Evaluation of an interview technique for the detection of child sexual abuse

Evaluation of an interview technique for

the detection of child sexual abuse

A master thesis by

MSc in Forensic Science University of Amsterdam

Supervised by K. van Ham, PhD Examined by R.R. van Rijn, Prof. Dr. Performed at Academic Medical Center (AMC)

36 ECT January 22nd 2018 – July 27th 2018 Date of submission: 27-07-2018

2

Abstract

The detection of child sexual abuse (CSA) is essential for prevention of further harm to the child and providing the appropriate care. However, to determine whether the child has been exposed to CSA is challenging, because physical indicators are not always present in the victim and psychosocial symptoms appear non-specific for CSA. In many cases the verbal disclosure of the child may be a last resort, but a limited amount of methods is available to elicit and assess a verbal disclosure and all of them involve interviewing techniques. The wielding of an interview technique is precarious, because the quality of the information provided by the child in an interview setting is highly dependable on the conductance of the interview and the type of questions asked, which may lead to false allegations or reduced credibility of the child. The quality of a semi-structured interview technique of one of the available methods, the Sexual Knowledge Picture Instrument (SKPI) method is assessed in this study. The composition of the interviews is explored by scoring five types of utterances; invitational, facilitative, directive, leading (introducing new information) and suggestive (signaling a desired answer). Additionally, it is investigated whether the composition of interviews differs between a control group and a group with children with suspected exposure to CSA. The results showed more than half of the interview consists of directive questions and roughly 12% of the interview involves leading or suggestive utterances. No differences were found between the distributions in the two groups, implying that the knowledge of the interviewers does not affect their performance. This study thusly sets a baseline measurement and shows consistent performances, indicating an adequate quality, but also room for improvement. More and expanded evaluations of other protocoled interview techniques are recommended for the future to provide our society with a well-founded approach for the detection of CSA, which is urgently needed.

Key words: Child sexual abuse, forensic interview, SKPI, semi-structured protocol, suggestibility, verbal disclosure.

Introduction

Child sexual abuse (CSA) is a large problem in our society. A recent national report on sexual violence against children reveals that the Netherlands 32% of all children experience sexual violence throughout their life [1]. Many studies have shown that adverse childhood experiences can have far-reaching consequences, both on short term, as well as on long term [2,3,4], including physical health problems, such as hypertension, cardiovascular diseases and type 2 diabetes [5,6,7], and mental health problems [8,9]. Detection of these children who have been exposed to CSA is crucial to protect them for further harm and to prevent or limit the adverse effects of their exposure. This generally requires the involvement of professionals from both the legal field and the medical field.

3

However, while the distinction between sexually abused and non-abused children is essential, it is a complicated process. The signals of exposure to sexual violence have to be recognized as such, while these signals are often nonspecific [10]. The occurrence of CSA may be established through video recordings, witnesses or the confession of an offender, but often this kind of substantiation is not available. From the portion of court cases that are dismissed due to technical dismissals, a percentage of 82% is on the grounds of lack of evidence [1]. Another contributing factor is the dearth of medical evidence, such as genital injuries or sexually transmitted diseases which are found only in a minority of cases [11,12]. Also other laboratory evidence, such as semen, is rarely found even within 24 hours after the assault and is even more rare in children younger than 10 years old [13,14]. Evidently, physical evidence oftentimes does not provide a definite answer.

Due to the absence of physical indicators in a majority of the children exposed to sexual violence, determining that the child has been sexually abused is often dependent on non-physical indicators, which applies to both the legal and the medical context. Apart from various nonspecific psychosocial symptoms [10,15], CSA may be indicated by verbal disclosure as well. However, eliciting a verbal disclosure is a delicate process, because it is a laborious experience for children and various emotions can restrain the child from disclosing the abuse [16,17,18]. Consequently, the probability of evoking a verbal disclosure is affected by multiple factors and varies between 24% and 96% [19]. Research has been performed to uncover how interviewing skills can contribute to the amount and accuracy of information provided by the child. Subsequently, it has been described what could be an effective approach to obtain accurate information [20,21]. Incidents in the past have taught us that children are sensitive to various suggestive interviewing techniques, because the quality of the information provided by the child becomes affected [22,23]. This deterioration of quality may lead to false allegations, but may also lead to reduced credibility of true statements [23]. Contrarily, the instructions given to the child can positively affect their performance, as can the type of questions asked during the interview, for these factors have a major impact on the accuracy of information provided by the child [18]. Biases can be introduced by questions which contain presuppositions or by closed-end questions, because children tend to supply unelaborated answers in response to these type of questions [18]. Avoiding these questions, despite awareness, is difficult and the risk of unintentionally wielding suggestive techniques during an interview is undeniably present. This has caused forensic child interviewing to become a controversial topic.

Nevertheless, in many cases a verbal disclosure may be the only meaningful evidence available. Thus, it would not be desirable to disregard a verbal disclosure, because it can contain possibly crucial information. For this reason, correct interviewing is necessitated to do justice to a verbal disclosure and facilitate a reliable outcome. The question arises what should qualify as correct

4

interviewing and which standards should be used to measure this, to then determine how an interviewing technique can be developed which complies with the standards. In response to the hardship of refraining from suggestibility, multiple investigative interviewing techniques have been developed. A protocol that has been used worldwide, for forensic interviewing was developed by the National Institute of Child Health Care and Development (NICHD) [24]. This protocol specifies how the child should be instructed and directs the types of questions that should be asked. This approach indeed improves the quality of the interview, because the children are allowed to recall information freely, without suggestions of the interviewer [25]. Moreover, interviews performed according to a structured protocol contain more open-ended questions and a lower percentage of focused utterances, which is important because open-ended questions yield longer and more detailed responses from children [26,27]. Another available protocol is the CornerHouse Forensic Interview Protocol; RATAC, which is bolstered by scientific literature and legal support all throughout the USA [28]. Despite differences between these protocols, the correspondence illustrates that many guidelines are already generally accepted. The aspects on which the protocols agree include documentation, setting and developmental appropriateness [21,29,30]. A fundamental feature of both protocols, is the flexibility of the interview, which allows the interviewers to modify their approach to each individual child [29].

In the Netherlands, several methods have been developed to assist in the process of diagnosing CSA in health care and psychological care, which also (partly) adopted these guidelines [31,32]. In one of these methods, the conversation is facilitated by combining an interviewing technique with an instrument, called the Sexual Knowledge Picture Instrument (SKPI) [33,34]. Originally this method was based on research that sexual knowledge of the child and the emotional response to this knowledge might serve as alternative indicators for CSA [35] and thus, with this method a conversation between a young child and a health professional is entered to appraise these signals. The outcome of this interview will help with the assessment of the probability that the child has been exposed to CSA. Aware of the difficulties with regard to interviewing techniques, the SKPI method adopted the coherent guidelines from the aforementioned protocols to ensure the quality of the interview.

However, the implementation of some of the guidelines is impeded by multiple factors. Some aspects of the SKPI method diverge from previous protocols, because the SKPI method is ultimately intended to determine the wellbeing of the child and not intended for criminal investigations. Where other techniques directly pursue to elicit information about past events, the SKPI method aims to assess the sexual knowledge of the child. For this purpose, general questions are asked and only after remarkable statements the interviewer will inquire details about personal experiences of the child,

5

because this may offer a possible explanation for deviant knowledge. Also, the emotional response of the child is assessed, but this will not be mentioned further in this study. The second way in which the SKPI method differs is that the method utilizes a visual aid. Although the RATAC protocol incidentally uses visual aids as well, this picture instrument is different and used for another purpose. Nevertheless, the manner of interviewing has been adapted, because the pictures are often addressed during the interview.

Hence, the question is raised which requirements should apply to this method. Although this instrument is intended for the context of healthcare, the content of the interview might become forensically relevant due to legal interference and so there is a strong coherence between the two fields. Simultaneously, truth-finding also has medical importance, although more diagnostic in nature, due to its implication for appropriate care. Additionally, even a social purpose can be raised for providing a definite answer, as this knowledge may aid the emotional processing of the child, its family and other social surroundings. Thus, it can be argued that the SKPI method should be able to meet the forensic standards and that the interview should be conducted according to the applicable guidelines. Much alike the NICHD and RATAC protocol, the SKPI method takes into consideration the guidelines regarding flexibility, setting, documentation and developmental appropriateness. Furthermore, a proficient manner of questioning can be characterized by a proper amount of objectivity, which implies that the interviewer does not add information to the conversation, even though he may have acquired information from other sources already. Also, he maintains a neutral attitude towards the answers of the child, to prevent the child from adjusting its story. In practice this means that the types of utterances by the interviewer should be objective of nature and consistent over multiple interviews, regardless of the circumstances.

To answer the question whether the current practice of the SKPI method suffices regarding this manner of questioning, and in order to assess and optimize the interview technique, critical appraisal of the earlier performed interviews is required. This study will therefore examine the consistency of types of utterances by interviewers using the SKPI method. Utterances will be categorized which will provide a baseline distribution for the interview and allow comparison between interviews from different groups. By means of these comparisons, various influences on the distribution will be examined, such as age and sex of the child, prior knowledge of the interviewer and individual interview styles. In order to support the feasibility of this study, first a pilot study will be carried out. The aim of this pilot study is to verify that the designated classification system for utterances is convenient for the objective of this study. The variability within interviewers will be inspected, for it should be sufficiently small to minimize the probability of interference with

6

variability caused by groups. Furthermore, the performance of the classification system by multiple observers will be gauged by calculating the agreement percentage.

Methods

PICAS study

This study was executed within the Picture Instrument Child Abuse Screening (PICAS) study [36]. The PICAS study investigates the validity of the SKPI method as a diagnostic tool for CSA in a case-control design where the outcome of the interview with sexually abused and non-abused children is compared. For the PICAS study, the Interviews were conducted with children between the ages of three and eight years old. They were appointed to two different groups: a control group or the suspicion group. The suspicion group was recruited through the pediatric outpatient department from the Emma Children’s hospital at the Academic Medical Center in Amsterdam, while the control subjects were recruited via primary schools. Primary schools in the area of the Academic Medical Center were approached and asked to distribute information to the parents so the parents could subsequently contact the investigators. Parents of subjects with an existing suspicion of CSA who visited the outpatient clinic to seek medical care, were told about the research during their visit to the hospital. After this visit they received a telephone call and were informed about the study and asked whether they would like to participate. If the parents agreed to the participation of the child, they signed an informed consent and an appointment was made for the interviewer to visit, preferably at home or another location that was comfortable for the child.

The interviews were conducted according to the semi-structured protocol that was developed for the SKPI method [Appendix I]. The SKPI consists of 15 pictures depicting varying scenes, including everyday situations and anatomical drawings, as well as scenes depicting situations with intimacy between children and adults. During the interview, the child was asked to describe what it sees, if they recognize the situation from their own life, and to name various parts of the body and tell their function. Follow up questions were asked on occasion of remarkable statements by the child, at the discretion of the interviewer. The interviews were video-recorded with a small portable camera and analyzed afterwards.

In accordance with the PICAS protocol, the parents of the child were asked to fill in a survey. Based on this survey, the probability of having fallen victim to CSA was deemed negligible in the control group.

7 Current study

The PICAS study had included 17 subjects up until the recruitment of the current study that were suitable to include in the suspicion group of the current study. To keep the conditions of the control group and suspicion group uniform, the control group included 17 children with approximately the same age distribution. In total 34 interviews were included in this study and both groups were interviewed by multiple interviewers.

Five interviewers participated in this study. The interviewers that were not yet experienced in interviewing, received training prior to the conductance of interviews. The training session included a brief history of forensic child interviewing, the development of the instrument and instructions on how to use the SKPI and explanation on the protocol. Subsequently, the trainees had to perform and video-record several exercise interviews with children and received feedback on these interviews. When performance was sufficient, they received approval to start performing real interviews. The experienced interviewers also attended these training sessions, but did not perform additional exercise interviews.

From the selected 34 interviews, samples were taken by selecting video-fragments regarding five specific pages of the picture book, including one introduction-picture, two anatomical drawings and two intimacy-pictures [Appendix II], which were believed to be both representative and potentially most relevant for the outcome of the interview. These parts of the interview were transcribed and the utterances of the interviewer were extracted and subsequently categorized. This categorization of utterances was performed according to a classification system based on the classification by Lamb and colleagues, which is specifically appropriate for the substantive part of an interview [37]. The instructions for the observers following from this classification system were adapted and expanded and are enclosed in the appendix of this report [Appendix III]. Categorization was performed by one observer, but an overlap of 10% was performed by an independent second observer to monitor the agreement percentage.

Utterances could be scored in five different categories; invitational, facilitative, directive, leading and suggestive, which were respectively characterized by open-ended questions, encouraging remarks, closed questions regarding aspects already mentioned by the child, focused questions regarding a new topic raised by the interviewer and questions which signaled an expected answer to the child. After the utterances were categorized for every interview, a total number of utterances was calculated. Subsequently, a percentage per utterance type was computed per interview. The average percentage of utterance types between the suspicion group and the control group were examined and compared by means of a chi-square test. The distributions of utterance types were compared between interviewers and between the sex and various ages of the

8

interviewees. The mean percentages of utterance type were also calculated for the five distinct pictures and compared with a chi-square test. When differences were found, a post-hoc analysis was performed. Additionally, the mean of the absolute number of utterances per interview was examined and differences between groups were inspected by performing a t-test and an ANOVA or the non-parametric alternatives if necessary. The results are discussed in the results section.

Pilot study

A pilot study was performed by gathering data from six interviews, performed by the same interviewer. The mean age of the children was seven years, as well as the median with a range of two years and they were all appointed to the control group. The composition of the interviews did not seem to vary strongly, as the variability of utterance types were described with standard deviations ranging between 2,2% and 5,9%. Also the classification system was assessed on the inter-observer reliability. The utterances typing of two observers was compared and the average agreement percentage was calculated and amounted to 73,6%. The observer instructions were optimized and the study was continued.

Results

The interviews included in this study took place between January 2017 and May 2018. Four of the interviewers had a background in medical science, while one of them received an education in child psychology. All of the interviews were performed by a single interviewer, except for two interviews where two interviewers collaborated. In these cases, the second interviewer was responsible for 33% and 35% of the utterances during the interview. Of the 34 children that were interviewed, 23 were girls and 11 were boys. The mean age was 5.5 (± 1.8) years over all the groups, with a median of 6 years and a range of 3 years. The same mean was described for age in the control group, 5.5 (± 1.8) with also the same median and range of years, while the suspicion group had an average age of 5.4 (± 1.9) years, a median of 6 years and a range of 4 years.

The observers that categorized the utterances, had a background in forensic or medical science, but no previous experience with the classification system. Consensus between the observers on the instructions was reached during the preparation phase and the pilot study. The agreement between the observers (84.0%) was substantial, κ = 0.75. The interviews comprised in total 63.9 utterances on average. Also, the mean distribution of utterances during the interviews was observed. The values of the mean distribution are given in table 1 and visualized in fig. 1.

Although the distribution of utterances did not differ between the control and suspicion group (see fig. 2), the absolute amount of total utterances did (see fig. 3). The interviews in the

9

suspicion group contained significantly more prompts (72.3 ± 22.5) than the control group (55.5 ± 6.8) t(18.86) = 2.939, p = 0.008, while equal variances were not assumed. All of these value are presented in table 1. Then distributions were compared between the sexes and between age groups. No significant differences were detected between the distributions of interviews with boys and girls (G2(4) ≥ 1.791, p = 0.79). Age groups did not cause differences in the composition of the interviews as well (G2_{(20) ≥ 14.057, p = 0.83). These and supplementary results are displayed in table 2.} Comparison of absolute amount of utterances between the groups with boys and girls did not show any differences (U = 106, p = 0.45). The absolute amount of utterances in the various age groups, did also not differ (χ2(5) = 5.022, p = 0.41).

Figure 1. The average

distribution of

utterance types over all of the interviews.

The proportions of utterance types are shown as percentages. The directive prompts make up more than half of the interviews, followed by the facilitative utterances, the leading, invitational and suggestive utterances.

GROUPS

UTTERANCES

Absolute Invitational Facilitative Directive Leading Suggestive

Mean SD Mean

(%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD

All 63.91 18.45 5.61 3.51 28.96 10.47 53.16 9.86 9.62 7.23 2.65 2.68

Control 72.29 22.53 6.23 3.89 24.55 7.84 58.01 7.75 8.34 4.06 2.87 2.95

Suspicion 55.53 6.76 4.98 3.07 33.37 11.11 48.31 9.51 10.90 9.37 2.44 2.46

Table 1. Results over all the interviews and for the control and suspicion group. The mean of

the absolute amount of utterances during the interviews is given with the corresponding SD. Additionally, the proportions of utterance types in the interviews are presented. The distributions of utterance types in the suspicion group and the control group did not differ from each other (G2_{(4) ≥} 2.752 (p = 0.6)). 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Suggestive Leading Directive Facilitative Invitational

10

Figure 3. The absolute amounts of utterances in the interviews on average in all of the interviews and in the control and suspicion group. Error bars display the

SD. Comparison of the suspicion group with the control group showed a significant difference with a p-value smaller than 0.05.

Figure 2. The average distributions of utterance types in the control and suspicion group. The proportions of utterance types are

shown as percentages. No significant differences were found between the two groups.

GROUPS

UTTERANCES

Absolute Invitational Facilitative Directive Leading Suggestive

Mean SD Mean

(%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD

Girls 59.09 11.95 4.67 2.59 25.72 9.12 55.34 5.77 9.94 6.93 4.33 3.60 Boys 66.22 20.71 6.05 3.84 30.51 10.91 52.11 11.27 9.47 7.51 1.85 1.68 3 years 61.13 18.11 7.10 3.76 27.02 9.35 53.11 8.07 10.68 10.40 2.09 3.52 4 years 59.75 12.87 4.53 0.75 30.82 10.61 58.14 9.67 4.56 2.72 1.95 2.26 5 years 51.75 9.74 3.64 2.30 29.27 12.22 58.88 10.62 6.16 2.72 2.06 1.54 6 years 65.83 22.52 7.18 3.08 25.99 13.02 51.62 13.51 11.74 5.30 3.47 2.56 7 years 74.50 23.62 4.14 2.18 31.07 6.13 47.03 4.79 14.73 7.43 3.03 1.08 8 years 66.00 16.69 5.54 5.52 30.97 14.41 53.76 11.24 6.66 4.66 3.07 3.95

Table 2. Absolute amounts and proportional values for utterance types in age groups and in sexes. Both the means of the absolute amount of utterances and the distribution of utterance

types is displayed with corresponding SD. No differences were detected in the absolute amounts for sex (U = 106, p = 0.45) or age (χ2_{(5) = 5.022, p = 0.41). Also, no differences were detected} between the group with girls and the group with boys (G2_{(4) ≥ 1.791, p = 0.79) and any of the} age groups (G2_{(20) ≥ 14.057, p = 0.83).} 0 20 40 60 80 100 All 0 20 40 60 80 100 Control Suspicion p = 0.008* 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Control Suspicion Suggestive Leading Directive Facilitative Invitational

GROUPS

UTTERANCES

Absolute Invitational Facilitative Directive Leading Suggestive

Mean SD Mean

(%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD Mean (%) (%)SD

Interviewer A 57.46* 15.52 4.26 3.83 30.20 7.83 55.17 9.52 7.47 6.77 2.91 2.78

Interviewer B 76.38 24.99 4.09 2.08 30.53 11.21 44.67 7.50 16.65* 7.07 4.07 2.61

Interviewer C 54.86* 5.90 8.92 1.42 20.53 8.59 58.55 7.21 9.89 4.76 2.12 2.98

Interviewer D 83.33* 5.03 4.83 1.30 42.66* 13.13 49.73 10.51 1.60 0.68 1.17 1.19

Interviewer E 60.33 9.71 8.56 4.35 25.39 6.18 57.94 11.27 7.64 3.63 0.47 0.81

Table 3. Absolute amounts and proportional values for utterance types for individual interviewers. The distribution of utterance types is displayed per individual interviewer including

the corresponding SD. Significant differences were detected (G2_{(16) ≥ 34.232, p = 0.005) and are} indicated.

Figure 4. The average distributions of utterance types for individual interviewers.

Significant different values were detected in interviewers B and D for consecutively the leading and facilitative category.

Figure 5. The absolute amount of utterances on average by individual

interviewers during the interview.

Differences were detected between interviewer D and A and D and C with consecutive p-values of 0.03 and 0.04.

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% A B C D E Suggestive Leading Directive Facilitative Invitational * * 0 20 40 60 80 100 120 A B C D E p = 0.03* p = 0.04*

12

GROUPS

UTTERANCES

Absolute Invitational Facilitative Directive Leading Suggestive

Mean SD Mean (%) Mean (%) Mean (%) Mean (%) Mean (%)

Picture 2 16.76 6.23 4.74 35.96 42.63 12.28 4.39

Picture 4 8.62 4.59 7.17 29.01 54.61 5.46 3.75

Picture 7 19.50 10.19 2.11 25.19 58.82 12.37 1.51

Picture 11 8.09 3.29 11.64 29.45 51.64 6.55 0.73

Picture 13 10.94 5.96 5.91 32.26 48.39 11.56 1.88

Table 4. Absolute amounts and proportional values for utterance types for interview segments regarding separate pictures. The distribution of utterance types is presented for

every picture that was used during the interview Analysis with a Kruskal-Wallis test revealed differences in the absolute amount of utterances between pictures (χ2_{(4) = 66.613, p < 0.01). The} distribution of utterance types did not differ between the pictures.

Figure 7. Differences in absolute amounts of utterances for interview segments regarding separate pictures. Picture 2 and

picture 7 differ significantly (p < 0.01) in absolute amount of utterances from the other pictures.

Figure 6. Distributions of utterance type for the interview segments regarding separate pictures. No significant differences were detected

between the distributions for the separate pictures.

PICTURE 2 PICTURE 4 PICTURE 7 PICTURE 11

U-statistic p-value U-statistic p-value U-statistic p-value U-statistic p-value

PICTURE 4 U = 135.5 p < 0.01*

PICTURE 7 U = 509 p = 0.396 U = 146.5 p < 0.01*

PICTURE 11 U = 97 p < 0.01* U = 559 p = 0.815 U = 106.5 p < 0.01*

PICTURE 13 U = 260.5 p < 0.01* U = 440 p = 0.89 U = 247.5 P < 0.01* U = 413 p = 0.042

Table 5. Post-hoc testing for differences in absolute amounts of utterances between interview

segments regarding separate pictures. The location of the differences in absolute amount was determined

by performing Mann-Whitney U tests and the results are displayed in this table. Both the U-statistic and the p-value are given.

0 5 10 15 20 25 30 35

PIC. 2 PIC. 4 PIC. 7 PIC.11 PIC. 13

p < 0.01 p < 0.01 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

PIC. 2 PIC. 4 PIC. 7 PIC. 11PIC. 13

Suggestive Leading Directive Facilitative Invitational

The distributions of utterance types were calculated for individual interviewers (see fig. 4) and compared. Note that not all interviewers conducted interviews in both the suspicion and control group. Differences were detected (G2(16) ≥ 34.232, p = 0.005) with a p-value < 0.05 (αoriginal) and post-hoc analysis with a Bonferroni correction for the p-value, revealed a significant higher percentage of leading utterances in interviewer B (Z = 3.18, p = 0.0015) with a p-value < 0.002 (αcorrected). Interviewer D used a significant higher percentage of facilitative utterances (Z = 3.13, p = 0.0017), also with a p-value < 0.002. Absolute total amounts of utterances by the interviewers were analyzed and differences were detected (Welch’s F(4, 8) = 13.342, p = 0.001). Post-hoc analysis showed that interviewer D used significantly more utterances during interviews, compared to interviewer A (p = 0.03) and interviewer C (p = 0.04). These results are summarized in table 3 and visualized in fig. 5.

Furthermore, the distribution of utterance types did not differ between the five pictures that were used during the interview (G2(16) ≥ 22.113, p = 0.14). Absolute amounts of utterances varied between the pictures (χ2(4) = 66.613, p < 0.01). Table 4 gives the exact values that were observed for the five pictures and fig. 6 visualizes the separate distributions that do not differ significantly from each other. Significantly more utterances were used to discuss picture 2 and picture 7 (p < 0.01), compared to the conversations concerning the other three pictures. The post-hoc testing results that revealed the differences between the amount of utterances are presented in table 5 and displayed in fig. 7. Interpretation of the results will be discussed in the next section.

Discussion

This study pursued to provide a baseline measurement of utterance types during the interview of the SKPI method, which was yielded. On average, the interview samples contained roughly 64 utterances and were composed for the most part of directive utterances, focused questions that directed the attention of the child to something he mentioned before or to the picture being discussed with the child. Roughly a quarter of the interview the interviewer was encouraging the child with facilitative utterances. Invitational prompts were only occasionally used during the interviews. Leading and suggestive utterances covered a considerable part of the interview, which was more than 10% in total. Completely eliminating suggestive and leading utterances during an interview may not be realistic, as it is recognized in literature that the ‘perfect’ interview does not exist [30]. However, the amount of prompts in these categories that is considered acceptable should be discussed. The numbers of leading and suggestive utterances do not seem to differ from the study performed by Lamb et al., where also large proportions were measured for suggestive and here named option-posing utterances [37]. The proportions of option-option-posing and suggestive utterances amounted to 8.13. and 3.93, consecutively. This was measured after the interviewers received training.

14

The amount of directive utterances by the interviewer in this study is considerable, as they made up half of the interview. However, this was an expected result, as the method is characterized by the use of an visual aid, the SKPI, and interviewers will refer to the description of the picture the child provided him or her with. Although the article of Lyon advocates the use of open-ended questions, carefully formulated focused questions can also elicit accurate information in children and directive prompts may be necessary to inquire the sexual knowledge of the child [18]. Nonetheless, the personal experiences of the child should be inquired using invitational utterances. Ideally, the parts of the interview that inquire the knowledge and the personal experience of the child would be disaggregated to measure the performance separately. However, those two aspects are strongly cohesive and can therefore not be observed independently.

Facilitative utterances were also vastly represented and scored as an independent category, contrarily to the study of Lamb and colleagues where these utterances were considered a reinforcement of the preceding utterance [36]. Previous research has described that, although a large risk for shaping children’s responses lies in the type of questions asked, this is not the only way to do so. Wood & Garven described an overview of multiple forms of how improper interviewing can affect the answers provided by children [38]. The child may be persuaded due to social pressure when the interviewer tells the child what he or other people believe to be true, which is considered an influence. Also reinforcements can be used to steer the child’s story, which involve everything between praising and criticizing the child and issue sanctions. Improper interviewing can be expressed by removal from direct experience as well, where the child is encouraged to speculate, pretend or engage in imaginative play [38]. These forms may be used in-explicitly, which is expected to make them difficult to recognize. Nevertheless, if they occur, they would be most likely to be disguised as facilitative utterances. For example, because of considerations about the developmental appropriateness the interviewers in the current study were instructed to reward the effort of telling, instead of rewarding the answers given by the child. Moreover, the correctness of the answers were deemed irrelevant, because the child is considered the expert concerning his sexual knowledge and personal experiences. Despite the instructions, facilitative utterances occasionally included praise to the child (‘Very clever’) and also feedback on the correctness of the answer was repeatedly provided by the interviewers (‘Very good’, ‘That’s right’ or ‘I think so too’). Although the interviewers did not intentionally disregard the instructions and were not deliberately looking for a particular outcome, these utterances may already be considered reinforcements or influences and may affect the children’s (further) responses. Here, additional research is required. The content of facilitative utterances should be examined to establish the prevalence of exerting unintentional improper interviewing amongst trained interviewers and the extent of its effect on the information provided

15

by the child. Encouraging the child to speak, while avoiding improper interviewing is crucial. Therefore, an approach should be cogitated and implemented into the instructions of the interviewers.

Besides examining the typical composition of the interview technique of the SKPI method, the groups with and without suspicion were compared. No differences were found in the distributions of utterance types, which indicates that the interview composition is not influenced by the knowledge of the interviewer. This consistency of composition over groups, but also in general may be due to the (semi-)structure of the protocol, which dictates the main part of the interview and provides the interviewer with questions to ask. This means the interviewer can focus his attention on the story the child is telling and devise additional questions of added value. The benefit of a structured protocol has already been illustrated by Orbach et al. [26], but is emphasized by these results. The only difference detected between the suspicion group and the control group, was the significantly higher absolute amount of utterances in the suspicion group. A possible explanation for this difference can be the heightened inquisitiveness of interviewers, being aware of the suspicion of CSA. However, it can also be explained by an increased amount of opportunities in the suspicion group for the interviewer to ask follow up questions, as the probability of the child making remarkable statements is expected to be higher in this group. One might fear that the more utterances are used during an interview, the more chances will be present to ask leading questions or use suggestive prompts. Nevertheless, the composition of the interview remained unaffected, regardless of the greater amount of utterances.

The fact that no differences were found between the interview compositions in age groups and sex of the interviewees, does also substantiate the point of view that the interview protocol provides consistency in interviews. However, it should be noted that the age groups, although evenly distributed, were small. Previous research has found differences in responses of children of different ages [26]. This could have an effect on the utterances and therefore it is important to monitor the results in various age groups. Even though this previous study included children with a wider range of ages, it should be kept in mind that differences can occur, also between the ages ranging from three to eight years old. To detect diversity in age groups, larger groups per age category are required and these were not available in this study.

Significant differences were revealed in composition of the interviews of individual interviewers as well as in the amount of utterances. Most of the differences were detected in interviewer D, who used an unexpected amount of facilitative utterances and significantly more utterances compared to interviewers A and C. The proportion of leading utterances used by interviewer B also diverged from other interviewers. Although these results are on an individual level

16

of the interviewer, these findings illustrate that interviewers have their own style of interviewing which can cause aberrancies when compared to other interviewers. Not all variations are considered meaningful, for example facilitative utterances may be unimportant for the outcome of the interview. Still, too much diversity in leading or suggestive prompts can indicate ineffective interviewing.

The comparison between the conversation regarding the five different pictures included in this study, yielded interesting results. The distribution of the questions were similar for all of the pictures. The two pictures differing in absolute amount of utterances is compatible with the expectations, because the instructions imply the most questions for these pictures. Although the use of visual aids during forensic interviewing has been considered risky [39], it is essential in the SKPI method to offer concrete stimuli [34]. Moreover, the use of drawings in the SKPI method was distinct from usage in previous studies. Previous research used drawings as a touch inquiry to be provided with more details [28, 40,41], whereas the SKPI method merely uses them as an incitement for conversation with the young child. The composition of the interviews in general proves that the use these pictures does not necessarily increase suggestibility of the interview, as the proportion of the suggestive and leading prompts in interviews without visual aids by Lamb et al. [37] correspond with the results in this study. The compositions remaining unaffected by the diverse pictures means that the pictures themselves do not give cause for an adapted questioning style, even though the displays vary within a spectrum of situations, from commonplace to intimate scenes. This makes it less probable that the use of these pictures would increase suggestibility, thus clearing an obstacle for the SKPI method.

Strengths and limitations

The concept of assessing the quality of a forensic interview technique is relatively new and thus has not been performed much, apart from evaluations by the research group of Lamb [25,26,27]. For this reason, a pilot study added value to this study by being able to indicate problems early on in the process. Few difficulties were encountered during this pilot phase and therefore the study could be continued. Confounding factors that were anticipated, were counteracted if possible. Children of different ages were evenly spread over the suspicion and control group and both groups were interviewed by multiple interviewers. However, only one of the interviewers conducted interviews in both groups, so it cannot be concluded that individual interviewers will not be influenced by foreknowledge. Because no differences in composition were found between the groups, it can be concluded that the performance of the interviewers as a group did not alter the compositions of the interviews as a consequence of possessing prior knowledge. Differences between interviewers may

17

also have occurred in this study, due to discrepancies in the training which they received. The interviewers attended the training session at different moments in time while the training protocol had not been formalized yet, therefore monitoring the consistency of the instructions was not possible.

Furthermore, the classification of utterances may not be an optimal tool to consider the interview qualities of interviewers. The level of inter-observer reliability which was described by Cohen’s κ as substantial and is deemed acceptable. However, only a small percentage of the interviews was covered by multiple observers and the measurement might therefore deviate from the true inter-observer reliability. For future research it is recommended to increase the proportion of observations by a second observer. Nonetheless the average agreement percentage seemed to be slightly improved with respect to the preliminary result of the pilot study, possibly due to the modifications of the instructions for the observers.

Another important remark about the method of this study is the selection of samples. This study chose to delimit the inquiry to five inconsecutive pictures of the SKPI as a representative part of the whole interview. This, however, makes it difficult for observers to distinguish between new information introduced by the interviewer (which should be classified as leading) and information already introduced by the child the interviewer is referring to (which should be classified as directive). This may have caused some hindrance in the accurate categorization of utterances. Nevertheless, more interviews could be included due to this choice and the amount of interviews analyzed contributed to the variety of interviewees, making the analysis more informative. However, other methods for evaluation of forensic interviewing quality are lacking. Consequently, in future research the evaluation should be improved by including a larger set of samples, preferably of complete interviews.

A more general problem for the evaluation of interview techniques comprises the building of a consistent framework to score the prompts of interviewers. This is a complicated task, because the communication has to be reduced to text. Aspects of communication that are transmitted in other ways, in a non-verbal manner or by intonation, cannot be included when solely the textual component is analyzed. Moreover, the intention of the prompt might diverge from the literal statement; the message may be implicitly present. Typically this is dependent on the context of an utterance, but this is difficult to quantify and substantiate and thus hard to include in a classification system. How these complications can be addressed, will be discussed in the next section.

18 Considerations for future research

Not many studies have evaluated interview techniques by comparing distributions of utterance types and therefore this evaluation method still has to be optimized. As long as alternative methods to appraise interview proficiency remain lacking, the development of adequate evaluation methods deserves scientific attention. Meanwhile, this method requires thorough instructions to guide observers in the process of classification. The training sessions should be extended and include feedback to improve the performance of the observers and increase the agreement percentage between their observations.

The results of this study call attention to points for improvement of the SKPI method as well. Dissimilarities were found between the compositions of interviews by different interviewers, so the foremost desired adaptation to the current state would be ameliorating the training program. The training program for interviewers has not been formalized yet and the congruence of the training content may become affected by time elapsing or the change of instructors. Therefore, the inconsistencies between interviewers might be reduced by means of an improved training program. Moreover, suggestibility of interviewers may decrease to even lower levels. It is therefore expected that a lot of quality can be gained by improving the training course of the SKPI method. Training techniques have been studied extensively and the review of Powell formulates the critical elements of a training course, including continuity of training sessions, effective ongoing practice, expert feedback and regular evaluation of interviewer performance [42]. Implementing and extending these aspects in the training course of the SKPI method is therefore highly recommended.

The SKPI method is still in its infancy and is therefore still moldable. The results of this study and the recommendations it provided, will advance the SKPI method and bolster its use. However, to gain a complete view of the quality of interviewing, more data is needed. This applies not only to the SKPI method, but also to other protocoled interview techniques, as empirical evidence is absent in many methods. The substantiation by literature is a start, but eventually not sufficiently reliable, as practice may diverge from theory. This can be illustrated by the noteworthy discrepancies between the perceived and the actual performance of interviewers in the study of Powell & Wright [43]. Instruments that are used for such delicate matters as the detection of CSA, require profound validation and should not be accepted due to mere lack of a better alternative.

The results of this study are specific to the SKPI method, which is applied in the medical context. Whether these results are conformable for the legal context is yet to determine. However, additional research should focus its attention especially on the application of such methods in the medical field. The methods for conducting interviews in a legal setting have been formalized and have become more and more substantiated over the years [44], but the funnel model described by

19

literature shows that a criminal investigation is started in only a minority of CSA cases [1]. Health care may be more accessible to CSA victims, which indicates that more victims could benefit from well-grounded methods to detect CSA in the medical context. In the distant future, perhaps even the hesitance to bring a case to court will decrease, because a dependable method might provide reasonable grounds for this and encourage caretakers to take measures. For this reason, further research may have a large impact on society.

Conclusion

This research has provided insight in the performance of the interview technique that is used in the SKPI method. The results show that the distributions of utterances during the interview are consistent, regardless of the degree of suspicion of exposure to CSA. Although the interviews contain a small proportion of leading and suggestive utterances, the amount is considered acceptable. Nevertheless, it would be preferable to reduce the percentage for these types of utterances by means of improved training sessions for the use of the SKPI method. This adjustment might also improve uniformity of the interviews over various interviewers, as some dissimilarities were detected between individuals. The strengths and limitations of this evaluation were discussed, subsequently formulating recommendations for the SKPI method and future research. More interview techniques should be evaluated and it might be beneficial to pay special attention to the methods which are applied in the medical context. Advancements in this field will contribute to society by offering reliable methods for detection of CSA, so that the victims can be protected from further harm.

Acknowledgements

I would like to very warmly thank my supervisor, Kirsten van Ham, for her close involvement during the process of writing my thesis. Also, I would like to thank Rian Teeuw for her valuable contributions during the process in the form of comments and feedback and Sonja Brilleslijper-Kater for sharing much of her knowledge about the SKPI method and how it came to be. A special thank you for the whole Social Pediatrics Department at the AMC, allowing me to watch them work was really inspiring. Furthermore, I would like to thank Marjan Sjerps, who was very willing to provide me with methodological and statistical advice. And lastly, I want to thank my examiner Rick van Rijn, for assessing my work.

20

References

[1] Nationaal Rapporteur Mensenhandel en Seksueel Geweld tegen Kinderen. (2014) Op goede grond. De aanpak van seksueel geweld tegen kinderen. Den Haag: Nationaal Rapporteur.

[2] Beitchman, J. H., Zucker, K. J., Hood, J. E., & Akman, D. (1991). A review of the short-term effects of child sexual abuse. Child abuse & neglect, 15(4), 537-556.

[3] Briere, J. N., & Elliott, D. M. (1994). Immediate and long-term impacts of child sexual abuse. The future of children, 54-69.

[4] Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., et al. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American journal of preventive medicine, 14(4), 245-258.

[5] Huang, H., Yan, P., Shan, Z., Chen, S., Li, M., Luo, C., ... & Liu, L. (2015). Adverse childhood experiences and risk of type 2 diabetes: a systematic review and meta-analysis. Metabolism-Clinical and Experimental, 64(11), 1408-1418.

[6] Suglia, S. F., Clark, C. J., Boynton-Jarrett, R., Kressin, N. R., & Koenen, K. C. (2014). Child maltreatment and hypertension in young adulthood. BMC public health, 14(1), 1149.

[7] Thurston, R. C., Chang, Y., Derby, C. A., Bromberger, J. T., Harlow, S. D., Janssen, I., & Matthews, K. A. (2014). Abuse and subclinical cardiovascular disease among midlife women: the study of women’s health across the nation. Stroke, 45(8), 2246-2251.

[8] Spataro, J., Mullen, P. E., Burgess, P. M., Wells, D. L., & Moss, S. A. (2004). Impact of child sexual abuse on mental health: prospective study in males and females. The British Journal of Psychiatry, 184(5), 416-421. [9] Maniglio, R. (2010). Child sexual abuse in the etiology of depression: A systematic review of reviews.

Depression and anxiety, 27(7), 631-642.

[10] Vrolijk-Bosschaart, T. F., Brilleslijper-Kater, S. N., Widdershoven, G. A. M., Teeuw, A. H., Verlinden, E., Voskes, Y., ... & Benninga, M. A. (2017). Psychosocial symptoms in very young children assessed for sexual abuse: a qualitative analysis from the ASAC study. Child abuse & neglect, 73, 8-23.

[11] Christian, C. W., Lavelle, J. M., De Jong, A. R., Loiselle, J., Brenner, L., & Joffe, M. (2000). Forensic evidence findings in prepubertal victims of sexual assault. Pediatrics, 106(1), 100-104.

[12] Heger, A., Ticson, L., Velasquez, O., & Bernier, R. (2002). Children referred for possible sexual abuse: medical findings in 2384 children. Child abuse & neglect, 26(6-7), 645-659.

[13] Girardet, R., Bolton, K., Lahoti, S., Mowbray, H., Giardino, A., Isaac, R., et al. (2011). Collection of forensic evidence from pediatric victims of sexual assault. Pediatrics, 128(2), 233-238.

[14] Young, K. L., Jones, J. G., Worthington, T., Simpson, P., & Casey, P. H. (2006). Forensic laboratory evidence in sexually abused children and adolescents. Archives of pediatrics & adolescent medicine, 160(6), 585-588.

[15] Kendall-Tackett, K. A., Williams, L. M., & Finkelhor, D. (1993). Impact of sexual abuse on children: a review and synthesis of recent empirical studies. Psychological bulletin, 113(1), 164.

21

[16] London, K., Bruck, M., Ceci, S. J., & Shuman, D. W. (2005). Disclosure of child sexual abuse: What does the research tell us about the ways that children tell?. Psychology, Public Policy, and Law, 11(1), 194. [17] Duffy, C., Keenan, M., & Dillenburger, K. (2006). Diagnosing child sex abuse: A research challenge.

International Journal of Behavioral Consultation and Therapy, 2(2), 150.

[18] Lyon, T. D. (2014). Interviewing children. Annual Review of Law and Social Science, 10, 73-89.

[19] London, K., Bruck, M., Wright, D. B., & Ceci, S. J. (2008). Review of the contemporary literature on how children report sexual abuse to others: Findings, methodological issues, and implications for forensic interviewers. Memory, 16(1), 29-47.

[20] Cronch, L. E., Viljoen, J. L., & Hansen, D. J. (2006). Forensic interviewing in child sexual abuse cases: Current techniques and future directions. Aggression and Violent Behavior, 11(3), 195-207.

[21] Olafson, E., & Kenniston, J. (2008). Obtaining information from children in the justice system. Juvenile and Family Court Journal, 59(4), 71-89.

[22] Schreiber, N., Bellah, L. D., Martinez, Y., McLaurin, K. A., Strok, R., Garven, S., & Wood, J. M. (2006). Suggestive interviewing in the McMartin Preschool and Kelly Michaels daycare abuse cases: A case study. Social Influence, 1(1), 16-47.

[23] Anne Tubb, V., Wood, J. M., & Hosch, H. M. (1999). Effects of suggestive interviewing and indirect evidence on child credibility in a sexual abuse case. Journal of Applied Social Psychology, 29(6), 1111-1127.

[24] La Rooy, D., Brubacher, S. P., Aromäki-Stratos, A., Cyr, M., Hershkowitz, I., Korkman, J., ... & Stewart, H. (2015). The NICHD protocol: A review of an internationally-used evidence-based tool for training child forensic interviewers. Journal of Criminological Research, Policy and Practice, 1(2), 76-89.

[25] Lamb, M. E., Orbach, Y., Hershkowitz, I., Esplin, P. W., & Horowitz, D. (2007). A structured forensic interview protocol improves the quality and informativeness of investigative interviews with children: A review of research using the NICHD Investigative Interview Protocol. Child abuse & neglect, 31(11-12), 1201-1231.

[26] Orbach, Y., Hershkowitz, I., Lamb, M. E., Sternberg, K. J., Esplin, P. W., & Horowitz, D. (2000). Assessing the value of structured protocols for forensic interviews of alleged child abuse victims. Child abuse & neglect, 24(6), 733-752.

[27] Lamb, M. E. (1996). Effects of investigative utterance types on Israeli children's responses. International Journal of Behavioral Development, 19(3), 627-638.

[28] Anderson, J., Ellefson, J., Lashley, J., & Miller, A. L. (2009). The cornerhouse forensic interview protocol: RATAC. TM Cooley J. Prac. & Clinical L., 12, 193.

[29] Toth, P. (2011). Comparing the NICHD and RATAC child forensic interview approaches—Do the differences matter. APSAC Advisor, 23, 15-20.

[30] Newlin, C., Steele, L. C., Chamberlin, A., Anderson, J., Kenniston, J., Russell, A., ... & Vaughan-Eden, V. (2015). Child forensic interviewing: Best practices (pp. 1-20). US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention.

22

[31] Methodisch kind-interview. (n.d.). Retrieved June 28, 2018, from https://hetlock.nl/opleiding/methodisch-kind-interview-2017/

[32] De Ruiter, C. (n.d.). Training NICHD interviewprotocol. Retrieved June 28, 2018, from

https://www.conflictscheiding.eu/training-nichd-interview-protocol/

[33] Brilleslijper-Kater, S.N. (2005). Beyond words: Between-group differences in the ways sexually abused and nonabused preschool children reveal sexual knowledge. Utrecht: Febodruk.

[34] Brilleslijper-Kater, S.N., & Baartman, H.E.M. (2000). What do young children know about sex? Research on the knowledge of sexuality of children between the ages of 2 and 7 years. Child Abuse Review, 9(3), 1-17. [35] Brilleslijper-Kater, S. N., Friedrich, W. N., & Corwin, D. L. (2004). Sexual knowledge and emotional reaction

as indicators of sexual abuse in young children: theory and research challenges. Child Abuse & Neglect, 28(10), 1007-1017.

[36] Van Ham, K., Brilleslijper-Kater, S.N., van der Lee, J. H., van Rijn, R. R., van Goudoever, J. B., & Teeuw, A. H. Beyond words. A protocol to validate the Sexual Knowledge Picture Instrument (SKPI) as a diagnostic instrument for child sexual abuse. (Unpublished manuscript).

[37] Lamb, M. E., Sternberg, K. J., Orbach, Y., Hershkowitz, I., Horowitz, D., & Esplin, P. W. (2002). The effects of intensive training and ongoing supervision on the quality of investigative interviews with alleged sex abuse victims. Applied Developmental Science, 6(3), 114-125.

[38] Wood, J. M., & Garven, S. (2000). How sexual abuse interviews go astray: Implications for prosecutors, police, and child protection services. Child Maltreatment, 5(2), 109-118.

[39] Russell, A. (2008). Out of the woods: A case for using anatomical diagrams in forensic interviews. Update, 21(1), 2-6.

[40] Aldridge, J., Lamb, M. E., Sternberg, K. J., Orbach, Y., Esplin, P. W., & Bowler, L. (2004). Using a human figure drawing to elicit information from alleged victims of child sexual abuse. Journal of Consulting and Clinical Psychology, 72(2), 304.

[41] Brown, D. A., Pipe, M. E., Lewis, C., Lamb, M. E., & Orbach, Y. (2007). Supportive or suggestive: Do human figure drawings help 5-to 7-year-old children to report touch?. Journal of Consulting and Clinical

Psychology, 75(1), 33.

[42] Powell, M. B. (2008). Designing effective training programs for investigative interviewers of children. Current Issues Crim. Just., 20, 189.

[43] Powell, M. B., & Wright, R. (2008). Investigative interviewers' perceptions of the value of different training tasks on their adherence to open-ended questions with children. Psychiatry, Psychology and Law, 15(2), 272-283.

[44] Faller, K. C. (2014). Forty years of forensic interviewing of children suspected of sexual abuse, 1974–2014: Historical benchmarks. Social Sciences, 4(1), 34-65.

23

Appendices

Appendix I – Protocol of the SKPI

Handleiding SKPI

(Sexual Knowledge Picture Instrument, Brilleslijper-Kater, 2005)

Inhoud

1.1 INTRODUCTIE ...24

1.2 ALGEMENE AANWIJZINGEN BIJ AANVANG VAN HET GESPREK ...24

1.3 ALGEMENE AANWIJZINGEN TIJDENS HET DOORNEMEN VAN DE PLATEN ...25

2. SCORING KENNIS PER PLAAT ...26

2.1PLAAT 1:INTRODUCTIEPLAAT I(GEZIN AAN TAFEL) ...26

2.2 PLAAT 2:INTRODUCTIEPLAAT II(SMARTPHONES / BEELDVORMING) ...27

2.3PLAAT 3:AANGEKLEDE KINDEREN...27

2.4PLAAT 4:BLOTE KINDEREN (VOORKANT) ...28

2.5PLAAT 5:BLOTE KINDEREN (ACHTERKANT) ...29

2.6PLAAT 6:AANGEKLEDE MAN EN VROUW ...29

2.7PLAAT 7:BLOTE MAN EN VROUW (VOORKANT) ...30

2.8PLAAT 8:BLOTE MAN EN VROUW (ACHTERKANT) ...32

2.9PLAAT 9:KUSSENDE MAN EN VROUW ...32

2.10PLAAT 10:BLOOT OP ELKAAR LIGGENDE MAN EN VROUW ...33

2.11PLAAT 11:(SEKSUEEL)KINDERSPEL ...34

2.12. PLAAT 12:KIND BIJ EEN DOKTERSFIGUUR...34

2.13PLAAT 13:VADERFIGUUR OVER MEISJE IN BED HEEN GEBOGEN ...35

2.14PLAAT 14:BLOOT JONGETJE IN DE DOUCHE MET MOEDERFIGUUR ...36 2.15PLAAT 15:MOEDERFIGUUR MET HUILEND KIND IN BED ...37

3. AFSLUITEN VAN HET INTERVIEW ...37

4. SCORING NON-VERBAAL...38

24

1.1 Introductie

Aan de hand van deze semi-gestructureerde vragenlijst worden de platen van het Sexual Knowledge Picture Instrument (SKPI) met het kind doorgenomen.

Doelstelling is om per plaat een scoring te geven aan het kind. Dit gebeurt door middel van observaties, waarbij wordt gelet op zowel nonverbale signalen als de verbale kennis en uitlatingen van het kind.

Kinderen zijn normaliter open en onbevangen en zien vrijwel alles op de platen als normaal. Zij zullen veelal open ingaan op de vragen die gesteld worden door de interviewer. Ook als het kind iets niet lijkt te willen vertellen is dit dus iets opvallends.

Het is van belang om aandacht te hebben voor: - WAT het kind vertelt.

- HOE het kind het vertelt.

- IN WELKE CONTEXT het kind het vertelt. - Wat het kind NIET vertelt.

Voorafgaand aan het voeren van het gesprek met het kind is het raadzaam om de handleiding en het platenboek goed door te nemen, zodat de onderzoeker kennis heeft van de aanwijzingen en te stellen vragen bij elke plaat.

1.2 Algemene aanwijzingen bij aanvang van het gesprek

Bij aanvang van het gesprek is het van belang het kind uit te leggen wat er gaat gebeuren, en wat je van het kind verwacht.

Geef eerst een korte, neutrale inleiding:

‘Ik heb hier een boekje met tekeningen.’

‘Ik ga jou alle tekeningen laten zien en dan mag jij daarover vertellen.’ ‘Ik ga je ook vragen stellen.’

Daarnaast is het van belang een aantal zaken aan het begin van het interview goed aan het kind duidelijk te maken:

- Benadruk dat je als onderzoeker ‘onwetend’ bent

Vermeld daarom dat het kind het NIET fout kan doen, en er dus geen sprake is van een goed of fout antwoord op je vragen. (dus niet zoals op school als de juf/meester een vraag stelt).

Doe dit als volgt:

‘Het gaat erom wat JIJ weet en wat JIJ mij wil vertellen.’

‘Alles wat je zegt is altijd goed. Je kunt dus geen foute antwoorden geven (zoals op school, als de juf of meester je iets vraagt.)’

- Niet weten – instructie

‘Als je het antwoord op de vraag niet weet, mag je dat gewoon zeggen.’

Toets hierna of het kind dit begrepen heeft, door bijvoorbeeld de volgende vraag te stellen:

‘Dus als ik jou vraag: Hoe heet mijn hond? Wat zeg je dan?’ Eventueel na antwoord: Dat kun je ook niet weten, want dat heb ik je niet verteld.

25

‘Als je de vraag niet snapt, mag je dat gewoon zeggen.’

Toets hierna of het kind dit begrepen heeft, door de volgende vragen te stellen:

‘Dus als ik jou vraag: Wat is jouw geslacht? Wat zeg je dan?

‘Dat komt omdat geslacht een moeilijk woord is. Dan zal ik het op een andere manier vragen, bijvoorbeeld: ‘ben jij een jongen of een meisje’?

(Houd eventueel een 2e voorbeeld paraat voor het geval het kind een antwoord gokt).

1.3 Algemene aanwijzingen tijdens het doornemen van de platen

Open vragen stellen

Bijvoorbeeld: ‘Wat zie je hier?’

Dus NIET: Suggestieve vragen, of gesloten vragen, dus die waarop alleen met ‘ja’ of ‘nee’ geantwoord kan worden.

NB: gesloten vragen zijn niet altijd suggestief. Soms zelfs minder dan een open vraag, bijvoorbeeld vergelijk: Met wie heb je dit besproken/ Heb je dit besproken?

NB: Uit onderzoek blijkt dat kinderen de neiging hebben om laatste antwoordalternatief te kiezen, houd ook hier rekening mee.

Niet weten of niet willen vertellen?

Als het kind niets zegt, of zegt het niet te weten, echter het voor de onderzoeker lijkt of het kind het niet WIL vertellen (dit merk je bijvoorbeeld doordat het kind ‘weet niet’ zegt, en wegkijkt, de interviewer duidelijk probeert af te leiden, o.i.d.), dan vragen:

‘Weet je dat niet, of wíl je dat niet vertellen?’

Bij antwoord ‘dat wil ik niet vertellen’, één keer vragen waarom het kind het niet wil vertellen (kan je mij zeggen waarom je dat niet wil vertellen?), Dan respecteren (dus: ‘o.k.’ zeggen en niet

doorvragen.)

Overnemen van de bewoordingen van het kind.

Neem tijdens het gesprek met het kind zoveel mogelijk de bewoordingen van hem/haar over, van de personen op de tekeningen, en de benoemde lichaamsdelen (m.n. genitalia) en voorwerpen. Het kind zal bijvoorbeeld de volwassenen op de tekeningen ‘papa en mama’ noemen, en de genitalia

benoemen op zijn/haar eigen wijze.

Doorvragen

Kinderen hebben meestal nog een beperkt vocabulaire, en dus niet het vermogen om alles goed te kunnen verwoorden. Zij zullen daarom soms andere woorden voor iets gebruiken. Vraag daarom naar eigen inzicht door bij de antwoorden die het kind geeft waarbij het mogelijk een ‘andere betekenis’ zou kunnen hebben voor het kind.

Bijv. bij plaat 14, als het kind zegt ‘die moeder is het jongetje aan het soppen’, vragen ‘Vertel eens,

wat is dat dan, soppen?’ en: ‘Doet jouw moeder dat ook wel eens bij jou?’ Aanmoedigen

Let op: Niet kind aanmoedigen door alleen iets als 'goed zo' te zeggen (er is namelijk geen goed of fout in de antwoorden van het kind, want alles wat het kind wel/niet zegt is goed). Wel het kind zo nu en dan stimuleren door dingen te zeggen als 'wat kan jij veel vertellen, je doet/vertelt het heel duidelijk goed, je doet goed mee, etc.

26

Opvallende uitlatingen van het kind

Indien het kind tijdens het doorlezen van de SKPI bij een of meerdere platen een voor de onderzoeker opvallende uitspraak doet, is het allereerst belangrijk hierop non-verbaal zo normaal /neutraal mogelijk te reageren. Daarnaast kan de onderzoeker hier kort en middels een open vraag op in gaan, door (een of meerdere keren) te vragen:

‘Vertel eens…?’ of: ‘Vertel daar eens wat meer over?’

Het staat hierna het kind vrij hierover meer te vertellen. Eventuele vervolgvragen, als het kind meer vertelt:

‘En wat gebeurde er toen?’ ‘En verder?’ etc.

Als het kind hierop een vrij duidelijke uitlating doet hierop door vragen:

‘met wie was dat?’ en ‘waar was dat precies?’

Vervolgens zal de onderzoeker het verdere onderzoek op normale wijze hervatten.

2. Scoring kennis per plaat

De interviewer stelt per plaat een aantal vragen en scoort vervolgens de kennis en eventuele uitlatingen van het kind.

Bepaalde antwoorden:

- Als een kind (bij doorvragen) zegt: ‘ik weet het niet’ of ‘gewoon’, dan scoren als NEE/NIET GOED (kennisvragen). Bij de inschatting gevoel vragen, dit scoren bij: anders, nl… ‘weet ik niet’.

- Als iets niet is gevraagd, dan scoren als: anders, nl… en dan invullen ‘niet gevraagd’.

- Bij herhaalde vragen zoals gender identiteit/genitalia functies, iedere keer opnieuw scoren, niet de score meenemen uit eerdere vragen. Uitzondering: als bijvoorbeeld oren, of andere niet relevante lichaamsdelen die als voorbeeld gelden, al eerder aan bod zijn gekomen.

- Als er 2 antwoorden gegeven worden, waarvan er een goed en een fout is, dit goed rekenen. (bijvoorbeeld: functie mannelijk genitaal: poepen en plassen)

- Altijd maar 1 antwoord aankruisen (wat het dichtste in de buurt komt) meerder antwoorden kunnen niet verwerkt worden

- Als het antwoord niet gezegd wordt, maar duidelijk wordt uitgebeeld (bijvoorbeeld functie handen: kind gaat klappen), dit goed rekenen.

Let op! Naast het toetsen van de kennis is het belangrijk dat wordt gelet op denon-verbale reacties die het kind vertoont tijdens het interview. Deze kunnen gelijktijdig (of evt achteraf, indien video opnames gemaakt worden) per plaat worden gescoord in de scoretabel. Zie tabel hoofdstuk 4. In hoofdstuk 5 wordt tot slot ook de algemene indruk van het kind en eventuele opvallendheden tijdens het interview gescoord.

2.1 Plaat 1: Introductieplaat I (Gezin aan tafel)

Vertel eens / leg eens uit: -Wat zie je hier?

-Wie zijn dat? -Wat doen ze?

Doorvragen tot er een kort ‘verhaaltje’ uitkomt, bijv. ‘Een vader en moeder en kind, ze zijn aan het