The past two decades have seen the development of research-based recommendations for the conduct of forensic interviews with typically developing (TD) children who have been witness to, or victims of, crimes (Lamb, Hershkowitz, Orbach, & Esplin, 2008). Less attention has been given, however, to particular groups of vulnerable witnesses, including those with intellectual disabilities (also referred to as learning difficulties, developmental delays, developmental disabilities, learning disabilities, or mental retardation) and those with other disorders usually first diagnosed in infancy, childhood, or adolescence (e.g., pervasive developmental disorders such as autism, and attention deficit and disruptive behavior disorders such as attention-deficit/hyperactivity disorder [ADHD]). Children with disabilities are a particularly vulnerable group of witnesses. They are both more likely to experience or witness abuse (Balogh et al., 2001; Crosse, Kaye, & Ratnofsky, 1993; Goldman, 1994; Hershkowitz, Lamb, & Horowitz, 2007; Randall, Parrila, & Sobsey, 2000; Reiter, Bryen, & Shachar, 2007; Sedlak & Broadhurst, 1996a, 1996b; Sobsey & Doe, 1991; Sobsey & Mansell, 1994; Sobsey, Randall, & Parrila, 1997; Sullivan & Knutson, 1998, 2000; Verdugo, Bermejo, & Fuertes, 1995; Vig & Kaminer, 2002; but see also Jaudes & Mackey-Bilaver, 2008) and yet less likely to report their abuse or to have their complaints investigated (Goldman, 1994; Reiter et al., 2007; Sharp, 2001) in a developmentally appropriate manner (Cederborg & Lamb, 2008), or have their capacities and limitations recognized in court (Cederborg & Lamb, 2006; Westcott & Jones, 1999). Thus, researchers and practitioners in a number of relevant fields (e.g., law, social services, policing, psychology) are increasingly recognizing the need for empirical research to provide an evidence base from which to (1) inform expectations of these witnesses; (2) guide the conduct of interviews that facilitate reporting without compromising reliability; and (3) develop resources, guidelines, and education for the legal system to improve access for alleged victims or witnesses who are both young and intellectually challenged.
There is a widespread perception that children with intellectual (or learning) disabilities (CWID) are even less able to provide meaningful accounts of their experiences than typically developing children (Aarons & Powell, 2003; Aldridge & Wood, 1998; Ericson, Perlman, & Isaacs, 1994; Henry, Bettenay, & Carney, 2011; Nathanson & Platt, 2005). Indeed, cognitive impairment is a central diagnostic feature of intellectual disability, and comorbid communication deficits are not uncommon. Police officers often feel they have insufficient skills, resources, and support when interviewing witnesses with intellectual disabilities, perceiving them as difficult interviewees as a result of behavioral difficulties and cognitive, communicative, and attentional limitations (Aarons & Powell, 2003; Aarons, Powell, & Browne, 2004; Milne, 1999; Sharp, 2001). Negative perceptions about the reliability and suggestibility of witnesses with intellectual disabilities appear to be widespread among police officers, legal professionals, and mock jurors (Aarons & Powell, 2003; Nathanson & Platt, 2005; Peled, Iarocci, & Connelly, 2004; Stobbs & Kebbel, 2003), meaning that cases are less likely to be investigated because successful outcomes (i.e., guilty verdicts) are deemed unlikely (Aarons & Powell, 2003; Aarons et al., 2004). Nevertheless, this group increasingly does participate in forensic interviews and court trials in a number of countries (e.g., Cederborg, Danielsson, LaRooy, & Lamb, 2009; Cederborg & Lamb, 2008; Cederborg, LaRooy, & Lamb, 2008; Connolly, personal communication, June 2011; Hanna, Davies, Henderson, Crothers, & Rotherham, 2010), despite the concerns outlined above. Indeed, 4% of the children testifying as witnesses in New Zealand recently had an intellectual disability (Hanna et al., 2010), and between August 2009 and June 2011, 215 applications were made for registered intermediaries to support child witnesses in the United Kingdom (Connolly, personal communication, 2011). Furthermore, whether a case ultimately reaches court or not, CWID are likely to be “interviewed” in a number of contexts, both informal (e.g., by parents, caregivers, or the persons they first disclosed to) and formal (e.g., child protection workers, investigators, attorneys). Thus, evidence-based information about how CWID narrate their personal experiences and the interviewing strategies that may enhance or detract from the accuracy of their accounts is sorely needed.
Even when cases involving CWID reach court, procedures and attitudes undermine their ability by seldom acknowledging or accommodating witnesses’ intellectual difficulties (Cederborg & Lamb, 2008; Kebbell, Hatton, & Johnson, 2004; O’Kelly, Kebbell, Hatton, & Johnson, 2003). Although complex, directive, and suggestive questions abound, judges tend not to intervene to reduce the potentially harmful impact of such questions on the witnesses’ reliability and credibility (Kebbell, Hatton, Johnson, & O’Kelly, 2001).
There are many possible reasons why intellectual disabilities may compromise children’s abilities to provide meaningful and reliable eyewitness testimony (Henry et al., 2011). Slower information processing, poorer comprehension of events, and more specific deficits (particularly in working memory and executive control) may reduce the amount of information that is encoded (Clements, 1998; Henry, 2001; Milne & Bull, 1999; Swanson, 1990; Swanson & Trahan, 1990; Vicari, 2004). Communicative deficits may mean that CWID are less able to report what they do recall (Clements, 1998; Ericson et al., 1994; Moss, 1998). Social demands during the interview or court testimony may make these children more susceptible to suggestive techniques (Sigelman, Budd, Spanhel, & Schoenrock, 1981), and cognitive limitations may allow them to be confused more easily by lawyers’ questioning styles (Ericson et al., 1994). In reality, any or all of these processes may be at play, because CWID do not comprise a homogenous group, even when they share diagnostic labels (Cederborg & Lamb, 2008; Clements, 1998; Hatton, 1998; Vicari, 2004).
Despite widespread perceptions that CWID make less able witnesses, the findings are inconsistent and vary depending on question type and the type of competency assessed (e.g., recall vs. suggestibility). When information is elicited with open questions or during free recall, for example, CWID have been shown to provide as much information as typically developing (TD) children (chronological age- [CA-]matched: Agnew & Powell, 2004; Dent, 1986, 1992; Henry & Gudjonsson, 1999, 2003; mental age- [MA-] matched: Agnew & Powell, 2004; Gordon, Jens, Hollings, & Watson, 1994; Henry & Gudjonsson, 1999; Jens, Gordon, & Shaddock, 1990; Michel, Gordon, Ornstein, & Simpson, 2000), less information than CA-matched children (Henry & Gudjonsson, 2004, 2007; Michel et al., 2000), and more information than MA-matched children(Henry & Gudjonsson, 2003). When asked closed or specific questions, CWID provide as much information as MA-matches (Henry & Gujonsson, 1999; Jens et al., 1990; but see Gordon et al., 1994) and less than CA-matches (Dent, 1986, 1992; Henry & Gudjonsson, 2003, 2004, 2007). Findings concerning suggestibility are similarly inconsistent, with some studies revealing no differences between CWID and CA- or MA-matched children (Henry & Gudjonsson, 1999, 2004; Jens et al., 1990; Robinson & McGuire, 2006) and others showing heightened suggestibility relative to CA-matches (Gudjonsson & Henry, 2003; Henry & Gudjonsson, 1999, 2007; Michel et al., 2000; Young, Powell, & Dudgeon, 2003). Almost without exception, however, researchers have shown no differences in the overall accuracy of the accounts provided by CWID and TD children responding to open questions (Agnew & Powell, 2004; Henry & Gudjonsson, 2003). As with TD children, the amount and quality of information elicited from CWID is affected by the way in which they are interviewed (Brown & Lamb, 2009; Brown, Lamb, Pipe, & Orbach, 2008).
Cross-study differences, however, limit the extent to which existing research informs interviewing practices in the forensic context. For example, some studies focus on event memory in children who have specific developmental or learning difficulties but average cognitive abilities (e.g., autism spectrum disorders: Bruck, London, Landa, & Goodman, 2007; McCrory, Henry, & Happé, 2007; specific learning disability: Nathanson, Crank, Saywitz, & Reugg, 2007), whereas others have examined transcripts of interviews with CWID to evaluate interview dynamics but cannot elucidate accuracy (e.g., Cederborg et al., 2009; Cederborg & Lamb, 2008; Cederborg et al., 2008). Still others have explored the effectiveness of using different interview techniques with CWID (Dent, 1986, 1992; Milne & Bull, 1996; Robinson & McGuire, 2006). We have identified only 12 empirical studies in which the performance of CWID was compared with that of TD children, and these varied considerably on a number of dimensions that might also affect performance (e.g., age; severity of intellectual disability [ID]; whether the CWID were compared to childrenmatched for MA, CA, or both; the event-to-be-recalled; analysis of suggestibility vs. reliability; question type; delay), as more fully explained below.
Most studies have only included children within a single narrow age range, with different control groups (matched for mental and/or chronological age) and little consideration of the severity or type of learning disability (but see Agnew & Powell, 2004; Henry & Gudjonsson, 2003), despite concerns that CWIDs do not comprise a homogeneous group (e.g., Beail, 2002; Clare & Gudjonsson, 1993; Milne, 1999). We included two groups of CWIDs—those with “mild” or “borderline” intellectual disability (IQ = 55–78), and those whose disability fell within the moderate range (IQ = 44–53), and we included MA- and CA-controls for each CWID participant. We also included a wide range of ages within our CWID sample (7–12 years) so that we could explore the competencies of younger children than those who have typically been studied.
TD children recall personally experienced events better than observed events (e.g., Baker-Ward, Hess, & Flannagan, 1990; Jens et al., 1990; Murachver, Pipe, Gordon, Owens, & Fivush, 1996), so caution is needed when generalizing to the forensic context from studies using other types of stimulus events, including those in which to-be-remembered “events” were observed (Beail, 2002; Gudjonsson & Henry, 2003; Henry & Gudjonsson, 2003). Only three studies with CWIDs have involved personally experienced stimulus events (Gordon et al., 1994; Jens et al., 1990; Michel et al., 2000). Our study used a novel, rich, and interactive event that allowed children to provide a wide range of information when interviewed.
Range of Competencies Under Investigation
Many studies of CWID have focused on their suggestibility and acquiescence (Agnew & Powell, 2004; Gudjonsson & Henry, 2003; Henry & Gudjonsson, 1999, 2003; Milne & Bull, 1996; Sigelman et al., 1981) and have highlighted the dangerousness of certain strategies (e.g., suggestive questions) but have not elucidated the capacities of CWIDs interviewed in a neutral or supportive manner. We thus need more studies exploring the conditions under which recall may be enhanced. In the present study, we explored the effects of age and severity of intellectual disability on different memory processes and indices of competency (e.g., completeness vs. accuracy vs. suggestibility) to advance our understanding of memory development in CWIDs.
We know that children are less accurate when responding to suggestive questions for both social (e.g., demands of the interview context, desire to please the interviewer, or acquiescence to the perceived authority of the interviewer) and cognitive (e.g., source-monitoring difficulties, weak memory traces) reasons (Ceci & Bruck, 1998). Studies of witnesses who have learning disabilities have not been able to identify which of these processes explains their heightened suggestibility (e.g., Gudjonsson & Henry, 2003). Zigler, Hidgen, and Stevenson (1958) showed long ago that CWID were more sensitive to social reinforcement when performing cognitive tasks than were MA-matched TD controls, and many other psychologists have suggested that people with intellectual disabilities are more vulnerable witnesses because they are especially eager to please questioners (e.g., Milne & Bull, 1999). Although the severity of intellectual disability may be associated with decreased accuracy, few researchers have asked whether these problems can be ameliorated by appropriately supportive interviewing (Robinson & McGuire, 2006). In the current research, we asked whether CWIDs of different ages and levels of intellectual disability might benefit from the supportive conditions built into the National Institute of Child Health and Human Development (NICHD) Investigative Interview Protocol (Lamb et al., 2008). We also sought to examine CWID’s recall in response to different types of questions (open vs. closed), as well as a series of suggestive questions that varied in format (open vs. closed) and content (leading vs. misleading).
In previous studies, the gaps between target events and interviews have been minimal, typically 1 day, although some studies have included a second interview 2–6 weeks later (Gordon et al., 1994; Henry & Gudjonsson, 2003; Michel et al., 2000). Most forensic interviews involve delays of weeks or months, and some for even longer (Hershkowitz, Horowitz, & Lamb, 2005). Several additional months may pass before investigations reach court. A survey of young witnesses in the United Kingdom showed delays averaging 11.6 months (Plotnikoff & Woolfson, 1995), for example, and similarly long delays have been found in the United States also (e.g., Pipe, Orbach, Lamb, Stewart, & Abbott, 2008). Accordingly, we examined CWID’s memories for personally experienced events when interviewed for the first time after a short (1-week) delay with those interviewed after a longer (6-month) delay. We also examined recall across repeated interviews to determine whether the CWID’s recall and reporting were affected by repeated interviewing; these data are the focus of another report and are not described here. The current study examined recall and reporting of a personally experienced event in CWID of varying severity (Mild vs. Moderate) and in comparison with TD children matched for both MA and CA. Children were recruited from special schools or identified during brief cognitive assessments in mainstream schools. We excluded children with diagnosed syndromes (e.g., William’s syndrome) and pervasive developmental disorders (e.g., autism spectrum disorders) to enhance the homogeneity of our sample and because the excluded children often have specific information processing deficits and neuropsychological characteristics (e.g., Henry, 2001; Vicari, 2004). Some of the children included in the Mild ID group had estimated IQ scores that fell within the Borderline range (n = 20, IQ range = 72–78), but because these children were attending special schools and thus had well-documented cognitive impairments, we included them in the study as other authors have done (Agnew & Powell, 2004; Murfett, Powell, & Snow, 2008). Children took part in a 45-min-long staged event (an interactive presentation about first aid and safety) at their school, modeled after an event used successfully in previous studies (Brown & Pipe, 2003a, 2003b). Half of the children in each group were interviewed 1 week later, with the remaining children interviewed 6 months after the event. All children were interviewed using the NICHD Investigative Interview Protocol (Lamb et al., 2008; Lamb, Orbach, Hershkowitz, Esplin, & Horowitz, 2007), which is consistent with best practice guidelines for the conduct of forensic interviews with children. At the conclusion of the NICHD Investigative Interview Protocol, the childrenwere asked a series of suggestive questions that varied in content (central vs. peripheral detail), suggestiveness (leading vs. misleading), and style (open vs. closed).
Based on previous findings, we expected that CWID would report as much information as MA-matched controls, and that children in both groups would report less than those matched for CA. We expected all children to provide similarly accurate accounts. We expected highly suggestive questions to have a heightened (negative) impact on the accuracy of the responses by CWID and MA-matched children. Because few researchers have specifically examined reports made by children with moderate intellectual disabilities, we expected them to perform less well than the CA-matched children but made no predictions regarding their performance relative to MA-matched children. Because the delays were longer than in previous studies, we made no specific predictions regarding group differences in recall, but expected that delay would affect the amount and accuracy of information reported by all children.
Children (n = 206; 86 female) were recruited from four mainstream schools and five schools for children with intellectual disabilities. The timing of the first interview was a between-subjects design: Approximately half (n = 112) were interviewed for the first time at 1 week, with the remainder (n = 94) interviewed for the first time 6 months after the event. Table 1 presents the descriptive data regarding the composition of each group with respect to sample size, gender, age, and estimated IQ scores.
ccp-80-5-829-tbl1a.gif Characteristics of the Sample (Collapsed Across Children Interviewed at 1 Week or 6 Months)
To confirm that children in the MA group were indeed younger than those in the other conditions, and that there were no significant chronological age differences across the remaining conditions, a univariate analysis of variance (ANOVA) on age (months) was conducted. This showed a significant main effect for condition, F(3, 198) = 103.78, p < .001, ηp2 = .61; Tukey tests showed that children in the MA group were significantly younger than all others, who did not differ (all ps < .001). There was no effect for the timing of the first interview and no interaction between timing and age.
Children were categorized into four groups on the basis of their performance on four subtests (Picture Completion, Information, Block Design, and Vocabulary) of either the Wechsler Preschool and Primary Scale of Intelligence—Third Edition, U.K. Version (WPPSI-III-uk; Wechsler, 2003) or the Wechsler Intelligence Scale for Children—Third Edition, U.K. Version (WISC-III-uk; Wechsler, 1992), and, in the case of the intellectual disability groups, in conjunction with additional information reflecting adaptive function deficits or poor academic achievement consistent with a low level of intellectual function (as indicated by either attendance at a special school or targeted teacher aid assistance provided through mainstream schooling). Children were placed in the CWID (mild) group if their estimated IQ score fell below 80. A number of the children (n = 20) had scores that fell within the Borderline range of intellectual function, but because the overall IQ scores were indicative and the childrenhad well-documented cognitive and adaptive functional impairments (and so attended special schools), these children were included in the mild intellectual impairment group. Children were allocated to the CWID (moderate) group if their estimated IQ score fell within the range of 40–55. To be included in the study, the children had to be capable of basic verbal communication (minimum phrase-based speech), confirmed in consultation with the child’s teacher. Those with ID arising from organic syndromes (e.g., Down’s syndrome) and those with diagnoses (confirmed or pending) of autistic spectrum disorder were excluded. Children were also excluded if they had comorbid conditions (e.g., ADHD, conduct disorder) or histories of infections, trauma, or brain injuries contributing to their cognitive deficits. Children were included in the TD group if their estimated IQ scores fell within the average range. Univariate analysis of estimated IQ scores for the four groups revealed a significant main effect of condition, F (3, 198) = 384.84, p < .001, ηp2 = .85; Tukey tests indicated CWID (Moderate) had lower IQ scores than CWID (Mild), who in turn differed from children in both of the TD groups (all p s < .001). Equal numbers of children with consent to participate were allocated to each delay group following the initial event but some were unavailable for interview 6 months after the event because they had moved out of the area or were absent from school due to illness or family holidays. This was a particular problem with children in the moderate ID group, of whom there were fewer available for recruitment in the first instance. No effects of the timing of the first interview or interaction between timing and group were evident in analyses of the IQ scores.