

the courtroom. Chichester: Wiley. Roman, J., Walsh, K., Lachman, P., & Yahner, J. (2012). Post-conviction DNA testing and

wrongful conviction. Washington, DC: Urban Institute. Ross, L., & Anderson, C. A. (1982). Shortcomings in the attribution process: On the origins and

maintenance of erroneous social assessments. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases (pp. 129–152). Cambridge: Cambridge University Press.

Rossmo, D. K. (2009). Criminal investigative failures. Boca Raton, FL: Taylor & Francis. Simon, H. A. (1956). Rational choice and the structure of the environment. Psychological Review,

63, 129–138. Snook, S. A. (2000). Friendly fire: The accidental shootdown of U.S. Black Hawks over Northern

Iraq. Princeton, NJ: Princeton University Press. Staw, B. M. (1981). The escalation of commitment to a course of action. Academy of Management

Review, 6, 577–587. Stelfox, P., & Pease, K. (2005). Cognition and detection: Reluctant bedfellows? In M. J. Smith &

N. Tilley (Eds.), Crime science: New approaches to preventing and detecting crime (pp. 191–207). Cullompton: Willan.

Police Practice and Research: An International Journal 227

Stubbins, D., & Stubbins, N. (2009). On the horns of a narrative: Judgment, heuristics, and biases in criminal investigation. In D. K. Rossmo (Ed.), Criminal investigative failures (pp. 99–140). Boca Raton, FL: Taylor & Francis.

Taroni, F., Aitken, C., Garbolino, P., & Biedermann, A. (2006). Bayesian networks and probabi- listic inference in forensic science. Chichester: Wiley.

Willmer, M. A. P. (1970). Crime and information theory. Edinburgh: Edinburgh University Press. Wright, M. (2013). Homicide detectives’ intuition. Journal of Investigative Psychology and

Offender Profiling, 10, 182–199.

228 D.K. Rossmo

Copyright of Police Practice & Research is the property of Routledge and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.