Received: 8 May 2019 Revised: 20 December 2019 Accepted: 8 January 2020
DOI: 10.1002/acp.3637
R E S E A R CH A R T I C L E
Cross-cultural differences in eyewitness memory reports
Nkansah Anakwah1,2 | Robert Horselenberg1 | Lorraine Hope2 |
Margaret Amankwah-Poku3 | Peter J. van Koppen1,4
1Department of Criminal Law and Criminology,
Maastricht University, Maastricht, The Summary
Netherlands Increasingly, investigators conduct interviews with eyewitnesses from different cul-
2Department of Psychology, University of
tures. The culture in which people have been socialised can impact the way they
Portsmouth, Portsmouth, UK
3Department of Psychology, University of encode, remember, and report information about their experiences. We examined
Ghana, Legon, Ghana whether eyewitness memory reports of mock witnesses from collectivistic (sub-
4Department of Criminal Law and Criminology, Saharan Africa) and individualistic (Northern Europe) cultures differed regarding quan-
VU University, Amsterdam, The Netherlands
tity and quality of central and background details reported. Mock witnesses (total
Correspondence N = 200) from rural Ghana, urban Ghana, and the Netherlands were shown stimuli
Nkansah Anakwah, Faculty of Law, Maastricht
University, Bouillonstraat 3, Maastricht 6211 scenes of crimes in Dutch and Ghanaian settings and provided free and cued recalls.
LH, The Netherlands. Individualistic culture mock witnesses reported the most details, irrespective of detail
Email: nkansah.anakwah@
maastrichtuniversity.nl type. For each cultural group, mock witnesses reported more correct central details
when crime was witnessed in their own native setting than a non-native setting,
Funding information
Erasmus Mundus Joint Doctorate Program The though for different recall domains. The findings provide insight for legal and investiga-
House of Legal Psychology (EMJD-LP), Grant/ tive professionals as well as immigration officials eliciting memory reports in cross-
Award Numbers: 2013-0036, 532473-EM-
5-2017-1-NL-ERA MUNDUS-EPJD cultural contexts.
K E YWORD S
cultural differences, eyewitness memory reports, individualism–collectivism, interview
1 | INTRODUCTION witnesses in such settings may find it challenging, particularly if
insight into culturally determined reporting norms of the witnesses
An international criminal tribunal, The Special Court for Sierra is limited.
Leone, put Charles Taylor, a former president of the West African Aside from international criminal settings, the increase in interna-
state of Liberia, on trial in The Hague. Taylor was accused of war tional migration has made it more likely that legal and investigative
crimes, crimes against humanity, and violations of international professionals in different countries will need to obtain eyewitness
human rights law during the civil war in Sierra Leone. He was alleged memory reports in cross-cultural contexts. For instance, police detec-
to have supplied arms to rebel groups in Sierra Leone in exchange of tives are increasingly likely to interview eyewitnesses from cultural
diamonds and also to have been involved in the massacre of many backgrounds different to their own. In other contexts, immigration
innocent people. In the legal proceedings that ensued in his trial, officials typically interview asylum seekers from different cultures
eyewitness evidence from sub-Saharan African witnesses was about their recollections of events and locations in order to verify
instrumental (Keith, 2012). As in the trial of Taylor, eyewitness their claims (van Veldhuizen, Maas, Horselenberg, & van Koppen,
memory reports in international criminal settings are crucial in pros- 2018). Irrespective of the case type, such interviewees will have been
ecuting alleged atrocities. However, due to the cross-cultural con- socialised into their respective cultures, and embedded in these cul-
text of international criminal settings, investigators who interview tures are norms (Hofstede, 2001). Various cultural norms may have
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2020 The Authors. Applied Cognitive Psychology published by John Wiley & Sons Ltd.
504 wileyonlinelibrary.com/journal/acp Appl Cognit Psychol. 2020;34:504–515.
ANAKWAH ET AL. 505
implications for how people view, remember, and report about their tendency for familiar environments to modulate the processing of visual
experiences and how they behave in the course of cross-cultural stimuli (Epstein, Higgins, Jablonski, & Feiler, 2007). According to Epstein
interactions (Kastanakis & Voyer, 2014; Wang, Song, & Kim Koh, et al. (2007), people activate long-term representations of spatial struc-
2017). Hence, it is entirely possible that witnesses, victims, and other tures of familiar environments to aid recall. Therefore, it is plausible that
interviewees reflect culturally determined reporting norms when eyewitnesses are likely to have superior performance when attending to
being questioned in legal and forensic contexts. Therefore, an scenes in their native cultural environment (own-setting effect) than
increased understanding of the impact of cross-cultural differences on scenes located in a different cultural environment.
interviews in forensic settings is vital (see Hope & Gabbert, 2019). Consistent with these perspectives, research suggests that individ-
The culture in which people have been socialised has been shown uals' cultural orientation can bias their perceptual processing and content
to impact both behaviour and psychological processes (Schwartz, of their reports (Boduroglu et al., 2009; Istomin, Panáková, & Heady,
Boduroglu, & Gutchess, 2014; Wang, 2004). The individualism– 2014; Masuda & Nisbett, 2006). For example, in a study comparing chil-
collectivism cultural dimension has been particularly influential in dren from three Siberian cultures, Istomin et al. (2014) found children
research exploring cross-cultural differences across various social phe- from the two cultures with holistic perception included more contextual
nomena (Triandis, 2001; Triandis, Bontempo, Villareal, Asai, & Lucca, information in their drawings than those from the culture with analytic
1988). Individualism refers to a cultural orientation where the ties perception. They also found that children from a collectivistic cultural ori-
between individuals in a society are relatively loose, whereas collectiv- entation tend to draw background objects before drawing focal objects,
ism refers to a cultural orientation where a person is embedded in a whereas the reverse was true for those from individualistic cultural ori-
complex web of social relationships (Hofstede, 1983). The entation. Istomin et al. (2014) attributed these findings to differences in
individualism–collectivism cultural dimension may lead to biases in what attention that the different cultures accord to contextual information.
is considered worthy and informative to report when people from these However, other results have been inconclusive with respect to
cultures are exposed to similar scenes (Boduroglu, Priti, & Nisbett, cultural differences in memory reporting. For example, Wong, Yin,
2009). For example, drawing on the individualism–collectivism dimen- Yang, Li, and Spaniol (2017) compared Canadian and Chinese partici-
sion, Markus and Kitayama (1991, 2003) proposed independent– pants with respect to memory for individual and background objects
interdependent cognitive styles. According to Markus and Kitayama of picture scenes. Participants were exposed to picture drawings con-
(1991, 2003), an independent construal of the self is a characteristic of taining focal and background scenes and later reported whether they
individualistic societies and features the self as having significant dispo- attended to the focal or background scene. Irrespective of partici-
sitional attributes, and as being more autonomous and independent. For pants' cultural background, participants reported attending more to
that reason, individuals with an independent self-construal become focal details than background details, and there was no difference in
more perceptually oriented towards the properties of an object than memory for focal objects between cultures. However, Canadian par-
the context (analytic perception). Accordingly, they become more prone ticipants reported attending more to background scenes than Chinese
to attend to the properties and characteristics of an object and as a participants did. Thus, there seem to be mixed findings on research on
result, narrow their attentional resources to focal objects at a visual field the influence of culture on memory.
(Boduroglu et al., 2009). In contrast, an interdependent construal of the
self, whereby individuals view the self as integrated with (i.e., not sepa-
rate from) the social context, is proposed as a characteristic of collectiv- 1.1 | The current research
istic cultures. Markus and Kitayama (1991) argued that due to the
interdependent self-construal, people from collectivistic cultures Increasingly, investigators interview witnesses from diverse cultural
become more perceptually oriented towards a broader visual field backgrounds, and given that cultural norms may influence the nature or
(holistic perception) and, as a result, are more likely to allocate their content of the information reported in such interviews, this may have
attentional resources broadly. Applying Markus and Kitayama's (1991) implications for the criminal justice system. Criminal justice profes-
framework, it might be predicted that reports about events by people sionals can be confronted with challenges when they lack the relevant
from individualistic and collectivistic cultures may differ as their cultural awareness, knowledge, and training about cultural differences in eye-
background biases them to be either analytically or holistically oriented. witness memory reports. To date, research in this area has largely been
Aside from an individual's cultural background, it has also been conducted using Western, Educated, Industrialised, Rich and Demo-
suggested that the characteristics of a cultural setting could direct atten- cratic (WEIRD; Henrich, Heine, & Norenzayan, 2010) samples, with little
tion (Masuda & Nisbett, 2006; Miyamoto, Nisbett, & Masuda, 2006). consideration of cross-cultural factors or comparisons. There have been
Proponents of that perspective have argued that irrespective of their cul- calls for cross-cultural research to go beyond Western borders to
tural background, individuals are likely to detect changes to focal objects enhance our understanding of cultural variations in behaviour (Brady,
of scenes from individualistic cultures than scenes from collectivistic cul- Fryberg, & Shoda, 2018; Gelfand, Harrington, & Jackson, 2017) and,
tures. Conversely, they argue that individuals, regardless of their cultural more specifically, to appreciate cultural differences relevant for the field
background, are more likely to detect changes to contextual objects for of investigative interviewing (Hope & Gabbert, 2019).
scenes from collectivistic cultures than scenes from individualistic cul- Consequently, efforts are being made in psychological science to
tures (Masuda & Nisbett, 2006). Other researchers have observed a explore other non-WEIRD samples. However, a recent meta-analytic
506 ANAKWAH ET AL.
review revealed that, even for the small proportion of non-WEIRD for inclusion is consistent with previous research (Hofstede, 1983,
populations studied in cross-cultural research, the majority of these 2001).1 Out of the 207 participants recruited, seven were excluded.
non-WEIRD populations were from East Asia (collectivistic culture), These participants were excluded because they did not follow instruc-
with only 0.63% of the non-WEIRD sample populations from Africa tions (n = 2), viewed only three out of the four scenes (n = 4), and had
(Veillard, 2017). Hence, in the current study, we sampled participants East Asian parents although born in the Netherlands (n = 1). Our final
from sub-Saharan Africa (typifying collectivistic culture) and Western sample comprised 200 participants (103 males and 97 females;
Europe (typifying individualistic culture). Within the collectivistic cul- Mage = 28.44, SD = 12.43). The urban sample (n = 70; Mage = 26.39,
ture, we were also interested in comparing rural and urban cultures, SD = 10.79) in Ghana was recruited in the capital city, Accra, whereas
as the latter tends to be less collectivistic than the former (Rooks, the rural sample (n = 75; Mage = 31.61, SD = 14.29) was recruited in
Klyver, & Sserwanga, 2016). This difference is likely due to the fact Akim Aduasa, a farming community in the Eastern Region of Ghana.
that urban centres are prone to cultural infiltration, and there is Participants from the Netherlands (n = 55; Mage = 26.78, SD = 10.96)
greater exposure to western cultural values in urban areas than in were recruited in Maastricht, a provincial capital in the south of the
rural areas (Ma, Pei, Jin, & De Wit, 2015). To date, the literature on country. Student participants in the Netherlands were awarded course
cross-cultural cognition has rarely made the distinction between rural credits, whereas nonstudent participants received a €5 shopping
and urban dwellers in collectivistic societies. To address this issue in voucher. Student and nonstudent participants from Ghana received a
the current research, we compared eyewitnesses from Western GHС| 5 voucher for phone credit.
European culture with eyewitnesses from urban and rural sub-Saharan The design for the study was a 3 (cultural group: rural Ghana,
African cultures. urban Ghana, The Netherlands) × 2 (crime setting: Ghanaian setting,
Mock witnesses from sub-Saharan Africa and Western Europe Dutch setting) mixed factorial design. The between-group variable
viewed stimuli scenes presenting African and European settings and was cultural group, and the within-group variable was crime setting.
reported what they saw in a free recall test. Afterwards, they were The dependent variables were correct, incorrect, and withheld (Don't
asked cued recall questions that focused on both central and back- know) details, for both central and background information.
ground details of the scenes. Drawing on theory and previous findings,
we predicted cultural differences in the types of details reported by the
cultural groups. Specifically, we expected Western European mock wit- 2.2 | Materials
nesses to report more focal details about the crime scene than sub-
Saharan African mock witnesses. Conversely, we expected sub-Saharan 2.2.1 | Stimuli
African mock witnesses to report more contextual details than Western
European mock witnesses. Among sub-Saharan African mock wit- The stimuli used were eight photographs rich in central and back-
nesses, we expected differences between mock witnesses from rural ground details. The photographs depicted four crime scenarios (theft,
and urban areas in the type of details reported. Specifically, we assault, accident, and robbery). Each of these crime scenarios was
predicted that those from rural sub-Saharan Africa would report more photographed in a Ghanaian setting as well as in a Dutch setting. For
contextual details than those from urban sub-Saharan Africa. Mock wit- example, for a crime depicting, a theft in a Ghanaian setting, the same
nesses from urban sub-Saharan Africa were expected to report more crime was depicted in a Dutch setting. Each participant viewed four
focal details than those from rural sub-Saharan Africa. We also of these stimuli (two stimuli each for Dutch and Ghanaian settings).
expected that cultural setting would play a role in the memory reports The stimuli were prepared in the Netherlands and Ghana. Scenarios
of mock witnesses of all cultural groups. Specifically, we predicted that were prepared with a very clear central event that was distinct from
mock witnesses across cultures would report more central details about the background.
Western European cultural settings than sub-Saharan African cultural Two of the stimuli (one Ghanaian setting and one Dutch setting)
settings. Mock witnesses across cultures were also predicted to report were piloted in the respective countries. A total of 14 participants
more background details for sub-Saharan African settings than Western (nine males and five females, M = 24.07, SD = 3.20) from Ghana and
European settings. Finally, we expected mock witnesses from sub- 15 participants (four males and 11 females, M = 30.40, SD = 13.12)
Saharan Africa to report more central and background details about from the Netherlands provided ratings, using a 5-point Likert scale.
sub-Saharan African settings than Western European settings, whereas They rated the extent the stimulus (a) represented their native setting
we expected the reverse for mock witnesses fromWestern Europe. and (b) represented a crime scene. Consistent with Paz-Alonso, Good-
man, and Ibabe (2013), the mid-rating score was used in deciding
whether a stimulus received sufficient rating. The stimuli settings
2 | METHOD were rated by participants to adequately represent settings in their
respective countries (Ghanaian stimuli—M = 3.79, SD = .97; Dutch
2.1 | Participants and design stimuli—M = 3.33, SD = .62) and reflect plausible crime scenes
(Ghanaian stimuli—M = 3.43, SD = 1.28; Dutch stimuli—M = 3.47,
A total of 207 participants were sampled from Ghana (nrural Ghana = 78; SD = .83). The pilot study also determined which details participants
n = 73) and the Netherlands (n = 56). The selection of countries regarded as central and background details in each scene. To establish
urban Ghana
ANAKWAH ET AL. 507
stimulus centrality, the participants were asked two open-ended ques- Participants received the same instructions for all tasks. Participants'
tions: “What do you regard as the central event in the picture?” and responses were audio recorded. After completing the procedures,
“What do you regard as background event(s) in the picture?.” All par- they were thanked and debriefed. The test session took approxi-
ticipants identified the central and contextual events in a manner con- mately 60 min per participant. The study received ethical approval
sistent with our intended central and contextual elements when from the Ethics Review Committee Inner City faculties, Maastricht
constructing the stimuli (with the exception of one participant who University, and the Ethics Committee for the Humanities, University
did not identify central event for the Ghanaian stimuli as such). of Ghana.
Results from this pilot informed the development of the remaining
stimuli with Ghanaian and Dutch settings, which were developed to
have a clear central event distinct from the background. The stimuli 2.4 | Coding
are available on Open Science Framework at https://osf.io/t89hu/?
view_only=59e038117b2d4d5588e00c804de3539a Verbal responses were transcribed. The interviews conducted in Twi
in rural Ghana were translated into English during the transcription by
one of the research assistants indigenous to the region. A detailed
2.2.2 | Cultural orientation scale coding template for each of the stimulus scenes was developed by the
first author and was adapted from previous research (Gabbert,
We used the cultural orientation scale (Triandis & Gelfand, 1998) to Hope, & Fisher, 2009; Wright & Holliday, 2007). For the purposes of
measure self-reported individualism and collectivism of participants. our study, details provided by participants were classified as either a
That scale has 16 items with a 9-point Likert scale (1 = never or defi- background detail or central detail, in both free and cued recall, adher-
nitely no and 9 = always or definitely yes). It has four subscales: vertical ing a coding manual prepared in advance.4 An item was coded as cor-
individualism (VI), horizontal individualism (HI), vertical collectivism rect if it was present in the stimuli scene and given a correct
(VC), and horizontal collectivism (HC).2 Sample items on the scale description. Incorrect items were also coded and scored accordingly.
include VI—“winning is everything”; HI—“I often do my own thing”; Vague responses (e.g., it was a red or green bag) or subjective infer-
VC—“Parents and children must stay together as much as possible”; ences (e.g., the car belonged to the woman lying on the floor) were
and HC—“If a co-worker gets a prize, I would feel proud.” The coeffi- not coded. “Don't know” responses were coded as withheld details. A
cient alphas of the subscales range from .62 to .75 (Soh & second coder coded 20% of the transcripts, which were randomly
Leong, 2002). selected to check for coding consistency. We found high intercoder
reliability (intraclass correlation coefficient) for free recall with regard
to correct central details (r = .97) and correct background details
2.3 | Procedure (r = .95). The details provided by participants were collated across all
stimuli, and analysis was based on data for all scenes.
All participants in the study were tested individually. After consenting
to participate, participants completed the cultural orientation scale
and a short demographic questionnaire. Participants then viewed the 3 | RESULTS
stimulus scenes, one at a time. Consistent with previous research
(e.g., Wang & Pomplun, 2012), participants viewed each scene for Analyses were conducted using a mixed factorial analysis of variance
5 seconds. After viewing a scene, participants worked on a distractor (ANOVA), except analysis on type of detail that dominated in the
task (mathematical problems) for 5 minutes. Participants were then memory reports of the cultural groups, where repeated measures
instructed to provide a verbal free recall describing what they could ANOVA was used. Where significance difference existed, we used
remember about the scene they viewed. Participants were asked to Games–Howell multiple comparisons test as this post-hoc test is suit-
be as detailed and accurate as possible in their reports about the able for comparison groups of unequal size (Lee & Lee, 2018). We
scene. Participants had up to 6 minutes to provide that account. applied a Bonferronni correction (.017) to control for increased error
After the free recall task, participants answered 20 cued recall rates arising from multiple tests.
questions about central and background events or items in the stimu-
lus (e.g., “How was the attacker dressed” and “Can you describe the
colour of the building?”). The order of questions alternated between 3.1 | Free recall
questions on central and background details. The instructions and
questions for some participants in rural Ghana were given in the local 3.1.1 | Central details
language (Twi) as these participants had a low level of English
comprehension.3 Cultural group had a significant main effect on the number of correct
After completing both recall tasks, participants saw the next central details reported, F(2, 197) = 43.02, p < .001, η 2p = .30. Partici-
scene, and the procedure was repeated until they had viewed all four pants from the Netherlands reported significantly more correct central
scenes. The presentation of the scenes was counterbalanced. details than participants from urban Ghana (p = .003), who also reported
508 ANAKWAH ET AL.
significantly more correct central details than participants from rural from rural Ghana (p < .001; see Table 1). There was also a significant
Ghana (p < .001; see Table 1). We also found a significant main effect main effect for crime setting, F(1, 197) = 38.03, p < .001, η 2p = .16. Par-
for crime setting on correct central details, F(1, 197) = 8.78, p = .003, ticipants reported more correct background details for crime scenes
η 2p = .04. Participants reported more correct central details when the with Dutch settings (M = 9.22, SD = 5.94) than Ghanaian settings
crime scene was a Ghanaian setting (M = 15.91, SD = 7.50) than when it (M = 6.93, SD = 4.81). However, the interaction between cultural group
was a Dutch setting (M = 14.54, SD = 7.35). There was no significant and crime setting was not significant, F(2, 197) = .94, p = .39, η 2p = .01.
interaction effect between cultural group and crime setting, F Cultural group had no significant main effect on incorrect back-
(2, 197) = 3.28, p = .04, η2p = .03. In order to test evidence in favour of ground details reported, F(2, 197) = .47, p = .62, η
2
p = .01. Crime set-
the null, we proceeded with a Bayesian ANOVA analysis using JASP ting also had no significant main effect on incorrect background
(Wagenmakers, 2007). The analysis yielded a Bayes Factor of details reported, F(1, 197) = .33, p = .57, η 2p = .00. The interaction
BF10 = 2.35 × 10
14. According to Raftery (1995), Bayes factor of between cultural group and crime setting on incorrect background
150 and above is indicative of very strong evidence in favour of the details was also not significant F(2, 197) = 1.13, p = .33, η 2p = .01.
alternate hypothesis. A planned comparison revealed both participants
from rural Ghana (p = .019) and urban Ghana (p = .001) significantly
reported more correct central details for Ghanaian crime settings than 3.1.3 | Type of detail reported
Dutch crime settings. Participants from the Netherlands, however, did
not significantly differ in correct central details reported for Ghanaian We examined the total (correct and incorrect) amount of central and
and Dutch crime settings (p = .770). Results are shown in Figure 1. background details reported by each group. Participants from the
There was a significant main effect of cultural group on the num- Netherlands reported more central details than background details, F
ber of incorrect central details reported, F(2, 197) = 9.27, p < .001, (1, 54) = 93.25, p < .001, η 2p = .63. A similar pattern was found for
η 2p = .09. Participants from rural Ghana reported significantly fewer participants from urban Ghana who also reported more central details
incorrect central details than participants from the Netherlands than background details, F(1, 69) = 100.85, p < .001, η 2p = .59. Partici-
(p = .001). Participants from urban Ghana and the Netherlands did not pants from rural Ghana also reported more central details than
significantly differ in incorrect central details reported (p = .055). Par-
ticipants from rural Ghana and urban Ghana also did not significantly
25
differ in incorrect central details reported (p = .146; see Table 1).
Crime setting did not have a significant effect on incorrect central 20
details F(1, 197) = 3.80, p = .05, η 2p = .02. The interaction effect for
15
cultural group and crime setting for incorrect central details was not
significant, F(2, 197) = 2.05, p = .13, η 2p = .02. 10
5
3.1.2 | Background details 0
Rural Ghana Urban Ghana The Netherlands
Cultural Group
There was a significant main effect of cultural group on the reporting of
Ghanaian Setting Dutch Setting
correct background details F(2, 197) = 45.35, p < .001, η 2p = .32. Partici-
pants from the Netherlands reported more correct background details F IGURE 1 Mean correct details for different crime settings
than participants from urban Ghana (p = .002). Participants from urban reported across cultural groups under free recall. Error bars represent
Ghana also reported more correct background details than participants 95% confidence intervals
TABLE 1 Mean (standard deviation)
Rural Ghana Urban Ghana The Netherlands
correct, incorrect, and withheld central
Free recall Correct Central 9.71 (6.58) 15.57 (6.53) 20.39 (5.78) and background details reported in free
Background 3.99 (4.76) 8.36 (4.69) 11.87 (4.75) and cued recall by cultural groups
Incorrect Central 1.13 (1.39) 1.51 (1.34) 2.17 (1.33)
Background .78 (1.13) .91 (1.09) .96 (1.11)
Cued recall Correct Central 13. 47 (5.72) 18.06 (5.69) 22.77 (5.71)
Background 4.59 (3.38) 6.26 (3.35) 10.73 (3.34)
Incorrect Central 5.59 (2.25) 5.15 (2.26) 6.10 (2.30)
Background 4.27 (2.51) 4.27 (2.51) 5.73 (2.52)
Withheld Central 6.25 (4.16) 5.54 (4.18) 3.75 (4.15)
Background 10.54 (3.64) 10.13 (3.68) 7.18 (3.63)
Mean correct details
ANAKWAH ET AL. 509
background details, F(1, 74) = 156.35, p < .001, η 2p = .68 (see Table 3). significant main effect on the central details withheld by participants,
Although central details dominated in the memory reports of all cul- F(1, 197) = .90, p = .34, η 2p = .01. The interaction between cultural
tural groups, there was a significant difference in the total amount of group and crime setting on withheld central details was also not sig-
central details reported across cultural groups, F(2, 197) = 43.09, nificant, F(2, 197) = 1.29, p = .28, η2p = .01.
p < .001, η2p = .30. Participants from the Netherlands significantly
reported more central details than participants from urban Ghana
(p = .002), who also reported more central details than participants 3.2.2 | Background details
from rural Ghana (p < .001; see Table 3).
There was a significant main effect of cultural group on correct back-
ground details reported in response to questions about background
3.2 | Cued recall details, F(2, 197) = 55.59, p < .001, η 2p = .36. Participants from the
Netherlands reported more correct background details than participants
3.2.1 | Central details from urban Ghana (p < .001) and rural Ghana (p < .001). Participants
from urban Ghana also reported more correct background details than
There was a significant main effect of cultural group on correct central participants from rural Ghana (p = .004; see Table 1). The main effect of
details reported in response to cued recall questions focused on central crime setting on correct background details reported was significant, F
details, F(2, 197) = 42.66, p < .001, η 2p = .30. Participants from the (1, 197) = 130.51, p < .001, η
2
p = .40. Participants reported more cor-
Netherlands reported more correct central details than participants rect background details when crime setting was a Dutch setting
from urban Ghana (p < .001), who also reported more correct central (M = 8.95, SD = 4.38) than when it was a Ghanaian setting (M = 5.44,
details than participants from rural Ghana (p < .001; see Table 1). There SD = 3.68). The interaction between cultural group and crime setting
was also a significant main effect of crime setting on correct central was also significant, F(2, 197) = 15.23, p < .001, η 2p = .13. A planned
details reported, F(1, 197) = 5.82, p = .017, η 2p = .03. Participants comparison revealed participants from rural Ghana reported more cor-
reported more correct central details when the crime scene was a rect background details for Dutch settings than Ghanaian settings
Dutch setting (M = 18.66, SD = 7.21) than when it was a Ghanaian set- (p < .001). Participants from urban Ghana also reported more correct
ting (M = 17.55, SD = 6.08). The interaction between cultural group and background details when crime scene was a Dutch setting than Ghana-
crime setting was not significant, F(2, 197) = 2.85, p = .06, η 2p = .02. We ian setting (p < .001). We found a similar pattern for participants from
proceeded with a Bayesian ANOVA to test for evidence for the null. the Netherlands, who reported more correct background details when
We found the Bayes Factor to be BF10 = 7.964 × 10
12, indicative of crime setting was a Dutch setting than when it was a Ghanaian setting
very strong evidence (Raftery, 1995) in favour of the alternate hypothe- (p < .001). The interaction effect for correct background details could
sis. A planned comparison revealed participants from rural Ghana did be accounted for by the magnitude of the simple main effect. This is
not significantly differ on correct central details reported for Ghanaian because, for all cultural groups, the slopes of the simple main effect of
and Dutch crime settings (p = .91). Participants from urban Ghana also crime setting have the same direction. See Table 2 for descriptive statis-
did not significantly differ on correct central details reported for the tics on interaction between cultural group and crime setting.
two cultural settings (p = .36). However, participants from the Nether- There was also a significant main effect of cultural group on incor-
lands reported more correct central details for Dutch crime settings rect background details, F(2, 197) = 6.81, p = .001, η 2p = .07. Partici-
than they did for Ghanaian crime settings (p = .01; see Figure 2). pants from urban Ghana reported few incorrect background details
The main effect of cultural group on incorrect central details than participants from the Netherlands (p = .009). Participants from
reported, in response to questions focused on central details, was not rural Ghana also reported few incorrect background details than partici-
significant, F(2, 197) = 2.66, p = .07, η 2p = .02. There was, however, a pants from the Netherlands (p = .005). Participants from urban Ghana
significant main effect of crime setting on incorrect central details and rural Ghana did not differ in incorrect background details reported
reported, F(1, 197) = 10.16, p = .002, η 2p = .05. Participants reported (p = 1.00; see Table 1). Setting of crime had a significant main effect on
more incorrect central details when crime setting was a Ghanaian set- incorrect background details reported, F(1, 197) = 15.29, p < .001,
ting (M = 6.02, SD = 2.97) than when it was a Dutch setting (M = 5.21, η 2p = .07. Participants reported more incorrect background details for
SD = 2.83). The interaction between cultural group and crime setting Dutch crime settings (M = 5.22, SD = 3.25) than Ghanaian crime settings
was not significant, F(2, 197) = .36, p = .699, η 2p = .00. (M = 4.30, SD = 2.83). The interaction effect between cultural group
Cultural group had a significant main effect on the central details and crime setting on incorrect background details reported was not sig-
withheld by participants, F(2, 197) = 5.97, p = .003, η 2p = .06. Partici- nificant, F(2, 197) = 1.50, p = .23, η
2
p = .02.
pants from rural Ghana withheld more responses for questions about The analysis also revealed that the main effect of cultural group
central details than participants from the Netherlands (p = .004). Par- on background details withheld by participants was significant, F
ticipants from urban Ghana also withheld more central details than (2, 197) = 15.06, p < .001, η 2p = .13. Participants from urban Ghana
participants from the Netherlands (p < .00). Participants from rural withheld significantly more responses for questions on background
Ghana and urban Ghana did not significantly differ in central details details than participants from the Netherlands (p < .001). We also
withheld (p = .619; see Table 1). Crime setting did not have a found a similar pattern for participants from rural Ghana, who
510 ANAKWAH ET AL.
TABLE 2 Mean (standard deviation) correct and incorrect details reported in free and cued recall for cultural groups by crime setting
Rural Ghana Urban Ghana The Netherlands
Ghanaian Dutch Ghanaian Dutch Ghanaian Dutch
setting setting setting setting setting setting
Free recall Correct Central 10.56 (5.45) 8.87 (5.02) 16.93 (7.80) 14.21 (7.38) 20.25 (9.08) 20.53 (9.29)
Background 3.20 (2.68) 4.79 (4.52) 7.06 (4.86) 9.66 (6.76) 10.55 (6.63) 13.20 (6.40)
Incorrect Central 1.29 (1.55) .97 (1.00) 1.46 (1.73) 1.56 (1.97) 2.53 (2.40) 1.82 (1.86)
Background .63 (1.17) .93 (1.26) .97 (2.02) .86 (1.12) .95 (1.15) .96 (1.41)
Cued Correct Central 13.51 (5.70) 13.44 (6.70) 17.69 (6.30) 18.44 (7.72) 21.45 (5.85) 24.09 (6.70)
recall
Background 3.72 (3.10) 5.45 (3.86) 4.81 (3.17) 7.71 (4.06) 7.78 (4.54) 13.67 (5.17)
Incorrect Central 5.85 (3.28) 5.33 (2.96) 5.61 (2.47) 4.69 (2.61) 6.60 (3.20) 5.60 (2.86)
Background 3.59 (2.60) 4.96 (3.02) 4.06 (2.76) 4.49 (3.15) 5.25 (3.06) 6.20 (3.37)
Withheld Central 6.45 (11.26) 6.05 (3.05) 4.81 (2.52) 6.27 (2.47) 3.56 (2.49) 3.93 (2.36)
Background 11.96 (4.64) 9.39 (4.16) 10.74 (4.17) 9.51 (3.69) 8.55 (4.61) 5.81 (3.53)
TABLE 3 Mean (standard deviation) of amount of central vs background details for cultural groups under free and cued recall
The Netherlands Urban Ghana Rural Ghana
Central Background Central Background Central Background
M SD M SD M SD M SD M SD M SD
Free recall 45.13 19.21 25.65 11.70 34.16 14.27 18.54 12.34 21.69 9.41 9.55 6.66
Cued recall 57.75 10.97 32.91 9.89 46.43 11.83 20.97 8.02 38.13 12.24 17.64 8.59
30 3.2.3 | Type of detail reported
25
The total (correct and incorrect) amount of details reported for central
20
and background details for each group was compared to find out the
15 type of detail that dominated in their reports. Participants from rural
10 Ghana significantly reported more central details than background
details, F(1, 74) = 304.58, p < .001, η 2p = .81. Participants from urban
5
Ghana also significantly reported more central details than background
0
Rural Ghana Urban Ghana The Netherlands details, F(1, 69) = 370.02, p < .001, η
2
p = .84. We found the same pat-
Cultural Group tern for participants from the Netherlands who also significantly
Ghanaian setting Dutch setting reported more central details than background details, F
(1, 54) = 334.83, p < .001, η 2 = .86 (see Table 3). Notwithstanding the
F IGURE 2 Mean correct details for different crime settings p
reported across cultural groups under cued recall. Error bars represent observation that in all cultural groups central details dominated in the
95% confidence intervals memory reports, the cultural groups significantly differed in amount of
central details reported, F(2, 197) = 44.11, p < .001, η 2p = .31. Partici-
withheld significantly more responses to questions on background pants from the Netherlands significantly reported more central details
details, than participants from the Netherlands (p < .001). No signifi- than participants from urban Ghana (p < 001), who also reported more
cant difference was observed for withheld responses for participants central details than participants from rural Ghana (p < .001).
from rural Ghana and urban Ghana (p = .781; see Table 1). The setting
of crime also had a significant main effect on background details with-
held by participants, F(1, 197) = 54.54, p < .001, η 2p = .22. Participants 3.3 | Self-reported cultural orientation
withheld more background details for Ghanaian crime settings
(M = 10.33, SD = 4.53) than Dutch crime settings (M = 8.24, SD = 3.81). We conducted an exploratory analysis on the self-reported cultural
The interaction effect between cultural group and crime setting for orientation of participants from the cultural groups. The analysis rev-
background details withheld by participants was not significant, F ealed that the cultural groups did not differ on horizontal collectivism,
(2, 197) = 2.47, p = .09, η 2p = .02. F(2, 197) = .69, p = .50, η
2
p = .01, but did differ on vertical collectivism,
Mean correct details
ANAKWAH ET AL. 511
F(2, 197) = 8.30, p < .001, η 2p = .08. Participants from rural Ghana for this finding could be elaboration differences due to socialisation
(M = 29.72, SD = 6.87) significantly scored higher on vertical collectiv- affordances (Peterson, Sales, Rees, & Fivush, 2007). Such a differ-
ism than participants from the Netherlands (M = 26.20, SD = 4.67; ence is conspicuous in childrearing practices, where it has been
p = .002). Participants from urban Ghana (M = 30.01, SD = 4.99) also observed that parents from individualistic cultures provide much
scored higher on self-reported vertical collectivism than participants more feedback to their children in conversations than those from
from the Netherlands (p < .001). There was no significant difference collectivistic cultures (Wang, 2004). It may be the case that differ-
between participants from rural Ghana and urban Ghana on vertical ences in linguistic elaboration are transmitted to children and persist
collectivism (p = .95). to later adulthood. Consequently, although eyewitnesses from col-
There was also a (marginally) significant difference between the lectivistic cultures report details about a crime scene, they may not
cultural groups on horizontal individualism, F(2, 197) = 3.05, p = .05, spontaneously provide a detailed elaboration in their memory narra-
η 2p = .03. Participants from rural Ghana (M = 26.83, SD = 6.45) and tives. This speculation fits with assertions that individuals from col-
the Netherlands (M = 25.84, SD = 4.78) did not differ on scores on lectivistic cultures report less specific and more generic details than
horizontal individualism (p = .57). There was also no significant differ- individuals from individualistic cultures (Millar, Serbun, Vadalia, &
ence between participants from rural Ghana and urban Ghana Gutchess, 2013; Wang & Ross, 2005). Similar results have been
(M = 28.29, SD = 5.21) on self-reported horizontal individualism observed in research on deception detection, showing interviewees
(p = .29). However, there was a significant difference in self-reported in individualistic cultures typically report more explicit details than
horizontal individualism between participants from urban Ghana and interviewees from collectivistic cultures (Leal et al., 2018). Leal et al.
the Netherlands (p = .02). Participants from urban Ghana gave higher (2018) argued that interviewees from collectivistic cultures tend to
ratings than participants from the Netherlands on horizontal individu- leave many things unsaid, allowing the context to communicate
alism. The cultural groups significantly differed on self-reported verti- what is implied, whereas in individualistic cultures, the communica-
cal individualism F(2, 197) = 14.86, p < .001, η 2p = .13. Participants tion style tends to be more explicit. Therefore, during investigative
from rural Ghana (M = 24.52, SD = 6.80) reported higher scores on interviews, it may be necessary to prompt and encourage eyewit-
vertical individualism than participants from the Netherlands nesses from collectivistic cultures to elaborate further on the initial
(M = 17.98, SD = 7.26; p < .001). Participants from urban Ghana information they provide.
(M = 23.04, SD = 6.83) also significantly gave higher ratings on vertical Apart from the possibility of elaborative differences, it may be the
individualism than participants from the Netherlands (p < .001). There case that individuals from collectivistic cultures have a tendency to be
was no significant difference between participants from rural Ghana more modest or restrained when providing their memorial accounts
and urban Ghana on self-reported vertical individualism (p = .40). than those from individualistic cultures. Cultural differences in self-
effacement and self-enhancement have been documented, with self-
effacement attributed to collectivistic cultures and self-enhancement
4 | DISCUSSION attributed to individualistic cultures (Takata, 2003; Yamagishi et al.,
2012). Such differences may reflect cultural disparities in the
We examined eyewitness memory reports of individuals from differ- independent–interdependent construal of the self (Markus &
ent cultural groups thought to typify individualistic (Western Europe) Kitayama, 1991). Individuals from cultures with independent construal
and collectivistic (sub-Saharan Africa) cultures. The results appear to of the self are more likely to emphasise the unique attributes of a per-
reveal a tendency towards the underreporting of details by sub- son. This tendency may be reflected in their self-presentation in
Saharan African mock witnesses. In addition, central details dominated regard to expressing themselves, as they may be inclined to empha-
in the eyewitness memory reports provided across cultures. The sise their positive attributes (self-enhancement; Takata, 2003). In con-
results also showed that in free recall, sub-Saharan African mock wit- trast, individuals from collectivistic cultures, in comparison with
nesses reported more correct central details when the crime scenario individuals from individualistic cultures, have a tendency to be self-
was witnessed in their own native setting than when it was witnessed critical and modest about emphasising their unique attributes (self-
in a non-native setting. Western European mock witnesses also effacement; Heine, Lehman, & Takata, 2000). Therefore, individuals
reported more correct central details in cued recall when the crime from collectivistic cultures are more likely to be modest in terms of
scenario was witnessed in their own native setting than a non-native self-presentation and expression (Wise, Gong, Safer, & Lee, 2010).
setting. Mock witnesses from sub-Saharan Africa reported more back- These concepts have been identified as powerful determinants of
ground details about a non-native setting than they did for their own behaviour, especially within a social context (Brown & Gallagher,
setting under cued recall. Crime context did not appear to affect the 1992). It is possible for a witness from a collectivistic culture to self-
nature of correct background details that Western European Mock efface when being interviewed, by being modest in terms of the
witnesses reported in free recall. However, they reported more cor- extent of the personal memory narrative provided (i.e., providing a
rect background details when crime was witnessed in their own native less elaborative or detailed account spontaneously). However, it is
setting than a non-native setting in cued recall. worth noting that this tendency to self-efface may attenuate when
The differences between cultural groups with respect to the the implications or stakes of self-effacing are high (Yamagishi et al.,
amount of reported details is noteworthy. One possible explanation 2012). Future research should explore whether this tendency is
512 ANAKWAH ET AL.
attenuated when investigators emphasise the importance of providing when a crime occurs is likely not the same as any ordinary or neutral
details to pursue an investigation. everyday scene. For example, in a robbery, the threatening and
In the current study, mock witnesses from the collectivistic cul- unusual nature of the scene will make it more likely for people at the
tural groups provided more “Don't Know” responses than those from scene to attend to this focal event than other activities that may be
the individualistic cultural group. Thus, in this study at least, partici- going on at the background. The tendency to attend more to notice-
pants from collectivistic cultures might have applied a relatively strict able details at a visual field is well-documented (Loftus & Mackworth,
criterion for reporting, and withheld details they remembered but 1978; Masuda & Nisbett, 2006, Experiment 3; Wang &
were not confident about (Cai, Brown, Deng, & Oakes, 2007). This Pomplun, 2012).
pattern aligns well with the self-effacing tendency of collectivistic cul- It is also worth noting that past research on culture and visual
tures. In a study on self-effacement and self-enhancement among attention focused mostly on comparing East Asian and other Western
Canadians and Japanese participants, Heine et al. (2000) found that cultures. Hence, even though African cultures are regarded as collec-
although the former were confident they performed well on a test, tivistic, the findings for East Asian cultures may not be generalisable
the latter were reluctant to admit that they had performed better. It to sub-Saharan Africa. Studies in cross-cultural cognition have largely
may be that when sub-Saharan mock witnesses were not confident studied East Asian cultures, and it may be that the collectivistic self
about memory for certain details, they simply decided not to report (interdependent self-construal) may not be a one-size-fits-all phenom-
them. Consistent with this notion is the observation that participants enon for all collectivistic cultures. This conclusion is consistent with
from Western Europe, who tend to be more assertive and expressive the notion that collectivism is not a context-free construct (Triandis,
than people from collectivistic cultures (Matsumoto et al., 2008), pro- 2001). As such, the self-construal for collectivistic cultures may be
vided more inaccurate responses than participants from sub-Saharan context-specific. For example, it has been argued that the
Africa, which suggests Western European mock witnesses had a interdependent self-construal among Africans does not suggest a total
looser threshold for reporting accurate details. Future research should loss of the independent self in the collective (Adams & Dzokoto,
examine the extent to which there are cultural differences in the 2003), and there may be different variations of the interdependent
reporting of low-confidence memories. self-construal among collectivistic cultures. In that vein, the holistic–
The social dynamics during the interview may have also played a analytic categorisation of visual attention across cultures may be rela-
role in the amount of information mock witnesses reported, particu- tive. Future research should explore differences between and within
larly those from sub-Saharan Africa. Individuals from sub-Saharan different collectivist cultures.
Africa have been shown to be high on the cultural dimension of power The current results suggest that the cultural setting in which a
distance (Hofstede, 1983). Power distance, another dimension in crime is witnessed may also be important when considering eyewit-
which cultures differ, is the extent to which a society endorses hierar- ness reports. Mock witnesses reported more correct central details
chy in social relationships (Oyserman, 2006). High power distance for Ghanaian crime settings than for Dutch crime settings for free
(endorsement of hierarchy in social relationships) may inhibit free and recall. When cued recall questions were asked, mock witnesses
spontaneous communication when an individual is in a social interac- reported more correct central details for Dutch settings than Ghana-
tion with an authority figure (Ghosh, 2011). Consistent with this spec- ian settings. That finding partially aligns with the results of previous
ulation, in the present study, sub-Saharan African mock witnesses research. For example, Masuda and Nisbett (2006) found that both
endorsed more hierarchy in social relationships (vertical collectivism) participants from individualistic (North America) and collectivistic
than Western European mock witnesses. Therefore, there is a possi- (Japan) cultural groups detected focal changes to North American
bility that the mere fact of reporting to an authority or expert (i.e., a stimuli scenes quicker than they did for Japanese stimuli scenes. In
researcher) may have produced cultural differences in the amount of the current research, sub-Saharan African mock witnesses reported
details provided. Future research should explore the impact of this more correct central details in free recall, when reporting about crime
dimension further to (a) determine whether in an interview context, witnessed in their own native setting than when it was witnessed in a
the presence of an authority figure plays a culture-related role in the non-native setting. This superior performance for crime witnessed in a
amount of information reported by witnesses and (b) explore how native setting was not observed when cued recall questions were
such differences might be attenuated. asked. However, Western European mock witnesses reported more
None of the cultural groups appear to have processed back- correct central details when the witnessed crime was in their own
ground information deeply (cf. central details; Wong et al., 2017) as, native setting than a non-native setting in cued recall, but not for free
regardless of cultural background, central details dominated in the recall. The own-setting effect for central details observed for the cul-
memory reports provided. This finding does not align with previous tural groups is consistent with work that shows familiar environments
research suggesting collectivistic cultures attend holistically to a visual have the tendency to modulate the processing of visual details
field (Istomin et al., 2014). However, it is worth noting that the stimuli (Epstein et al., 2007). However, that explanation does not fit for cor-
used in our study were crime scenes and quite different to the stimuli rect background details witnessed by sub-Saharan African mock wit-
used in previous research. Previous studies used stimuli such as pic- nesses when crime setting was considered, as sub-Saharan African
tures from the physical environment and artistic representations mock-witnesses reported more contextual information about a non-
(Boduroglu et al., 2009; Miyamoto et al., 2006). The focus of attention native setting than they did for their own setting in cued recall. We
ANAKWAH ET AL. 513
suspect that because the non-native setting was an unfamiliar setting, CONFLICT OF INTEREST
participants from sub-Saharan Africa may have attended more to con- The authors declare no conflict of interest
textual information in that setting than they did for their own setting.
Future work should pursue the issue of crime context and how this ACKNOWLEDGEMENTS
relates to reporting in cross-cultural contexts. This research is supported by a fellowship awarded from the Erasmus
There are some limitations associated with the current research. Mundus Joint Doctorate Program The House of Legal Psychology
The first limitation relates to some unavoidable differences in the edu- (EMJD-LP) with framework partnership agreement (FPA) 2013-0036
cation levels for one of the cultural group samples. Although the and specific grant agreement (SGA) 532473-EM-5-2017-1-NL-ERA
Dutch and urban Ghanaian samples comprised mainly university-level MUNDUS-EPJD to Nkansah Anakwah. We thank Manouk Vrouch,
students with a similar age range and were, as such, well-matched George Ofori, and Melody Konadu Frempong for their assistance with
with respect to education level, this was not the case for the rural data collection. We also thank Michael Obiri Yeboah for his assistance
Ghanaian sample. Participants from rural Ghana had a minimal level of in translating the study instructions into Twi.
education and were relatively older. Both of these factors may have
affected the performance of this group relative to the other experi- ORCID
mental groups—although it is also worth noting that it would likely be Nkansah Anakwah https://orcid.org/0000-0002-4271-4728
impossible to recruit university-level educated sample in rural Ghana.
Similar issues relating to the difficulty of matching samples across dif- ENDNOTES
ferent cultures are common in the cultural literature (Buil, De 1 Hofstede's individualism–collectivism index indicates the extent to
Chernatony, & Martínez, 2012). A second possible methodological which countries are individualistic and collectivistic. On Hofstede's index
concern relates to the test language. As the study instructions were (ranging from 0 to 100), the Netherlands is associated with an individual-
ism index of 80, whereas Ghana is associated with an index of 14, where
translated for participants in rural Ghana who lacked adequate com- a higher score reflects greater individualism.
prehension of the English language, we do not rule out the possibility 2 Vertical individualism refers to individualistic cultures where hierarchy is
that the translation into a different language may have in some way emphasised in social relationships; horizontal individualism refers to indi-
affected the outcomes for the rural sample. Finally, we acknowledge vidualistic cultures where equality is emphasised in social relationships;
that the static nature of the stimuli used limit generalisability to the vertical collectivism refers to collectivistic cultures where hierarchy is
emphasised in social relationship; and horizontal collectivism refers to
eyewitness context. Typically, crime events involve dynamic move-
collectivistic cultures where equality is emphasise in social relationships
ment and action, and the reporting of such information may also vary (Triandis & Gelfand, 1998).
culturally. Although static images might be a useful starting point to 3 A PhD student in Linguistics with expertise in the Ghanaian language
examine reporting from memory, future research should adopt the translated the protocol. The interviewer who also had a good command
more typical mock witness paradigm using recorded or live events. of the local language explained the study instructions to these partici-
pants thoroughly and also read the questions out to such participants in
the Twi language.
4 Classification of central and background details in this coding manual
5 | CONCLUSION was based on stimulus centrality established in the pilot study earlier
reported.
In this research, we sought to take the first steps in addressing an
important gap in the eyewitness literature. Specifically, drawing on
samples from sub-Saharan Africa and Western Europe, we examined DATA AVAILABILITY STATEMENT
eyewitness memory reports for differences predicted by theory in the The data that support the findings of this study are openly
cross-cultural literature. Our results show that individuals from indi- available at https://osf.io/9gs78/?view_only=135f537f7bf8437998
vidualistic cultures provide more details in their account of crime 906da24a4184c5
scene information, irrespective of type of detail. We also found evi-
dence that regardless of the cultural background of eyewitnesses, ORCID
central details dominate in their reports of crime scene information. Nkansah Anakwah https://orcid.org/0000-0002-4271-4728
Finally, we found evidence that the cultural setting in which a crime is
witnessed may play a role in eyewitness memory reports. These find- REFERENCES
ings not only identify important routes for future research in this area Adams, G., & Dzokoto, V. A. (2003). Self and identity in African
but also highlight the importance of considering the cultural back- studies. Self and Identity, 2, 345–359. https://doi.org/10.1080/
714050252
ground of the witness when eliciting memory reports. As such, these
Boduroglu, A., Priti, S., & Nisbett, R. E. (2009). Cultural differences in allo-
findings should be informative for legal and investigative professionals cation of attention in visual information processing. Journal of Cross-
working in international criminal justice settings, border and security Cultural Psychology, 40, 349–360. https://doi.org/10.1177/
practitioners interviewing in asylum, migration and intelligence- 0022022108331005
Brady, L. M., Fryberg, S. A., & Shoda, Y. (2018). Expanding the interpretive
gathering contexts, and law enforcement personnel who regularly
power of psychological science by attending to culture. Proceedings of
interview witnesses from different cultural backgrounds.
514 ANAKWAH ET AL.
the National Academy of Sciences of the United States of America, 115, Ma, Q., Pei, G., Jin, J., & De Wit, H. De. (2015). What makes you generous?
11406–11413. https://doi.org/10.1073/pnas.1803526115 The influence of rural and urban rearing on social discounting in China.
Brown, J. D., & Gallagher, F. M. (1992). Coming to terms with failure: Pri- PLoS One, 10, 1–11. https://doi.org/10.1371/journal.pone.0133078
vate self-enhancement and public self-effacement. Journal of Experi- Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for
mental Social Psychology, 28, 3–22. https://doi.org/https://doi.org/10. cognition, emotion, and motivation. Psychological Review, 98,
1016/0022-1031(92)90029-J 224–253. https://doi.org/10.1037/0033-295X.98.2.224
Buil, I., De Chernatony, L., & Martínez, E. (2012). Methodological issues in Markus, H. R., & Kitayama, S. (2003). Culture, self, and the reality of the
cross-cultural research: An overview and recommendations. Journal of social. Psychological Inquiry, 14, 277–283. https://doi.org/10.1080/
Targeting, Measurement and Analysis for Marketing, 20, 223–234. 1047840X.2003.9682893
https://doi.org/10.1057/jt.2012.18 Masuda, T., & Nisbett, R. E. (2006). Culture and change blindness. Cognitive
Cai, H., Brown, J. D., Deng, C., & Oakes, M. A. (2007). Self-esteem and cul- Science, 30, 381–399. https://doi.org/10.1207/s15516709cog0000_63
ture: Differences in cognitive self-evaluations or affective self-regard? Millar, P. R., Serbun, S. J., Vadalia, A., & Gutchess, A. H. (2013). Cross-
Asian Journal of Social Psychology, 10, 162–170. https://doi.org/10. cultural differences in memory specificity. Culture and Brain, 1(2–4),
1111/j.1467-839X.2007.00222.x 138–157. https://doi.org/10.1007/s40167-013-0011-3
Epstein, R. A., Higgins, J. S., Jablonski, K., & Feiler, A. M. (2007). Visual Miyamoto, Y., Nisbett, R., & Masuda, T. (2006). Culture and the physical
scene processing in familiar and unfamiliar environments. Journal of environment. Psychological Science, 17, 113–119. https://doi.org/10.
Neurophysiology, 97, 3670–3683. https://doi.org/10.1152/jn.00003. 1111/j.1467-9280.2006.01673.x
2007 Oyserman, D. (2006). High power, low power, and equality: Culture
Gabbert, F., Hope, L., & Fisher, R. P. (2009). Protecting eyewitness evi- beyond individualism and collectivism. Journal of Consumer Psychology,
dence: Examining the efficacy of a self-administered interview tool. 16, 352–356. https://doi.org/10.1207/s15327663jcp1604_6
Law and Human Behavior, 33, 298–307. https://doi.org/10.1007/ Paz-Alonso, P. M., Goodman, G. S., & Ibabe, I. (2013). Adult eyewitness
s10979-008-9146-8 memory and compliance: Effects of post-event misinformation on
Gelfand, M. J., Harrington, J. R., & Jackson, J. C. (2017). The strength of memory for a negative event. Behavioral Sciences & the Law, 31,
social norms across human groups. Perspectives on Psychological Sci- 541–558. https://doi.org/10.1002/bsl
ence, 12, 800–809. https://doi.org/10.1177/1745691617708631 Peterson, C., Sales, J. M., Rees, M., & Fivush, R. (2007). Parent-child talk
Ghosh, A. (2011). Power distance in organizational contexts—A review of and children's memory for stressful events. Applied Cognitive Psychol-
collectivist cultures. Indian Journal of Industrial Relations, 47, 89–101. ogy, 21, 1057–1075. https://doi.org/10.1002/acp
Retrieved from. http://0-search.ebscohost.com.library.regent.edu/ Raftery, A. E. (1995). Bayesian model selection in social research. Sociologi-
login.aspx?direct=true&db=bth&AN=69815239&site=eds-live cal Methodology, 25, 111–163. https://doi.org/10.2307/271063
Heine, S. J., Lehman, D. R., & Takata, T. (2000). Beyond self-presentation: Rooks, G., Klyver, K., & Sserwanga, A. (2016). The context of social capital:
Evidence for self-criticism among Japanese. Personality and Social Psy- A comparison of rural and urban entrepreneurs in Uganda. Entrepre-
chology Bulltin, 26, 71–78. https://doi.org/doi.org/10.1177% neurship: Theory and Practice, 40, 111–130. https://doi.org/10.1111/
2F0146167200261007 etap.12107
Henrich, J., Heine, S. J., & Norenzayan, A. (2010). The weirdest people in Schwartz, A. J., Boduroglu, A., & Gutchess, A. H. (2014). Cross-cultural dif-
the world? Behavioral and Brain Sciences, 33, 61–83. https://doi.org/ ferences in categorical memory errors. Cognitive Science, 38(5),
10.1017/S0140525X0999152X 997–1007. https://doi.org/10.1111/cogs.12109
Hofstede, G. (1983). Dimensions of national cultures in fifty countries and Soh, S., & Leong, F. T. L. (2002). Validity of vertical and horizontal individualism
three region. In J. Deregowski, S. Dzuirawiec, & R. Annis (Eds.), Explo- and collectivism in Singapore. Journal of Cross-Cultural Psychology, 33,
rations in cross-cultural psychology (pp. 335–355). Lisse: Swets and 3–15. https://doi.org/https://doi.org/10.1177/0022022102033001001
Zeitlinger. Takata, T. (2003). Self-enhancement and self-criticism in Japanese culture.
Hofstede, G. (2001). Culture's consequences: Comparing values, behaviors, Journal of Cross-Cultural Psychology, 34, 542–551. https://doi.org/10.
institutions and organisations (2nd ed.). Thousand Oaks, CA: Sage. 1177/0022022103256477
Hope, L., & Gabbert, F. (2019). Interviewing witnesses and victims. In Triandis, H. C. (2001). Individualism-collectivism and personality. Journal of
N. Brewer & A. B. Douglass (Eds.), Psychological science and the law Personality, 69, 907–924. https://doi.org/10.1111/1467-6494.696169
(pp. 56–74). New York: The Guildford Press. Triandis, H. C., Bontempo, R., Villareal, M. J., Asai, M., & Lucca, N. (1988).
Istomin, K. V., Panáková, J., & Heady, P. (2014). Culture, perception, and Individualism and collectivism: Cross-cultural perspectives on self-in-
artistic visualization: A comparative study of children's drawings in group relations. Journal of Personality and Social Psychology, 54,
three siberian cultural groups. Cognitive Science, 38, 76–100. https:// 323–338. https://doi.org/10.1037/0022-3514.54.2.323
doi.org/10.1111/cogs.12051 Triandis, H. C., & Gelfand, M. J. (1998). Converging measurement of hori-
Kastanakis, M. N., & Voyer, B. G. (2014). The effect of culture on perception zontal and vertical individualism and collectivism. Journal of Personality
and cognition: A conceptual framework. Journal of Business Research, 67, and Social Psychology, 74, 118–128. https://doi.org/10.1037/0022-
425–433. https://doi.org/10.1016/j.jbusres.2013.03.028 3514.74.1.118
Keith, K. (2012). Blood diamonds and war crimes: The case against Charles van Veldhuizen, T. S., Maas, R. P. A. E., Horselenberg, R., & van
Taylor. Southern Cross University Law Review, 15, 99–110. Retrieved Koppen, P. J. (2018). Establishing origin: Analysing the questions asked
from. http://classic.austlii.edu.au/au/journals/SCULawRw/2012/7.pdf in asylum interviews. Psychiatry, Psychology and Law, 25, 283–302.
Leal, S., Vrij, A., Vernham, Z., Dalton, G., Jupe, L., Harvey, A., & Nahari, G. https://doi.org/10.1080/13218719.2017.1376607
(2018). Cross-cultural verbal deception. Legal and Criminological Psy- Veillard, N. (2017). WEIRD sampling in cross-cultural psychology, should it
chology, 23, 192–213. https://doi.org/10.1111/lcrp.12131 not be less WEIRD and more representative?: The over-representation
Lee, S., & Lee, D. K. (2018). What is the proper way to apply the multiple of individuals from Western, educated, industrialized, and democratic
comparison test? Korean Journal of Anesthesiology, 71, 353–360. countries as sample populations in cross-cultural psychology research
https://doi.org/doi.org/10.4097%2Fkja.d.18.00242 (master's thesis). Retrieved from https://openaccess.leidenuniv.nl/
Loftus, G. R., & Mackworth, N. H. (1978). Cognitive determinants of fixa- handle/1887/56361
tion location during picture viewing. Journal of Experimental Psychol- Wagenmakers, E.-J. (2007). A practical solution to the pervasive problems
ogy: Human Perception and Performance, 4, 565–572. https://doi.org/ of p values. Psychonomic Bulletin & Review, 14, 779–804. https://doi.
doi/10.1037/0096-1523.4.4.565 org/https://doi.org/10.3758/BF03194105
ANAKWAH ET AL. 515
Wang, H., & Pomplun, M. (2012). The attraction of visual attention to texts Cross-Cultural Psychology, 49, 1–14. https://doi.org/10.1177/
in real-world scenes, 12, 1–17. https://doi.org/10.1167/12.6.26 0022022117748763
Wang, Q. (2004). The emergence of cultural self-constructs: Autobio- Wright, A. M., & Holliday, R. E. (2007). Enhancing the recall of young,
graphical memory and self-description in European American and Chi- young–old and old–old adults with cognitive interviews. Applied Cogni-
nese children. Developmental Psychology, 40, 3–15. https://doi.org/10. tive Psychology, 21, 19–43. https://doi.org/10.1002/acp.1260
1037/0012-1649.40.1.3 Yamagishi, T., Hashimoto, H., Cook, K. S., Kiyonari, T., Shinada, M.,
Wang, Q., & Ross, M. (2005).What we remember and what we tell: The effects Mifune, N., … Li, Y. (2012). Modesty in self-presentation: A compari-
of culture and self-priming on memory representations and narratives. son between the USA and Japan. Asian Journal of Social Psychology, 15,
Memory, 13, 594–606. https://doi.org/10.1080/09658210444000223 60–68. https://doi.org/10.1111/j.1467-839x.2011.01362.x
Wang, Q., Song, Q., & Kim Koh, J. B. (2017). Culture, memory, and narra-
tive self-making. Imagination, Cognition and Personality, 37, 199–223.
https://doi.org/10.1177/0276236617733827
Wise, R. A., Gong, X., Safer, M. A., & Lee, Y. T. (2010). A comparison of How to cite this article: Anakwah N, Horselenberg R, Hope L,
Chinese judges' and US judges' knowledge and beliefs about eyewit- Amankwah-Poku M, van Koppen PJ. Cross-cultural
ness testimony. Psychology, Crime and Law, 16, 695–713. https://doi.
differences in eyewitness memory reports. Appl Cognit
org/10.1080/10683160903153893
Wong, B. I., Yin, S., Yang, L., Li, J., & Spaniol, J. (2017). Cultural differences Psychol. 2020;34:504–515. https://doi.org/10.1002/acp.3637
in memory for objects and backgrounds in pictures. Journal of