Appendix C — Further information on the sample

This Appendix provides further information on the process of sampling, and on the eventual sample itself.

C.1 Variations on ‘innocence’ in sampling

Chapter 3 described how one of the original sampling criteria was dropped, because it proved difficult to identify prisoners who were maintaining innocence. This Appendix describes how that became clear, and sets out some variations in how the term ‘innocence’ was understood by me and a prison psychologist working at Swaleside.

While I was in the early stages of constructing the sample, and still intending to try and apply the original eligibility criteria described in Chapter 3.3.1, a psychologist at Swaleside kindly offered to help me gauge how feasible it would be to use all six criteria, by generating two random dummy samples of twenty-six men from the 191 there who met criteria 1, 5 and 6, and then reviewing the OASys assessments for these men to attempt a classification. This felt unduly complicated, but it would have been churlish to decline the offer so early in the fieldwork. In the event, the exercise was unexpectedly useful. It clarified that trying to exclude potential participants who were ‘maintaining innocence’ or ‘not taking responsibility’ was futile, since these were highly nuanced issues, in ways the risk assessment documents were largely indifferent to.

This became clear through differences of interpretation between me and the psychologist, as we worked through the dummy samples. Table C.1 presents a rough-and-ready inductive typology of what (to me) were different attitudes to the index offence, all of which she saw as communicating a claim of ‘innocence’. By her assessment, twelve and seventeen members respectively of each dummy sample of twenty-six fell under this description. By mine, only three or four people in each group fitted the ‘Denial’ description in Table C.1, the only stance to which I would have intuitively applied the label. Moreover, in the OASys records, prisoners’ reported views often differed markedly from the factual evidence the records described.¹

Table C.1: Typology of styles of ‘innocence’ evident in the dummy sampling exercise

Description	Typical narratives about the offence
Denial	I wasn’t there It was a case of mistaken identity I won’t talk about it with anyone The police framed me The victim isn’t actually dead
Deflection	I was there and I was involved, but my co-defendant(s)/others did it I was there and there was a fight but they were alive when I left them and I don’t know what happened
Reinterpretation	I did it, but it wasn’t murder It wasn’t as bad as they said
Dissociation	I was so intoxicated (or so psychiatrically unwell) that I might have done it, but I can’t remember it or don’t recognise these as my actions

In short, it became clear that a strict approach to the original sample eligibility criteria was unworkable. First, there was no way of confidently estimating from prison records whether someone ‘maintained innocence’; partly because, second, it seemed possible that some maintained this stance habitually, for the benefit of the prison, while possibly making different claims to other audiences. Third, neither ‘innocence’ nor ‘responsibility’ were obviously meaningful distinctions by which to divide the sample. To my mind, of the attitudes described in Table C.1, ‘deflection’, ‘reinterpretation’ and ‘dissociation’ all conveyed meaningful information about a response to censure, and only ‘denial’ suggested a flat denial of any ‘moral responsibility’, still less a claim that the speaker was ‘innocent’.

C.2 The sample’s representativeness of the eligible populations at both prisons

The sample was, broadly, representative of the populations of the two fieldwork prisons. This can be seen in Table C.2, which summarises how the final sample and the eligible populations at the two sites compared on their sentence characteristics. The point to note here is that the samples at both prisons came close to the populations for the age and sentence characteristics shown.

Table C.2: Sentence characteristics of the sample and population at both sites. Source: P-NOMIS data supplied by each prison

	Statistic		Age when sampled	Age when sentenced	Tariff (years)	Years served	Years served ÷ tariff (as %)
Swaleside	Minimum	Sample Population	22 21	16 16	12 7	2 2	7% 6%
	Average (s.d.)	Sample Population	43.8 (12.9) 40.2 (12.2)	33.4 (12.8) 31.9 (11.6)	20.0 (5.2) 20.0 (5.3)	9.8 (6.1) 7.8 (4.8)	55% (41%) 43% (33%)
	Maximum	Sample Population	77 77	63 63	30 35	27 27	180% 180%
Leyhill	Minimum	Sample Population	31 31	17 17	8 8	7 7	66% 67%
	Average (s.d.)	Sample Population	50.7 (14.1) 54.7 (12.4)	32.7 (14.7) 32.6 (12.9)	14.3 (3.2) 14.5 (3.2)	17.5 (8.2) 21.7 (9.0)	125% (59%) 153% (72%)
	Maximum	Sample Population	82 91	74 82	20 21	35 44	250% 420%

Table C.3, meanwhile, compares the per-prison samples and the populations of the two prisons in terms of ethnicity and age. The point to note here is that the sample at Swaleside was broadly representative of the population in terms of its ethnic composition. At Leyhill, this was not true because of the deliberate oversampling of non-white prisoners described in Chapter 3.

Table C.3: Descriptive statistics showing ethnicity of the samples and eligible populations at both sites. Source: P-NOMIS data supplied by each prison.

Ethnicity	Swaleside		Leyhill
	sample	population	sample	population
Asian or Asian British	4 (13%)	17 (9%)	—	—
Black or Black British	9 (29%)	64 (33%)	3 (17%)	5 (8%)
Mixed	1 (3%)	6 (3%)	—	—
Chinese or other	—	1 (1%)	—	—
White	17 (55%)	102 (53%)	14 (78%)	60 (91%)
Unknown or not stated	—	2 (1%)	1 (6%)	1 (2%)
All	30 (100%)	191 (100%)	18 (100%)	66 (100%)

Both tables show that the sample was broadly representative of those eligible to participate at each site.

Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder D  Further information on documents and risk assessments used in the research B  Interview materials Frontmatter Welcome Abbreviations Epigraph Thesis 1  Punishment and moral obligation 2  Ethical life and the long prison sentence 3  Methods and ethics 4  Becoming ‘a murderer’ 5  The life course and the life sentence 6  Risk and self-governance 7  Conclusions Backmatter References Appendices A  Changes in murder sentencing since the formalisation of the ‘tariff’ system B  Interview materials C  Further information on the sample D  Further information on documents and risk assessments used in the research Appendices C  Further information on the sample

Appendix C — Further information on the sample – Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder Appendix C — Further information on the sample – Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder Appendix C — Further information on the sample – Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder Moral messages, ethical responses: punishment and self-governance among men serving life sentences for murder

---

1. For example, one dummy sample member who maintained a strong claim of innocence admitted being at the scene, chasing and threatening the victim with a knife (the notes said this had been caught on CCTV), but claimed that he had broken off the pursuit; someone else must have stabbed the victim. Another in the same category maintained that the victim wasn’t dead, though couldn’t explain (so said the notes) how it came to be that a body had been found and identified by family members.