This report seeks to examine the aims and practices of home educators in the UK

Chapter 9: Social and Psychological Data - Children's Assertive Behaviour Scale (CABS), Revised Rutter Scale (RRS), Goodman Strengths & Difficulties Questionnaire (Goodman SDQ)

Abstract

This element of the research involved 103 home-educating children aged from 4 to 11 years old. Three instruments were used: the Children's Assertive Behaviour Scale (CABS), the Revised Rutter Scale (RRS) and the Goodman Strengths & Difficulties Questionnaire (SDQ) in a total of 136 social and psychological based assessments. Questionnaires were either self-rated or parent-rated but referred to the children. The purpose of conducting these assessments was to establish whether the home-educated children experienced social or behavioural problems beyond the 'norm'. Some critics have suggested a relationship between home-educated children, social ineptness and behavioural problems. Results confirmed that the home-educated children were socially adept and did not display behavioural problems beyond the norm.

9 Social and Psychological Data Results: overview

A total of 103 respondents provided 136 assessments, namely, 43 CABS, 51 SDQ and 42 Revised Rutter Scale[1]. None of those invited to participate either refused or failed to respond. Except for seven child-rated SDQ questionnaires (completed by 11 year olds) all assessments were parent rated[2]. Beyond the initial questionnaire survey, 64 (62.13%) of the 103 participants had assisted with other areas of the research[3]. As far as could be known, none of the children assessed in this section had attended school within the three months prior to assessment.

The following sub-sections discuss the results for each of the assessments and are followed by a discussion of the combined results from this chapter.

9.1 CABS Results

The Children's Assertive Behaviour Scale (CABS) [4] contains 27 items that can be grouped into five sub-categories; ‘Positive’ (how a respondent manages in situations involving opportunities for positive expression); ‘Negative’ (how a respondent manages in situations involving opportunities for negative expression); ‘Request’ (measures the ability to formulate and react to, a request); ‘Conversation’ (examines the facility to make and/or be involved in, conversation); and ‘Feeling’ (explores the capacity to express one's own and understand another’s feelings). For each item, participants could pick from answers a-e which were scored from -2 to +2.

The instrument was chosen because Shyers (1992) had used it with a home-educated sample. Therefore, Shyers procedure was adopted to allow comparisons to be made. However, when this writer later acquired a copy of Michelson et al's (1983) work on CABS, it transpired that the original authors had used a method of analysis that differed from that used by Shyers. Since both methods were valid and focused on differing aspects of social behaviour, both CABS' analyses are used here.

The difference between Shyers' and Michelson et al's analyses is as follows:

a) In looking at participants' 'Total Difficulties' scores (the addition of scores from all bar the Prosocial category) Michelson et al (1983) summed individual's scores as absolute values, that is, they ignored the + and - signs. Thus, a zero score represented assertiveness and the further a participant's score was from zero, the less assertive they were. This applied whether the category scores showed a person to be aggressive (mostly + scores) or passive (mostly - scores).

b) Shyers (1992) used an 'all values' approach whereby a '+' or '-' score indicated whether a participant was mostly passive or aggressive. A zero score represented equally passive and aggressive behaviour with the implication that this was the ideal. Shyers explained:

'I was interested in the degree of passivity vs the degree of aggression, so the positive vs negative angle was best instead of absolute values.'

Email conversation with L. Shyers, 20.3.98

Taking a normal North American school-aged population of approximately eight to twelve years of age, Michelson et al. (1983) described an 'absolute value' mean of 13 with an SD of 7 as the general norm. Shyers (1992) found the 'all values' mean for his eight to ten-year-old home-educated sample (n=70) to be –7.79 and schoolchildren (n=70) as -6.10. Shyers concluded that the schoolchildren were slightly less passive than those homeschooled.

Table 9.1 gives results from five CABS studies including the present one[5]. The present sample's 'all values' mean was -6.11, thus supporting Shyers' finding that the home-educated children were passive rather than aggressive. Using Michelson et al's 'absolute value' approach the UK sample with a mean of 16.11 were far less assertive than Michelson et al's 'norm' mean of 13.

Table 9. 1: cabs results from four studies, displayed with present research data

Author	Wood, Michelson & Flynn	Michelson, Andrasic, Vucelic & Coleman	Rothermel absolute values	Shyers		Rothermel all values
Year	1978	1979	1999	1992		1999
Country	USA Florida	USA Pennsylvania	UK National	USA Florida		UK
Educated	School	School	Home	School	Home	Home
age	9	8-10	8-10	8-10	8-10	8-10
N	149	90	43	70	70	43
Mean	12.74	12.98	16.11	-6.10	-7.79	-6.11
St. Dev.	8.48	6.47	6.41	5.52	6.92	8.78
Min. Score	no data	no data	5	-21	-28	-30
Max. Score	no data	no data	33	5	11	10

Wood et al (1978) and Michelson et al (1979) are cited in Michelson et al. (1983)

Thus, the UK home-educated sample were less passive than Shyers' home-educated sample and considerably less assertive than either Wood et al. (1978) or Michelson et al (1979).

9.1.1 Confidence Interval

Using a confidence interval[6] it was possible to establish whether the means were similar between this study's results and those of Wood et al. (1978) and Michelson et al (1979) and between this study's results and those of Shyers (1992). The confidence interval assessed using 'absolute values' for this research was from 14.20 to 18.03, thus overlapping with Michelson et al.'s (1979) confidence band of 11.56 to 14.39. Shyers' 'all values' confidence band lay between -6.17 and 9.41 and so encompassed the current confidence interval of 8.74 to -3.48. So, the assertion that the UK home-educated sample was, where confidence intervals were used, as passive as Shyers' home-educated children and as assertive as Michelson et al's school children, thus, supported.

9.1.2 Scores by Sub-category

Graph 9.1 provides graphical breakdowns of the UK home-educated children's scores across the five sub-domains when individuals' 'all value' (Shyers 1992) scores were summed[7]. The 5 graphs suggest, by the amount of bars in the negative zones, that overall, the children responded passively.

Graph 9. 1: individual scores across the five domains

The 'Positives' sub-category, shows that in situations involving an opportunity for positive expression, the larger proportion of the home-educated children would react with moderation; this can be seen in the high degree of 0 scores, 28 out of a possible 43. Asked about situations involving complaint ('Negatives') there were diverse reactions, ranging from very aggressive to very passive. The ability to make and accept requests was competent, as was the capacity to express their own and understand other’s, feelings. The ability to converse ('Conversations') tended towards passivity whilst remaining proficient, assuming that zero was equal to the optimum reaction.

In summary, the domain graphs show that the children coped best where there was an opportunity for positive expression whilst needing, perhaps, to improve their skills in making and receiving complaints.

9.1.3 Participants' Comments

In contrast with the present generally unassertive and passive home-educated sample, one mother of five made the comments quoted below about her school attending child. The two older children now at school had previously been home-educated throughout their primary years, whilst the three younger primary aged children were currently home-educated. The child referred to had completed a CABS questionnaire[8], as had his home-educated sibling. In analysis, the child's responses were aggressive yet unassertive, his overall assertiveness score being 28, where the group mean was 16.11 and the 'norm' 13 (Michelson and Wood 1981). His passive-aggressive score (Shyers 1992) was +6 where the group mean was -6.11. The relevance of the child's score was that he had, up until a year previously, been home-educated since birth. He wrote, in answer to some questions:

'Shut up' (4[9]) // 'Say OK, talk quiet and then get noisier' (13) // 'Say nothing and push them out the way' (14) // 'Chin them' (15)'

His mother wrote:

'Jak started school in October 1996. Some of Jak's answers would have been different if this had been done before he started school especially things like 'chin him'. Sometimes, the social side of school life is with children you wouldn't have anything to do with.'

Despite Jak's mother explaining his responses, analysis revealed his younger, home-educated brother, Ty, to have gained an assertiveness score of 20 and a passive-aggressive score of 10. Thus Ty, though a little more assertive than his elder sibling, was actually more aggressive. The difference between brothers appeared to be in their vocabulary rather than their actual responses. It was possible that Ty, who during interview had conveyed himself as gentle and considerate, wanted to be like his older brother, and so gave answers that he regarded as socially acceptable. Whilst the family had in the past lived an itinerant life, continuing to have traveller families to stop over on their land, they were, as far as could be ascertained, a gentle family.

However, the idea that schoolchildren were more aggressive was not that of Jak's mother alone; Mrs Bolan wrote of her home-educated son:

'George, unknown to me at the time, took the CABS paper to the sports club and said that his research showed that schoolchildren were more aggressive, choosing to call names and saying that if threatening to punch was an option, they would tick that'

In the context of the study at hand, test fallibility arose, not from a question of instrument robustness, nor even from questions over the different forms of analysis employed previously, but from the attitudes of the participants themselves. Some children went out of their way to provide answers that differed from those choices provided. A mother wrote:

'We have managed to complete your questionnaire. I must say, I did not like it, but have persevered by allowing our alternative 'f'. Some alternatives were so amusing (to Rick) that I think he chose them on that basis. I expect you'll be able to transpose Rick's answers into the choices offered. I valued his willingness to be involved with the project too much to confine him to the choices offered: indeed, his own ideas will surely tell you more about him than the restrictive choices offered (as I'm sure you appreciate). I know it’s extra work for you, but you will be able to utilise it better than conformist answers!'

Avoiding the a-e scale, one lad had answered two of the questions in his own words, as follows:

(4 [PR1] [10]) 'it depends on what I have forgot and who I am talking to'

(24) 'depends what the other person did'

In response to the researcher's request that the child provide answers within the 5 point answer scale, his mother commented:

'If you could ask the reason for giving the chosen answer you would get a much clearer picture.'

Analysis of each child’s response in detail including their rationale, had been discussed during consideration of the methodology to be used in assessing social skills. It would have been advantageous and illuminating to gather this additional data, but the practicality was that the CABS questionnaire had been chosen as a screening instrument, employed to identify a general pattern amongst a sample of home-educated children. It was not selected as a tool for identifying individual needs. However, as with the other measures used for this research, comments made by participants were noted for the light they shed on how the group as a whole reacted to the CABS questionnaire.

Rick was not the only participant who had found the questionnaire amusing. Several others did and this could, in part, be attributed to the use of American language throughout the questionnaire, e.g. the word 'dumb' was substituted to read 'stupid', following comments that some children might not understand the word 'dumb'. The other cause of amusement cited by those whose views were invited prior to administration, was that by inviting participants to say what they would do in situation 'X' the youngsters would take this further by jesting over the answers before writing down what they would actually do. Speaking of amusement one mother wrote:

'John was very interested and amused to do the questionnaire, so here are his replies:…'

Whilst another family, Jehovah's Witnesses, remarked:

'We laughed till we dropped'

One participant, seemingly pleased with himself, simply drew a picture:

Image 9. 1: one playful participant

A father spoke of the family's enjoyment with the questionnaire and with the way they had used it to explore the issues further:

'Just half an hour each on their own, asking intimate, trusting, non judgmental, searching questions was really fun. They - and I loved it. Far from very demanding it felt really nice - we could happily do one of these every day.'

The extent to which families did add their comments can be seen below, in Image 9.2.

Image 9.2: illustration of the extent to which parents and participants added comments to the questionnaires

One mother, writing about her child diagnosed with Semantic Pragmatic Disorder, wrote:

'I hope I haven't messed this up for you, but as it is not a reading test, I read the first three questions to Danny to explain what to do and later on, at his request, read some more. I did talk to Danny about being honest, but autistic children have very little, if any, understanding of this concept. In Danny's case this understanding is very literal and does not, to my knowledge, cover hypothetical situations. Danny was, in effect, saying what he thinks the right thing is, rather than what he would actually do. It is interesting that in some cases he would actually say out loud, 'I want c' or 'I want d'. Generally I would say Danny made himself slightly more passive and a lot more socially adept than he actually is.'

The idea of a 'right' answer was also referred to by another mother:

'My partner thinks 'social desirability' may have influenced both of them quite a bit.'

Overall, the results, concurring largely with both Michelson and Wood (1981) and Shyers (1992), suggested that the CABS portrayal of the group was a valid one.

9.1.4 Summary (CABS)

In conclusion, using the confidence interval for reference, the UK home-educated children were as passive as Shyer's (1992) home-educated sample and as assertive as Michelson et al.'s (1979) school sample. Thus it can be argued that the social skills of the home-educated sample represented here did not differ in any important respect from those reported in previous studies.

Overall, CABS appeared to be a robust test of social skills. From the 43 CABS assessed children, 29 had assisted with other areas of this research and aspects of them were known, beyond both the CABS measure and the initial questionnaire survey. Further, the researcher had the opportunity to observe and interact with 25 of the CABS participants in their home surrounding, during interview. It was therefore, possible to say, at least for those 25 children, that the analysis of their social skills according to the CABS concurred with the researcher's personal assessment of their social skills. The child-centred approach of home-education (Rothermel 1998) was perhaps, substantiated by the sample's difficulty with handling complaints and in their marginally higher than the normal level of unassertiveness. Such children may not need, whilst learning at home, to manage conflict to the same degree as schoolchildren do (Sluckin 1981).

9.2 Revised Rutter Scale (RRS) Results

9.2.1 Revised Rutter Parent Scale for School-Age Children

The total number of RRS child 'participants' was 42; 22 girls and 20 boys, aged from five to eleven-years-old with a mean age of 7.1 years. The RRS asks parent raters to respond on a 3 point scale to 50 statements that might apply to their child, such as, 'Cries easily'. The scores allotted are '0' for 'does not apply, '1' for 'applies somewhat' and '2' for 'certainly applies'. Specified items categorise into 5 groups, the main one being 'Total Difficulties', where the score range is 0-52. The other 4 categories and score ranges are, 'Emotional Difficulties' 0-12 (5 statements); 'Conduct Difficulties' 0-12 (5 statements); 'Hyperactivity/Inattention' 0-6 (3 statements); and 'Prosocial' 0-20 (10 statements). Parents reported on their child's behaviour during the previous three months.

Table 9.2 shows the mean scores for the children across the main 'Total Difficulties' and sub-score systems. The cut-off point for 'Total Difficulties' was 11, with the inference that any person with a higher score experienced behavioural difficulties. The home-educated sample mean of 8.38 was well below this level.

Table 9. 2: combined mean scores for the five domains of the rrs (n=42: age 5-11 years)

Domains	Total Difficulties	Prosocial	Conduct Difficulties	Emotional Difficulties	Hyperactivity /Inattention
Mean Scores	8.38	15.5	1.69	1.80	1.21
Max.Score Possible	52	20	12	12	6
Max. Scored	24	20	6	5	5
Stand. Dev.	5.17	3.24	1.63	1.68	1.38
Cut-off point	11	N/a	N/a	N/a	3
Participants above cut off	10	N/a	N/a	N/a	4

Ten home-educated children (23.80%) scored above 11. Where this happens, the area of specific difficulty is defined by the child's highest score in either the 'Emotional Difficulties' or 'Conduct Difficulties' domain (Elander and Rutter 1996). Four children scored above the 'Hyperactivity/Inattention' cut-off point of 3 (Elander and Rutter 1995) and of these, three had also scored over 11 points on the 'Total Difficulties' scale. Examples of difficulties defined by the RRS in each of the 3 difficulty domains are provided in Box 9.1.

Box 9.1 RRS difficulty domains

Domain of Difficulty	RRS Item
'Emotional Difficulties'	· Cries/ refuses to go to school (a zero score unless withdrawn) · Often worries · Often miserable, unhappy, tearful, distressed · Fearful/afraid in new situations · Often complains of aches & pains
'Conduct Difficulties'	· Tells lies · Bullies · Disobedient steals · Fights or quarrels with other children
'Hyperactivity/Inattention'	· Restless · Fidgets · Cannot settle

Graph 9.2 represents the grading of the children with 'Total Difficulties' scores exceeding 11 points, together with their scores in the 'Emotional Difficulties' and 'Conduct Difficulties' domains. The graph shows that for three children their difficulties lay in the 'Emotional Difficulties' domain, whilst for another four they lay in the area of 'Conduct Difficulties'. A further three children displayed undifferentiated behavioural difficulties.

Graph 9. 2: 'Total Difficulties' scores combined with specific difficulty scores (n=10)

Of the ten children represented in Graph 9.2 with a 'Total Difficulties' score over 11, six were from three families. Two of the children with 'Emotional Difficulties' had been withdrawn from school within the previous six months, whilst another child had been described by her parents as having a very high IQ and another had been diagnosed as having special needs within the autistic spectrum. All ten of these children had been interviewed for this research. Of the three families with two child participants each, one family was a one parent [father] family, the second lived a 'New Age', itinerant (inhabiting a variety of houses) lifestyle and the third family moved often owing to the father's work and he, in turn, was suffering from depression[11]. It was thus possible that the difficulties reported in the children may have been a response to the problems in the family.

Whist it was not surprising to find that the two children with SENs (autistic and gifted) were classed through the questionnaire as exhibiting difficult behaviour, it was unexpected to find the other 8 children so classified: their behaviour during interview had given no such indication; for example, two of the children from one family were extremely confident and articulate, but perhaps, as was noted during observation, overly so for their parent who had completed the questionnaire. This may have been the case for the other two families who interpreted both their children's behaviour as difficult, since, although to lesser degree, their children were also confident and articulate. The remaining two children, who had been withdrawn from school after experiencing bullying, were also lively and articulate. It was conceivable that the difficulties portrayed by their single parent father revealed more about his personal difficulties than about those of his children (see Sclare 1997, Section 4.8.1). For the remaining thirty-two children, with 'Total Difficulties' scores of less than 11, their scores in the other difficulty domains were generally low, in keeping with their low 'Total Difficulties' scores.

The 'Prosocial' mean score of 15.5 for the group indicated a strong degree of this characteristic amongst the sample. Summarised examples of 'Prosocial' behaviour were (Box 9.2):

Box 9.2 prosocial item examples

· Fair in games

· Considerate to others

· Helps someone who is hurt

· Volunteers at home

· Kind to younger children

· Comforts distressed child

· Tries to stop quarrels/fights

· Shares with friends

· Helps children who feel ill

· Kind to animals

Graph 9.3 illustrates the children's 'Prosocial' scores. The mean is represented by a thick red horizontal line. A dashed-red horizontal line is drawn 1.5 SD from the mean at 10.7[12] points. Three children (7.1%) fell short, scoring 7, 8 and 9 on the 'Prosocial' scale and 11, 24 and 16, respectively, on the 'Total Difficulties' rating (cut-off point 11). One of the three children had been diagnosed with Autism.

Graph 9. 3: prosocial scores[13] (n=42)

Following the above analysis as detailed by the scale's author, the data was compared to that of Rutter et al (1974) as shown by Table 9.3. The RRS question numbers relating to each item are provided in Appendix 9.2.

Table 9.3: comparisons in behaviour difficulties (in %), between two studies using rrs

Source	*Rutter et al.* (1974)**		Rothermel (2001) (this paper)
Year	1970		1998
Place	England		United Kingdom
Informant	Parents		Parents
Age group (years)	10-12		5-11
Population	British		British
Sex	Boys	Girls	Boys	Girls
Number	1564	1500	20	22
Aggressiveness	15.2	5.3	20	22.7
Anxiety	35.4	39.2	25	40.9
Hyperactivity	32.0	25.5	20	18.1
Lying	16.1	9.7	15	0
Nail-biting	28.8	33.1	10	4.5
Phobias	25.2	24.8	0	0
Stuttering	3.5	1.5	5	4.5
Theft	5.7	2.6	10	4.5
Thumb-sucking	6.2	13.0	10	9
Tics, twitches	5.9	2.9	0	0

Table 9. 3 shows that the home-educated sample demonstrated more signs of aggressive behaviours than the schoolchildren from the Rutter et al study, particularly for home-educated girls where aggressiveness was at 22.7% as opposed to 5.3% for Rutter's girls. This contradicts the CABS data whereby the UK home-educated sample were judged as passive. Table 9.3 also indicates that whilst the home-educated boys were less anxious than their school counterparts, the girls, home and school, were similarly anxious. However, the home-educated sample did not suffer from tics or twitches. Theft amongst the home-educated boys was substantially higher than for the schoolchildren. 'Hyperactivity', 'Lying' and 'Nail-biting' were far less common amongst the home-educated sample than Rutter et al's schoolchildren. The frequency of lying amongst the home-educated girls was 0% contrasting with 9.7% amongst the schoolgirls. Phobias had been expressed by the RRS item relating to a child's fearfulness at new things or new situations and it was particularly notable that whilst 'phobias' were rated zero amongst home-educated children in the sample they had been particularly prominent amongst the school sample. Overall 23.8% of the home educated sample had been above the RRS 'Total Difficulties' score cut-off point and a further 4.7% on it. The norm according to Sclare (1997) was between 10-25%.

9.2.2 Participants' Comments

Some RRS participants wrote their comments on the returned papers, or, in one case, emailed their reaction. These commentaries provided a useful backdrop to the families' approaches in terms of psychological health. For example, there was plenty of evidence of the acceptance within families of imaginary companions. When responding to the item, 'Blames others' Jane Woolas, indicating the presence of imaginary friends, had replied:

'The Nungs!'

The theme of imaginary friends emerged often during this research and will be discussed further in the post-thesis Interview report.

The mother of Jane Moore, aged five-and-a-half-years-old, made several comments. Asked whether Jane was much liked by others she replied:

'I sometimes see children who are very 'conventional' in their upbringing shying away from Jane's openness. Generally she is very popular.'

And asked whether Jane helped out at home:

'I expect very little of this - whatever comes is freely volunteered'

The statement, 'tends to be on own, rather solitary', was considered one of the 'Total Difficulties' items. In response to this item, Mrs Moore said:

'Plays loads and loads with other children. Very occasionally wants to be alone and gets that space easily, then goes back to playing intensively'

Answering the item, 'Is often disobedient' she wrote:

'Doesn't get corrected'

On the subject of theft within the last three months, Mrs Moore replied:

'Only one occasion'

And on her daughter's attempt to prevent quarrels or fights, if at all:

'More in the role of 'defending' younger children - if she knows them well. Generally keeps well clear of conflict.'

Referring to tearfulness or fear at new situations, she replied

'No more than I see as healthy'

Final general comments from Mrs Rook and Mrs Moore respectively were:

'This would have been much easier using a continuum rather than 3 discrete categories'

'Sitting at the kitchen table filling this form in - I've found it confusing at times. Gary suggested notating it. Hope that's useful. If not - ignore it. The other form was easy!'

A mother, whose children scored above the 'Total Difficulties' cut-off point, commented:

'I hope this is useful. I shall look up Rutter and find out what it is all about. The questions are very hard 'Often tells lies' What sort of question is that? If I say DNA[14] he would be a paragon. If I say CA he would be unliveable with, so it has to be AS. I suppose it is designed to make us make awkward choices. And yes, I recognise that some of the questions seem to come up with contradictory answers!'

What all these comments indicate is the complexity involved with placing children's behaviour onto a three point scale, or indeed, any scale. Whilst the scale results provide a clear-cut portrayal of the group's behaviour, the comments highlight the complications involved, on an individual basis. Normally, such rating scales might be both parent and teacher completed so that comparisons can be made. However, with home-educated children no such corroboration of behaviour, as determined by those familiar with the child on a daily basis, can be made.

9.2.3 Summary (RRS)

The percentage of children identified as exhibiting a behavioural difficulty of some kind, was at 23.8%, marginally higher than the 22.3% prevalence found amongst a pre-school sample by Thompson et al. (1996); more than the 10% of ten to eleven-year-olds said by Rutter et al. (1970) to have shown signs of behavioural problems and greater also than the 18% found by Minde in 1977 (cited by Elander and Rutter I996) to have behavioural problems. All these samples were apparently 'normal' and the problems identified ranged from mild to severe. Rutter et al. (1975), however, with their inner city London sample of children from West Indian backgrounds, found at 25%, a prevalence exceeding that of this study. Nevertheless, just over 76% of the home-educated sample were rated as normal and on the 'Prosocial' scale, 93% scored as 'normal'. Overall, it was found that the home-educated sample exhibited far fewer behavioural problems than those represented in Rutter et al.'s (1974) London sample. Placing aside the 4.7% of home-educated children who were borderline cases, then 71.43% of the sample demonstrated their behaviour, according to the RRS, to be within accepted parameters. Parents generally found it difficult to categorise their children's behaviour and because the children were home-educated there was no way of cross checking outcomes with teacher ratings as there might have been had the children been in school. Previous literature however, suggests a low parent to teacher rating overlap (Rutter, Tizard and Whitmore 1970).

9.3 Strengths & Difficulties Questionnaire (SDQ) Results

9.3.1 SDQ Overview

The 25 SDQ items are divisible into six categories. The main category is 'Total Difficulties', requiring summation of all the 'difficulty' items. There are four 'difficulty' sub-categories; 'Emotional Symptoms Scale'; 'Conduct Problems Scale'; 'Hyperactive Scale'; 'Peer Problems Scale'; and one further sub-category giving the 'Prosocial Behaviour Score'. Earlier research, Goodman (1997) identified the percentages of children expected to fall into each of three score bands, 'normal', 'borderline' and 'abnormal'[15]. Table 9.4 provides the percentages of the home-educated children whose behaviour, as interpreted by their mothers, placed them into each of these three score bands. Underneath the three score band headings are the percentages of children who would normally be expected to be within each score band. Where the percentage of home-educated children was larger than expected, the amount has been highlighted in a bold font.

Table 9. 4: percentages of individuals in each score category when viewed by domain

(n=41)	Total Difficulties	Prosocial	Emotional	Hyperactive	Conduct	Peer Problems
Frequency	%	%	%	%	%	%
Normal (80%)	92.7	19.5	95.1	85.4	97.2	70.7
Borderline (10%)	4.9	19.5	2.4	7.3	7.3	7.3
Abnormal (10%)	2.4	61	2.4	7.3	0	22
Totals	100	100	100	100	100	100

As Table 9.4 illustrates, in marked contrast to the RRS results[16], the home-educated children here emerged as mostly 'Abnormal' in terms of their 'Prosocial Behaviour'. The related items were about trying to be nice to others, sharing, being helpful when someone is ill, kind to little ones and offering to help others. Abnormal here would have indicated less willingness in the these behaviours. Whilst the RRS had contained 10 prosocial items, the SDQ used only 5 and this, together with the type of items chosen, may have accounted for the difference. In terms of the 'Peer Problems Scale', although 70.7% of the children fell into the 'normal' bracket, 22% were termed 'Abnormal'.

In terms of reliability the test identified one child with Asperger's Syndrome[17] as having 'abnormal' behaviour across the board, whilst showing as 'abnormal' three children who, by their parent's estimation, experienced no behavioural difficulties. Moreover, the only fully autistic child[18] in the sample produced a 'normal' score pattern similar to many of the children without SEN.

Whether the assessment was not as efficient as had been reported earlier (Goodman 1997), or whether the sample in question were in some way a 'special' sample, was impossible to determine without further evaluation of a kind not within the mandate of this research. Overall however, the assessment was useful in providing information that supplemented other data presented throughout this thesis.

Advantage was taken of an offer by two parents from one family to independently evaluate the behaviour of their three children[19]. Their data provided an indicator of the level of agreement that might be expected between parents. There was agreement in all areas other than 'Conduct Difficulties' where the mother placed two children into the 'normal' and one into the 'borderline' band, whilst the father placed all three in the 'normal' bracket. There was agreement in the 'Prosocial' (abnormal), 'Emotional Difficulties' (normal), 'Hyperactivity' (2 'normal', 1 'borderline') and 'Peer Problems' ('normal') categories.

Seven home-educated children, aged 11 years, were evaluated by their mothers and also by themselves[20]. The results are provided in Table 9.5.

Table 9. 5 mother reported versus self-reported (n=7)

Total Difficulties

Prosocial

Emotional

Hyperact-ivity

Conduct

Peer Problems

% moth

% self

% moth

self

% moth

self

% moth

self

% moth

self

moth

self

Normal (80%)

85.7

85.7

28.6

42.9

85.7

100

85.7

100

85.7

100

71.4

14.3

Border (10%)

14.3

14.3

28.6

14.3

85.7

Abnor. (10%)

42.9

57.1

14.3

Table 9.5 shows consensus only in the 'Total Difficulties' domain. Most notable of the differences was that represented by the 'Peer Problems' domain where the mothers, in terms of the SDQ rating, perceived their children to be more 'normal' than the children in fact felt themselves to be. In the 'Emotional Difficulties', 'Hyperactivity' and 'Conduct Difficulties' domains it was the children who believed themselves more 'normal' than their mothers had reported them to be. This finding was supported by the conclusion of Ekblad (1990) that mothers were critical of their children's behaviours.

9.3.2 Comparison between the RRS and the SDQ

Copies of both instruments appear in Appendices 4.14 (RRS), 4.15 (SDQ) and 4.18 (comparison of RRS and SDQ items). A sample of twelve children participated in both the RRS and SDQ, four of these also having completed a CABS questionnaire.

Amongst eleven of the twelve children there were no 'Total Difficulties' scores over the cut-off point on either the RRS or the SDQ. One boy's RRS score, however, fell upon the RRS cut-off point, whilst his SDQ score lay within the borderline category. The one child identified by the SDQ as having behavioural problems was also one of those identified as such by the RRS. Therefore, 'Total Difficulties' interscale reliability, in respect of these twelve children, was 100%.

A review of the RRS results in the 'Prosocial' area, however, revealed that only two of the ten children with a good 'Prosocial' score[21] on the RRS had scored as 'normal' on the SDQ. Five children were deemed 'abnormal' on the SDQ whilst having achieved good to high 'Prosocial' scores on the RRS. Just one child whose score was 'abnormal' on the SDQ had scored more than 1.5 SD's below the mean on the RRS. It appeared, therefore, that in the 'Prosocial' domain, interscale agreement may have existed in only three cases out of twelve. The SDQ identified a further four 'borderline' cases, however, and these could not be compared with the RRS where no such band had been calculated. Further, an analogy between scales over the SDQ 'Peer Problems Scale' could not be drawn since this domain was not identified by the RRS.

Apart from there being fewer statements to consider throughout the SDQ than the RRS, some of the items may have triggered different responses, discussed below, to those expected by the scale's author. This being so, the reduction in items could have adversely affected the outcome of this scale. The concept of sharing for example, cited in the 'Prosocial Scale' was an issue raised often during the interviews, many parents viewing sharing as maturational rather than taught. Home-education it seemed, created an environment whereby such development could unfold 'in its own time', thus it was found that many of the children did not share their toys until far later than one might normally see in children who attend nursery or school. The same theory was occasionally voiced when speaking of the amount of assistance expected of the children; a comment to this effect was given in the section on the Revised Rutter Scale (see p. 23). This was not to say that all families followed such a philosophy, but the inclusion of those who did, may have led to the lower scores for these items than was anticipated by Goodman (1997).

The 'Peer Problems Scale' reinforced other issues that had emerged during the interviews. For many home-educating families, the child that was 'solitary' and tended 'to play alone' was a desirable child: having one good friend may not have been considered essential, particularly in larger families: being 'liked by other children' was not particularly important for many families, particularly where there were a number of siblings. Finally, for many home-educated children, more so, perhaps, for single children, getting on 'better with adults than with other children' was an everyday reality and was often viewed by families as an asset enabling their children, unlike schoolchildren they knew, to exude confidence in all types of company.

9.3.3 Further analysis of the SDQ

Graph 9.4 provides information about how the group answered each SDQ question[22]. The method used here to analyse the data is contentious and serves only to illustrate how the results appeared when, instead of taking the child as the unit of analysis, the group are used. The bar chart values refer to the group's overall score out of a possible 100 marks for each question; thus, for item 25[23] 'Gets on better with adults than with other children' the group scored 20.7 marks. Boxes to the right of the bars note the banding values for each domain[24]; the 20.7 for item 25, fell therefore, within the 'Peer Problems Scale' 0-30 'normal' banding.

Graph 9. 4: SDQ scores given as marks out of 100 with each item, for the whole sample combined (n=41)

What is interesting about Graph 9.4 is that, unlike Table 9.4 above, it shows the children as a group, to have fallen within the 'normal' band for each domain. Table 9.4 represented the children as measured individually, so that where a participant had scored, for example, 'somewhat true' for all five items, their score would have been 5 points and therefore, 'borderline'. However, if those marks were then combined with those for all the group's responses to that question, whereby a number of participants may have scored between 6 and 10 points, this 'borderline' response would have been absorbed into the group's answers, as a whole. Since the SDQ was chosen for its value as a screening instrument and also because of the interest in the group's inclusive responses, the system of evaluation used in Graph 9.4 is substantiated.

The score for item 25, 'Gets on better with adults than with other children', though within the 'normal' banding was relatively high, as was item 21, 'Rather solitary, tends to play alone'. As indicated above, this was, perhaps, to be expected, given the sample. The same might also be argued for item 17, 'Generally obedient', where the group score lay only marginally within the 'normal' band: some families might have believed that by following their child's lead, the need for obedience was reduced in proportion to the corresponding lack of disobedience. One parent during interview had remarked, when asked about punishment:

'We do not order our child to do anything, therefore, he cannot be found wanting in obedience.'

A final comment on the SDQ came from a parent whose two children had participated, one with the RRS and one with the SDQ:

'Here is Janie's life on the line. Perhaps it's a good idea that the questionnaires are different so I don't go about comparing them, but I do any way. I've tried to be honest but I wonder how much painting over the cracks goes on, not just for me but for everyone talking about their children.'

9.3.4 Summary (SDQ)

Although the percentage of 'Total Difficulties' identified by the SDQ and the RRS, differed substantially overall, there was 100% agreement in the cross-scale sample of twelve children. In the 'Prosocial' domain, the SDQ differed significantly, both overall and within the cross-scale sample from the RRS. Socially, the SDQ found 61% of the home-educated children to exhibit 'abnormal' social behaviour, whilst only 7.1% of the RRS sample were thus identified. The SDQ 'Prosocial' and 'Peer Problems Scale' portrayal of home-educated children as lacking socially, very much reflected some of the views provided in Chapter 3 Section 3.4.2 (i.e. Shearer 1999; Hastings 1998) and it was easy then, using scales such as the SDQ, to understand how such sentiments could arise, the point being, that diagnosis is definition dependent.

The home-educated sample had been assessed as 'normal' by CABS analysis of their social skills where such skills have been interpreted by Michelson et al. (1983) as encompassing social behaviour, assertiveness and social competence. However, the Rutter and Goodman instruments (the Goodman questionnaire in particular) appear to take a different approach to social behaviour in that their focus is more on socially acceptable behaviour, ie. behaviour accepted as the 'norm' by school-going society, and this becomes clear through the items included in these instruments. Whilst the Rutter Scale items are broad in scope, the shorter Goodman questionnaire offers, it appears, brevity at the expense of diversity. Judging from the interview data, a number of the Peer Problems and Prosocial items were, for the home-educating families, controversial insofar as the way in which the scores were awarded. The values of the home-educating families differed from those generally expected of school-educating families, where concepts such as sharing, playing in groups and being popular, are vital to the school ethos. Nevertheless, according to the SDQ, whilst home-educated children appeared 'abnormal' in some categories, as a group they rated as 'normal' and actually made up a larger 'normal' sample than Goodman would have expected (92.7% as opposed to the expected 80%).

Overall, the number of participants taking these social and psychological assessments was small. The three instruments provided for analysis from three 'points' (see Section 4.3) and as such assisted in providing a general view of the cohort. Whilst the RRS and SDQ were parent-rated and focused on children aged 4 to 11 years, the CABS centred on 8-10 year olds and was child-rated.

Social and Psychological Endnotes