Introduction

The accuracy and completeness of witnesses’ reports are crucial when witnesses are alleged victims of child sexual abuse because they are often the only possible sources of information about the alleged incidents. For this reason, many researchers and expert professional groups have explored how investigative interviewers can obtain more abundant and accurate information from young alleged victims. For example, allowing children to say ‘I don’t know’ and ‘I don’t remember’ or to correct an interviewer if required increases the accuracy of children’s accounts (Lyon & Saywitz, 1999). Establishing rapport with alleged child victims at the beginning of the interview helps create a relaxed and supportive environment, and can thus alleviate anxiety or discomfort which could negatively affect the informativeness and the accuracy of information provided (Roberts, Lamb, & Sternberg, 2004; Sternberg et al., 1997). Moreover, episodic memory training, designed to familiarize children with the task of describing past experiences in response to open-ended questions during the pre-substantive part of the interview increases the accuracy and informativeness of child victims’ accounts (Roberts et al., 2004; Sternberg et al., 1997). Since it makes children aware how much detail is expected of them when allegations are later discussed, such practice can enhance the amount of event-specific information provided (Lamb, Hershkowitz, Orbach, & Esplin, 2008; Roberts, Brubacher, Powell, & Price, 2011).

The types of questions asked by interviewers during investigative interviews play a significant role in enhancing the likelihood that accurate and informative accounts will be elicited from child victims. Suggestive questions in which interviewers imply the expected response can decrease the accuracy of children’s accounts or even create false memories (Cassel, Roebers, & Bjorklund, 1996; Lamb & Fauchier, 2001). Option-posing prompts, defined as questions only requiring yes, no, or one-word answers, also impair the accuracy of children’s accounts because children may guess or thoughtlessly choose one of the options provided (Lamb & Fauchier, 2001). By contrast, open-ended prompts yield responses that are longer and more detailed than responses to directive, option-posing or suggestive questions (Davies, Westcott, & Horan, 2000; Dent & Stephenson, 1979; Lamb, Orbach, Sternberg, Esplin, & Hershkowitz 2002; Orbach et al., 2000; Sternberg et al., 1997). Responses to open-ended questions are also considered more reliable because they are not contaminated by information introduced by interviewers (Milne & Bull, 2002). Therefore, guidelines based on the professional consensus regarding high quality interviews have especially put emphasis on the importance of narrative responses elicited using open-ended prompts (e.g., Anderson et al., 2010; APSAC, 1990; Fisher & Geiselman, 1992; Home Office, 1992, 2007; Lamb, Orbach, Hershkowitz, Esplin, & Horowitz, 2007; Saywitz & Camparo, 2014; Saywitz, Lyon, & Goodman 2011; Yuille, 2002).

However, previous studies have shown that police officers in many countries underutilize the recommended question types in their investigative interviews with child witnesses. Investigative interviewers rarely use open-ended questions and rely excessively on such focussed questions as option-posing or suggestive utterances (Cederborg, Orbach, Sternberg, & Lamb, 2000; Hershkowitz, Lamb, Sternberg, & Esplin, 1997; Lamb, Sternberg et al., 1996; Orbach et al., 2000; Sternberg, Lamb, Davies, & Westcott 2001; Sternberg et al., 1996; Thoresen, Lonnum, Melinder, & Magnussen 2009).

Because investigative interviewers appeared to have difficulty following recommended guidelines even when they believed they were adhering to them, researchers at the National Institute of Child Health and Human Development (NICHD) developed structured interview guidelines designed to translate professional recommendations into operational procedures and concrete guidance (Orbach et al., 2000). The NICHD Protocol (Lamb et al., 2007) covers all phases of the investigative interview, including the ‘introductory’ (e.g., introducing him/herself, clarifying the ground rules etc.), ‘rapport-building’ (e.g., establishing a relaxed and supportive environment), ‘episodic memory training’ (e.g., giving children practice describing past experiences in response to open-ended questions)’, ‘substantive’ (e.g., investigating possibly abusive incidents using open-ended prompts as much as possible), ‘break’ (e.g., allowing the child a break while the interviewer reviews missing information and plans the rest of the interview) and ‘closure’ (e.g., allowing the child to volunteer additional information before thanking him/her and shifting the discussion to a neutral topic) phases of the interview. It has been convincingly shown in different countries that interviewers elicit more informative accounts from alleged victims when they follow the NICHD Protocol (Cyr & Lamb, 2009; Lamb et al., 2009; Orbach et al., 2000; Sternberg, Lamb, Orbach, Esplin, & Mitchell 2001).

In South Korea, the NICHD Protocol (Lamb et al., 2008; Orbach et al., 2000) was recently introduced to the police officers at ‘One-stop Support Centers’ where all alleged child sex abuse victims have been interviewed, examined, assessed and treated since 2010 (Ministry of Gender Equality & Family, 2010). Previous guidelines for interviewing sexually abused child victims (Ministry of Justice, 2003; Korean National Police Agency, 2008) put more emphasis on video-recording procedures to ensure its legal admissibility but paid less attention to practical interview strategies designed to elicit accurate and abundant information from child victims. Therefore, police officers failed to follow the relevant practice guidelines (Ministry of Gender Equality and Family, 2010) and there was a clear need for operational guidance. This need was met by introduction of the NICHD Protocol. However, although the police officers have been encouraged to follow the NICHD Protocol since 2010, police officers’ understanding of recommended practices, their perceptions of how well they adhere to the recommended techniques, and their actual adherence to the Protocol have not previously been assessed. Accordingly, there is an urgent need to examine Korean police officers’ understanding and use of the Protocol.

Two studies have shown that investigative interviewers often perceive their interviews with witnesses and alleged victims positively. Dando, Wilcock and Milne (2008) surveyed 221 inexperienced police officers’ perceptions of their use of the ‘Cognitive Interview’ (CI) components of the PEACE interview protocol with adult witnesses. The police officers reported that they used some of the Cognitive Interview retrievals more frequently and more effectively than others and that they frequently established rapport, let the interviewees provide uninterrupted accounts, and explained the interview process to them. While Dando and her colleagues (2008)’ study focused on inexperienced interviewers’ use of the CI with adults witnesses, La Rooy, Lamb and Memon (2011) surveyed 91 Scottish police officers about their interview practices with suspected child victims. Interviewers believed that they were able to obtain full and complete accounts of the events being investigated. Most reported that they established rapport, explained the ground rules, and explored whether the children understood the difference between truth and lies. Most of them indicated that they always or almost always used specific wh- questions and fewer than half said that they always or almost always used open prompts. Interestingly, nearly all reported that they never or rarely asked leading or suggestive questions. Of course, the previously cited studies in which real interviews were examined suggested that police officers rarely perform in this way, and the discrepancy underlines the need for research directly comparing police officers’ perceptions of their practice with actual performance.

Accordingly, in this study, we surveyed police interviewers using a questionnaire soliciting self-evaluations of their interviews as well as assessments of their interviewing effectiveness and of how well they adhered to the NICHD Protocol. Transcripts of actual interviews were also analysed not only to describe how closely the officers adhered to the guidelines but also so that we could compare the self-reported and actual performances of these officers. Unfortunately, we could not make direct comparisons between self-reports and actual performance since the questionnaires were completed anonymously but we were able to compare mean levels of performance on the two types of measures. This study therefore addressed two questions: (1) How do Korean police officers perceive recommended practices and their interview behaviors? and (2) How well do they adhere to the recommended techniques?

Methods

Sample

Thirty-two out of 76 police officers working at One-stop Support Centers agreed to complete questionnaires asking about their investigative interview practices. All respondents were women who averaged 36.6 years of age (SD =7.73, range =26 to 50 years) and had experience interviewing alleged child victims of sexual abuse. The mean length of service was 25.10 months (SD =35.17, range =1 to 120 months) and the mean number of interviews they had conducted previously was 30.81 (SD =33.66, range =1 to 100). The police officers had been given the Protocol to use during their interviews but without explicit training.

Participants were also asked to submit verbatim transcripts of interviews they had conducted with alleged victims of sexual abuse who were under 12 years of age. Fifty-one transcripts were submitted by the thirty-two police officers, with 19 police officers (59.4%) submitting two of their interview transcripts while 13 of them (40.6%) submitted one. However, six were excluded (three were only partial transcripts; two were interviews of adults and one involved an adult with learning difficulties), resulting in a usable sample of forty-five transcripts, involving forty-one girls (73.1%) and four boys (8.9%). Four of the children (8.9%) were 3 to 5 years old, 19 (42.2%) were 6 to 9 years old and 22 (48.9%) were 10 to 13 years old. Three (6.7%) of the alleged perpetrators were immediate family members, two (4.4%) were other family members, 18 (40.4%) were non-relatives familiar to the alleged victims, and 22 (48.9%) were unfamiliar to them. Three (6.7%) of the alleged incidents involved exposure, 10 (22.2%) fondling over clothes, 26 (57.8%) fondling under clothes, and 6 (13.3%) penetration. Single incidents were reported by 33 alleged victims while 12 (26.7%) reported they had been abused more than once.

The questionnaires were distributed and collected between February and March 2011. Most interviews were conducted contemporaneously but some interviews were conducted, at the longest, approximately 10 months earlier than the questionnaires were completed. Thirty (66.7%) of the submitted interviews were conducted from January to March 2011 while 3 (6.7%) interviews were conducted from October to December 2010, 8 (17.8%) were from June to September 2010 and 4 (8.9%) dates were unknown. Because the questionnaires were completed anonymously, it was not possible to directly compare the respondents’ reported and actual interviewing behaviors.

Procedure

The questionnaire was modelled after those that have been used in previous studies asking police officers about their interview practices (Dando et al., 2008; La Rooy et al., 2011). The questionnaire had four sections: (1) General information: age, gender, working experiences and length of service; (2) Investigative interview experience: number of interviews previously conducted; perceptions of interviewing skills using a 5-point Likert scale (e.g. ‘I fully understand the guidelines for interviewing alleged victims’, ‘I can effectively interview alleged child victims with intellectual and communicative difficulties.’); (3) interview practice: use of recommended interview practices using a 5-point Likert scale (e.g. ‘I introduce myself before I start investigative interviews with alleged child victims’, ‘I make clear to children that they need to describe events in detail’, ‘I make clear to children that they are obliged to tell the truth’); multiple choice questions probing the types of questions asked during the substantive portions of interviews and the types of questions they found most useful; (4) Comment: description of difficulties encountered when interviewing children and training courses they would like to attend.

Coding the Transcripts

Two Korean native speakers independently reviewed transcripts of the introductory, substantive, and closing portion of the submitted interviews. In the introductory part of the interviews, they noted whether or not the investigative interviewers explained the ground rules, established rapport, and conducted episodic memory training. While analysing the closure phase of the interview, they also noted whether or not the interviewers thanked the children for participating and gave them opportunities to ask questions. The coders then classified interviewer utterances during the substantive portion of the interview into five categories: invitations, summaries/facilitators, directive utterances, option-posing utterances and suggestive utterances using the coding system and definitions introduced by Lamb and his colleagues (Lamb et al., 2007, 2009). Invitations are input-free utterances, including questions, statements or imperatives prompting free-recall responses from the children. Summaries accurately restate what the child has just said in the form of a statement, without any explicit request for information or responses. Facilitators provide non-suggestive encouragement to continue with on-going responses to the previous utterances (e.g. ‘Ok’, ‘Uhum’ or the repetition of the last few words spoken by the child). Directive utterances refocus the child’s attention on details or aspects of the alleged incident that the child has already mentioned, thus using recall memory processes (e.g., “What colour was his shirt?”, “when was it?”). Option posing utterances involve yes/no questions or explicit options from which the child must choose but do not imply that a particular response is expected (e.g., “Was that over or under your clothes?”, or “ Is it always at the weekend?”). Suggestive utterances are stated in such a way that the interviewer strongly communicates what response is expected or they assume details that have not been revealed by the child.

The numbers of substantive details provided by the children were then tabulated. Following Lamb, Hershkowitz et al. (1996), details involved any information relating to the investigated incidents, including naming, identifying or describing individual(s), object(s), events(s), and action(s). They also included expressions of emotion(s), thought(s) and sensation(s) related to the alleged incidents but not references to the children’s mental or emotional state at the time of the interview. Details were only counted when they supplied new information. A negative response to a yes/no question was counted as a detail, but claims of lack of knowledge (e.g. “I don’t know” or “I don’t remember”) were not counted as details.

Inter-Rater Reliability

Ratings of the utterance types used by interviewers and of details elicited by children were conducted by the first author and a graduate student who had been educated with the NICHD Protocol and analysing types of questions and children’s responses more than 30 hours. During the training, they reviewed the coding manuals and practiced with sets of transcripts which were not involved in this study until they agreed with each other more than 80% of the time. When 20% of the transcripts were coded by two raters independently to assess reliability, inter-rater reliability in the identification and classification of interviewer utterances was high, Kappa =0.853 (p < .001). Intra-class correlation coefficient (ICC) assessing agreement between the two raters regarding details provided by the children was 0.97 (p < .001).

Results

Explaining the Interview’s Purpose and Ground Rules

Most police officers reported that they frequently explained the role of the interviewer and ground rules while the transcripts revealed that few of them actually did (see Table 1). Specifically, most officers indicated that they almost always introduced themselves, anyone else in the interview room and explained their role as interviewers. However, in practice, 60% of the officers actually introduced themselves or anyone else and only a few explained their role as interviewers. Similarly, most officers said they almost always or always explained the ground rules but in practice, very few explained the ground rules in their interviews.

Table 1 Self-reported and actual behaviors in the introductory phase
Full size table

Rapport-Building and Episodic Memory Training

Almost all (90.9%) of the police officers said they built rapport, reportedly spending an average of 13.78 minutes (SD =9.77, range 3 to 30 minutes) doing so. However, the transcripts showed that only 31.1 % (n =14) made any effort to build rapport. Officers on average asked 4.38 questions (SD =1.85, range 1 to 7) to establish rapport, predominantly using directive (M =1.85, SD =1.41) and option-posing (M =2.00, SD =1.93) questions, while using invitations (M =0.71, SD =1.68), facilitators (M =0.14, SD =0.36), and suggestive questions (M =0.14, SD =0.36) much less often. Rapport-building topics were: ‘today’ (n =5), ‘things children like to do’ (n =4), ‘school’ (n =3), ‘a recent event’ (n =3), ‘Family’ (n =2), and ‘Food’ (n =2). No interviews included any episodic memory training.

The Substantive Phase of the Interview

On the questionnaires, respondents indicated that they used invitations and directive utterances most and suggestive utterances least during the substantive part of the interview. In addition, 90.5% said that invitations were the most useful types of prompt and none identified option-posing or suggestive utterances as useful utterances types. Analysis of the transcripts (see Table 2), however, showed that they actually used option-posing and directive utterances most and almost never used invitations.

Table 2 Use of difference types of utterances in the substantive portion of the interviews
Full size table

The number of details was also analysed. Overall, interviewers elicited an average of 261.3 details (SD =181.8) from alleged child victims using directive utterances, followed by invitations, option-posing questions, suggestive prompts and summaries/facilitators. The average invitation elicited 17.4 details (SD =28.4) while directive utterances tended to elicit considerably less information (M =3.2, SD =2.2). Option-posing questions, which were the most commonly used prompts, only elicited 1.5 (SD =1.0) details per question on average. A one-way within subjects ANOVA assessing differences in the quantity of details elicited revealed a significant effect of question type, Wilks’ Lambda = .480, F(4, 14) =3.791, p =.027 (see Table 3).

Table 3 The number of details elicited using different utterance types
Full size table

Breaks and Closure

Although participants were not asked how they closed their interviews, an examination of the transcripts revealed that 62.2% included closing comments which let the children know the interview was ending, with 48.9% of the interviewers asking children whether they had additional things they wanted to say. Few officers thanked the children for their efforts (n =8, 17.8%) and even fewer talked about a neutral topic before ending the interview (n =3, 6.7%). Few (13.3%) of the interviews included a break.

Perceived Effectiveness of Interview Practices

As shown in Table 4, approximately half the respondents agreed that they understood and followed the guidelines, and were aware of the factors that may impair children’s performance. However, fewer than a third of the officers believed that they could distinguish between truthful and false allegations, interview alleged victims effectively, effectively interview alleged child victims with intellectual and communicative difficulties, effectively interview alleged victims under six years of age and/or effectively interview alleged victims who were reluctant to disclose abuse.

Table 4 Perceived skill as interviewers
Full size table

Difficulties

Forty-four comments were made about difficulties the interviewers had experienced when interviewing children. Seven (15.9%) officers indicated that they had difficulties interviewing children who were reluctant to disclose or were very young. Five (11.4%) reported difficulties interviewing children with intellectual and communicative difficulties. Four reported using improper types of questions excessively (n =4, 9.1%), being interrupted by parents and guardians while questioning children (n =4, 9.1%) and that children were unable to express themselves and to concentrate (n =3, 6.8%). Other officers reported difficulties assessing children’s credibility (n =3, 6.8%), interviewing children with whom they could not establish rapport or were nervous (n =2, 4.6%), self- generated pressure (n =2, 4.6%), interruptive investigative environments (n =2, 4.6%), or a lack of knowledge about children’s development (n =2, 4.6%).

Training Course

Police officers reported that they would benefit from training that included practice with actual cases (n =10, 27%), information about child development (n =8, 21.6%), case-specific interview techniques (e.g., interviewing children who do not disclose, interviewing children with intellectual and communicative difficulties, interviewing very young children) (n =6, 16.2%), information about improving their abilities to communicate with children (n =6, 16.2%), training using a standardised interview protocol (n =5, 13.5%) and receiving feedback (n =2, 11.8%).

Discussion

This study was conducted to compare Korean police officers’ perceptions of their interview practices with their actual interviewing behaviors. Most respondents believed that they were adhering to the NICHD Protocol whereas examination of their interviews suggested otherwise. Also, as in similar studies in the U. K. (Dando et al., 2008; La Rooy et al., 2011), we found that Korean police officers perceived their interview practise more positively than their practices warranted. The majority of officers said that they almost always explained the purpose of the interview, spelled out ground rules, and made efforts to build rapport. However, only about 30% of them actually explained the purpose of the interview or established rapport and fewer than 20% explained ground rules before the substantive interview started. Moreover, the police officers excessively relied on option-posing or directive questions and asked very few invitations during the substantive part of the interview. Such findings are consistent with the results of previous studies, showing forensic interviewers in many countries seldom use open-ended questions (e.g., Cederborg et al., 2000; Lamb, Sternberg et al., 1996; Sternberg et al., 1996; Sternberg, Lamb, Orbach, et al., 2001), but was especially surprising in the present study because the Korean police officers reported that invitations were the most useful types of questions and that they often used invitations! The officers also reported that they knew the Protocol sufficiently well and adhered to it. Thus, although these interviewers understood the importance of adhering to the NICHD Protocol, they failed to adhere to it or to self-monitor their performance.

One possible explanation may be that the police officers do not understand the recommended guidelines. When they are asked about perceived skill as interviewers, only just below half (46.7%) agreed or strongly agreed that they fully understand the guidelines but 33% of them chose the neutral response option and 20% of them disagreed or strongly disagreed. In South Korea, the first NICHD Protocol training courses were given to all police officers at One-stop Support Centres in April 2011. Because the questionnaires were completed before the training was held, no police officers had been formally taught to use the NICHD protocol. Protocol training along with feedback sessions might give them the necessary opportunities to understand both the guidelines and their own interviewing practices.

Surprisingly, none of the participants provided interviewed children with episodic memory training although they believed they were fully following the NICHD Protocol. Research shows that when children are trained to respond to open-ended prompts in pre-substantive phases of the interview, they respond more informatively to open-ended prompts later in the interview (Sternberg et al., 1997; Roberts et al., 2004). It is possible that episodic memory training is such a new concept for Korean police officers that they do not understand its practical implications although the Protocol clearly illustrates how it should be conducted. Perhaps it is difficult to understand and follow this guidance without intensive and on-going training. It would be informative to determine whether providing training in the use of episodic memory training would improve the accuracy and informativness of the accounts South Korean police officers actually elicit.

Lastly, our results showed that invitations elicited approximately six to eleven times more information than other types of prompts, thereby replicating previous findings regarding the value of open-ended prompts (e.g. Lamb, Hershkowitz, et al., 1996; Sternberg, Lamb, Hershkowitz, et al., 1996; Sternberg et al., 2001; Cederborg et al., 2000). Especially, because no previous study focused on the forensic interviewing of children in East Asia countries, the study has important implications in showing the value of using open-ended prompts when interviewing children from different ethnic backgrounds.

Although most police officers completed the questionnaires around the same time that they conducted the submitted interviews, some had conducted the recorded interviews some time earlier. It is thus possible that the differences between these Korean police officers’ reported and observed interviewing behaviors were attributable in at least some cases to actual changes over time in their practices.

In conclusion, this study provided insight into Korean police officers’ actual and perceived practices when interviewing alleged child victims. The results clearly showed that Korean police officers were not adhering to the NICHD Protocol although they believed they were doing so. These results suggest that investigative interviewers need assistance improving their investigative practices. Since there had been no intensive NICHD training courses for the police officers at One-stop Support Centres in South Korea, they may have had difficulty understanding and following the Protocol. Previous studies have shown that it is possible to effect major improvement in the quality of forensic interviewing only when interviewers are trained to follow a very detailed and specific interview protocol and are given continuing supervision and feedback on simulated and actual forensic interviews (Lamb, Sternberg, Orbach, Esplin, & Mitchell, 2002; Orbach et al., 2000; Sternberg, Lamb, Orbach, et al., 2001). Therefore, further research is needed to determine whether training in use of the NICHD Protocol along with continued supervisions and feedback might improve the ability of Korean police officers to interview child victims and to evaluate their practices more accurately.