Eliciting an offence narrative: what types of questions do forensic mental health practitioners ask?

Abstract

Maximising the accuracy and detail of information elicited through a clinical–forensic interview may increase the reliability and validity of an individual’s assessment. Despite this, there is little empirical research on what questions forensic mental health practitioners employ, and whether these correspond with empirically established interviewing strategies. In this study, 22 forensic mental health practitioners participated in a mock interview of a young person referred for a sexual risk assessment. The results highlighted that participants asked very few ‘open’ questions, over-relied on ‘specific’ questions and an average of 13% of questions were leading. Finally, practitioners predominantly used ‘yes/no’ questions when exploring the young person’s thoughts, feelings, and physiological responses. Overall, the study demonstrated that empirically supported interviewing techniques were not commonly employed and highlighted the need for further professional development and training around clinical forensic interviewing strategies that best elicit the information needed to inform risk assessment.

Recent attention has been drawn to the lack of literature informing clinical forensic interviewing strategies (Logan, 2018). While experts in the area have suggested frameworks and approaches for undertaking these interviews (Day et al., 2019; Medoff & Kinscherff, 2006; Shea, 2017), research on investigative interviewing has demonstrated that knowledge and awareness of best practice approaches do not necessarily result in applied skills (Powell, 2005). It is important, therefore, to explore how practitioners undertake interviews in practice, the extent to which best practice approaches are utilised, and what additional considerations (e.g. professional development or training) may be required.

In this paper, we briefly outline the empirical literature on effective clinical and forensic interviewing skills, with a focus on children and young people. We have focused on this age group for two reasons. First, children and young people are more susceptible to errors in response to poor interview techniques (e.g. leading questions) than mainstream adults (Bowles & Sharman, 2014). Second, there is extensive research exploring the most effective methods for eliciting reliable and detailed information from children so best practice has been empirically validated (National Children’s Advocacy Centre, 2019) and has informed the basis of this study.

Following an outline of the current best practice principles for clinical forensic interviews with youth, we present the current study that explored the question types used by forensic mental health practitioners undertaking a mock clinical forensic interview. We conclude with a discussion of how these question types compare to both recommended best practice and empirically supported interviewing questions that best elicit accurate, comprehensive, and specific offence narratives.

Clinical forensic interviews combine aspects of both clinical interviewing and forensic interviewing approaches. The purpose of a clinical interview is to develop an understanding of an individual’s presenting mental health issues, which may include eliciting information on the nature, frequency, and severity of symptoms and the subjective experience of the individual. In contrast, the purpose of a forensic interview is to elicit detailed facts and objective information pertaining to a specific legal issue or criminal behaviour. Frequently, the forensic interviews are undertaken by police or child protection personnel. Therefore, the purpose of a combined (or ‘hybrid’) clinical forensic interview approach is to develop an understanding of a person’s overall presentation and functioning, and how this relates to their engagement in illegal behaviour. This is done through exploring the ‘matters relevant to the reason for their detention in a secure facility or the restrictions legally imposed upon them’ (Logan, 2018, p. 300). Clinical forensic interviews therefore often explore both a person’s subjective experience and the objective facts pertaining to their specific legal issue.

In Australasian practice, clinical forensic interviews are conducted by a wide range of disciplines in a diverse set of contexts, including child and youth forensic mental health services. They may be conducted by psychiatrists, psychologists, social workers, and/or other mental health practitioners in prisons, in-patient units, or community-based mental health or justice settings. Logan (2018) argued that practitioners likely need ‘a hybrid of clinical and forensic interview skills, strategies and techniques’ (p. 299) to undertake their role. However, due to the diverse training and experience of practitioners conducting interviews, there is unlikely to be one standard approach to this important aspect of assessment.

Leach and Powell (2020) explored what a hybrid approach to clinical forensic interviewing may involve by reviewing the literature on forensic investigative interviewing and clinical suicide risk interviews. Their paper highlighted that there are many synergies between the approaches, including the emphasis on preparation, the need to outline the purpose of the interview, and the importance of building rapport. It was identified that there was also a need to ensure that accurate information was elicited from interviewees to maximise the validity of the information elicited.

Literature on both clinical and forensic interviewing highlights the importance of maximising the use of ‘open’ questions and avoiding the use of ‘leading’ questions (Leach & Powell, 2020). Open questions are framed in a way that encourage elaborate detail without specifying what type of detail is required, for example ‘What happened next?’ or ‘Tell me more about that’. In contrast, specific questions dictate what precise information is required, for example ‘What did she look like?’ or ‘Did you close the door?’. Specific questions need to be avoided until after the narrative is exhausted, when the interviewer seeks to elicit necessary details that were not provided earlier. Leading questions can presume information that has not already been disclosed or suggest the desired answer to the interviewee and can be either ‘open’ or ‘specific’ in nature. Ideally, leading questions should be avoided, particularly when discussing the issue in contention.

There is a wealth of literature that has demonstrated that ‘open’ questions elicit the most accurate and detailed information from interviewees (Benia et al., 2015; La Rooy et al., 2015; Memon et al., 2010; Powell, 2005; Powell & Brubacher, 2020; Powell & Snow, 2007). Importantly, interviewees also feel more heard and understood when ‘open’ questions are utilised. Brubacher and colleagues (2019) explored the experiences of children (aged 7–12 years) after they completed two interviews about a short film they had watched. One interviewer asked only open questions, and the other asked only specific questions. The majority of the children reported that the ‘specific’ questions were easier to answer, yet the ‘open’ interviewer rated significantly higher on variables exploring rapport and content. Across the sample, a majority of the children reported that the ‘open’ interviewer appeared more interested, listened more, and elicited more information than the interviewer that only asked specific questions. This is important insight in the context of clinical forensic interviews, whereby perceptions of the interviewer may impact on an individual’s willingness to accept the recommendations of the assessment and participate in related therapeutic interventions. It appears that ‘open’ questions have the advantage of eliciting more information, enhancing rapport between the interviewer and interviewee, and inducing feelings of validation for the interviewee.

Historically, investigative interviewing research has focused on eliciting detailed factual knowledge, with less attention paid to the subjective experience of the informant, which is more relevant in a clinical forensic interview. Children rarely provide memories of emotional experiences spontaneously in forensic interviews (Katz et al., 2016; Lamb et al., 1997; Westcott & Kynan, 2004) yet researchers in this area have acknowledged that expressions of emotional experiences may enhance the perceived credibility of the victim (Castelli & Goodman, 2014). Katz and colleagues (2016) reviewed the transcripts of 97 children interviewed with the National Institute of Child Health and Development (NICHD) Protocol, which is an evidence-based protocol for investigative interviews that emphasises the use of ‘open’ questions to elicit an accurate and detailed narrative. Their study demonstrated that in over one third of interviews, children did not use emotional language when providing a narrative of their abuse. Of the children that did include emotional language in their narrative of the offence, there was an average of 3.88 (SD = 3.07) instances of emotional language. Children aged 10–14 years used a higher average of emotional words than children aged 4–6 years, which is understood as developmentally normative.

Given the low rate of emotional language in interviewee’s narratives, Katz et al. (2016) concluded that interview protocols may need to be revised to enhance disclosures of emotional experiences. This may be even more relevant for children who are still undergoing developmental maturation in their cognitive, social, emotional, psychological, and communication functional domains. This is even more relevant in the context of clinical forensic interviewing, where the emotional experience of the young person at the time of offending is integral to formulating their offence and future risk.

In the absence of guidance on how to elicit accurate emotional language, it appears that interviewers most frequently use ‘specific’ questions (including ‘yes/no,’ ‘forced-choice’, or ‘cued-recall’), or clinically informed ‘leading’ questions to prompt the subjective experiences of an interviewee at the time of an event (Lyon et al., 2012). A retrospective study of 80 child-witness trial testimonies explored how such ‘evaluative’ content was elicited from child witnesses, and defined this as any information pertaining to the emotional, cognitive, or physiological responses of the child at the time of offence (Lyon et al., 2012). Consistent with the results described above, the study found that children rarely provided spontaneous ‘evaluative’ content if the question did not specifically direct them to provide an ‘evaluative’ response, with only 2% of responses including an unprompted description of thoughts, feelings, or physiological responses. Interestingly, even when attorneys specifically asked for ‘evaluative’ responses, only 23% of answers from children included the requested evaluative content. However, this was also impacted by the question type. Specifically, the most frequently used ‘option-posing’ (i.e. yes/no or forced-choice questions) ‘evaluative’ questions resulted in an evaluative response only 10% of the time, while the rarely used ‘how-’ (i.e. specific cued-recall questions) questions resulted in an ‘evaluative’ response around 75% of the time. The authors concluded that the infrequently used ‘cued-recall’ questions (such as ‘How were you feeling?’) may be required to elicit more individuals’ thoughts, feelings, and physiological responses when recalling abuse narratives (Lyon et al., 2012).

To address this, more recent iterations of investigative interviewing protocols encourage interviewers to specifically elicit emotional information (with questions such as ‘How did you feel?’ and ‘You said you felt [emotion], tell me more about that’), which has demonstrated a significant increase in the expression of emotion by alleged victims (Karni-Visel et al., 2019). These recent adaptations may be particularly useful in the context of a clinical forensic interview, where the subjective experience of the individual is critical to case conceptualisation and management. Interestingly, the Lyon et al. (2012) study found that ‘how-’ (i.e. cued-recall) questions elicited all types of ‘evaluative’ responses, including thought processes and physiological responses, which is relevant for clinical forensic interviewing. It could be argued that ‘cued-recall’ questions should be adopted by forensic mental health practitioners undertaking clinical forensic interviews, to elicit more reliable responses beyond the closed ‘yes/no’ or ‘forced-choice’ question types because they ask the interviewee to generate, as opposed to recognise, a response and thereby elicit more elaborate information.

One aspect of interviewing that is more difficult to resolve with a ‘hybrid’ clinical forensic interviewing approach is the use of ‘leading’ questions. The problem with leading questions is two-fold. First, leading questions increase the likelihood that inaccurate information will be provided (Melnyk et al., 2007). Second, interviewees sometimes incorporate misinformation into their memories, resulting in inaccurate recall of an event into the future (Loftus, 2005). Sharman and Powell (2012) found that participants were susceptible to these effects when the leading questions were either ‘open’ or ‘specific’ and included a large amount of detail. Investigative interviewing protocols now encourage interviewers to be mindful of leading questions and give interviewees an opportunity to confirm or deny the accuracy of new information introduced by the interviewer (Benia et al., 2015; Powell & Brubacher, 2020).

In contrast to the discouragement of leading questions by investigative interviewing researchers, experts in clinical interviewing skills advocate the judicious use of leading questions to facilitate disclosure by clients around sensitive topics (Shea, 2017). For example, Shea (2009) encouraged interviewers to use ‘gentle assumptions’ to explore usually taboo topics (i.e. assume the presence of normative behaviour, which may otherwise be stigmatised). For example, he would encourage a practitioner to ask ‘How often do you masturbate?’ rather than ‘Do you masturbate?’ Similarly, ‘symptom amplification’ is a strategy aimed at reducing informants’ tendency to minimise disturbing behaviour. With this strategy, an interviewer would intentionally overstate the likely frequency of a behaviour. For example, Shea would encourage clinicians to ask, ‘Would you say you masturbate ten or fifteen times per day?’ rather than ask ‘How often do you masturbate?’. In the latter example, the interviewee may minimise the frequency, under-report the behaviour, or try to provide a more socially desirable response.

Based on the literature described above, it appears that an effective ‘hybrid’ clinical forensic interviewing approach would preference the use of ‘open’ questions to elicit a detailed and reliable narrative of an offence and build a rapport with the interviewee. Further, forensic mental health practitioners should explore subjective experiences of an individual using ‘cued-recall’ questions (such as ‘How did you feel at that moment?’) to encourage them to provide ‘evaluative’ content. The appropriateness of using ‘leading’ questions when interviewing around sensitive topics is not yet clear, although it is likely that there may be multiple categories of leading questions, some of which should be avoided and some of which may be used judiciously. It is likely that leading questions that presume new material or indicate a desired response should be avoided. Conversely, there may be an ongoing place for leading questions that provide ‘gentle assumptions’ or ‘symptom amplification’ when discussing sensitive or taboo topics to maximise disclosure of critical information.

Given the diversity in training and experience of many forensic mental health practitioners undertaking clinical forensic interviews, it is unclear whether practising practitioners employ best practice question types. Due to the established literature regarding how to elicit a reliable and detailed narrative, it may be expected that interviewers utilise open, non-leading questions when eliciting an offence narrative. What is known, however, is that without adequate training many interviewers in other fields frequently utilise poor questioning techniques (Powell, 2005). For example, field recordings of UK police and social workers interviewing alleged victims of child abuse found that there was an average of 6% open questions in interviews and an average of 5% leading questions across interviews (Sternberg et al., 2001). A similar distribution was found in an Australian study of teachers who undertook a mock interview of a student, whereby they asked an average of 42.2 questions about a short scenario, which resulted in an average of 13% open and 33% leading questions (Brubacher et al., 2014). To date, there has been no comparable study exploring questions employed in a mock interview using a ‘hybrid’ clinical forensic interview approach, which is best reflected during the elicitation of a narrative around a sensitive subject matter (i.e. a sexual offence).

To address the gaps in research on clinical forensic interviewing approaches, the purpose of this study was to explore the question types used by practising forensic mental health practitioners when interviewing a young person. As an exploratory study, it is anticipated that the findings will provide insight into any gaps between the state of the research and skills in practice, which may inform both trainers and practitioners in regards to current training needs. This exploration may prompt researchers to further explore the role of leading questions in clinical forensic interviews and the most effective method to elicit thoughts and feelings within an offence narrative.

There were three key research questions:

1. What are the types of questions that forensic mental health practitioners use to elicit a young person’s offence narrative of a sensitive incident (i.e. sexual offence)?

2. To what extent was the use of ‘leading’ questions instrumental or clinically informed (e.g. ‘gentle assumptions’ or ‘symptom amplification’)?

3. What types of questions do forensic mental health practitioners use to elicit ‘evaluative’ information (i.e. thoughts and feelings) about the offence?

Method

Design

This was a mixed-methods exploratory study that primarily utilised content analysis to investigate the question types used by participants during a mock clinical forensic interview. For the first research question, a well-established coding protocol was applied, which has been used extensively in previous empirical training research (Benson & Powell, 2015; Brubacher et al., 2022). For the second and third research questions, there were no established coding protocols, so a unique protocol was developed based on the literature (see coding schedule). The study also included a validity check of the mock interview scenario, and participant responses were thematically analysed.

Sample

Participants were multi-disciplinary practising forensic mental health practitioners recruited via email through the researchers’ Australasian professional networks. This included forwarding emails through three youth forensic mental health services (across three jurisdictions), the Australasian section of Child and Adolescent Forensic Psychiatrists, and a network of practitioners undertaking assessment for a Children’s Court. There were a total of 22 practitioners who consented to participate, with their demographics outlined in Table 1. There was an average of 10.77 years of post-graduate experience (range = 2−34 years).

Table 1.

Demographics of sample.

Demographic variable	Participants (N)
Jurisdiction of service
New South Wales	2
New Zealand	5
Queensland	13
Victoria	2
Gender
Male	9
Female	13
Other	0
Discipline
Mental health nurse	1
Psychiatrist	6
Psychologist	11
Social worker	4

Note: n = 22.

Youth forensic mental health is a niche area in Australia and New Zealand. It is difficult to estimate the exact population of clinicians working in this area; however, in 2018 there were 572 endorsed forensic psychologists (Allan et al., 2018) in Australia, and approximately 200 psychiatrists across Australia and New Zealand have completed a formal sub-specialty training in Forensics since 2004 (Ellis, 2020). Taken together with other non-endorsed practitioners or disciplines, it may be estimated that there are approximately 1000 practitioners working in forensic mental health across Australia and New Zealand. A majority of these practitioners work in adult services, due to the significantly larger population of adults involved in the correctional system. For example, on 30 June 2021 there were 42,970 adults in custody, and this same quarter (April-June 2021) there was an average of only 819 young people in custody per day (AIHW, 2021). Taken together, it may be conservatively estimated that no more than one fifth of all forensic mental health professionals in Australia and New Zealand work with young people, therefore the total pool of practitioners may be around 200. A sample of 22 practitioners (approximately 10% of the estimated population) was determined to be a reasonable sample within this region.

Coding protocol

Research Question 1

For Research Question 1, the coding protocol has been used extensively within empirical studies of investigative interviewing (Benson & Powell, 2015). Each question posed by the participants during the mock interview was coded as one of six mutually exclusive question types based on those found to have empirical support in the literature. Specifically, three question types were ‘open-ended’ (breadth, depth, and descriptive questions), and three were ‘specific’ questions (cued-recall, yes/no, and forced-choice).

The first category was open-ended breadth questions, such as ‘What happened next?’, which encouraged the interviewee to provide a broad report of activities or events. The second category was open-ended depth questions, such as ‘Tell me more about when you filmed the woman’, which encourage the interviewee to elaborate in more detail about previously reported information. Both types of open-ended questions do not specify the type of information being requested. A third type of open-ended questions prompted the interviewee to provide descriptive information. These questions were coded as open-ended descriptive and included questions such as: ‘Tell me about the lady you saw swimming.’

The next three question categories pertain to specific questions that indicated the type of information required from the interviewee and usually elicit less detailed responses. Specific cued-recall questions included ‘who, what, where, when, why, and how’ questions. Specific yes/no questions were any question that dictated a ‘yes/no’ response. Finally, forced-choice questions offered alternative responses for the interviewee to choose from, such as ‘were you at the pool in the morning or the afternoon?’. Questions that began with ‘Can you tell me–’ were often coded as specific ‘yes/no’ questions (as per the investigative interviewing literature). This is in the context of younger children tending to take a literal understanding of the question and providing a yes/no response. Given that the mock interview was for a 15-year-old, we were more generous in our coding of these questions and coded them according to the remainder of the question. For example, ‘Can you tell me what happened then?’ would be coded as an open-ended breadth question, or ‘Can you tell me what she was wearing?’ would be coded as a specific cued recall question versus a strict ‘yes/no’ question.

All questions were double-coded as leading (1) or not leading (0). A question was coded as ‘leading’ if it introduced or assumed information that had not been mentioned by the interviewee, or indicated a desired response (for example ‘You were attracted to her, weren’t you?’).

Research Question 2

All questions that were coded as leading were reviewed and further coded into three categories for the variable leading type. This included ‘gentle assumptions’ (where information was assumed around a sensitive topic) and ‘symptom amplification’ (where the question was posed in a way to overstate socially undesirable behaviours). The remaining questions were ‘suggestive’, where the leading question suggested the desired response or assumed new information, without giving the interviewee an opportunity to confirm or deny it.

Research Question 3

Finally, all questions were coded to explore how participants asked about ‘evaluative’ content. The variable evaluative was created by dichotomously coding all questions as either narrative (0) or evaluative (1). All questions coded as evaluative were then further divided into either emotion-focused or thought-focused questions. Emotion-focused questions included questions about both emotional and physiological content (for example ‘How did you feel?’ and ‘Were you aroused?’ were both coded as emotion-focused).

Validity check

All participants were asked two questions to explore the ecological validity of the mock interview design. First, participants were asked how similar the young person was to young people that they would typically interview. Second, they were asked how similar the interview was to their usual style or approach. The initial response to each question was coded as (1) yes, (2) somewhat, or (3) no. Responses were coded as ‘yes’ where participants agreed the scenario was representative and did not provide a follow-up qualification or caveat to their agreement. Responses were coded as ‘somewhat’ when participants agreed it was representative but provided qualifications or caveats to their agreement. Responses were also coded as ‘somewhat’ when participants explained aspects that were both representative and not-representative of their real-life experiences interviewing young people. Responses were coded ‘no’ where participants clearly communicated that the interview was not representative.

All participants were asked follow-up prompts to explore their responses to the two validity questions in more depth. Responses to these questions were thematically analysed to identify limitations to the ecological validity of the study.

Procedure

Participants were invited to partake in the study to explore their process for interviewing young people during a sexual offence interview comprising a broader forensic risk assessment. They were informed that the study would include a brief conversation about their interview process and approach, followed by a short mock interview with an actor (played by an adult male practitioner with professional actor training and experience) in relation to a youth sexual offence, and would finish with a short debrief about their experience. The entire research session was scheduled for one hour, although participants were not informed of any specific time limits on their interview. Rather they were encouraged to continue the interview until they were comfortable that they had elicited all relevant details from the young person about the offence.

Participants conducted the interview either in-person with the actor and lead researcher present (n = 12), or via video-conferencing on the Microsoft Teams platform (n = 10). This dual-mode delivery was undertaken out of necessity to reduce barriers to participation of experienced practitioners located in other regions of Australia or New Zealand in light of travel restrictions following the COVID-19 pandemic. In addition to this, the dual-mode replicated the experience of practitioners during the COVID-19 pandemic where many undertook clinical forensic assessments via video-conference.

On the day of the interview, participants were provided with written instructions and a mock statement of facts. The instructions informed participants that they had already met with the young person (‘Nigel’) – who was easy to engage and open – and had already explored his personal history and other information related to his clinical presentation. The purpose of the mock interview was to explore the actual offence so that they could complete their risk assessment. The mock statement of facts was:

Victim reported that on 9.03.19 she was at Jackabees Swim Centre, 185 Longbay Road, Jackabee in a private shower cubicle in the women’s change room at approximately 1430. Victim had just exited the public swimming pool. Victim was facing away from cubicle door for approximately one minute and then turned around to wash her hair when she noticed a mobile phone being held under the wall of the cubicle to her right when facing the door. She reported observing that the screen was on and she could see movement on the screen, however was unable to confirm the image on the screen due to the steam in the shower. Victim also observed a hand to be holding the phone. Victim reported that she yelled out and kicked at the phone, at which time the hand and phone was removed. Victim reported that by the time she was able to exit the cubicle, the defendant had left the change room. Victim reported the incident to the Jackabees Swim Centre manager on duty Rebekah Kilmore, who reviewed the CCTV footage and observed the defendant entering the change room shortly after the victim. The defendant can then be observed running out of the change room several minutes later. Ms Kilmore was able to identify the defendant as Nigel Moore, who is a member at the Swim Centre. Ms Kilmore subsequently referred the matter to Police.

The defendant was interviewed about the offence on the 12.03.19 and admitted to entering the change room. He reported that he was confused about which change room he was in and then left when he heard the victim yelling. Defendant denied filming the victim. The defendant’s phone was seized and footage of the victim in the shower was located. Following this, defendant was re-interviewed on the 14.06.19 and he admitted to filming the victim. Defendant reported that he did not know why he filmed the woman, denied entering the change room with intent to film, and stated that he did it as a joke.

At the conclusion of the interview, participants were asked to discuss their experience and the validity of the mock interview. All interviews were audio-recorded, transcribed, and de-identified for analysis. For the purpose of this study, only the mock interview component of the study was coded. All interviews were coded by two authors, who completed basic training in the coding system developed by the Centre for Investigative Interviewing. During this process, the authors independently coded training transcripts, which were then reviewed by a trainer from the Centre, and feedback was provided. The authors then independently coded all interviews and cross-checked coding. Any discrepancies were resolved by reviewing the coding dictionary and reaching a consensus on how the question should be coded.

Analysis

Following the coding of the interviews, data were entered into SPSS-28 (IBM, 2021) and analysed. As an exploratory descriptive study the research questions were explored with descriptive statistics that included the range, mean, standard deviation, and percentage of each question type. The qualitative responses to the ecological validity question regarding the mock interview scenario were coded according to the coding schedule above, and discussions about the limitations of the scenario were analysed thematically. All responses to the validity question were entered into NVivo 12 software, and the first author applied the coding schedule developed and then identified the key themes around limitations. The results of the initial content analysis were summed and reported, along with a summary of the key limitations identified for each question.

Results

Research Question 1: What are the types of questions that forensic mental health practitioners use to elicit a young person’s offence narrative of a sensitive incident (i.e. sexual offence)?

Overall, the mean number of questions asked about the offence was 55.23 (SD = 22.64, range = 14–113). Participants asked few ‘open’ questions, with a mean of 4.27 (SD = 3.01) open questions per interview (range = 0–15), which was an average of 8.45% of all questions. All participants asked at least one ‘leading’ question, with a mean of 7.64 (SD = 6.01) leading questions (range = 1–23) resulting in an average of 13.30% leading questions per interview. Examples of leading questions included where the question suggested the desired response, such as ‘Oh, you weren’t thinking of posting it on social media or anything I suppose’ and ‘ . . . I’m guessing you haven’t done anything like this since then?’. Similarly, questions that inserted new information into the narrative were also coded as leading, such as ‘ . . . and did you know, I assume you knew the layout of the change rooms umm, wouldn’t, you’d know there were some shower units and things like that?’. A final example is questions with leading tags, such as ‘To see if you could just see a girl or some girls naked, yeah?’.

Table 2 shows the breakdown of question types. The most frequently used ‘open’ question was a ‘breadth’ question, such as ‘What happened next?’. The most frequently used question type overall was ‘yes/no’ questions, which accounted for almost two thirds of all interview questions asked. Overall, only one participant asked an ‘open’ leading question, whereas the remainder of leading questions were ‘specific’ questions, with the most frequently used leading question being a ‘yes/no’ question (Table 2).

Table 2.

Breakdown of question types by quantity, leading, and percentage.

Question type	Range	Total M	Total SD	Percentageof all questions	M leading	SD leading
Open
Breadth	0–14	3.41	2.97	6.35	0.05	0.213
Depth	0–3	0.82	1.14	2.04	0.0	0.0
Descriptive	0–1	0.05	0.21	0.06	0.0	0.0
Specific
Yes/no	9–77	33.14	16.33	59.03	6.95	5.33
Cued recall	3–25	13.91	6.21	25.88	0.27	0.63
Forced choice	0–11	2.95	3.47	4.45	0.59	1.22

Research Question 2: To what extent was the use of leading questions instrumental or clinically informed (e.g. ‘gentle assumptions’ or ‘symptom amplification’)?

The most frequent types of ‘leading’ questions were ‘suggestive’ where the interviewer either introduced new information into the narrative or asked the question in a way that indicated a desired response, with a mean 89.57% (SD = 15.41) of all leading questions falling into this category. Around 10% of all leading questions may be explained by the use of clinically informed interviewing techniques. Overall, an average of 8.39% (SD = 13.01) of leading questions could be considered an example of a ‘gentle assumption’. One such example of a gentle assumption is: ‘You were sort of drawn to her because she was attractive or you felt aroused?’ This question introduced new information into the narrative so would be considered leading; however, the content was centred around sensitive information that may have been difficult for the young person to disclose so was coded as a ‘gentle assumption’. Another example, which appeared to be a more conscious employment of a gentle assumption, is: ‘It’s normal for young people to masturbate and look at pornography. I’m just wondering what sort of pornography you might be looking at.’

There were very few examples of ‘symptom amplification’ questions, with only three questions across all interviews, which may have been considered an intentional exaggeration to facilitate disclosure. One example of this was when the participant was exploring how often the young person had previously seen the victim and asked ‘When you say a couple of times, like 4 or 5?’.

Research Question 3: What types of questions do forensic mental health practitioners use to elicit ‘evaluative’ information about the offence?

Across all interviews, participants asked a mean of 21.91 ‘evaluative’ questions (SD = 12.07), which accounted for a mean of 40.11% of all questions asked (SD = 14.72). When considering the question type, ‘yes/no’ questions were the most prevalent, with an average of 22.55% (SD = 9.00) of all questions asked being ‘yes/no’ questions with evaluative content. This was closely followed by cued-recall evaluative questions, which accounted for an average of 16.00% (SD = 9.50) of all questions. Forced-choice evaluative questions were rarely used and only accounted for a mean percentage of only 1.56% (SD = 2.08) of all questions.

When exploring the type of evaluative content that participants asked about, there was an average of 6.00 (SD = 4.62) emotion-focused questions (range = 0–15) and a mean of 0.91 (SD = 1.06) emotion-focused questions that were leading (range = 0–4). There was an average of 15.91 (SD = 8.88) thought-focused questions (range = 2–45) and a mean of 1.77 (SD = 2.37) leading questions that were thought-focused (range = 0−9). Table 3 provides the breakdown of what specific question types were used to elicit information on emotion and thoughts at the time of the offence. As with the overall pattern of question types, the most frequently used question required a ‘yes/no’ response, followed by cued recall (Table 3).

Table 3.

Question types used to elicit evaluative information about the offence.

Question type	Range	M	SD	Percentage of the meanby focus area
Emotion-focused
Yes/no	0–11	3.95	3.24	67.64
Cued recall	0–6	1.86	1.91	29.78
Forced choice	0–3	0.18	0.50	2.58
Thought-focused
Yes/no	1–28	8.68	6.50	51.76
Cued recall	1–13	6.36	3.79	43.83
Forced choice	0–5	0.86	1.36	4.41

Validity check

Most participants indicated that the actor in the mock interview was either representative (n = 11) or somewhat representative (n = 8) of young people that they have typically interviewed for this type of offence. Notwithstanding this, just under half of all participants (n = 10) commented that the actor was more articulate or forthcoming than an average 15-year-old that they have interviewed. For the two participants that indicated the actor was not representative, the primary reason was the lack of complexity in the scenario. These participants highlighted that their roles usually involved interviewing young people for more serious offence types, with compounding issues such as mental health disorders or neurodevelopmental issues.

Similarly, almost all participants indicated that their interview approach was either representative (n = 13) or somewhat representative (n = 8) of their usual approach or style of interview. Many participants expanded on this question to outline that what they found difficult about the mock interview scenario fell into three broad themes. The first theme related to the difficulty of undertaking an interview with limited collateral information and time for planning and preparation. Seven participants raised that this likely impacted how they approached the interview.

For the second theme, when participants discussed their usual interview style, their reflections pertained to the limited opportunity to develop rapport with the young person and explore their background history, prior to exploring the offence. This was raised as a limitation of the mock interview design by eight participants. For the third theme, seven participants highlighted the challenge of focusing exclusively on the offence, when they wanted to explore other topics in conjunction with the offence information. Examples of other topics they wanted to explore included other instances of offending and sexual development. Finally, three participants who undertook the mock interview through Microsoft Teams raised the challenges of this modality and how that may impact their usual interviewing style. Notwithstanding this, one participant highlighted how COVID-19 had resulted in increased telehealth assessments and felt that completing the assessment on Microsoft Teams had replicated their practice experience in the current environment.

Discussion

This was the first study to explore how forensic mental health practitioners question a young person in relation to a sexual offence in a mock clinical forensic interview. The results revealed that the practitioners predominantly utilised specific ‘yes/no’ questions and relatively few ‘open’ questions. Concerningly, around 13% of questions could be classified as ‘leading’ and were not otherwise accounted for by the use of typical clinically informed strategies. The findings suggest that practitioners may not be currently employing best practice strategies to elicit the most accurate and specific information from an interviewee, as it relates to their offending behaviour.

Overall, the findings are consistent with other research on multi-disciplinary professionals needing specific training in evidence-based interview techniques (Brubacher et al., 2014; Powell, 2005; Sternberg et al., 2001) and highlights the outstanding practice development needs for forensic mental health practitioners undertaking clinical forensic assessment interviews with young people.

Notwithstanding the concerning prevalence of ‘leading’ questions, the vast majority of questions were ‘yes/no’ questions – questions that are less likely to result in inaccurate recall than other leading specific questions such as ‘forced-choice’ questions (Gee et al., 1999). As the study was based on a mock interview, there was no ‘ground truth’ to explore whether the ‘yes/no’ questions risked misinformation effects, which occur when leading questions impact the memory of the witness so inaccurate information is incorporated into later recall (Sharman & Powell, 2012). Positively, there was only one instance of a leading ‘open’ question, which has the highest risk of eliciting a false belief. The use of ‘leading’ questions does not appear to be explained by intentional clinical strategies such as ‘gentle assumptions’ or ‘symptom amplification’, as these leading question types accounted for only 10% of all leading questions. This highlights the need for forensic mental health practitioners to undergo specific training around the implications of using ‘leading’ questions in standard interview practice with justice-involved young people. Continued use of ‘leading’ questions risks practitioners eliciting errors, which will lead to inaccurate estimations of risk for young offenders who are already facing a multitude of psychosocial disadvantage and stigmatisation. Further, misinformation effects may lead young people to incorporate the interviewers’ assumptions into their own recall of events and skew their own internal narratives of key life events.

In the study, the subjective experience of the young person was frequently explored by practitioners, who spent around one third of the interview asking questions with ‘evaluative’ content. While this is much higher than the 6% of ‘evaluative’ questions asked by attorneys in the Lyon et al. (2012) study, this is likely to be explained by the differing context whereby an attorney’s aim is to establish facts, while the participant’s aim was to understand the various factors contributing to the offence. Interestingly, participants focused more on cognitions than emotions or physiological experiences, with an average of 15.91 ‘thought-focused’ questions and 6.00 ‘emotion-focused’ questions across all interviews. The majority of questions were still ‘yes/no’ questions, with around 30% of ‘emotion-focused’ questions and 44% of ‘thought-focused’ questions asked as ‘cued-recall’ evaluative questions. The Lyon and colleagues study suggested that these questions were most likely to elicit detailed ‘evaluative’ responses. It is possible that forensic mental health practitioners may be more adept than attorneys at eliciting ‘evaluative’ information; however, the reliance on ‘yes/no’ questions suggests that there may be additional training needs in relation to maximising the recall of an individual’s cognitive, affective, and physiological processes at the time of an offence. This is critical to understanding the confluence of predisposing, precipitating, perpetuating, and protective factors that contribute to an individual engaging in a particular illegal behaviour, at a particular point in time and a particular place. Further research is needed to establish the best question types to elicit the most accurate ‘evaluative’ content from young people who have committed an offence.

Implications for practice

The findings of this study suggest that there were important gaps between the current practice and recommended interviewing approaches for forensic mental health practitioners undertaking clinical forensic interviews that involve eliciting an offence narrative (e.g. risk assessment interviews). While there is a lack of empirical research to guide practitioners towards what ‘hybrid’ clinical and forensic skills are involved in the combined interview approach, there is widespread agreement that interviewers should prioritise ‘open’ questions.

Importantly, the participants in this study came from a diverse range of professions and jurisdictions, and their real-world practice experience ranged from 2 to 34 years, yet there was a consistently low prevalence of ‘open’ questions across all interviews. This suggests that all practitioners would benefit from specific evidence-based skills training to improve their questioning technique, and that the underutilisation of best practice techniques was not explained by the differences in training or professional affiliations or orientations. Research has demonstrated that for sustained improvements in questioning techniques, practitioners should undertake training that includes opportunities to practise, such as through coding activities and interview skills practice (Brubacher et al., 2022). Overall, practitioners should actively look for opportunities to reflect or obtain feedback on their question types in order to increase their use of open questions and reduce their reliance on specific questions. This may be possible through supervision, watching, or listening to interview recordings, engaging in mock interviews with colleagues, or enrolling in evidence-based training courses (Benson & Powell, 2015).

It was beyond the scope of the study to determine whether clinically informed strategies such as ‘gentle assumptions’ and ‘symptom amplification’ are appropriate to use in the context of a clinical forensic interview. However, it was clear that very few participants utilised these strategies (whether consciously or not) as they only accounted for approximately 10% of all ‘leading’ questions. Rather, the prevalence of ‘leading’ questions highlights the gap between current versus best practice approaches for practitioners who have not undergone adequate training to mindfully avoid these problematic question types. Positively, experts have developed evidence-based training that may support practitioners to both improve their use of ‘open’ questions and reduce their use of ‘leading’ questions (Price & Roberts, 2011; Lamb et al., 2002; Powell, 2008). The current study highlights that practitioners would benefit greatly from completing professional development and training in the area of investigative interviewing, to improve their questioning techniques.

Limitations and future research

This study explored what ‘hybrid’ clinical forensic interviewing skills are spontaneously employed by forensic mental health practitioners in a mock offence interview. The nature of the study design gave rise to potential limitations. While a mock interview allows for a standardised approach and comparison between participants, it is also artificial in nature. Positively, the vast majority of the participants believed that the interview was, at least somewhat, representative of their usual style. Ecological validity could be improved with more opportunity for preparation, rapport building, and broader exploration of issues that were commonly mentioned as constraints on their usual style. Ideally, future research would endeavour to review field interviews by practitioners to determine whether the results are generalisable. Further, given the participants’ interest in asking questions beyond the offence to help formulate the event, it may be worthwhile to explore questioning style when exploring a young person’s history. In particular, it is possible that eliciting information on sexual development may give rise to the use of ‘gentle assumptions’ and ‘symptom amplification’. It could be informative to explore the approaches used by clinicians to investigate this usually taboo topic and how this impacts their questioning style.

The sample size in the study prevented exploration of some variables of interest, including whether video-conferencing impacted interviewer questioning style. When the study was designed, the use of video conferencing was to aid the inclusion of participants from multiple services and jurisdictions; however, with the advent of COVID-19 it provided increased ecological validity for assessment interviews conducted during the pandemic. In future studies with a larger sample size it may be helpful to compare results between in-person and video-conference interviews to ascertain whether the different modalities impact the number or type of questions asked.

Conclusion

The findings of this study have demonstrated that forensic mental health practitioners use few ‘open’ questions and some ‘leading’ questions when eliciting information about an offence in a mock interview with a young person. This is contrary to best practice recommendations in both the clinical and forensic interviewing literature, which encourage the use of ‘open’ questions and restrict ‘leading’ questions. There were few examples of clinically informed ‘leading’ questions (such as ‘gentle assumptions’ or ‘symptom amplification’), so the high prevalence of ‘leading’ questions cannot be explained by differences in training between clinical and forensic practice approaches. This highlights the need for training in effective interviewing techniques for forensic mental health practitioners. While participants frequently explored ‘evaluative’ information about thoughts, emotions, and physiological responses during the interview, their over-reliance on ‘yes/no’ questions indicated that training in effective interviewing techniques to elicit more individualised case information may be beneficial. In the absence of an evidence-based clinical forensic interviewing protocol or training, it appears that forensic mental health practitioners from all disciplines would benefit from undertaking evidence-based training to enhance their interviewing skills.

Ethical standards

Declaration of conflicts of interest

Chelsea Leach has declared no conflicts of interest.

Francesca Brown has declared no conflicts of interest.

Luke Pryor has declared no conflicts of interest.

Martine Powell has declared no conflicts of interest and would like to note she is the Director of a Centre that is a research and training hub in investigative interviewing.

Scott Harden has declared no conflicts of interest.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Children’s Health Queensland Human Research Ethics Committee (Reference Number: LNR/2019/QCHQ/59087) and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Informed consent

Informed consent was obtained from all individual participants included in the study