Chapter Six Methods
6.2 Hypotheses
6.4.4 Involvement with professional agencies
Although all the subjects had been referred to the Department o f Psychological Medicine at Great Ormond Street Hospital, only 54 (51.9%) were actually seen there. The reasons for the remaining 50 subjects not being seen were varied. Often, social services found alternative treatment centres for the subject’s care that were available sooner than GOSH, but some individuals simply failed to attend prearranged assessments. Occasionally GOSH professionals turned a referral down if appropriate criteria (e.g. age) were not met or if safety had not been assured for the subject.
Eighty subjects (76.9%) were referred by social services whilst 23 (22.1%) were referred by some other source including General Practitioners, Paediatricians and Psychiatrists fi’om other institutions. The source of referral could not be established for 1 subject (1%). Eighty-five subjects (81.7%) were referred as a direct result of their sexual victimisation although for 9 subjects (8.7%) the reason for referral could not be established. The remaining 10 (9.6%) subjects were either referred for other mental or physical health reasons or because they had already begun to act out sexually.
The mean age of the subjects at referral was 10.69 years, with the youngest being 4 years 7 months and the oldest 19 years and 3 months. The difference in age at referral between the two groups was not significant (t = .075, d f = 95, p>.05). The age o f 7 subjects (6.7%) at the time of their referral was not known.
6.4.5 Status o f subject at referral
Details o f the involvement o f social services, the courts and education authorities at any point in the subjects’ lives prior to age 16 years or the onset o f perpetration (whichever came first) are given in Table 6.1 below. The victim-perpetrators did not significantly differ from the victims-only with regard to any o f these variables.
Table 6.1: Details of professional involvement with subjects
No. of subjects Number (%) Number (%)
(% of total of Victim- of Victims-
sample) perpetrators only
On Child Protection 54 12 42 Register (51.9) (57.1) (50.6) On Care Order 43 7 36 (41.3) (33.3) (43.4) Ward o f Court 29 6 23 (27.9) (28.6) (27.7) Statement of Educational 19 5 14 need (18.3) (23.8) (16.9)
6.5 Procedures and M aterials
6.5.1 Process 1: Creation o f research tools
6.5.1.1 Generation o f the Computerised Trawling program
Social services and clinical case files were used as sources of the necessary data about the risk and protective factors that each subject had experienced. However these contemporaneous files were extremely large, often running to many volumes. It was therefore necessary to find a way o f condensing this information into a more digestible form for coding to subsequently take place. It was for this reason that the Computerised Trawling Program was designed: to facilitate the collection, storage and retrieval of case file material. The program was developed in a Windows 95 ACCESS format. For each subject, the program was used in the following way:
• Basic demographic information on the subject (e.g. name, date of birth etc) and the case file (e.g. start date and end date o f file, size o f file) was recorded in a computerised cover sheet.
• The researcher read the file until s/he reached a quote relating to risk or protective factors, which met the criteria for extraction described in the manual. This quote was then typed into the text box o f the Quote Screen. The risk or protective factors to which the quote referred was then recorded by clicking on one or more electronic tick boxes (known as ‘tagging’ the quote to a factor). The date o f the document from which the quote was extracted was also recorded.
• During trawling, all the extracted quotes referring to one factor could be retrieved by positioning the cursor on the factor name (e.g. experiencing physical abuse) and then double clicking. This allowed the trawler to establish what information had been trawled, and what still needed to be recorded.
• The process of reading and extracting quotes was continued until the file had been read in its entirety.
• The trawler would use the quotes extracted about the subject’s care history to generate a care table. Each change in care arrangements, which had lasted for a month or more, would be recorded in the care table along with supporting quotes. The introduction section to the Trawling and Coding Manual (Appendix 2) provides further details o f the procedure for constructing the care table.
A copy of the trawling program can be obtained from the author.
6.5.1,2 Development o f the maltreatment and protection measurement systems
The Maltreatment and Protection measurement systems were developed in parallel with the computerised trawling program described above. The conceptual principles for the measurement o f protection and maltreatment adopted in this thesis were described in the previous chapter (see section 5.7). The Trawling and Coding Manual (See Appendix 2) was created on the basis of these principles.
A first draft o f the manual was written on the basis of concepts drawn form the literature concerning the measurement o f maltreatment. The manual was refined using a reflexive process o f piloting and redrafting.
Piloting consisted o f the trawling and scoring o f Great Ormond Street Hospital clinical files held on cases referred to the Department of Psychological Medicine, but who were not included in the sample used in this thesis. These cases included sexually victimised females referred between 1980 and 1992, and both male and female victims referred after this time. The manual was generated and refined independently o f cases that were to be used to establish reliability. This was to ensure that the reliability was generalisable beyond the cases on which piloting took place.
The pilot cases were trawled and scored by each member o f the team, and the team would then meet to compare the scorings that had been made, and to discuss any difficulties. Matters that arose relating to risk factors were resolved by consensus decisions made within the whole team, and these were then incorporated into the new manual. Issues relating to the trawling or coding of protective factors were referred by the other team members to the author, who then made alterations to the manual where necessary. This procedure was repeated until no new major difficulties emerged.
The completed Trawling and Coding Manual (see Appendix 2) consisted of an introduction, a ‘Risk Factor’ section and a ‘Protective Factor’ section. In the introduction, general guidance was given about both trawling the files and subsequently coding the extracted quotes. The ‘Risk Factor’ section consisted o f a chapter focussed on each o f the maltreatment subtypes whilst the ‘Protective Factor’ section comprised guidelines for the extraction and coding o f quotes relating to the putative protective factors. The two latter sections are described in more detail below.
6,5.13 Risk factor section
The chapters relating to the putative risk factors were designed to enable the reliable extraction and coding o f information about those factors. This included sections concerning the following: definitions o f terms; the appropriate tagging o f quotes to factors; the criteria an individual had to meet to be considered a perpetrator o f a maltreatment subtype; and specific difficult issues relating to that factor.
For the purpose of this thesis, the main consideration was simply the presence or absence of any particular risk factor. Guidance on coding the presence o f a risk factor as a dichotomous variable was contained within a specific section in the chapter. The title o f this section depended partly on the factor under investigation. For example, within the chapter focussed on ‘experiencing physical abuse’, the title was “Dichotomous question: Did the subject experience physical abuse”. The risk factors included in the study were drawn fi'om the research literature, as summarised in Chapter 3. A table of the risk factors investigated, and the minimum criteria needed for their presence to be coded, is given below (see Table 6.2). For more details, please refer to Appendix 2.
Table 6.2: Minimum criteria for coding putative risk factors as present
Putative risk factor Minimum criteria for presence
Physical abuse Evidence o f minor marks on body, non-specific statement that subject had been physically abused Witnessing intrafamilial
violence
Subject witnessed violence between two persons, at least one of whom was a family member.
Neglect (failure to provide) Subject missed odd meals, presented as dirty, lived in squalor, failed to attend medical appointments
Neglect (lack of supervision) Inadequate supervision for up to a morning (<3 hrs) or for an unspecified duration
Rejection by carers Evidence o f hostility, degradation, blaming, harsh punishment, rejection, preference for others, threats to kill or remove
Discontinuity o f care Physical separation from a primary carer for 1 month or more, excluding holidays
Sexual abuse by a female Any sexual contact with female, either forced or with perpetrator who was two years older
6.5,1.4 Protective factor section
This section dealt exclusively with the trawling and coding of protection and comprised seven chapters, each of which focussed on either a single putative protective factor or a group of conceptually related factors. Brief definitions o f these factors are given in Table 6.3 below, although more detail is contained within the Trawling and Coding Manual (See Appendix 2).
The structure o f each chapter in the protective factor section was specifically designed to deal with the issues thought likely to arise with the particular protective factor under discussion. However, each one tended to start with definitions of the key terms followed by the coding questions. These questions were accompanied by explanations or clarifications concerning how they might be answered reliably. (See Appendix 2).
Table 6.3: Definitions of putative protective factors
Putative protective factor Definition Number of non-abusive male
primary carers prior to the subject’s 12* birthday
Number o f adult males who had lived for a year or more with the subject in a parenting role, and had not sexually or physically abused, neglected nor rejected the subject at any time
Number o f non-abusive female primary carers prior to the subject’s 12* birthday
Number o f adult females who had lived for a year or more with the subject in a parenting role, and had not sexually or physically abused, neglected nor rejected the subject at any time
Number o f sexually non- abusive male primary carers before subject’s 12* birthday
Number o f males who had lived for a year or more with the subject in a parenting role, and had not sexually abused the subject at any time
Number o f sexually non- abusive female primary carers before subject’s 12* birthday
Number o f females who had lived for a year or more with the subject in a parenting role, and had not sexually abused the subject at any time
Number o f years in non- abusive foster care
Total number of decimal years spent living with foster parents who had never sexually or physically abused, neglected nor rejected the subject
Number o f care units before subject’s 12* birthday
Number o f care arrangements lasting at least one month experienced by the subject up to the age o f 12 years - this factor is reverse scored
Continuity of care The longest uninterrupted period spent living with the same primary carer in decimal years
Good relationship with adult Evidence that at least one relationship with an adult is emotionally supportive for the subject or described as close/good, with no evidence o f abuse
Good relationship with sibling
Evidence that a relationship with a sibling is one of mutual cooperation and fiiendship or described as close/good, with no evidence of abuse
Good relationship with peer Evidence that a relationship with a non-family member under 18 years old is a reciprocal fiiendship or described as close/good, no evidence o f abuse Treatment Subject had at least one meeting, after his sexual
abuse, with a psychological health care worker Positive response to
disclosure
Evidence that a non-abusive primary carer believed the disclosure or acted to prevent further abuse
Perpetrator convicted Evidence that the perpetrator of the subject’s sexual abuse was convicted o f his offences in a court
A brief explanation for the inclusion o f some of these factors, as defined above, may be necessary, as well as the exclusion o f other possible protective factors. Many o f the included factors are designed to indicate the availability o f stable, non-abusive figures for the subject either to form attachments to or to model his behaviour on (e.g. continuity of care, good relationship with adult/sibling/peer). This is based upon evidence in the literature on protection against maltreatment, which suggests such factors are protective against certain adverse outcomes (e.g. Lam and Grossman, 1997; Lynskey & Fergusson, 1997; Spacarrelli & Kim, 1995). However, it might appear unclear as to how some of the factors could be protective. For example, the factors ‘number o f non-abusive male/female primary carers’ might be interpreted as more representative o f risk than of protection, since a higher number could indicate less stability in the subject’s care arrangements.
The reason for this approach being adopted relates to the nature o f the population fi’om which the sample was drawn and particularly the severity o f the maltreatment that its members have typically experienced. It was expected that a high proportion of the subjects in this sample would have experienced many care arrangements, with very few of them containing non-abusive carers. Therefore, it was hypothesised that the more non-abusive care figures in a subject’s life who modelled consistency and support, the more protective it would be for that subject. However, the extent to
which these particular fectors do or do not represent risk will be assessed statistically in the next chapter (see Section 7.4.1).
There are a number o f putative protective factors that were thought to warrant some investigation in this study on the basis o f the resilience literature, but are not in the list above. These included some measure o f intelligence, achievements at school and extra-curricular achievements. An attempt was made to establish the presence o f these Actors in the various files. However, the recording o f information about these 6ctors was so rare as to render meaningless any analysis based on these data.
6.5.1.4.1 Construction o f the Protective Index
Although the protective index was not specifically described within the manual, being a construction for the puipose o f the statistical analysis only, it seems appropriate at this point to give the details of, and account for, the selection o f fiictors that made up the index, and to describe how the index was calculated.
The protective index consisted o f the following variables: good relationship with an adult; good relationship with a sibling; good relationship with a peer; years spent in foster care; number o f non abusive male carers by age 12; number o f non abusive female carers by age 12; longest time the subject lived with the same primary carer; number o f care units by age 12 (reverse scored).
As stated above, the review o f the resilience literature highlighted the importance o f fectors indicating the strength o f the positive relationships between maltreated children and non-abusive individuals who were significant to them. The Actors listed above were selected for inclusion in the protective index, prior to any empirical analysis, on the basis o f their relatedness to this concept.
As detailed above, the protective index consisted o f a combination o f binary and continuous variables. In order to avoid undue weighting being given to continuous variables in the construction o f the index, all variables were standardised. These standardised values were then summed to create the protective index.
6.5.L 5 R eliability
Adequate inter-rater reliability was established for the five individuals who trawled the files and the four who coded the resultant data for the majority o f 6ctors. Details o f the method used to initially establish reliability o f trawling and coding, to re-establish the reliability o f coding and the results o f all these tests, are given in the appendix (See Appendix 1).
6.5.2 Process 2: Establishine presence o f predictor variables
6.5.2.1 Identification o f the Sam ple
The sample was generated on the basis o f records held at the Department o f Psychological Medicine, Great Ormond Street Hospital for Children, NHS Trust.
The sample selection process is represented in Figure 6.2 and the inclusion/exclusion criteria for the sample are given in Figure 6.3.
Ô.5.2.2 Location o f social service fU es
The Department o f Psychological Medicine, Great Ormond Street Hospital held clinical files on all the sample subjects and they were made available for this study. Initially, they were used to locate each subject’s social service file that, it was hoped, would contain more detailed information on risk and protection. The clinical files were examined therefore, and the following information extracted:
• Full names o f the subject, including all changes in name and variations in spelling • Date o f birth, including all recorded variations
• Address o f the subject, all 6m ily members. General Practitioner and the social services department involved in the case.
An attençt was made to locate the social service files o f subjects who met two criteria: the subjects would be aged 18 years or older on 01.05.99; there was sufficient information in their clinical file (subject’s first name, last name, date o f birth, or sibling
Figure 6.2: Selection of the sample
S ta g e a
S ta g e b
S ta g e c
Starting sangle N = 601 A ged> 18 (01.05.99) Aged < 18 (01.05.99) N = 449 N = 1 5 2Sufficient information Insufficient info. for tracing files for tracing files
N = 427 N = 22
S ta g e d
Social service depts. agree to take partN = 408
Social service depts. decline to take part
N = 19