Introduction
Eyewitness evidence can be a key factor in a jury’s decision making about a defendant’s guilt or innocence (Nicholson et al.
2014); if jurors do not find a witness to be credible, they are less likely to decide that the defendant is guilty (Pica et al.
2017). In judging the credibility of a witness, jurors consider several factors aside from the content of the witness’ testimony, including expression of emotion (Cooper et al.
2014; Wessel et al.
2013), eye contact (Field et al.
2010), confidence (Dodson and Dobolyi
2015), and surface features of speech (e.g., pause, intonation) (Ozuru and Hirst
2006). Jurors, therefore, rely on non- and para-verbal behaviours, opposed to focusing solely on what the witness has said, when making judgements about their credibility.
As autistic
1 individuals may be more likely to encounter the criminal justice system than those without autism (e.g., Lindblad and Lainpelto
2011; Turcotte et al.
2017; Woodbury-Smith and Dein
2014), it is crucial to assess how they are perceived in this context (for example, as witnesses). A growing body of empirical evidence has explored the performance of autistic individuals (largely those who do not have intellectual disabilities) when giving evidence in a criminal justice context. This illustrates how both children and adults on the autism spectrum tend to recall less information than their typically developing peers when free recalling information about an event (e.g., Bruck et al.
2007; Henry et al.
2017b; Maras et al.
2012; Mattison et al.
2015; McCrory et al.
2007), but can perform similarly in more structured interviews (e.g., Henry et al.
2017a; Maras and Bowler
2010), or when additional supports (e.g., more specific questioning, physical reinstatement of context, or concrete visual prompts) are provided at recall (e.g., Maras and Bowler
2012a; Maras et al.
2012; Mattison et al.
2016). Research has also highlighted how eyewitness information provided by children with autism can be as accurate as that of comparable peers (Bruck et al.
2007; Henry et al.
2017a,
b; McCrory et al.
2007); although it should be noted that accuracy levels can vary with interview type (Mattison et al.
2016) and findings are less consistent in autistic adults (e.g., Maras and Bowler
2010; Maras et al.
2012,
2013). Autistic children and adults may also display atypicalities in narrating memories of personally experienced events; for example, lacking a consistent high point in their narratives (e.g., Goldman
2008; McCabe et al.
2013). Importantly, however, there is no evidence supporting the notion that autistic people are more suggestible than non-autistic people (Bruck et al.
2007; Maras and Bowler
2011,
2012b; McCrory et al.
2007; North et al.
2008). This is despite many legal professionals believing this to be true (see, for example, George et al.
2018, for a survey study on this topic in UK barristers).
Although research suggests that autistic witnesses can give reliable and accurate eyewitness evidence, individuals diagnosed with autism may display atypical behaviours (e.g., unusual eye contact, repetitive body movements) that could result in their credibility being questioned (McCrory et al.
2007). This could be particularly relevant if the person’s autism diagnosis is not disclosed to the jury. Indeed, many witnesses and defendants may be reluctant to share information about their autism diagnosis with criminal justice processionals, for example, due to concerns about justice professionals’ perceived lack of autism of knowledge and awareness (e.g., Crane et al.
2016) or fears that their diagnosis may count against them (e.g., Cooper and Allely
2017). It is, therefore, vital to examine the effects of providing jurors with knowledge of a witness’ autism diagnosis, with or without the provision of additional information about the key characteristics of autism.
Previous research has shown that providing jurors with additional information about witnesses can help address unfair biases held about their credibility. This may be particularly the case for children, who—despite being able to provide accurate and forensically useful eyewitness testimony—often find their credibility questioned (e.g., Lamb et al.
2011). Goodman-Delahunty et al. (
2011), for example, found that jurors (in relation to child sexual abuse cases) were more likely to convict a defendant if information that challenged common misconceptions about children’s memory was provided by expert witnesses.
2 Further, Wadley and Haley (
2001) found that providing a diagnostic label (e.g., Alzheimer’s disease, major depression), in addition to information about a person’s condition, allowed others to attribute the person’s atypical behaviour to the diagnosis. Sasson and Morrison (
2017) also reported that while first impressions of autistic adults were rated less favourably, informing observers of their autism diagnosis resulted in significantly more favourable judgements about them. Such evidence (that observers can view a person’s behaviours more favourably if they know about a diagnosis) supports Kelley’s (
1971,
1972) discounting principle, in which one explanation is diminished when observers are alerted to an alternative one.
In this paper, we present a preliminary study of the perceived credibility of child witnesses on the autism spectrum. This was achieved by presenting mock jurors with an ecologically valid video-recorded interview (as would often be presented to jurors in cases involving child witnesses in England and Wales, as part of the range of Special Measures offered to vulnerable witnesses; YJCEA
1999). These videos were taken from a larger empirical study on the performance of child witnesses—with and without an autism diagnosis—who were interviewed about a staged event that they viewed at their school (see Henry et al.
2017a). Mock jurors rated videos of one of two autistic child witnesses using an 11-item credibility questionnaire, established as useful in previous research on child witnesses with and without intellectual disabilities (Henry et al.
2011; Peled et al.
2004). Prior to completing the questionnaires, we manipulated whether mock jurors were made aware of the children’s diagnoses, to test Kelley’s (
1971,
1972) discounting principle. We also manipulated whether providing jurors with additional information about the key characteristics of autism (in addition to the diagnostic label) affected credibility ratings. This is important given that lay people often hold clear, but not necessarily accurate, beliefs about autism (Gillespie-Lynch et al.
2015; Huws and Jones
2010). Following previous research (e.g., Wadley and Haley
2001), it was tentatively predicted that providing information about the child’s autism diagnosis together with information about autism may improve mock juror’s ratings of the child witnesses’ credibility. However, given the heterogeneity found across the autism spectrum, it was also predicted that this may be different for each child witness.
Results
Mean scores for the 11 credibility questions answered by mock jurors (in each of the three conditions) in relation to the video that they watched (Child A or Child B) are given in Table
4.
Table 4
Mean overall credibility ratings (on a 7-point Likert scale) for the children across the three conditions (Told AUT + info, Told AUT + no info, and Not told AUT + no info) (n = 120)
Mean credibility score (across all 11 items) | 4.45 (1.01) | 5.10 (0.78) | 4.51 (1.20) | 4.55 (0.92) | 3.86 (0.90) | 3.68 (1.17) |
How convincing was the child? | 4.60 (1.43) | 5.25 (0.85) | 4.70 (1.62) | 4.45 (1.10) | 3.75 (1.21) | 3.75 (1.25) |
How confident was the child in what they said? | 4.45 (1.47) | 5.05 (1.19) | 4.05 (1.57) | 4.45 (1.15) | 3.80 (1.28) | 3.70 (1.52) |
How confident was the child in their demeanour? | 3.90 (1.02) | 4.25 (1.07) | 3.45 (1.54) | 4.20 (1.28) | 3.65 (1.27) | 3.45 (1.43) |
How competent was the child? | 4.35 (0.87) | 5.05 (1.19) | 4.35 (1.09) | 4.45 (0.94) | 3.90 (1.07) | 3.55 (1.28) |
How honest was the child? | 5.20 (1.24) | 6.10 (1.02) | 5.50 (1.15) | 5.75 (1.02) | 5.05 (1.19) | 4.45 (1.19) |
How believable was the child? | 4.95 (1.32) | 5.50 (0.89) | 4.90 (1.48) | 4.70 (0.92) | 4.10 (1.21) | 3.95 (1.47) |
How complete was the child’s account? | 4.10 (1.74) | 4.50 (1.28) | 4.00 (1.49) | 3.75 (0.97) | 3.00 (0.79) | 3.20 (1.44) |
What was the child’s overall level of cognitive functioning? | 4.60 (1.54) | 4.90 (1.25) | 4.70 (1.49) | 4.40 (1.09) | 3.80 (0.83) | 3.75 (1.48) |
What was the child’s capability to testify? | 4.30 (1.52) | 5.15 (1.31) | 4.60 (1.63) | 4.45 (1.32) | 3.70 (1.30) | 3.50 (1.39) |
How good was this child overall? | 4.15 (1.50) | 5.20 (1.20) | 4.60 (1.50) | 4.50 (1.28) | 3.65 (1.27) | 3.50 (1.36) |
Prior to exploring the effect of providing jurors with information about autism and/or the child’s diagnostic label, a principle component analysis (PCA) using Oblimin rotation was conducted on the 11 credibility ratings (to reduce the dimensionality and noise in the data). The Kaiser-Mayer-Olkin measure verified the sampling adequacy for the analysis, KMO = 0.92, and KMO values for all individual items were > 0.8; well above the acceptable limit of 0.5 (Field et al.
2010). Bartlett’s test of sphericity, χ
2 (55) = 1157.28,
p < .001, indicated that correlations between items were sufficiently large for PCA. An initial analysis was run to obtain eigenvalues for each component in the data. One component had an eigenvalue over Kaiser’s criterion of 1 and explained 67.2% of the variance. The scree plot showed an inflection that would justify retaining component 1 (see Table
5 for factor loadings). All 11 items clustered on the same component, suggesting that component 1 represented ‘overall credibility’. This component was used as a dependent variable for subsequent analyses.
4
Table 5
Principal components analysis loadings (n = 120)
What was the child’s capability to testify? | 0.88 |
How good was the child overall? | 0.88 |
What was the child’s overall level of cognitive function? | 0.85 |
How accurate was the child? | 0.84 |
How believable was the child? | 0.84 |
How convincing was the child? | 0.84 |
How competent was the child? | 0.83 |
How complete was the child’s account? | 0.82 |
How confident was the child in what they said? | 0.80 |
How honest was the child? | 0.75 |
How confident was the child in their demeanour? | 0.65 |
To explore mock jurors’ perceptions of the witness’ credibility, two one-way between participants ANOVAs were used (one for each child) with Condition (‘Told AUT + info’, ‘Told AUT + no info’, ‘Not told AUT + no info’) as the independent variable and the ‘overall credibility’ component (extracted from the PCA) as the dependent variable. There was no significant effect of Condition for Child A, F(2, 57) = 2.55, p = .09, ηp2 = 0.08, although Condition was significant for Child B, F(2, 57) = 4.18, p = .02, ηp2 = 0.13. Bonferroni corrected post-hoc t tests revealed that Child B was judged as more credible in ‘Told AUT + info’ condition relative to the ‘Not told AUT + no info’ condition (p = .02), with all other comparisons between conditions non-significant (ps > .05).
Next, bivariate correlations were used to explore the relationship between mock jurors’ credibility ratings and their knowledge/experience of autism. These revealed that there was a significant positive relationship between scores on the ‘overall credibility’ component (extracted from the PCA) and mock jurors’ ratings of prior knowledge/experience of autism for those who viewed the video of Child B (r = .34, p < .05), but not for those who viewed the video of Child A (r = − .01, p = .93).
Discussion
This preliminary study examined how mock jurors perceived the credibility of two child witnesses on the autism spectrum and examined whether informing mock jurors about the child’s autism diagnosis (with or without the provision of additional information about autism) affected mock jurors’ credibility ratings. Results demonstrated that mock jurors’ credibility ratings were, to an extent (i.e., for only one child witness), influenced by the provision of the child’s diagnostic label and information about autism. Specifically, ratings of credibility were lower for Child B if no autism label or autism information had been provided, relative to both being provided. However, this was not the case for Child A (credibility ratings did not vary with the provision of information about autism or the child’s diagnostic label). It is important to consider why the results might differ for these two child witnesses on the autism spectrum.
One potential reason relates to the level of detail provided by each child during their investigative interview. Previous juror perception studies (e.g., Henry et al.
2011) have shown that witnesses who recall more information about an event are perceived to be more credible. Consistent with this suggestion, Child A (who was judged to be more credible on many items rated by mock jurors) recalled more details at interview (43) than Child B (27). Hence, there may have been less need to use Kelley’s (
1971,
1972) ‘discounting principle’ (that one explanation is diminished when observers are alerted to an alternative one) with Child A because his account was fuller and, possibly for this reason, more convincing. Importantly, for witnesses diagnosed with autism, a growing body of research has demonstrated that autistic children often recall less information than their non-autistic peers, largely during free recall (e.g., Bruck et al.
2007; Henry et al.
2017b; Mattison et al.
2015; McCrory et al.
2007). Whilst this is not always apparent during more structured questioning (e.g., Henry et al.
2017a), there is still significant variability in the amount that children (including those on the autism spectrum) recall: some individuals provide detailed and extensive accounts, whereas others provide sparser reports, even with more structured questioning (cf. Henry et al.
2017a). It is particularly important for jurors to be made aware of this prior to evaluating an autistic child’s testimony, since media representations of people on the autism spectrum commonly emphasise prodigious memory skills, potentially raising expectations of autistic people’s abilities to an unrealistically high level (Draaisma
2009).
The two children in the videos may also have differed in other important ways that had a bearing on mock juror credibility ratings. Whilst similar in age, gender, IQ, and receptive language, the two children may have differed in their manifestation of autism (i.e., the way in which their behaviours appeared to mock jurors). A key feature of autism is that different individuals display different behaviours and to different degrees (as emphasised in the information given to jurors in the ‘Told AUT + info’ condition); thus, it was not surprising that the two autistic children who featured in this study varied in their behaviour and demeanour. Our sample of 24 lay people who rated the videos on a series of behavioural characteristics highlighted several differences. While Child A was rated as significantly more monotonous than the Child B, Child B was rated as significantly less composed, coherent, and focused than Child A, whilst also being rated as showing significantly less appropriate use of vocabulary (the latter observation being in line with the poorer expressive language scores for Child B relative to Child A, on standardised tests). Again, Kelley’s (
1971,
1972) discounting principle may only have applied to Child B, who displayed atypical behavioural characteristics possibly associated with autism (i.e., being less focused, composed, coherent). This might link to our mock jurors’ ratings of their knowledge/experience of autism, which correlated with credibility ratings for Child B but not Child A. These findings suggest that knowledge/experience of autism may affect credibility judgements only when the child is displaying more atypical behavioural characteristics, which has implications for children who may not match the stereotypical view of an autistic child, such as some girls on the autism spectrum (Gould and Ashton-Smith
2011), an important area that warrants further investigation.
Overall, the results of this preliminary study highlight how care must be exerted regarding how information about autism is presented to juries. We suggest that, rather than providing generic information about autism, information specific to each autistic child should be presented; explicitly outlining how their autism manifests and how it might impact on their testimony (if at all). Surveys have highlighted how legal professionals (e.g., police officers, barristers), even with fairly good knowledge of the key characteristics of autism, often lack confidence in their abilities to make accommodations and adjustments to meet the needs of autistic witnesses and defendants (George et al.
2018; Maras et al.
2017). As such, they may be unsure how to present information about autism to a jury. In England and Wales, the Registered Intermediary (RI) role has been introduced; initially as a pilot scheme (in 2004) before being rolled out more widely in 2009 (Cooper
2016; Cooper and Wurtzel
2014; Plotnikoff and Woolfson
2015). RIs are impartial, trained professionals who facilitate communication between vulnerable witnesses
5 and criminal justice professionals. RIs are individually ‘matched’ to vulnerable witnesses based on the expertise of the RI and the needs of the witness (as such, an autistic child will likely be matched to an RI with expertise in both autism and working with child witnesses). As part of their wide-ranging role, the RI will conduct detailed assessments of the vulnerable witness, to identify how to facilitate fair and appropriate access to justice for that specific individual. An RI would, therefore, be in an excellent position to prepare individualised information about a witness for a jury, which the current research suggests is urgently needed. There is growing international interest in implementing RI schemes outside of England and Wales (Cooper and Mattison
2017; Plotnikoff and Woolfson
2015), with schemes being successfully implemented in other jurisdictions (Cooper
2016; Cooper and Wurtzel
2014). Until such a procedure exists in other countries, expert witnesses (or similar) may be able to provide such information to juries, in collaboration with the vulnerable individual (if possible) and those who know them well (e.g., family members, teachers, support workers).
Finally, it is important to highlight the strengths and limitations of this research. Most juror perception research uses transcripts of evidential interviews (e.g., Henry et al.
2011), enabling a high degree of experimental control but lacking real-world applicability. The current study presented ecologically valid video-recorded interviews, as would be presented to jurors in a real case. Yet this does come at a cost—whilst the results showed how credibility ratings differed for each child depending on the information provided to jurors, it was unclear exactly why these differences occurred. Within the constraints of this preliminary study, it was not possible to provide an objective measure of the children’s levels of autistic traits, nor explore precisely which aspects of the children’s testimony might have affected credibility ratings (e.g., by interviewing mock jurors about their rationale for their credibility judgements). It is important that future research (using a wider range of child witnesses) examines this more systematically.
Further, the videos shown to mock jurors were taken from an experimental study in which the children had viewed a video of a mild mock theft (as observers, rather than as active participants). In real-life, undertaking an investigative interview with a police officer (potentially about a serious or emotionally upsetting event) could cause high levels of anxiety and stress for a witness. This may be especially true of children on the autism spectrum, with one review suggesting that between 11 and 84% of autistic children experience impairing anxiety to some degree (White et al.
2009). This could affect how the child presents during interview, which may affect juror perceptions of credibility. Future work should, if possible, explore how jurors rate the credibility of autistic child witnesses in real-life evidential interviews.