Chapter 4: Methodology

4.1 INTRODUCTION

This chapter describes the multimethod approach used for this thesis. Five chapters make up the results.

· Initial Questionnaire (Chapter 5)

· PIPS Baseline Assessment (Chapter 6)

· Literacy (Chapter 7)

· Mathematics (Chapter 8)

· Social and Psychological Data (Chapter 9)

The methodology is described using the following general format:

· Choice of Methodology

· Design

· Materials

· Participant Selection

· Procedure

4.2 Overall Research Design

The research took the form of access to the home-educators, a questionnaire survey distributed to them, educational and psychological assessments of home-educated children and interviews with home-educating families. Questionnaires were analysed from 419 respondents[1] and 238 assessments were conducted. Whilst 100 families were interviewed, the results are not included in this research: analysis of interview data is taking place as part of a post-doctoral follow-up addendum report. This decision not to include the interview chapter was taken because of the limitations in terms of the time and word length of this PhD. However, field-notes from these interviews are referred to because of the qualitative background data that they add to the assessment programme and the PhD overall.

Below is an overview of the methodology.

Survey Data:

Educational Data:

Psychological Data:

And part of the overall research design but not included in this thesis: Interview Data: N=100 home-educating families.

4.3 Methodology Overview

This research used a multimethod approach that facilitated the quality and quantity of data necessary in order to gain a comprehensive portrayal of home-education in the UK. It enabled conclusions to be drawn through reference to a multiplicity of sources, methods and theories (Denzin 1989). It was also anticipated that by using different methods, interpretability would be enhanced whilst threats to validity[2] were kept to a minimum (Robson 1993).

In a discussion about multiple methodologies, Burgess (1982) referred to the process of ‘research triangulation’. This suggests that a field researcher must collect many sets of data, relating to the different phases of the research, the different settings and the different participants. This approach has drawn criticism from Gans (1991) for the vast amount of unmanageable data that can accrue. The quantity of data collected for this research was indeed vast and although it remained manageable, it was nevertheless, laborious to co-ordinate.

Guba and Lincoln (1985) describing elements that they considered vital to research, cited trustworthiness, transferability, dependability, confirmability, and internal[3] and external[4] validity. The trustworthiness of the research lies in the ability to compare and contrast data, and was enabled though the multimethod approach used. The research relied heavily upon tried and tested assessments, the validity of which had been established prior to this research: these assessments were combined with the voices of the home-educators to create a portrayal that was both subjective and objective. It was reasonable, from the sample size, to believe that generalisations could be made, although it was difficult to make any assumptions in the absence of definite knowledge about home-educators that remained unknown to any survey, authority or organisation. The attributes of the group studied were particular to themselves, therefore representativeness was not guaranteed. This research, most probably, could not be replicated, since the study was a product of its time: with the perceived advent of compulsory registration[5], many home-educators are increasingly less likely to accept such an intrusion into their privacy. Home-educators have become empowered: when the research began, there was just one home-educating organisation in the UK; six years on[6], there are many; the press cuttings were few then; now there are stories about home-educators almost weekly in the press. The advent of computer email lists for home-educators has created a vocal forum whereby home-educators have become a lobbying force[7].

All participants' real names have been altered. Pseudonyms have also been inserted into quotations, replacing any real names that had originally been used. Where scanned images appear to carry names, these too, have been digitally altered to pseudonyms.

4.4 Access to Home-Educators

As a result of reading several newspaper and magazine articles published during 1996 (Henson 1996; Midgley 1996), evidence of a British home-education movement became increasingly apparent. Reports referred to a national network of home-educators and suggested that the incidence of home-educating families had grown over recent years. A further literature search revealed that Meighan (1995) had presented a case arguing the effectiveness of home-education, while Knox (1988) had raised awareness of school phobia and Holt (1981) reflected a growing apprehension towards the school system. Petrie (1992) and Lowden (1993) documented the tension that existed between home-educators and their Local Education Authorities (LEAs). Almost a year of preliminary investigation prior to ‘official’ instigation of the present study showed that despite the work of the above mentioned writers, the available research was minimal, particularly in the UK where so little was actually known about home-educators. Access to home-educators was thus made gradually over a two year period^[8], through:

· Conversations with home-educating families.

· Observations (participant) of home-educators at local and national meetings, and at organised activities

· Interviews with LEAs

Initial contact was made by joining the national home-education organisation Education Otherwise ('EO'); this allowed the researcher to attend local and national meetings, besides possession of a contact list of members^[9]. Home-educating families were approached through support organisations, newsletters, curriculum suppliers and LEAs; other families were approached individually by the researcher, as and when such opportunity arose. Attendance at a national meeting of the 'EO' ruling council provided an opportunity to request that the initial survey questionnaire be inserted into the 'EO' newsletter. 'EO' membership and the questionnaire distribution initiated far wider access to home-educators. A splinter group from 'EO' formed the organisation Home Education Advisory Service (HEAS) and subscription to their newsletter also brought opportunities to gain access to home-educators. Observations at local and national meetings of home-educators enhanced the researcher's understanding of home-educators at grass roots level and gave broad access to many types of home-educator. Over the period of this research, such groups, initially set up from 'EO' contacts, have tended to take on their own identities, accepting home-educators with and without affiliations to any organisations and distributing their own newsletters. There are now many such groups throughout the UK. In a similar way Schoolhouse Education Association (SHEA) was started as a result of home-educators in Scotland realising that they needed their own national organisation in a country where the law on home-education differed slightly from that in England. Membership of SHEA^[10] and attendance at their Dundee conference permitted the researcher to network in Scotland. Contacts made through religious organisations further widened researcher awareness.

Once it became clear that many home-educators did not affiliate themselves to support organisations, LEAs were contacted and interviewed, by telephone, about their known home-education populations. The LEA experience and their knowledge of home-educators was often radically different from that gleaned through the support organisations; thus, contact with LEAs enabled a very different section of the home-education 'community' to be approached.

Networking in this way, home-educators became increasingly accessible. Access was nevertheless, a very long process that engendered many hours of research and time spent in establishing trust.

4.4.1 Ethics

A detailed discussion of ethical issues appears at Appendix 4.1. Many ethical issues however, are discussed in the text as they arise.

4.5 METHODOLOGY ADOPTED FOR THE Initial Questionnaire

4.5.1 Choice of Methodology (Chapter 5)

Decision to use a questionnaire

The overall research methodology was designed as a graduated sequence, moving from the collation of general background material towards family interviews and specific assessment based data. A questionnaire survey format met the research requirement: it permitted the solicitation of a broad spectrum of information from home-educating respondents, about whom little was known. For this reason too, open questions were considered to be more user friendly than a Likert scale. Furthermore, before the assessment and interview programme could proceed, it was necessary to have a pool of potential participants. The assessments needed to be conducted according to a strict timetable and so speed of delivery and response were also issues considered in the decision to adopt a questionnaire format for this initial stage of the research.

Bryman (1989, cited by Robson 1993)) believed that surveys provided a ‘snapshot’ of solicited views and practices whilst facilitating a focused content: both objectives met through the current research questionnaire survey. Moreover, the style adopted was one recommended by Mayberry, Knowles, Ray and Marlow (1995), who argued that a questionnaire survey should be designed to enable analysis, both quantitatively and qualitatively, whereby generalisations, trends and opinions could be established.

A questionnaire format would also provide the researcher with a tool for the gathering of superficial information on families who were subsequently to be interviewed, thus facilitating the interview time to be usefully employed in exploring pertinent issues.

Argument against using a questionnaire

A number of home-educators encountered informally expressed their view that they were too individual to be asked the same questions, an opinion apparently supported by Webb (1990). Endorsement of this was expressed in Robson (1993 p. 191), namely that representation of respondent’s idiosyncrasies and individuality might not be served or emphasised particularly well by use of a questionnaire. Meighan (1995) used individual differences between home-educators as an argument against using quantitative data, preferring to present examples case by case as social studies. However, this research attempted only to portray a general picture of home-education. It is because of the diverse nature of home-educating families that case study research, whilst illuminative, is only ever the story of those families and there was no intention here to highlight each individual family's practice. The questionnaire survey data, qualitative in nature, yet quantifiable, could for example, assist in identifying universal features, beyond the scope of an ethnographic instrument. Meighan (1995) had suggested that at any time ‘EO’ draws at least a third of its membership from teachers^[11] and based upon her study of twenty families, Webb (1990) stated that ‘almost all’ home-educators have been influenced by John Holt. Clearly, quantifiable data was necessary in order to substantiate these claims.

Lowden (1993) stated that his decision not to send out questionnaires to home-educators was based on the predicted poor rate of returns, an ‘EO’ questionnaire in 1988 having received a 30% return rate and Grant having experienced a 20% rate of return in 1983. Mayberry et al. (1995) defended their own questionnaire response rate of 25% as 'excellent', arguing that where a population is apprehensive of official intervention (as they might interpret it), a very low return is to be expected. However, they stressed that since so little was known about home-educators, there was every chance that the returns received could make a valuable contribution towards understanding at least some of the people whose pursuits would otherwise, remain non-quantifiable phenomena. Lowden used interviews to explore home-education but appeared to have made little use of the data from these meetings and there was scant description of the details involved in them. He therefore relied heavily on the LEA returns and on his own review of the law^[12] as it related to home-education.

Lowden (1994) criticised questionnaire surveys of home-educators because of the poor return rate that resulted in non-representation. However, in home-education studies, returns of 20 - 30% have been considered valid, particularly when the information from a collection of studies is assembled together (Van Galen and Pitman, 1991).

In response to anticipated criticism that questionnaire respondents might exhibit ‘social responsibility response bias’ (Robson 1993), that is, they may say they do one thing and in practice do another, it was conceded that while such a possibility existed, precautions were taken to limit it; anonymity was invited and the questionnaire was designed not to include any leading questions. Without knowing who the sample would be, it was not easy to anticipate in advance just what effect ‘social responsibility response bias’ could have on the study. It appeared reasonable to consider this issue retrospectively, upon analysis of the questionnaires. At that stage hints of ‘political correctness’ and ‘social responsibility response bias’ might show themselves more obviously. This might be detected through, for example, repetition of certain phrases from a variety of respondents, indicative perhaps, of social bias and influence.

4.5.2 Design of the Questionnaire (chapter 5)

Construction of the early versions

The initial questionnaire contained 80 questions that related to many aspects of the family such as income, political belief, religion. The original questionnaire was later amended to contain 50 questions and finally shortened to 35 items (attached at Appendix 4.2). With the assistance of Ms Mills^[13] and the 'EO' newsletter’s Co-ordinator, many of the questions were rephrased so as not to alienate potential respondents. The Co-ordinator explained that many home-educators did not ‘send’ their children anywhere and neither did they ‘teach’ them: children from these families tended to ‘go’ to wherever and to ‘learn’ rather than be ‘taught’: therefore questions such as,

'Might you send your children to school in the future?'

needed to read:

'Might your children go to school in the future?'.

The significance of words used in questionnaires has been discussed in Robson (1993). The initial stages of modification involved a general softening of the overall tone of the instrument to include phrases relating to feelings: the earlier version had invited objective answers using straightforward direct phraseology but this, the researcher was advised, was inappropriate for many home-educators. The change in tone was justified on the basis that only the more formal home-educators would respond to the initial version, whilst both they and their informal peers would be attracted to the more subjective approach.

Following earlier amendments Dr Petrie^[14] and Professor Aubrey^[15] assisted with final modifications. It was anticipated that the questions used in the administered version would resonate positively with respondents, whose profile^[16] was taken into consideration. The questionnaire included questions covering family background data, their motivation to home-educate, the significance that home-education had for the family, the structure adopted for the children's education, the materials used and the parent's attitude towards socialisation. There was also a question relating to the family's relationship with their LEA.

The questionnaire was piloted^[17] with three home-educating families and three non home-educating families, and input was also received from various professionals. A larger scale pilot was considered unwise at the time owing to the complexities in accessing home-educators. Robson (1993) explained that where interviews were concerned there can be justification for not running a pilot, owing to the effort necessary in building up a relationship of trust and acceptance with interviewee participants. This same rationalisation applied to the study at hand. Time was taken to gain access to possible respondents and acceptance within ‘EO’ to permit the questionnaire to be distributed amongst their members. In consideration of the inaccessible nature of home-educators in the UK, the involvement of potential questionnaire respondents with a widespread pilot version might have limited even more the small numbers who were thought likely to respond to the questionnaire. It was important that the maximum number of respondents were 'naïve', unlike those referred to in Webb (1990), where the sample of twenty participant families had included five families who had already written books about their experiences. Additionally, its distribution as a centrefold to the newsletter and its association with ‘EO’ might have jeopardised the study by creating a situation whereby members of ‘EO’ could have discussed and questioned the merits of the questionnaire. Contact with both ‘EO’ and non ‘EO’ home-educators strongly indicated that the families involved were sensitive to any move that might be deemed ‘official’^[18]. Minor alterations were made to the original questionnaire distributed through 'EO' in 1997, as highlighted in the following section.

Subsequent amendments

After the initial distribution of 2,500 questionnaires through the 'EO' newsletter, a sample of 50 questionnaires was analysed and a small study produced, thus creating a pilot analysis (as opposed to a pilot distribution). Having been led to believe that home-educators were sometimes wary of any enquiry into their decision to home-educate, many personal questions such as occupation and religion had been excluded^[19]. It soon became evident through the sample analysis, that home-educators would respond to questions on occupation and religion and therefore related questions were adopted into a modified version of the questionnaire. The wording on some questions was also altered: there had been one question on the meaning of home-education and another on the family's motivation: these were merged as it became clear that the responses could be separated at the point of analysis, whilst the space left on the questionnaire allowed for extra questions relating to occupation, religion and parenting styles.

The 34 question version had included leading examples (highlighted here in bold font) such as:

What was/is your family’s motivation to home educate? e.g., ideology, bullying, special needs, question of faith, morality in society, school phobia, always intended to do so……

These were removed and as with the merging of the 'motivation' and 'meaning' questions, simplified to read, for example:

'What does home education mean for your family and what motivated the decision to home educate?'

Other modifications included:

'How do you feel about the idea that children educated at home miss out on the opportunity children at school have, to socialise with their peers?'

This was reduced to:

'What is your view on child socialisation?'

The questionnaires analysed initially were absorbed into the main sample, together with all the questionnaire responses that had been returned by that time. The difference in questionnaire formats between the earlier and later version was small but effective. Since the research exercise was to explore home-education where little previous UK research existed, the decision to adapt the questionnaire seemed an appropriate reaction to an increased understanding of the situation.

The modified questionnaire contained 35 open questions. Below each question was a space of between 3 and 6 typewriter carriage returns in anticipation that respondents would write their answers in that area. The questionnaire was designed on 4 sides of A4 paper and reduced to fit entirely on both sides of one A4 sheet of paper. Modifications continued to be made to subsequent editions but these were generally the result of space limitations for the different distribution formats used.

In an attempt to eliminate the exclusion of families who objected to or found difficulty with, questionnaire formats, potential participants were invited to use plain paper and describe why they home-educated and what it meant to their family. Families were asked to provide their names and addresses if they were willing to participate further or be contacted for any reason; thus families wishing to remain anonymous were not deterred from replying.

4.5.3 Sample Selection (Chapter 5)

Targets for the questionnaires were home-educating families. A participant for the purposes of this study, was defined as either a family or a respondent.

Initially the questionnaire was sent out through the ‘EO’ newsletters and it was assumed that most respondents would be members of EO: thus a defining characteristic of that group would be that they were probably associated with a membership organisation. During the latter part of this study’s first year and the whole of its second year, the questionnaire distribution broadened to take in a further batch of 'EO' members and families contacted through other organisations^[20], word of mouth and LEAs^[21]. In some cases it was possible to speculate about the type of potential respondent, such as in the case where a number of questionnaires were distributed though a home-educating mother’s Christian distribution business and it was thus reasonable to assume that the participants would themselves be Christians. Where LEAs assisted with distribution, there was diversity amongst families although the more ‘wary’ families tended to be less common in this group, having decided to avoid becoming known to the LEA.

Reference to previous studies, (e.g. Webb, 1990; Meighan, 1995; Mayberry et al., 1995) and attendance at home-education meetings indicated that subjects would be well educated and literate, possibly environmentally attuned and possibly religious. No further assumptions could be made.

The sample was considered representative of many home-educators throughout the UK. Discussions between this writer, home-educating non-‘EO’ members and the ‘EO’ database secretary, indicated that home-educating families often joined ‘EO’ for support when they began home-educating their children, but chose not to renew membership once they had used that connection to make contact with other families and local groups. This use of ‘EO’ was evidenced by membership turnover^[22], where families were often believed to continue home-educating after lapsing their membership of ‘EO’. Therefore, it could be postulated that many home-educators, other than those who were allied with specific 'closed' religious groups, were or had once been, represented by the ‘EO’ membership. There were no parameters set for respondents in relation to education, age, ethnicity, nationality, or location.

Figure 4.1 shows the methods of distribution and the numbers of analysed returns. The first year brought 279 returns (all analysed) and second year many more than anticipated (140 of which were analysed[23]). Precision over numbers 'in' per year was not possible, bearing in mind there was no clear cut line between years one and two distribution and returns, or between the second and subsequent years. Questionnaires 'dribbled' in almost continually and were still arriving in March 2002 by post and electronically, despite not having been distributed since 1998! The high number of returns in the second year (over 500) appeared to be the result of both the expansion in distribution methods and because home-educators who had seen but not completed the questionnaire in year 1, did so in year 2. The decision to analyse the rather odd number of 419 came about because the intention to analyse all the questionnaires was thwarted after 419 analysed questionnaire data analysis sheets (sample at Appendix 4.3) were sent off to the data entry team at University of Newcastle with an intention to send on the remainder once they were analysed. The invitation to have the hand-written coded data entered onto computer at Newcastle came from Professor Tymms, Director of the PIPS Project. However, this researcher, realising that time was 'running out', and setting a limit of 500 questionnaires to be analysed in total, decided to enter the remaining data herself (81 questionnaires) - a task that itself became over complex owing to personal stresses at that time. Thus the decision was taken to keep with the 419 questionnaires whose data had been entered professionally. At the time this appeared to be the most responsible and appropriate way forward.

During the period of distribution just 3 families responded twice to the questionnaire whilst a further family responded with a letter during year 1 and a questionnaire in year two. Where the second response complemented the initial reply, the data were combined.

Figure 4. 1: methods of distribution 1997-98 and number of returns

Contact	Type of Organisation	Distribution	Number Out	Anal-ysed	Expected Characteristics
'EO' (year 1) 1997	Paid Membership	Inserted in newsletter	2,500	168	'EO' members - Diverse UK Sample
'EO' (year 2) 1998	Paid Membership	Inserted in newsletter	2,500	131	'EO' members - Diverse UK Sample
LEAs (year 1)	Local Government	Distributed by post or hand	Approx. 85+	31	Diverse Sample that may exclude families wary of intervention
SHEA	Membership with or without a donation	Distributed by hand and post	25	6	Scottish Families
CompuServe bulletin board (year 1)	Computer Server	Requested through the Internet(on the Web Site)	1	4	Access to the Internet. CompuServe subscribers World wide
'EO' Website	World Access Internet site	Invitation on-line to Internet site visitors	1	28	Access to the Internet
HEAS	Subscription to quarterly Bulletin	Requested by letter in Bulletin	12	5	UK. More likely families who have removed children from school^[24]
Christian organisations	Subscription and membership	Requested/distributed by hand and email	Unknown	7	UK Christians
Local organisations and groups	Informal groups	Requested & distributed, by hand and post	100+	17	UK Member of or contributors to those organisations or groups
HERALD^[25]	Paid Membership	Requested	6	3	UK Members of Herald
Computer requests	Casual contacts made through email	Email from researcher inviting them to respond	50	13	Access to a Computer. UK and Overseas
Personal Contact	None		Approx. 40	6	UK
Total	-	-	-	419	-

Initially, owing to the difficulty in contacting home-educators, it was anticipated that the survey would be distributed once only, through the 'EO' newsletter and all the returns collated and analysed. However, once contacts were established within the home-educating community and related LEA personnel, it became clear that a far wider sample of home-educators could be reached. Hence, further questionnaires were dispersed to a broad range of home-educators as distribution snowballed with organisations and individuals passing copies on. This was beneficial in creating a clearer picture of the status quo amongst contemporary home-educators but signified an ever expanding volume of data.

Although only 419 of the questionnaires were analysed in full, all questionnaires returned were read by the researcher. From the reading it was possible for the researcher to gain an overall picture of the respondents and thus, conclude that those analysed were representative of all those participating in the research. Further, the researcher's intention was that the research would extend beyond the current study and therefore incoming questionnaires beyond the cut-off point were encouraged.

It was not possible to define a 'rate of return' since no attempt was made to distribute a pre-determined number of questionnaires. The distribution aim was to send out as many questionnaires as possible in order to obtain information about home-education. Without knowing how many home-educators there were, calculations of rates of return were further complicated. For example, the request in the HEAS newsletter brought 12 responses and yet overall, the number of families who named their affiliation with HEAS totaled 113. This was further complicated by HEAS who, over the 3-year period, were unable to provide any indication of their subscriber numbers. Also, membership numbers of 'EO' did not indicate whether families were practising home-educators^[26]. Families may have had grown children, pre-school aged children or children moving in and out of school; they may also have been people with an interest only.

4.5.4 Initial Questionnaire Procedure (Chapter 5)

Procedure Outline

Initial contacts were made concurrently with 'EO' and LEAs. The LEAs were approached either by telephone, email or letter. Encountering the officer in charge of home-education was often an arduous task, many switchboards not understanding the term 'home-education'; this resulted, very often, in being passed from office to office only to find that the person responsible was on leave, out or in one case, no one knew who the person was. Another LEA stated that there were no home-educators in their area at a time when there were at least five known to this researcher.

On the 22^nd January 1997, the first Metropolitan LEA officer to assist with the distribution of questionnaires confirmed that it would be possible to distribute the home-education questionnaires during visits to the home-educating families in the region. Visits were due to begin the following week and an initial supply of ten questionnaires was requested. These were distributed to the families visited, together with ten letters for the families concerned and ten stamped addressed envelopes. An example of the prototype letter that accompanied all LEA distributed questionnaires appears at Appendix 4.4.

On the 25^th January 1997, the AGM of the organisation ‘E.O.’ authorised distribution of questionnaires through the February edition of their bi-monthly newsletter to members. The questionnaire was printed by the 'E.O.' printers^[27] and enclosed by them^[28], as a centrefold, into newsletters: 3,000 copies were made, 2,500 of which were going to active members. In 1998 approximately 2,500 more questionnaires were distributed in this way. During 1997 (no 1998 data available), 'EO' had 1191 new members and had lost 791^[29], therefore, it can be surmised that approximately 1100 new families received a copy of the questionnaire. The issue of the 'EO' newsletter that followed each of the mass distributions carried a short article urging more families to respond and thanking those who had. A copy appears at Appendix 4.5.

On the 30^th January 1997, a second Metropolitan LEA requested 9 questionnaires that they had agreed to distribute, together with a covering letter that they, the LEA, had prepared. The LEA required both the stamps for the letters being sent out and stamps for the questionnaire returns.

The general pattern in LEA distributed questionnaires, as described above, continued over a period of approximately eighteen months from 22^nd January 1997 to mid 1998. Eight local authorities assisted from about thirty who were approached initially. Only one LEA refused to assist, on the basis that they had too many home-educators and addressing the envelopes would take up too much time. Others did not assist for a number of routine reasons such as, agreement in principle but then the appropriate officer took sick leave; or the time scale offered was longer than that requested; during school holidays the home-education inspectors were re-deployed or given leave; letters that had been invited during telephone conversations were not answered; and there were changes in staff.

The ways in which LEA officers assisted varied greatly. Some required a set number of questionnaires, others made photocopies from one copy supplied, there were those who knew the quantity they would be sending out and those who were unable to provide figures. Several officers delivered questionnaires whilst on visits, others posted them. There were those officers who required postage and those who did not. Some officers asked for letters to their Directors, others made their own decisions. Since the purpose was to distribute as many questionnaires as possible, the actual numbers delivered was not as important^[30] as the agreement from an LEA that they would support this research.

Besides distribution through 'EO' and LEAs, questionnaires were distributed by a number of means, as described above in Figure 4.1. An example of the letter published in the HEAS Bulletin, for example, is contained at Appendix 4.6. Whilst distribution continued for eighteen months, returns arrived over the two-year period identified, and beyond.

On receipt, each questionnaire was numbered and where questionnaires were returned anonymously, a note was made of the place of posting^[31].

Analysis of the Questionnaires

A pilot analysis of 50 questionnaires was undertaken using a Microsoft Excel database to code respondents' answers, numerically, directly onto the computer. Answers to each question were coded and allocated a number according to the type of response^[32]. Upon return of the completed questionnaires, the responses to each question and subsequent analysis were subject to interpretation. Owing to the possible distortion of data that could occur, every precaution was taken by the researcher to prevent this. One method of prevention was to keep as closely as possible, if not exactly, to the phrases and words of respondents, and to read each answer in the context of that respondent’s responses to other questions. To test for reliability[33] at this stage, selected questionnaires were re-examined by a second analyst (every fifth questionnaire). Initially, reliability was between 40 and 50% but the process allowed remedial procedures to be established at this early stage: these are described below.

Following the pilot analysis it was decided to adopt a different method of data recording for the main sample^[34]. There were two reasons for this: the direct entry method was open to typing errors and researcher ‘database blindness’, also, in view of the increasing numbers of questionnaires being returned, it became clear that a more efficient and sophisticated database was needed.

Changes were as follows:

· Data entry sheets were printed. An example is provided at Appendix 4.3. These contained 90 fields where codes could be written. This meant that questionnaires could be coded away from the computer at any convenient moment and that backtracking and rechecking was made simpler.

· Coding sheets were used, where phrases were provided with number classifications. An example of a coding page appears at Appendix 4.7. The question responses were read and answers categorised according to the response offered. Category sizes were defined according to how many different answers were given: they were however kept to a minimum. Every effort was made to represent the feeling implied by the respondent. An example of classification and interpretation would be for instance, that under the question inquiring about the advantages of school, many respondents replied with the wording ‘free childminding service’ whilst others used the wording, ‘free time for parents’. These replies were seen as reflecting similar ideas about the benefits of school and were accordingly made into one category.

· An assistant, Dr Massey^[35], was enlisted to assist with the code allocation task. Initially, assistant and researcher worked together through a batch of 15 questionnaires. The discrepancy between analysts was minimal and worked through until a level of total agreement was reached[36]. Many of the questions required factual information and differences in interpretation were thus kept to a minimum and chiefly concerned answers to such questions as, ‘What was your motivation to home-educate (e.g. did the 'religious' family mean that they home-educated because they believed they could 'best provide for the children', or was the decision based upon religious motives)?' For this reason, as far as was possible, the categories used under each data-field number adopted the wording of the respondents, with some classifications eventually containing a subset of several phrases that appeared to represent the same overall sentiment.

· The statistical database package SPSS was used. This meant that the data could be analysed statistically with ease.

· Data was transferred from the hand written data sheets to an SPSS database by assistants at the University of Newcastle^[37].

Although it was reasonable to expect some errors in data entry, none have been found. A small number of discrepancies^[38] were encountered but these related to errors in data entry by the researcher and were rectified. The three-tier system of questionnaires, hand completed data entry sheets and computerised database, created an efficient system, whereby backtracking could allow for clarification as necessary.

Feedback and Analysis of Non Respondents

Two participants wrote that they had found the questionnaire to be very long and difficult, one of these choosing instead to reply with six pages of typed notes. Another such submission came from a lady who commented that the questionnaire had forced her to think through and write down her ideas for the first time; she too, had found that the questions were limiting, choosing instead the freestyle approach. A few other respondents, chiefly those who preferred to use separate paper, spoke of their enjoyment at using the questionnaire as a basis for thinking through and writing about their ‘philosophy’. One non-respondent commented that she found the questionnaire to be very complex, awkward and far too long.

It appeared that many of the initial non-respondents contacted though the 'EO' newsletter became respondents at a later date, either because of the second distribution through the newsletter or because they had been re-approached through their LEA, friends or another organisation that had co-operated with this research. Many non-respondents were probably parents with families who simply 'never got around' to completing the questionnaire: such an occurrence was clear from the number of questionnaires returned with comments like:

'it's taken me 7 months to do, but here it is, hope it helps'.

Where postage was included, such as with a number of questionnaires distributed though LEAs, SHEA and HEAS, questionnaires were more likely to be returned, even when not completed, as in the case of a number of families who had sent their children to school or otherwise changed their circumstances. Similarly, where participants emailed their responses, the ease of the medium meant that completion and return were made simple, thus encouraging responses.

There were families who did not reply because of their philosophical belief that questionnaires were invasive and there were others who were suspicious of any interest in their situation.

There were indications that few home-educators were completely isolated. Most home-educators it appeared were known, if not to their LEA, then to other home-educators, associated membership organisations and or to their religious organisations. Some religious organisations were difficult to approach and whilst several anonymously returned questionnaires appeared to come from affiliates of such organisations, it was recognised that many more such families would not have had access to the questionnaire or if they had, had chosen not to return it.

Two parents said they had not replied because they did not know what answers to give. This reaction may have affected a number of the non-returnees. There was little remedy for such a reaction. The questionnaire’s introductory paragraph made it clear that all home-educators had something to contribute. This type of response may have been an indication that some non-respondents were unsure of themselves; to minimise this problem the 'EO' newsletter follow-up letter urged families to respond (see Appendix 4.5). Where questionnaires had been distributed by other means, such an approach was not always possible.

4.6 Methodology used for the Baseline Assessment

4.6.1 Choice of Methodology (Chapter 6)

Selection of the PIPS Baseline Assessment (Start and End of Reception)

An assessment was sought for this research that would evaluate in some way the attainment of home-educated children. Early in the research planning in 1996, during discussion with Professor Peter Tymms (Director of the PIPS Project), at the University of Durham, it emerged that the Performance Indicators in Primary Schools (PIPS) Start and End of Reception Baseline assessment would be an appropriate measure to use.

Aspects of PIPS 'Reception' that made it suitable were that it was designed to capture pre-school style learning, as opposed to being formulated with the National Curriculum in mind. Although the 'End of Reception' measure was very much school based, its place at the start of the school 'ladder' meant that National Curriculum related items were kept to a minimum. The significance of the PIPS 'End of Reception' instrument was that it permitted this research to explore the value-added aspect of the home-educated children's learning. This was vital, since it meant that the programme was not just conducting a 'spot check' on the home-educated children, but was able to evaluate their learning over a 'year' thus introducing a longitudinal element to the research. The growing popularity of the PIPS 'Reception' assessment amongst schools and LEAs also made it an attractive choice. For example, in 1999, 3208 schools and 89,571 children participated (PIPS Project 1999).

Assessment Programme Design

Initially the questionnaires[39] completed by the 35 families participating in the PIPS 'Reception' assessments were analysed to extract background information relating specifically to these families.

The assessment was devised in two parts, to be given at the start and end of what would be the 'Reception Year'. PIPS 'Start of Reception' (PIPS Project 1997a) was administered in September, soon after school entry, followed by PIPS 'End of Reception' (PIPS Project 1998a) ten months later. The 'End of Reception' provided an extension to the 'Start of Reception' test and involved re-administering those parts of the 'Start of Reception' assessment that a child had not been able to complete, together with the follow-on 'End of Reception' measure. The PIPS Baseline was designed to monitor children's progress as they enter and pass through their initial year in full-time school. The 'Start of Reception' instrument tested for 'rhymes', 'early reading' and 'early maths', whilst the 'End of Reception' assessment evaluated performance in these areas, together with 'attitude' and 'self esteem'. Appendix 4.8 provides a list of the specific test areas. Whilst the 'Start of Reception' assessment aimed to study children who had not started formal schooling, the PIPS 'End of Reception' was directed at children who had spent a year in school and the instrument was designed to test the type of knowledge commonly expected to be acquired during that year. The 'Start of Reception' and 'End of Reception' tasks were designed to provide a value-added measure of progress and performance^[40].

In order to provide a rich backdrop to the quantitative data collected, semi-structured interviews were conducted with each family, in their own home, at the start and end of the assessment programme. Each interview lasted approximately 1.5 hours and involved speaking with all members of the family present, in a variety of locations within the family home and garden.

Materials used for the PIPS Baseline Assessment

For the assessments the following materials were used:

· PIPS 'Start of Reception' Assessment Booklet (PIPS Project 1997a)

· PIPS 'Start of Reception' Parent's Instructions (PIPS Project 1997a)^[41]

· PIPS 'End of Reception' Assessment Booklet (PIPS Project 1998a)

· PIPS 'Start of Reception' Parent's Instructions (PIPS Project 1998a)

To assist with analyses of the data the following publications were used:

· Performance Indicators in Primary Schools: Technical Report 1999 (PIPS Project 1999)

· PIPS Baseline Report 1997/98 (PIPS Project 1997b)

· Using the PIPS Scheme 1997/98 (PIPS Project 1997c)

· End of Reception Assessment: How to calculate initial feedback. PIPS Project (1998c)

4.6.2 Sample Selection (Chapter 6)

Selection was determined according to whether a home-educating family included at least one child aged approximately four years^[42]. Amongst the 212 questionnaires returned four months after distribution, 45 families met this criterion and were approached with a request that they participate in a further stage of the study. Families were members of the home-education organisation 'EO'^[43] (although by the time of 'End of Reception' administration, 8 of the 35 original families were no longer members of the organisation, 'E.O.') and had each returned a research questionnaire that had been distributed some months earlier through that organisation's bi-monthly newsletter. Prior to selection further characteristics could not be known, although it could be conjectured that such families would probably have at least one child who had never attended school. During the selection of families to be approached in connection with the Baseline assessment, questionnaire data relating to those families was available to the researcher, but it had not at that time been analysed, beyond identifying prospective subjects and noting that these were spread around the United Kingdom.

Whilst it is not possible to know the number of four-year-olds in 'E.O.' during 1997, in 1998 119 children were known to fall within the age range sought the previous year: this suggested that the response rate from families with four-year-olds (45 completed questionnaires) may have represented close to 40% of those 'E.O.' members who had at least one four-year-old in 1997.

All 45 families were approached by letter and from these, 35 participated in the initial baseline assessments. Of the ten families who did not participate, four were necessarily abandoned because of their location, one family declined to assist further^[44], three could not be contacted, one was moving house and the other family were on holiday at the time of the visit. One family agreed to participate but on the day arranged for the assessment, the child participant was absent, having been removed without warning, by an estranged parent^[45].

By the time of the second interview, of the initial 35 families, one had placed their children in school and a second family had moved, leaving 33 families to be assessed during the second phase of the PIPS 'Reception' assessment.

4.6.3 Procedure for Administering the PIPS Baseline Assessment

During February/March 1997, approximately 2000 questionnaires were distributed through the bi-monthly newsletter of 'E.O.'. By June 1997, 212 questionnaires had been completed and returned and 45 home-educating families were identified from this batch, as described in Section 4.6.2 above. These families were approached by letter and subsequently, where possible, by telephone: 35 families agreed to assist with the PIPS Baseline assessment.

The assessment materials were supplied to the researcher by the PIPS Project co-ordinator and the home-education research programme was issued with an identification number, used for reference purposes by the PIPS Project team. The cohort was named, 'Hill Top School' and provided with an identification number, thereby eliminating any possibility that members of the PIPS Project team, beyond several key people, would be aware of the actual nature of the cohort.

The participating families were visited in their own homes during the month of September. The researcher spent between 1.5 and 2 hours with each family, during which time a semi-structured, but generally informal, interview took place. At an appropriate time, the family's four-year-old child was assessed. The assessment was either conducted by the researcher, or by a parent^[46], under researcher supervision, dependent on the child's preference. Initially, most parents (20), following written instructions, read out aloud to the child, what needed to be done, whilst the researcher sat beside the child providing the correct materials at the appropriate time: it was far easier for the researcher to take notes when not also administering the assessment. During the remaining interviews, where the assessment was administered solely by the researcher, some parents watched the procedure, some went about their 'chores' and others chose to leave the child and researcher alone: likewise with siblings. In this way, most visits took the form of entry, a cup of tea offered to the researcher and in some cases breakfast, with the informal 'chat' gradually becoming a notebook interview. The assessment was generally administered after the preliminary interview, followed by a post-test discussion and the broader interview. The assessment formed an excellent basis for further discussion, opening the way to parental analysis of their child's abilities that might not otherwise emerged.

Interviews were, owing to time constraints on the PIPS assessment[47], mostly conducted in batches of 3-6 per day, dependent on the distances to be travelled throughout the UK, by the researcher. Families were interviewed within a circle that stretched, roughly, from Dundee to Exeter and from Aberystwyth to Norwich. Some assessments took place as early as 7am and some as late as 8pm. The visiting times arranged were worked around children's 'liveliest' hours of the day, parental convenience and within the realms of what the researcher could manage. During the PIPS related interviews, the opportunity was taken to interview many other families who lived 'en route' and who had responded to an initial questionnaire (data from these interviews appears in the separate Interview report). Whereabouts, within households, the interviews and assessments took place, depended very much on what was happening in the house, the time of day and the weather. Some assessments were undertaken at tables, some on the floor, several in the garden and a couple over the breakfast table.

Following the elapse of a ten month period, the equivalent of the school 'year', 33 families were re-visited and PIPS 'End of Reception' assessments were conducted.

Once the assessments and interviews were completed, the marks for each participant, as scored per section undertaken, were entered, using pseudonyms, on to a score sheet, as directed by the PIPS Project. These were then forwarded to Christine Merrell, the PIPS project's co-ordinator at the CEM Centre, University of Durham for initial analysis (comparison with PIPS national data). The formal scores, raw and standardised, were then returned to the researcher, together with relevant graphs and tables, normally distributed to participating schools. The interviews had been recorded using pen and paper and from these notes, typed transcripts were produced and analysed using the computer programme QSR NUD.IST[48]. Sections of the interviews relating directly to the PIPS 'Reception' programme are included in the PIPS Baseline Chapter.

4.7 INSTRUMENTS: Literacy & Mathematics ASSESSMENTS

4.7.1 Choice of Instruments (Chapter 7 & Chapter 8)

Decision to use the PIPS Year 2 Assessment and National Literacy Project Assessments for Years 1, 3 and 5

Following the initial discussion with Professor Tymms (Section 4.6.1) that had led to use of the PIPS 'Reception' assessment, subsequent discussion led to the suggestion that the PIPS Year 2 would be a useful instrument to include, with its focus on slightly older children. The PIPS Year 2 (PIPS Project 1998b) measure was particularly attractive owing to its seemingly non-curricular design in terms of its Maths, Reading, Picture Vocabulary and Non Verbal Ability components. Furthermore, having been created by the PIPS Project team, who had developed the PIPS 'Reception' instrument (PIPS Project 1997a, 1998a), the suggestion of design continuity was appealing.

Discussion with PIPS Research Associate, Christine Merrell, about the possibility of adopting a further measure that could be used to both assess other home-educated 'Year' groups and also to provide comparative data for enhanced evaluation, led to her introducing the idea of using the National Literacy Project assessments. PIPS staff (CEM Centre[49]) had designed these tests in collaboration with the National Literacy Project, the Education Department at Newcastle University and the School Curriculum and Assessment Authority[50] (SCAA).

'The Patterns in Language assessments are intended for use in schools taking part in the National Literacy Project. They have been designed to relate to the project's 'Framework for Teaching' and are intended to support that programme. They are meant to provide information about pupils and schools within the project and to evaluate the impact of the project as a whole.'

CEM Centre (1998a)

Following Professor Tymms' approval, permission for use of the NLP assessments was also sought from Mr David Hawker, then of the SCAA, who agreed to use of the test in the current research, provided that he was kept informed of the project.

Design of PIPS Year 2 Assessments and National Literacy Project Assessments for Years 1, 3 and 5

The PIPS Year 2 measure was designed by the PIPS Project team, 'for professional use in order to improve education' (PIPS Project 1998b) and included assessments in 'Maths', 'Reading', 'Non-Verbal Ability', 'Picture Vocabulary' and 'Cultural Capital'^[51]. Chapter 7^[52] provides a number of scanned images from the assessment that provide an idea of the assessment layout. Appendix 4.9 provides a summary and explanation of each assessment component.

Each of the three NLP assessments consisted of a number of exercises that tested a variety of literacy skills. Appendix 4.10 details the NLP assessment sections for each 'Year' group.

4.7.2 Materials Used (Chapter 7 & Chapter 8)

For the assessments and analyses the following materials were used:

· National Literacy Project: Patterns in Language Year 1: Pupil Booklet. (CEM Centre 1998b)

· National Literacy Project: Patterns in Language Year 3: Pupil Booklet. (CEM Centre 1998c)

· National Literacy Project: Patterns in Language Year 5: Pupil Booklet. (CEM Centre 1998d)

· National Literacy Project: Patterns in Language Year 1: Teacher Administration Instructions. (CEM Centre 1998e)

· National Literacy Project: Patterns in Language Year 3: Teacher Administration Instructions. (CEM Centre 1998f)

· National Literacy Project: Patterns in Language Year 5: Teacher Administration Instructions. (CEM Centre 1998g)

· National Literacy Project: Patterns in Language: interpreting and using the test results. (CEM Centre 1998a)

· Assessment 2: Performance Indicators in Primary Schools (student booklet) (PIPS Project 1998b)

· Assessment 2: Performance Indicators in Primary Schools (Administration instructions) (PIPS Project 1998d)

· Assessment Report: Year 2 to Year 8. (PIPS Project 1998e).

4.7.3 Sample Selection (Chapter 7 & Chapter 8)

Several months before the time period prescribed by the PIPS Project team for administration of the assessments, those families represented amongst the sample of completed home-education questionnaires returned at that date, who had appropriately aged children, were selected and separated into assessment groups. There were, 89 year 1 children, 74 year 3, 61 year 5 (potential NLP candidates) and 62 year 2 children (potential PIPS Year 2 participants). Each questionnaire was numbered and a computer randomising program used to generate numbers from each group: 20 families were selected for each of the NLP tests and 25 for the PIPS Year 2 assessment. Professor Tymms had recommended groups of at least 20-30 participants: in total there were to be 60 NLP participants (3 groups each with 20 children in each year) and 60 PIPS participants (1 PIPS Baseline group of 35 and a PIPS Year 2 with 25). Where possible, families were then contacted by telephone to request their consent to assist with this further stage of the research. The initial questionnaire had asked participants to provide their contact details if they were prepared to assist further with the research; therefore it seemed appropriate to make this secondary contact by telephone. Three of the families had been interviewed previously, as part of the 'PIPS Start and End of Reception' assessment project. Where no telephone number was given participants were contacted by letter. All those asked to assist with the NLP assessments agreed (although one child did not complete his assessment it was therefore not included in the analysis - see Section 7.4.5). Two families who initially consented to participate with the PIPS Year 2 assessment, later sent back the assessments not completed, with letters providing their reasons for doing so: one boy had 'not felt like doing it' after all and another parent felt that it was too demanding for their child. Four PIPS Year 2 participants returned their booklets completed, but too late for inclusion in the assessment process at Durham University's CEM Centre: the qualitative data, however, was made use of. One family, when reminded, had simply forgotten to return the completed booklet.

Table 4. 1: number of participants in each of the academic assessments

Assessment	No. of Participants	Age
NLP 1	17	6 years
PIPS Year 2	18^[53]	7 years
NLP 3	15	8 years
NLP 5	17	10 years
Total	67	6-10 years

Table 4.1 provides data concerning the number of participants involved in the literacy and mathematics assessments. Owing to the different age orientated assessments, no participant represented in the results data took part in more than one assessment.

4.7.4 Procedure for the Literacy & Mathematics Assessments

Following initial contact to obtain consent, letters containing the assessments were distributed to the participants by post, together with stamped addressed envelopes for their return. An example of a letter relating to the NLP assessment is provided at Appendix 4.11.

Parents, according to the instructions supplied to them, administered all assessments. These instructions were those provided by the CEM Centre/PIPS Project to the researcher, but modified to reflect the home, as opposed to classroom, environment.

Once completed, assessments were returned to the researcher. From the NLP cohort, 5 children[54] were re-assessed by the researcher, prior to any marking, to ascertain inter-rater reliability. The time-lapse between initial and subsequent administration varied between 3 and 5 weeks. The results of this second assessment can be seen at Appendices 7.1, 7.2 and 7.3. The differences revealed remarkable consistency between parent administered and researcher administered tests, as illustrated by Table 4.2.

Table 4. 2: score difference between parent & researcher administered assessments

Parent administered assessment result (standardised scores)	Researcher administered assessment result (standardised scores)
>130	129
126	>130
107	107
71	71
95	95

Owing to the text-length of the PIPS Year 2 tests and the comments received regarding the stress that the assessment had placed on several of the children, it was decided on ethical grounds not to re-administer this instrument. Some of the families involved with PIPS Year 2 had also participated with the interviews (7 of 18) and were, therefore, known to the researcher. Personal awareness of some participants assisted in assessing validity both with PIPS Year 2 and with the NLP assessments. The commentaries that accompanied returned assessments were also very useful in this respect. Whilst the system was not foolproof, every possible precaution was taken to ensure validity, bearing in mind that the nature of the cohort set restrictions on what, appropriately, could be asked of them.

Returned PIPS Year 2 assessments were copied out by hand and participants given pseudonyms to protect their anonymity. Booklets were delivered to the CEM Centre at the University of Durham for marking. Identifying information and annotated commentaries from parents and children were not passed on. Written comments on the assessments did however, provide rich qualitative data that assisted with drawing conclusions about the participants. Once marked, the CEM Centre returned the results for analysis by this researcher.

Completed NLP booklets were scored by the researcher. Instructions for marking were supplied by the CEM Centre: additional telephone assistance was also provided as and when questions arose. Once the CEM Centre had marked all school scripts, they forwarded a summary of the national data to assist with the researcher analysis of the home-educated cohort's scores.

4.8 INSTRUMENTS FOR COLLATING THE Social and Psychological Data (Chapter 9)

4.8.1 Selection and Design of Instruments (Chapter 9)

Once the decision to study home-educated children’s social and psychological skills had been taken, a judgement had to be made as to the type of measure to be used. With an inherent link between the acquirement of social skills and one's ability to operate in wider society, questions arose about the psychological state and behaviour of children considered to be 'missing out' (Shearer 1999; Walker 1998) through home, as opposed to school, education. Owing to the 'normal' nature of the home-educated sample, it was necessary to adopt measures that were not specifically tied to any one diagnostic category, but that provided instead, a general indication of the existence of psychological difficulties.

A number of measures were investigated. The Revised Conners Parent and Teacher Rating Scales[55] are currently used for the detection of Attention Deficit Disorder, hyperactivity and aggression and thus seemed inappropriate for a 'normal' population. Achenbach and Edelbrock's Child Behaviour Checklist referred to by Sclare (1997), measuring both competencies and behaviour, appeared to be more complex than was considered necessary for this research. The Eyberg Child Behaviour inventory (Buros 1961), used particularly to detect conduct problems was also considered, but attempts to contact the author in North America proved unsuccessful. The Piers-Harris Self Concept Scale (Shyers 1992), with its focus on physical appearance and attributes, anxiety, intellectual and school status, behaviour, happiness and satisfaction and popularity was, owing to its use by Shyers (1992), a suitable instrument, but ordering delays, length, complexity and cost excluded it. Discussion of these instruments is elaborated upon in Appendix 4.12.

Rating scales that could be administered by post, for completion by parents and or children, were the most suitable measures: however, problems highlighted by Sclare (1997) that might be associated with rating scales included:

· Those making the judgements can only report behaviour that they are aware of.

· Children's behaviour is often context-dependent

· Decisions on children's behaviour are often based upon comparison with other children.

· The effect of the rater's mental state on how they interpret and answer the questions.

· Ratings may be influenced by the 'halo' effect whereby the child is judged according to the overall impression that the rater has.

· Potential problems mean that it is preferable to gather data from more than one rater and in more than one location.

The current research took these points into account, considering that:

· The parent raters in this case would be home-educators and for the most part, continually within physical reach of their children: these were the people most exposed to their children's total behaviour repertoire.

· In the absence of the ability to make comparisons with other children, in that most other children would be school-children with their own culture of acceptable and non acceptable behaviour, it was reasoned that the parents would judge their children in accordance with adult standards of behaviour^[56], as indicated by Shyers (1992).

· It was recognised that, as in any population, some parents would be experiencing symptoms that could be regarded by doctors as evidence of minor psychiatric disorder. However, there was no reason, from reference to other research (e.g. Rutter, Tizard and Whitmore, 1970), to believe that this would invalidate the responses. Finally, the multimethod approach of this research enabled cross-checks (Denzin 1989) to be made on the children involved. At the lowest level, rating scale data could be checked with the introductory questionnaire survey data; at the highest level, rater attitudes could be cross-checked with interview, questionnaire and assessment information. This allowed the researcher to investigate any items of particular interest^[57].

Following deliberations as to the type of measures to be included, three emerged as optimum choices. They were:

Children's Assertive Behaviour Scale (CABS) (Michelson^[58] & Wood, 1981)

Revised Rutter Scale (RRS) (Rutter^[59], 1993)

Strengths & Difficulties Questionnaire (SDQ) (Goodman, 1994)

Samples of the instruments are attached at Appendices, 4.13, 4.14 and 4.15 respectively. The CABS Appendix 4.13 includes the scoring sheet.

Selection of the Children's Assertive Behaviour Scale (CABS)

CABS measures children’s social skills by questioning their assertiveness in face to face situations. Shyers (1992), in his North American comparative study of home and public schooled children, used CABS to assess social skills of both sets of children relative to each other. His research has been evaluated in the UK by Meighan (1995) and in North America by Mayberry et al. (1995).

CABS (Michelson and Wood 1981) was designed to consider the social skills of children on a sliding scale from passivity through assertiveness to aggression. The scale was first designed to be used as part of a clinical assessment procedure to screen and diagnose children in need of assertiveness and social skills training.

Michelson, Sugai, Wood and Kazdin, (1983) suggested that a child with good social skills would experience better interpersonal relations than a child who lacked such skills. They therefore justified the need for children’s social skills training and use of the CABS questionnaire as a screening instrument to identify children’s strengths and weaknesses.

Selection of the Revised Rutter Scale (RRS)

The RRS is a screening instrument, designed to distinguish between children with and without behavioural difficulties. In view of common conceptions about home-educated children's dispositions, this was an appropriate instrument to adopt; it has been suggested by Hersov and Berg (1980) and Rutter and Hersov (1985) that children not attending school could suffer separation anxiety, overprotection by their parents, school phobia or be similarly pre-disposed to another psychologically recognised condition: all conditions often linked to home-educated children, e.g. Wragg (1999) Shearer (1999) Hastings (1998). Galloway (1982) also used the Rutter Scale in his research involving absentees; an analogy was drawn earlier between absentees, as children with behavioural problems and home-educated children, thus reinforcing the use of the Rutter Scale with the current research.

Selection of the Strengths and Difficulties Questionnaire (SDQ)

The SDQ was considered suitable for the research at hand since it was designed as a screening questionnaire for use by researchers, clinicians and educationalists; it was short and appeared to complement the RRS by providing a second point of reference in terms of the Rutter Scale's evaluation of the 'Total Difficulties', 'Emotional Difficulties', 'Conduct Difficulties' 'Hyperactivity/Inattention' and 'Prosocial' dimensions. The SDQ also contained a 'Peer Problems Score' and a self-rated questionnaire for eleven to sixteen-year-olds^[60] that allowed them to make judgements about their own behaviour. Chosen for its similarly to the RRS, the SDQ added a further dimension to the study by providing a triangulation point (Robson 1993, p. 290; Lowden 1993; Taylor 1993) of reference for the psychological and social data. A final rationale for the use of both RRS and SDQ measures was that they might serve to detect common psychological denominators amongst home-educated children. Both the SDQ and the RRS were considered to offer a broader perspective on home-educated children than would have been gained through use of CABS alone.

The SDQ has been validated in several studies. Goodman (1994) first evaluated the SDQ with 320, five to seventeen-year-old hemiplegic^[61] schoolchildren, finding that it compared well with 'independent psychiatric evaluations'. Goodman (1997) further refined and evaluated the SDQ in research involving four to sixteen-year-olds, administering it together with the Rutter Scale to both parents and teachers of 403 children. About half the sample were children visiting a dental practice and the other half were registered at a psychiatric clinic. Goodman, wanting to evaluate the SDQ against the Rutter Scale, found that the scores from both questionnaires were highly correlated. The SDQ was found to differentiate between difficult and non-difficult cases as effectively as the Rutter Scale, besides providing data on inattention, peer problems and prosocial attributes.

Goodman, Meltzer and Bailey (1998) used the SDQ in a pilot study on the validity of the self-report version. They administered the self-report version to 83 unspecified youngsters and 116 mental health clinic attendees, all aged eleven to sixteen years. The researchers found that the questionnaire distinguished between the two samples to the extent that for the 'Total Difficulties' score the unspecified youngsters' mean was 1.4 standard deviations below that of the clinical sample. Goodman et al. (1998) also reported that correlations between parent and self-reports were 'favourable'.

In the final evaluation of the SDQ to date, Goodman and Scott (1999) compared the SDQ with the Achenbach's Child Behavior Checklist by using mother reports with samples drawn from both psychiatric and dental clinics. They found that the two measures correlated well and discriminated equally between the samples, with the SDQ better able to diagnose inattention and hyperactivity. They further reported that the SDQ was the more popular instrument among mothers of 'ordinary' children.

Design of the Children's Assertiveness Behaviour Scale (CABS)

The CABS questionnaire distinguished between 'very passive', 'mildly passive', 'assertive', 'mildly aggressive' and 'very aggressive' social behaviour. Michelson et al. (1983) suggested that passive individuals lack self-expression and control over their lives, whilst aggressive individuals undermine those around them. Assertiveness is seen as the optimum and involves having autonomy over one’s decisions, expressing oneself without causing offence and demonstrating respect for others. These skills, Michelson et al. (1983) argued, ensure ease of passage with friends, family, acquaintances and strangers.

The CABS scoring system uses a bipolar scale with a midway mark of zero for assertiveness. A high score (+ or -) in either direction is indicative of unassertiveness, with a high positive value (+) score implying aggression and high negative score (-) suggesting passivity. Scores are formulated by the addition of points awarded for each of 27 questions, scoring on a 5-point scale, -2 to +2. The questions subdivide into five categories. Briefly, these are:

‘Positive’:	relates to how a respondent manages in situations that involve opportunities for positive expression; eg. Someone says to you they think that something you did was terrific. You would usually:
‘Negative’:	relates to how a respondent manages in situations that involve opportunities for negative expression; eg. Someone often interrupts you when you’re speaking. You would usually:
‘Request’:	measures the ability to formulate and react to, a request; eg. Someone has something that you want to use. You would usually:
‘Conversation’:	examines the facility to make and/or be involved in, conversation; eg. Someone you have not met before stops and says “hello” to you. You would usually:
‘Feeling’:	Explores the capacity to express one's own and understand another’s feelings. eg. You bump your head on a shelf and it hurts. Someone says, “Are you all right?” You would usually:

These 5 categories, outlined in depth at Appendix 4.16, relate to specific questions on the CABS questionnaire that are designed to measure expressive and receptive responses in the same five social skill areas; a key to these is given in Appendix 4.17.

Design of the Revised Rutter Scale (RRS)

Elander and Rutter (1996) described the development of the teacher and parent behaviour measures known as the Rutter A and B scales, relating to both school and pre-school aged children and covering an age range of 3-16 years. Sclare (1997) subsequently referred to the scales as the Revised Rutter Scales. The Rutter A and B scales were designed as screening instruments for emotional and behavioural, or conduct, disturbances in middle childhood children and were based upon an earlier version of the Rutter Scale (Rutter 1967) but incorporating some modifications such as a prosocial element. Earlier versions of the Rutter Scales were used, in the Isle of Wight by Rutter and Graham (1966); Rutter et al. (1970); Rutter, Tizard, Yule, Graham and Whitmore (1976); and Graham and Rutter (1973): in inner London by Rutter (1973) and Rutter, Cox, Tupling, Berger and Yule (1975): in Aberdeen^[62] by Rutter (1967); and in Sheffield by Galloway (1982).

The Rutter scales have been most commonly used in longitudinal and epidemiological[63] survey research such as the Isle of Wight Study (Elander and Rutter (1996). Use with normal populations has further included those described by Zimmermann-Tansella, Minchetti, Tacconi and Tansella (1978), Venables, Fletcher, Dalais, Mitchell, Schulsinger and Mednick, (1983) and Ekblad (1990): these authors have all used the scales to screen for children who may have behavioural difficulties. Elander and Rutter (1996) suggested that the scales might further be used to study the relationship between behaviour problems and reading and cognitive development, and also to assess the effects of social and familial factors on child behaviour. Elander and Rutter's (1996) indication that the scale should be used as a screening instrument and not as an individual assessment tool, made the scale a particularly suitable measure to use with the present study.

The Revised Rutter Scales are included as part of a comprehensive professional assessment package (NFER-Nelson 1997). Thus, the popularity, durability and validity of the Rutter Scales made the Revised Rutter Scale^[64] (Rutter 1993) offered by NFER-Nelson (1997) a particularly appropriate choice for the current study.

The RRS requires a parent to complete a questionnaire containing 50 statements that might refer to their child's behaviour during the past three months. The questionnaire takes about ten minutes to complete and the parent can choose between three categories of answer, 'Does not apply', 'Applies somewhat', or 'Certainly Applies'; weighted '0', '1' or '2' respectively. Scores from selected statements are added together to produce a 'Total Difficulties' score with a range from 0-52. Specified statements are also allocated into four domains; 'Emotional Difficulties' (5 items), 'Conduct Difficulties' (5 items), 'Hyperactivity-Inattention' (3 items) and 'Prosocial' (10 items), respective example statements of which are, 'Often worried, worries about many things'^[65], 'Frequently fights or is extremely quarrelsome with other children'^[66], 'Cannot settle to anything for more than a few moments'^[67] and 'considerate of other people's feelings'^[68]. Scores for each of these domains are summed and analysed in the light of each participant's 'Total Difficulties' score. Where a child scores at or above the author suggested cut-off point of 11, the child can be said to display behavioural problems. The type of difficulty is established by reference to the 'Emotional Difficulties' and 'Conduct Difficulties' scores, whereby the higher of these two scores relates to the area of difficulty: same scores in these two domains indicates an 'undifferentiated' disorder. The 'hyperactivity' sub-scale has a cut-off point of 3.

According to Elander and Rutter (I996) the two domains of 'Conduct Difficulties' and 'Emotional Difficulties' are generally agreed to be distinct groups. They pointed out however, that the 'Hyperactivity/Inattention' domain was not so clearly defined as an independent dimension. In a discussion of validity, Elander and Rutter referred to Minde who in 1977 compared a sample of ordinary and reform schoolchildren. Minde found that 95.8% of the delinquent boys at reform school scored 9 or more on the Rutter Scale, whilst only 18% of the ordinary children were determined by the scale to have behavioural problems. Elander and Rutter (I996) further made a detailed comparison of the RRS with other instruments, discussing reliability in some depth: comment by Elander and Rutter (1996) provides further information in this respect.

Some modifications have been made to the RRS by the scale's author to accommodate distinct populations and translations have resulted in such phrases as, 'moving the limbs pretty fast in a funny way like the Konga monkey' to describe the item number 4, 'squirmy child'. The study at hand did not require the questionnaire to be altered but it was clear that different groups of people might interpret the items in very different ways and caution was therefore taken with placing too much weight on the outcomes for the home-educated sample beyond noting overall behavioural trends. Whilst a solitary child (item 15) or one who stares blankly into space (item 39) may be deemed by wider society to be showing signs of behavioural difficulties, the home-educating child might be seen by their parents as displaying desirable behaviour. Conversely, such interpretation of the questionnaire by participating parents may have contributed to their candid reporting of such behaviour. Elander and Rutter suggested that the 'Prosocial' element of the scale would make parents more likely to divulge their children's negative behaviours, thus reducing the risk of a 'halo' effect. The results detailed in Chapter 9, whereby some children were judged to display behavioural problems, suggested that the parent-supplied data was valid. A critique of Elander and Rutter's (1996) inherent definitions of appropriate and inappropriate behaviour is not provided here since the home-educated sample were, for this aspect of the research, being judged according to behavioural norms.

Elander and Rutter concluded that the scales were generally reliable, but more so for 'Conduct Difficulties' than 'Emotional Difficulties' and for teachers' rather than parents' ratings: the reliability of the 'Hyperactivity/Inattention' section in determining such behavioural difficulties was less certain. Discrepancies between parents' and teachers' responses were attributed, in part, to 'situation-specific behavioural disturbances' (Elander and Rutter 1996): the teachers' scales were those most in accord with clinical diagnosis and this, Elander and Rutter reported, may be because the teachers were able to observe a wider range of behaviours than the parents and because these diverse behaviours were taking place in a more standardised setting than in the many distinct home environments. Elander and Rutter further raised the point that mothers with psychological pathology were more likely to rate their children as displaying behavioural difficulties: mothers were also more likely to identify their children if the family was small as opposed to large. Overall, Elander and Rutter determined that the scales were more effective in identifying links between 'broad classes of variable' as opposed to pointing to precise mechanisms. Rutter (1967), describing his 1960's version of the scale as a 'simple and crude measure', voiced his view that the scale should be supplemented by other information. Finally, Fowler and Park (1979) warned that the structure of the scales might not be stable across populations differing in socio-economic or ethnic composition.

The RRS is used to identify problems in children's behaviour and Sclare (1997) quoted earlier research that defined a problem in this context as:

'a disturbance of function in one area of relationships, mood, behaviour or development of sufficient severity to require professional intervention.'

Wallace, Crown, Cox and Bewreger (1967) quoted in Sclare (1997)

Sclare (1997), citing Thompson, Stevenson, Sonuga-Barke, Nott, Bhatti, Price and Hudswell (1996) reported that 22.3% of pre-schoolers demonstrated behaviour problems. Whilst Rutter et al. (1970) found that in the Isle of Wight, 10% of ten to eleven-year-olds showed signs of behavioural problems, Rutter et al. (1975) concluded that amongst inner city London children the prevalence was 25%. Ellis (1998) further reported that 20% of children suffered mental health problems. Sclare (1997) suggested that such behavioural problems often endured, at least over the childhood years and that these problems were generally associated with an interaction of social, psychological and physical difficulties. Contrary to Sclare's viewpoint, however, Galloway (1987 p.60) cited research suggesting that only in a minority of cases did problems persist beyond childhood, stressing that much was dependent upon the nature of the problem; conduct disorders carrying a worse prognosis than other disorders. Although the time scale is not given, Sclare (1997) further cited evidence (Rutter and Smith 1995) that there had been an increase in the incidence of children's psychosocial problems.

Design of the Goodman Strengths and Difficulties Questionnaire (SDQ)

Goodman (l994) adapted the Rutter Scale^[69] to develop the SDQ. This added 19 items to the parent's scale, 10 of which referred to prosocial behaviour, 4 to positive conduct and 5 to inappropriate behaviours.

Whilst the original Rutter Scales did not include a prosocial element, Goodman (1997) suggested that the inclusion of a 'Prosocial Scale' would make parents more likely to complete the questionnaire than if they were faced only with negative questions. Elander and Rutter (1996) had concluded that prosocial behaviour was a dimension in its own right rather than merely the opposite of antisocial behaviour and thus was a suitable addition to the RRS.

The current Goodman (1997) SDQ consists of 25 statements divided between 5 scales of 5 questions each: 'Prosocial Scale' 'Hyperactivity Scale' 'Emotional Symptoms Scale', 'Conduct Problems Scale' and 'Peer Problems Scale'. Disregarding the 'Prosocial Scale', the remaining scores (0-2 for each statement), were summed according to the score key, to create a 'Total Difficulties Score'. An example of one statement in each of the five categories is as shown in Table 4.3.

Table 4. 3: examples of statements associated with each scale

Scale	Example Phrase
'Prosocial Scale'	'Kind to younger children'
'Hyperactivity Scale'	'Restless, overactive, cannot stay still for long'
'Emotional Symptoms Scale'	'Often unhappy, downhearted or tearful'
'Conduct Problems Scale'	'Generally obedient, usually does what adults request'
'Peer Problems Scale'	'Gets on better with adults than with other children'

Two versions of the SDQ^[70] were used: an informant-rated one and a self-rated version aimed at eleven to sixteen-year-olds but competed in this research by 7 eleven-year-olds whose parents had completed an informant-rated questionnaire.

Comparison between Goodman SDQ and RRS items

A number of items between the SDQ and the RRS were very similar. These are detailed in Appendix 4.18. The phraseology of these, so called, 'similar' items may have held significance for the home-educated participants, as is discussed later in Chapter 9.

4.8.2 Sample Selection (Chapter 9)

Families were selected for this aspect of the research by three methods.

· By letter to those families who were assisting with other aspects of this research. See Appendix 4.19 for an example of the letter sent.

· Verbal request to the family whilst visiting their home to interview them

· Through the Internet: families who completed questionnaires, electronic or otherwise were asked by email, to participate with this further aspect of the research.

The Social and Psychological data were collected during the second year of this research and the sample was selected from families whose children fell into the appropriate age categories and who were taking part in other aspects of the research. The advantage of using families already assisting with the project was that their data could be cross-referenced to assess reliability and any interesting factors that may have emerged could be further investigated. Where a child in these families was aged eleven years, he or she was asked to complete a self-report questionnaire.

Beyond this initial sample, some participants were sought from amongst those who simply returned a questionnaire during year two and whose children were of an appropriate age for the instruments.

All families asked to complete these psychological and social questionnaires did so; thus, the response rate was 100%. Agreement to complete the instruments was sought from the families prior to distribution. The families had already self-selected themselves by completing the original questionnaire survey^[71].

Table 4.4 shows that 103 participants assisted with this aspect of the research. A total of 136 tests were administered, 43 CABS, 51 Goodman SDQ and 42 RRS. A number of participants took part in more than one assessment and their data was used for same-subject comparisons.

Table 4. 4: number of participants in each of the social and psychological assessments

Assessment Combinations	No. of Participants	Age of Informant	Type of Assessment
CABS	22	Child aged 8-10 years	CABS questionnaires for completion by the children.
SDQ	21	Adult	SDQ questionnaires relating to children's behaviour: the children were aged between 4 & 11 years.
RRS	21	Adult	RRS questionnaires relating to children's behaviour: the children were aged between 5 & 11 years.
CABS/SDQ	8	Adult
CABS/RRS	9	Adult
CABS/SDQ/RRS	4	Adult
SDQ/RRS	8	Adult
SDQ dad	3	Adult
SDQ self	7	11 year olds	SDQ self-report questionnaires relating to the children's own behaviour
Totals	103	8-adult	All

4.8.3 Procedure for Collating the Social and Psychological Data

The three measures, CABS, RRS and SDQ questionnaires, were distributed over a 12 month period, beginning in November 1997. Initially the questionnaires were sent by post, together with a stamped, addressed envelope (SAE), to those families who had already agreed to participate in other areas of this research; later, families were recruited during interviews and the questionnaires were supplied at that time together with an SAE. During this second year of the research, a number of families who had returned questionnaires by email and whose children were of suitable age, were approached electronically: these families returned their questionnaires by email. The families came from a wide range of backgrounds and so far as it was possible to assess, no one group was represented over another by this selection method. The aim was to collect the maximum number of responses from as wide a range of home-educators as possible, during the time available: this objective was achieved. A small number of families whose children met the criteria for all three measures, completed the three.

Four CABS scales were completed in this researcher's presence.

Returned scales were scored according to each author's instructions. Reference to the individual scoring systems has been made above under Section 4.8.1. Precise direction for details of the RRS can be found in Sclare (1997); for the CABS instructions in Shyers (1992) and Michelson et al. (1993); and for the Goodman SDQ, in Goodman (1997). All the scales were analysed in 1999.

4.9 Overview of the methodology

This chapter has provided the details behind the research methodology together with the rationales involved in terms of selection and process. With such an unknown quantity (the home-educators) it was impossible to know what kind of outcomes to expect. It was clear that there could be no standardised conditions and no clear cut questions and answers. Thus, this report should be taken for what it is, an exploratory window on home-education.

Methodology Endnotes

---

[1] 1000+ were returned but time constraints meant not all could be fully analysed, therefore, it was decided to use only questionnaires with data entry undertaken by University of Newcastle staff.

[2] Validity is the extent to which a test measures what it is intended to measure.

[3] Internal validity - The degree to which the design of a study allows one to accurately attribute an observation to a specific cause rather than alternative causes.

[4] External validity concerns making generalisations about the results. That is, how well do the study's conclusions apply to other people in other places at other times.

[5] It is commonly believed that compulsory registration for home educators will be introduced.

[6] Study lasted from 1996 - 2001 (including a year of initial research).

[7] An example has been the mention, in discussion of the Crime and Disorder Bill [Lords] (June 1998), of home-educators as legally out of school. This followed heavy lobbying by campaigners and is one instance of the home educators increasing ability to affect legislation.

^[8] The year preceding the start of the main study was spent in building up contacts, as was the first year itself.

^[9] This was itself.never used as a research tool, but it was used on a personal basis to contact families about local events etc.

^[10] There is no membership fee and contributions are voluntary.

^[11] Meighan appeared to have taken the information from the 'EO' database in 1984. Since that time, to date, 'EO' have not had such information available and no other research in the UK has explored this issue.

^[12] Lowden was Law Lecturer working for the Police.

^[13] Home-educator and Lecturer in Alternative Education at the University of York.

^[14] Author of Petrie (1992).

^[15] Of the University of Durham, now a Professor at the University of Warwick.

^[16] As understood through conversations with long-standing ‘EO’ staff and members.

^[17] Pilot runs usually assist in assessing the instrument robustness, enabling final adjustments to be made so that the survey, once distributed amongst the target population, achieves maximum returns. An explanation of pilot studies was given in Robson (1993 p. 301).

^[18] This caution was vindicated: in 1999 on the UK Home Ed list (a computer email list with about 200 members) an undergraduate posted a request for assistance with her questionnaire. This was followed by an attack from one list member and supported by another, accusing the student of, perhaps, being an impostor. The member advised others not to respond to the questionnaire.

^[19] 'EO’ had, allegedly, previously been misled into providing confidential information. LEA’s have occasionally, particularly in the past (before de-registration, see Chapter 1), threatened parents with legal action and some home educators have been involved in court cases (Petrie 1992).

^[20] Both membership and non-membership

^[21] Using a list of all LEAs, contact was made according to whether a telephone number was provided and whether the telephone call or email was answered.

^[22] Eg. yearly turnover in 'E.O.' membership was almost 800 in 1997.

[23] 'EO' Year 2 (131)+ SHEA (6)+ email (3) = 140 year 2 questionnaires.

^[24] The November Bulletin stated (without evidence) that families who home-educated from birth were 'very much in the minority' - implying, perhaps, that this was the HEAS experience.

^[25] HERALD is a home education organisation that assists with curriculum management and offers support through a newsletter.

^[26] Membership of ‘E.O.’ on the 1^st February 1997 consisted of 2,209 families with compulsory school aged children of which 1,305 families were home educating 1,917 school aged children. The total number of children under 11 years of age was 2289 of which 452 were under 5 years old. The number of school aged children, home educated, in ‘E.O.’, aged between 5 and 11 years was 1027. The number of children ‘home educated’ in ‘E.O.’, aged under 5 was 33 (the total number of under 5s represented by ‘E.O.’ membership was in fact 452): new babies do not become known to ‘E.O.’ unless their parents inform them.

^[27] At a cost of £40 paid by the researcher.

^[28] Thus, the identities of the families who received copies of the questionnaire were unknown to the researcher: respondents were invited to remain anonymous unless, upon replying, they chose to supply their name and address.

^[29] In 1996 there were 800 new members and 726 lost; in 1995 736 new members joined and 918 left. Information provided by 'E.O.'s database secretary.

^[30] Clearly it would have been useful to have this information, but LEA officers were generally very difficult to contact since their jobs involved out of the office work. The researcher considered that their agreement to assist distribution had been sufficient contribution without further questioning.

^[31] This is stamped by the Post Office on the outside of the envelope.

^[32]Eg. asked for the ‘advantages of school’, responses ‘free time for parents’ and ‘free baby-sitting service’ would receive one code whilst ‘a good education’, would receive another.

[33] The reliability of a test is its consistency

^[34] The 50 questionnaires were also re-analysed using SPSS a statistical program.

^[35] University of Manchester.

[36] 5 response terms from a total of 15 questionnaires (90 items per questionnaire). Much of the information requested was factual, with few questions open to interpretation.

^[37] Using their Data Entry Service arranged by Professor Tymms, PIPS Project.

^[38] About 10

[39] A full description of the questionnaire data is described in Chapter 5.

^[40] Value-added is the learning achieved over the period between the two assessments, in relation to national predictions for the 'End of Reception' scores, ascertained from the overall 'Start of Reception' data and allowing for the effects of maturation. Children not progressing 'as expected', may be given a negative value-added score, despite having progressed in terms of maturation.

^[41] These appear in the assessment booklet, but for the test purposes had been extracted and adapted to remove mention of school, and reprinted by the researcher on a separate sheet.

^[42] The intention was to contact all families with at least one child who would be four on the first of September. On the questionnaire, families tended to supply their child's age rather than their month and year of birth, and therefore contact was made with some families whose children fell slightly outside this criterion. Having established contact with such families by letter followed up, where possible, with a telephone call, it was considered sensible to include such children, particularly since national PIPS 'R' data also included some children who fell either side of the desired age spectrum.

^[43] 'E.O.' was, at the time, the largest home-education support organisation in the UK, known to this researcher.

^[44] The family who declined to assist with the PIPS 'R' stage of the research did, at a later date, initiate contact and indicate that they were willing to be interviewed. An interview was undertaken as part of the larger scale research and the family further assisted with psychological testing using the Rutter (1993) and Goodman (1997) scales.

^[45] In view of the distance traveled by the researcher that day, the family was interviewed and other children in the family later went on to participate in other assessments connected with the research program.

^[46] Children were given a choice of researcher or parent administrator. This was to prevent children being disadvantaged by the researcher's dialect or through shyness with a stranger.

[47] The PIPS Baseline assessment needed to be administered within a few weeks of the start of the September school term.

[48] QSR NUD.IST 4 is a computer package designed to aid users in handling non-numerical and unstructured data in qualitative analysis. NUD.IST does this by supporting processes of indexing, searching and theorizing.

[49] Curriculum Evaluation and Management Centre at Durham University.

[50] Later re-named the Qualifications and Curriculum Authority (QCA)

^[51] 'Cultural Capital is a measure of the amount of educational support in the home' (PIPS Project 1998e).

^[52] Section 'Commentaries on the PIPS Year 2 assessments - author's and participant's'

^[53] One more home-educated child completed the PIPS Year 2 'Maths' section (n=18) than the 'Reading' section (N=17).

[54] 2 x Year 2, 1 x Year 3 and 2 x Year 5

[55] Revised by Goyette, Conners and Ulrich in 1978.

^[56] This concept of adult expectation had already been noted, in parental definitions of whether or not their children could read. Parents often seemed to feel that a child of, say, seven years, who could not read an adult text fluently, was a 'non-reader'. This point is discussed in Section 3.2.2.

^[57] One example in Chapter 9 is where the 'history' of children registering on the Rutter Scale as displaying behavioural problems could be explored by reference to their family questionnaire and interview data.

^[58] Professor Larry K. Michelson, Professor of Psychology and Co-Director of the Stress and Anxiety Disorders Institute, Pennsylvania State University.

^[59] Professor Sir Michael Rutter is currently the Honorary Director of the Medical Research Council.

^[60] Used in the current research with 11 year olds only

^[61] paralysis on one side of the body

^[62] Scotland

[63] Defined as the study of the distribution and determinants of disease

^[64] The Revised Rutter Parent Scale for School Aged Children known as the Revised Rutter Scale.

^[65] Item no. 12 'Emotional Difficulties': also described 'anxious-fearful', 'neurotic' (Elander & Rutter 1996)

^[66] Item no. 8 'Conduct Difficulties': also described 'antisocial', 'hostile-aggressive' (Elander & Rutter 1996)

^[67] Item no.34 'Hyperactivity/Inattention'

^[68] Item no 3 'Prosocial'

^[69] It appears that Goodman (1994) used the Rutter Scale developed prior to Rutter's (1993) Revised Rutter Scale, published in 1997 by NFER Nelson with some modifications.

^[70] There are various elements of the SDQ. The questionnaire is designed for use by both teachers and parents: there are school aged and pre-school aged versions besides informant and self-reported versions. Only the informant and self-reported scales were used with this research.

^[71] It is worth recalling that whilst the original sample was self-selected, return of the questionnaire was simplified for many participants, particularly those approached by their LEAs, who were provided with a stamped addressed envelope. Those families who replied through the Internet also had the task of returning the form made easier by simply having to email their response. Thus, the sample was drawn from diverse sections of the home-education community.