APPENDIX A
RESEARCH: THE EFFECT ON
MAGISTRATES OF KNOWING OF A
DEFENDANT'S CRIMINAL RECORD
A.1 In the consultation paper we referred to research into the effect on juries of knowing of a defendant's previous convictions conducted by Dr Sally Lloyd- Bostock, Research Fellow of the Centre for Socio-Legal Studies and of Wolfson College, University of Oxford. Dr Lloyd-Bostock has since completed a second project addressing how the same evidence affects magistrates. This research has been published in the Criminal Law Review. [1] We summarise it here.
OUTLINE OF THE STUDY
A.2 This project was modelled closely on its forerunner.[2] The study focused on how the knowledge that the defendant had a previous conviction affected magistrates and whether that effect differed depending on the nature of the conviction, including its similarity to the offence charged and the age of the conviction. The study also considered what assumptions magistrates made, on the basis of knowledge of a previous conviction, about the defendant's likelihood to reoffend. Magistrates were also questioned about whether they thought it was useful or appropriate to disclose previous convictions to fact-finders and what problems might arise for them from such disclosure. In total, 222 magistrates participated in the research.[3]
Methodology
A.3 The research was conducted by way of simulation experiment. Groups of magistrates viewed a video of a condensed reconstruction of one of two trials: one relating to a charge of handling stolen goods, the other to one of indecent assault on a woman. There were eight versions of the trial of each offence. In each there was a neutral voice-over, when the defendant came to the witness box. In each the judge in his summing up gave either no, or differing information about the defendant's criminal history as follows:
(1) no mention of previous conviction or good character (the base condition);
(2) good character;
(3) recent[4] similar conviction (to offence charged);
(4) old[5] similar conviction (to offence charged);
(5) recent dissimilar conviction;[6]
(6) old dissimilar conviction;
(7) recent conviction for indecent assault on a child; or
(8) old conviction for indecent assault on a child.
A.4 The judge gave the appropriate standard direction depending on whether a previous conviction was disclosed or not. Where it had been stated that the defendant was of good character, the judge gave the following direction:
You have heard that the defendant has no previous convictions. In deciding whether the prosecution has made you sure of his guilt, you should have regard to the fact that he is a man of good character. You should take it into account in his favour in the following two ways.[7]
A.5 Where nothing was disclosed about the defendant's character, the judge did not say anything about the defendant's character. Where a previous conviction had been mentioned, the judge said,
You have heard about the defendant's previous conviction for […]. You must not assume that the defendant is guilty or that he is not telling the truth because he has been convicted on a previous occasion. It is for you to decide the extent to which his previous conviction assists you.
A.6 The magistrates watched a single version of the video in groups of between 12 and 18. Immediately after watching it they completed Questionnaire 1 in which they expressed their initial verdict, a rating of confidence in that verdict, and two ratings (one on a five-point scale and one on a 0–100% scale) of the likelihood that the defendant committed the offence.
A.7 They were then divided into groups of three to arrive at a bench verdict. After group deliberation they then completed Questionnaires 2 and 3. Questionnaire 2 asked for the bench's decision, the individual magistrate's personal verdict and again, ratings of the likelihood that the defendant committed the offence. Questionnaire 3 asked the magistrates for (i) their impressions of the defendant (giving ratings for credibility, trustworthiness and propensity to commit particular crimes); (ii) their views on whether fact-finders should be told of previous convictions; (iii) their beliefs about people's propensity to re-offend or to lie in court depending on previous convictions and (iv) their experiences of discovering or realising that the defendant has previous convictions and the extent to which this is a source of difficulty for them.
A.8 The results were tested for statistical significance: that is the probability that such data would be produced by chance. In this study, as in that conducted with juries, a significance level of less than 0.05 (one in twenty) is used, meaning that a result which has a less than one in twenty likelihood of occurring randomly is treated as significant. Naturally, the significance level of particular results and confidence in the reliability of the findings is increased where analyses of different variables show a similar result or pattern of results. Where a large number of statistical tests are carried out, however, the probability of one reaching a significance level by chance increases.
SUMMARY OF RESULTS AND ANALYSIS
A.9 The first part of this analysis, concerning likelihood and belief in guilt, examines only the results of the groups that viewed the videos concerning previous conviction scenarios one to six. The results generated by the videos involving indecent assault on a child are discussed at paragraph A.24 below.
Likelihood scores
A.10 Below is a table of the average score (from one to a hundred) given by magistrates on their immediate individual analysis and after deliberation in a bench, of the likelihood that the defendant committed the offence.
Table 1
| Previous conviction Information |
Average likelihood rating Individual (predeliberation) |
Average likelihood rating Bench (postdeliberation) |
| Old similar |
65.5 |
65.3 |
| Recent similar |
60.9 |
59.1 |
| Old dissimilar |
56 |
58.5 |
| Base |
41.6 |
37.8 |
| Good character |
39.1 |
41.5 |
| Recent dissimilar |
37.4 |
34.9 |
A.11 The significance level between the sets of the differing conviction information only reached 0.05 in the three highest-scoring groups. The relationship between category of previous conviction and likelihood score is more significant after deliberation, following the same general pattern.
A.12 As occurred in the Jury Study, a similar previous conviction is associated with the higher perceived probability that the defendant committed the offence currently charged. In the present study both old and recent similar convictions have a significant effect. However, in the Jury Study, whilst recent similar convictions had the greatest effect on the likelihood score, old convictions made little or no difference to the likelihood estimates.
A.13 A recent dissimilar previous conviction appeared, as in the Jury Study, to be of positive help to the defendant. The dissimilarity neutralises the effect of a conviction, possibly invoking a stereotype of a person who commits that type of offence, not the type of offence charged. Unlike in the Jury Study, however, the small advantage of a recent dissimilar conviction over no information or good character is not significant.
Effect of previous conviction information on verdict
Individual magistrates
A.14 The table below shows the percentage of magistrates recording guilty verdicts and not-guilty verdicts (in brackets) before deliberation, after deliberation, and as a bench.
Table 2
| Conviction information |
Predeliberation [%] |
Postdeliberation [%] |
Bench verdict [%] |
| Old dissimilar |
24.2 (75.8) |
24.2 (75.8) |
20 (80) |
| Old similar |
23.1 (76.9) |
23.1 (76.9) |
12.5 (87.5) |
| Recent similar |
21.4 (78.6) |
21.4 (78.6) |
25 (75) |
| Good character |
19.2 (80.8) |
11.5 (88.5) |
0 (100) |
| Base |
17.2 (82.8) |
13.8 (86.2) |
0 (100) |
| Recent dissimilar |
11.1 (88.9) |
7.4 (92.6) |
0 (100) |
A.15 Unexpectedly, old dissimilar convictions were associated with the highest rate of guilty verdicts before and after deliberation. This was not the case in the Jury Study. This finding was consistent amongst the three groups that watched the video in question, but did not appear at a statistically significant level of variance. The "dissimilar conviction" was for a section 18 assault. It appears that a conviction for violent assault was associated with a consistently negative view of the defendant by magistrates.[8] Only where the conviction is recent does the dissimilarity outweigh this negative impression.
Bench verdict in relation to magistrates' belief in guilt.
A.16 The table on the following page demonstrates the relationship between the
likelihood scores and the bench verdict. The likelihood scores were recorded on a
five-point scale relating to belief in the defendant's guilt.
Table 3
| Belief (predeliberation) (likelihood score) |
Convictions (%) Pre-deliberation |
Convictions (%) Post-deliberation |
| Definitely committed offence |
88.26 |
71.4 |
| Almost certainly committed offence |
55.4 |
35.3 |
| Probably committed offence |
6.3 |
6.9 |
| Probably did not commit offence |
2.5 |
2.6 |
| Definitely did not commit offence |
0 |
11.1 |
Effect of nature of previous conviction on the perception of the defendant
A.17 The magistrates were asked to score (between one and five) their initial impressions relating to the defendant's honesty and propensity to commit different offences. These results may not entirely reflect the effect previous convictions have on magistrates' opinions of the defendant because some magistrates were reluctant to answer the questions, seeing this as an invitation to express prejudices, or did not answer at all. It is possible that others may even have underestimated their own unconscious prejudices. The results follow a similar trend to the previous analysis concerning likelihood of guilt. However, the magistrates consistently viewed good character favourably and any conviction, recent or old and of any nature, unfavourably.
A.18 The following tables show the average rating (between one and five) given to defendants with each type of conviction information.
Table 4
| Previous conviction information |
Belief in evidence as witness |
Propensity to commit type of offence charged |
Trustworthiness |
| Section 18 assault |
2.33 (1) |
2.72 (5) [1] |
1.87 (1) |
| Indecent assault on a child |
2.87 (2) |
2.57 (4) [1] |
2.23 (2) |
| Handing stolen goods |
2.87 (3) |
2.33 (2) [1] |
2.46 (4) [1] |
| Indecent assault on a woman |
3 (4) [1] |
3.29 (6) [1,2,3,4] |
2.29 (3) |
| Good character |
3.46 (5) [1] |
1.52 (1) |
3.18 (5) [1,2,3,4] |
| Base |
3.48 (6) [1,2,3] |
2.34 (3) [1] |
2.69 (6) [1,2] |
A.19 The scores are numbered in order by the numbers in round brackets. Significant variation between sets is denoted by the latter set's corresponding number in square brackets; this tended to occur in the highest rated categories. In interpreting the table it should be remembered that a previous conviction for handling stolen goods was used both as a similar conviction to the same charge and as a dissimilar offence where indecent assault was charged, whereas indecent assault was only used as a similar offence.
No information
A.20 Where no information was given about the defendant's previous convictions, the magistrates were asked how many convictions they assumed the defendant had. Just over half the respondents (55.2%) thought that the defendant would have one or two convictions and a further 14% thought that the defendant had more convictions. Around a third (31%) assumed the defendant had no previous convictions.
Other aspects of magistrates' impressions of the defendant
A.21 Magistrates viewed defendants with a conviction for section 18 assault as more deserving of punishment and more likely to have got away with criminal offences in the past than all other categories. This was seen as the most relevant conviction in terms of employment of the defendant, more so than a dishonesty offence, except where the job involved caring for children where the indecent assault convictions produced worse scores.
FUTURE OFFENCES
A.22 Magistrates were asked to assess the likelihood that the defendant would commit various offences in the future. These offences included those used as previous convictions and currently charged, plus some offences with common attributes (burglary, robbery with violence, indecent exposure to children and rape).
A.23 Those with a previous conviction for section 18 assault were seen as more likely than others to commit offences of burglary, robbery with violence, section 18 assault and, second only to the defendant with a previous conviction for handling, handling stolen goods. However, this defendant was regarded as least likely to commit a sexual offence.
A.24 A previous conviction for indecently assaulting a child made the defendant perceived as more likely to commit sexual offences or intentionally cause grievous bodily harm, but not commit burglary, handling stolen goods or robbery. The Jury Study found that such a conviction produced a consistent statistically significant increase in the likelihood of the defendant committing all other offences, but this was not the view of magistrates.
A.25 These results suggest that dishonesty, violence and sexual offending are seen as significant components of offences likely to be repeated. It is perceived that a previous conviction for one offence with such a component makes it more likely that other offences with the same component will be committed by that defendant.
Experiences and beliefs about previous conviction information
A.26 All the magistrates were asked whether they or juries should be told of previous convictions. A large majority of magistrates thought that previous convictions should never be disclosed to either magistrates or juries.[9] The primary reason for this was the risk of bias,[10] but some magistrates cited lack of relevance or importance of the information.[11] Around 10% of magistrates considered the fact that juries did not benefit from the training that magistrates had received to disregard certain evidence. This training and awareness of current debates might also explain some of their reservations about the disclosure of previous conviction information.
A.27 Fifteen magistrates thought that they ought to be told about the defendant's previous convictions and a further 55 thought that they should sometimes be told. Ten magistrates said that it depended on the type or gravity of offence and four that the information should only be revealed if relevant to the current case. Rather more magistrates, 35 in relation to telling magistrates and 34 in relation to juries, thought that in the case of crimes involving patterns[12] previous conviction information should be revealed. However, some limited this to cases where the information was of strict relevance.[13]
A.28 The magistrates were asked how many convictions a person would need before being described as a persistent offender. The magistrates' responses ranged from two convictions to ten, with the majority responding between three and six. [14]
A.29 The magistrates were also asked to consider how recent a conviction needed to be to be helpful to fact-finders. The responses were diverse: 64 said within three years, 56 said within five years, 19 thought within ten years, three thought within 15 years and three thought within 20 years.
ASSISTING IN REACHING VERDICT
A.30 Magistrates were asked whether the fact that they had been told about the defendant's previous conviction helped them arrive at their verdict.
Table 5
| Comment |
Frequency (Percentage) |
| Very much |
1 (0.6%) |
| Quite a lot |
10 (6%) |
| A little |
25 (14.9%) |
| Not at all |
132 (78.6%) |
| TOTAL |
168[15] |
A.31 It is possible – indeed likely – that previous conviction information would affect the judgment of magistrates in ways of which they are not aware. These results are significantly different from those in the Jury Study where the simulated jurors found the information significantly more helpful.
OWN EXPERIENCES
A.32 Whilst the study showed that it was common for magistrates to learn of the defendant's previous convictions, only 22% said that difficulties arose as a result. The magistrates were asked whether knowledge of the defendant's previous convictions means that magistrates should withdraw. Forty-six magistrates thought that this was always cause for withdrawal. Twenty-six regarded this as necessary only where it would cause bias, eleven thought it necessary if the clerk advised it and ten where the defendant was known or the crime affected the magistrate personally.
A.33 Only 10% of magistrates had never realised that a defendant had previous conviction from recognising the defendant or family from previous court appearances. Experiences of such recognition were wide-ranging: three magistrates had this experience in over 200 cases, six in 100–200 cases, 22 in over 51–100 cases, 41 between 21–50, 27 in 11–20, 40 in 6–10 cases, 39 in 2–5 and only two had recognised a defendant just once.
A.34 There is no direct parallel between the figures relating to magistrates knowing that a defendant has previous convictions and those relating to withdrawal due to such knowledge. Asked approximately how often they had withdrawn because of knowing or learning of the defendant's previous conviction, 161 magistrates (72.5%) said they had never withdrawn for that reason. Those that had had occasion to do so had withdrawn once (6.3%) or twice (9%).
CONCLUSION
A.35 Any previous conviction unfavourably affected the magistrates' impressions and verdicts, except where the defendant had a recent conviction dissimilar to that charged. The effects are not large but they are significant and follow a consistent pattern which broadly mirrors that found in the Jury Study.
A.36 The main difference between the results of the two studies relates to assumptions made by the mock jurors about a defendant with a previous conviction for indecent assault on a child and those made by magistrates in relation to a previous conviction for section 18 assault. The mock jurors consistently associated a previous conviction for indecent assault on a child with a negative evaluation of the defendant, including a greater perceived likelihood that the defendant would commit further offences, whether sexual in nature or not. The magistrates, however, had a less favourable impression of the defendant with a previous conviction for section 18 assault, also assuming that that defendant had a greater likelihood of commiting further offences. This virtually supplanted the effect of indecent assault on a child found in the Jury Study, although it was not as strong or consistent. This effect may be explained by the magistrates' greater awareness of the kind of offence a section 18 assault represents.
A.37 Other differences between the studies include the effect of good character, which impressed magistrates far more than the mock jurors. Magistrates were more influenced by old convictions than the mock jurors. Furthermore, magistrates tended to be more confident in their verdicts: the likelihood of guilt ratings altered less before and after deliberation.
A.38 In general the results indicate that information about previous conviction is likely to affect magistrates' decisions despite their awareness of the dangers and their efforts to avoid bias. These findings do not offer confidence that the rules on admitting previous convictions can be safely relaxed for magistrates any more than for juries.
Ý
Ü
Þ
