Testing the reflection assumption
| Date | 01 July 2018 |
| Author | Anna Carline,Nilda Karoğlu,Heather D Flowe |
| Published date | 01 July 2018 |
| DOI | 10.1177/1365712718782996 |
| Subject Matter | Articles |
The International Journal of
Evidence & Proof
Testing the reflection assumption:
2018, Vol. 22(3) 239–261
ª The Author(s) 2018
A comparison of eyewitness ecology
Article reuse guidelines:
sagepub.com/journals-permissions
in the laboratory and criminal cases
DOI: 10.1177/1365712718782996
journals.sagepub.com/home/epj
Heather D Flowe
School of Psychology, Centre for Applied Psychology, University of Birmingham, Birmingham, UK
Anna Carline
Leicester Law School, University of Leicester, Leicester, UK
Nilda Karog˘lu
University of Kent, Canterbury, Kent, UK
Abstract
In the US, experts are often called upon to provide evidence during criminal trials regarding
eyewitness identification research. A key factor is probative value: to what extent are findings from
laboratory studies generalisable to the real world? In order to answer this question, this article
explores the issue of eyewitness ecology, a term referring to the environmental context in which
people witness crimes, which includes characteristics of perpetrators and the viewing conditions,
as well as the identification context. Specifically, we explore the extent to which the typical eye-
witness ecology found in laboratory studies reflects or is similar to real-world conditions. We
coded the characteristics of the published literature on criminal identification in the laboratory (n
¼ 309), and the results were compared to the characteristics of a stratified random sample of
felony cases (n ¼ 721) obtained from a large metropolitan district in the United States. This analysis
demonstrated that in the criminal cases compared to the laboratory studies, duration of exposure
to the culprit and retention interval length were significantly longer, and weapons, violence and
showup identifications were more prevalent. Additionally, the laboratory studies and criminal
cases differed with respect to participant/witness race. These findings indicate a need to broaden
the range of conditions employed in the laboratory to increase the applicability of eyewitness
identification research to the legal system.
Keywords
archival studies, estimator variables, expert evidence, eyewitness ecology, eyewitness
identification, field studies, lineup
Corresponding author:
Heather Flowe, School of Psychology, University of Birmingham, Birmingham, UK.
E-mail: H.Flowe@bham.ac.uk
240
The International Journal of Evidence & Proof 22(3)
In the US, experts in psychology and the law are often called upon to share their knowledge and
understanding of the eyewitness identification literature with legal practitioners. For eyewitness experts,
this often means testifying for the defence regarding the factors that can negatively impact eyewitness
memory (e.g. Kassin et al., 1989, 2001; Pezdek, 2009), or making policy recommendations for the
handling and preservation of eyewitness testimony (e.g. Wells et al., 1998). Inherent in these applica-
tions is a reflection assumption: a term we use to refer to the presumption that the research conditions
found in eyewitness identification research reflects the eyewitness context in ‘real-world’ cases. Stated
another way, individuals summarising how different factors affect eyewitness performance are assuming
that the empirical findings, the vast majority of which come from laboratory studies, will generalise to
real-world cases (e.g. Pezdek, 2009; Yarmey, 2001). However, whilst the reflection assumption may be
reasonable, it has been given little systematic empirical treatment.
Many of the debates regarding the applicability of laboratory studies to the legal system have
revolved around the question of generalisability. Some have argued that we are in a position to generalise
because the laboratory methods used to study eyewitness identification are diverse enough to capture the
essential characteristics of real-world crimes (e.g. Deffenbacher, 1984; Haber and Haber, 2000; Loftus,
1983; Pezdek, 2009; Yarmey, 1997, 2001), whereas others have cautioned that we must be sensitive to
differences between the two settings with respect to the environmental and emotional context in which
eyewitness identifications are carried out (e.g. Elliott, 1993; Flowe et al., 2009; Konecˇni and Ebbesen,
1986; Malpass and Devine, 1981; Pachella, 1986; Yuille, 1993). For example, Weiner et al. pointed out
that some laboratory studies may oversimplify the visual behaviour of actual witnesses, as participant
witnesses have their attention directed to the perpetrator, making generalisations difficult. Yuille and
Wells (1991) have argued for the necessity for more comparisons between experimental research and
field contexts so that the similarities and differences between the two can be enunciated. Although we
would also want to know how memory performance between experimental research and field contexts
differs, a more basic and unanswered question is: How different are eyewitnessing conditions in the
laboratory and in actual criminal cases?
In this article, we examine eyewitness ecology, a term we coin here to refer to the context in which
people witness crimes and take eyewitness identification tests, both in the laboratory and in the real
world. The aim is to describe the ecological range that has been employed in laboratory research,1 and
compare it to those found in real-world cases. The primary reason to undertake this study is because the
results from laboratory research are often applied to eyewitnesses in the real world. First, a search of US
appellate cases on Lexis/Nexis database reveals that eyewitness experts testify in a wide range of
different types of cases. When experts testify in these cases, the underlying assumption is that the
eyewitness ecologies in the laboratory studies are similar enough to real-world eyewitness environments
to warrant theoretical generalisation. Our focus on eyewitness ecology is inspired by the work of Egon
Brunswik, who argued that researchers should particularise their experiments so that they are represen-
tative of the ecology, or habitat, to which generalisations are intended (Hammond and Stewart, 2001).
Second, procedural recommendations2 derived from laboratory research affect all cases in the legal
system once they are implemented, not just potentially problematic ones in which the witnessing
conditions are poor. Accordingly, it is important to compare the range of eyewitness ecologies in studies
with those found in real-world cases and assess the extent to which they overlap. In the main, we are
undertaking what Malpass et al. (2008) referred to as a study space analysis, wherein individual factors
in each of the studies (e.g. independent variables, dependent variables, methodological and procedural
strategies) are defined in a matrix. Subsequently, the individual study matrices are merged so that each
entry identifies the number of studies that utilised the corresponding characteristics.
1. Throughout, the laboratory studies will be interchangeably referred to as ‘laboratory studies’ or ‘the studies’.
2. Relating to, for example, best practice regarding the collection and preservation of eyewitness evidence.
Flowe et al.
241
The first objective of the research was to analyse the overall distribution of the conditions found in the
laboratory and ‘archival cases’: real-life felony cases that were referred for prosecution. Second, the
ecological conditions of laboratory and archival cases were compared to determine whether there were
areas in which additional research seemed warranted. For instance, how long are eyewitnesses exposed
to the culprit in the archives and in the laboratory? How often are eyewitnesses subjected to violence?
How often do eyewitnesses interact with the perpetrator before the onset of the crime? What are the
lower and upper limits for the retention interval between the crime and the identification test? Does the
retention interval length vary depending on the type of identification procedure employed? Answers to
such questions can offer some guidance to researchers who are interested in utilising real-world para-
meters to inform the types of procedures and methods they use to study eyewitness identification in the
laboratory. Our study space analysis revealed differences between the real-life cases and laboratory
studies, particularly with respect to the race of the participant/witness, the duration of exposure to the
culprit and retention interval length, and the use of weapons, violence and showup identifications.
Methodology
Laboratory studies
Sampling. A comprehensive literature search was performed using the Psychological Abstracts Informa-
tion Services (PsycINFO) and PsychARTICLES. PsychINFO covers publications from almost 2,500
scholarly peer reviewed journals from more than 50 countries and PsychARTICLES covers 110 journals,
including all APA published journals and Canadian Psychological Association journals. Our search
covered the years 1597–2014, which are dates that coincide with a recent National Academy of Sciences
report on eyewitness research (National Research Council, 2014). A selection of key words and title
word searches related to our project aim (e.g. eyewitness, lineup) were entered as appropriate. Our only
criteria for inclusion were that study participants were presented with a lineup or a showup task. We
chose to include only published laboratory studies, because courts, in determining reliability, may
consider whether the scientific evidence and methodology has been tested and subjected to peer review
or publication and generally accepted by the research...
To continue reading
Request your trial