Personality types and recidivism among Borstal trainees

Personality types and recidivism among Borstal trainees

Person. in&Cl Diff. Vol. 4, No. 2, pp Printed in Great Bntain 165-170. 0191-8869/83/020165-06$03,00/O 1983 PergamonPress PERSONALITY AMONG Ltd ...

480KB Sizes 8 Downloads 62 Views

Person. in&Cl Diff. Vol. 4, No. 2, pp Printed in Great Bntain

165-170.

0191-8869/83/020165-06$03,00/O

1983

PergamonPress

PERSONALITY AMONG

Ltd

TYPES AND RECIDIVISM BORSTAL TRAINEES*

BARRY J. MCGURK~ and ALEXANDER W. MCEWAN H.M. Prison Frankland, Finchale Ave. Brasside, Durham DHl 5SB, England and RAE E. MCGURK Social Services,

Sunderland.

England

Summary-A cluster analysis of the 16 Personality Factor Questionnaire, Hostility and Direction of Hostility Questionnaire and Psychological Screening Inventory scores from 181 delinquents at Deerbolt Borstal, England. revealed four personality types which were labelled ‘Withdrawn’, ‘Normal’, ‘Disturbed’ and ‘Truculent’. A 3-year reconviction follow-up showed that the groups differed in terms of recidivism. Sixty-four percent of the Withdrawn group reconvicted whilst 874;; of the Normal, 79”4 of the Disturbed and 76% of the Truculent group reconvicted. Controlling for the influence of number of previous convictions the Withdrawn group differed significantly from the other groups in terms of recidivism. The results are discussed with reference to similar results obtained in a previously reported study in a detention centre.

INTRODUCTION

In studies of recidivism the predictive ability of demographic and criminological variables has been established with both age (Hood and Sparks, 1970) and number of previous convictions being consistently related to recidivism. Conversely the evidence relating to the relationship between psychological variables and recidivism is equivocal. Studies attempting to identify the characteristics of recidivists by contrasting recidivists and non-recidivists using personality tests have generally produced negative results (Mack, 1969; Fraas and Price, 1972; Blum and Chagnon, 1967; Francis, 1970). Other studies, however, have adopted the strategy of determining whether delinquents of different personality types have responded differently to treatment programmes. These studies (Jesness, 1965; Grant and Grant, 1959; Palmer, 1973) have generally reported positive results and have almost exclusively been conducted in the United States. An attempt to follow a similar line of enquiry in England has recently been reported (McGurk, McEwan and Graham, 1981). A cluster analysis of responses to personality tests from detention-centre inmates revealed four lower rate groups and one group, which was labelled ‘Anxious’, reconvicted at a significantly than the other groups. It was reasoned that “detention centre regimes which have often been described in terms of providing a ‘short sharp shock’ could be expected to have the strongest deterrent effect on anxious individuals.” (p. 163) However, the suggestion was made also that anxious individuals may reconvict at a relatively lower rate irrespective of the nature of the regime experienced. The present study contributes to the literature by exploring the relationship between personality types and recidivism in a borstal with a regime emphasizing education and trade training in a relaxed environment within a secure perimeter. In addition the resultant personality types are contrasted on background and educational and criminological variables. *The opinions expressed in this paper are the authors’ PriSOn Department, tTo whom all reprint requests should be addressed.

and do not necessarily

represent

those

of the Home

Office,

I 66

BAKKY J. McGu~rt

PI cd

METHOD

The study was carried out at H.M. Borstal, Deerbolt, County Durham which is a custodial institution for approx. 300 young men aged between 15 and 21 years of age (x = 18.37. SD = 1.67) serving sentences of borstal training. This sentence, an intermediate length training sentence, is indeterminate--but not less than 6 months and not more than 2 years. The vast majority of trainees, however, spend 33 weeks (for administrative reasons) at the institution. Generally it tends to be used by the court for offenders for whom non-custodial measures and the shorter detention-centre sentence have failed. Procedure Of a sample of 250 receptions into the centre in 1975, 181 were able to complete the battery of tests in their first week at the institution; the remainder were unable to complete the tests either because they had a reading age lower than 10 years or because a small number were unavailable for testing due to other institutional requirements. The tests were administered in the following order: Raven’s Matrices (Raven, 1960), a specially devised test of literacy for use in prisons, the Hostility and Direction of Hostility Questionnaire (HDHQ; Caine, Foulds and Hope, 1967), the Psychological Screening Inventory (PSI; Lanyon, 1973) and the 16 Personality Factor Questionnaire, Form E (16PF; Cattell, Eber and Tatsuoka, 1970). In addition the following data were collected: number of previous convictions, record of previous institutional experience and the presence or absence of past offences in a standardized table of offence categories which is completed for all receptions received into the borstal. The raw scores of the Ss on the three personality tests were subjected to a cluster analysis using Ward’s (1963) method from the computing package ‘C‘LUSTAN’ (Wishart, 197X). This technique is based on a distance function which is the sum of the squared deviation of two series of scores. The smaller the value of the function the greater the similarity of the series of scores. Firstly, the pair of scores having the greatest similarity is found and the number of groupings is progressively reduced until only two groups remain. At each stage the next subject is chosen on the basis of the smallest increase in the total within-groups variation. Subsequent to the release of the Ss from the borstal a 3-year follow-up of reconvictions was carried out. The reconviction rates of the clusters which emerged from the cluster analysis were contrasted, firstly. using a Chi-square test and. secondly, to control for the influence of previous convictions. using a logistic regression analysis from the computer programme GLIM (Baker and Nelder, 1978). The clusters were then contrasted on the intelligence. educational and criminological variables and, finally. the cluster solution was subjected to a discriminant function analysis from the Stutisticul Puckug~ jbr the Sociul Sciencrs (Nie, Hadlai-Hull, Jenkins, Steinbrenner and Bent. 1975).

RESULTS

Examination of the error terms generated by the cluster analysis revealed the first large change to occur at the four-cluster level. Accordingly, this solution was adopted and Table 1 shows the mean scores of each cluster and a one-way analysis of variance across the clusters for the HDHQ, 16PF and PSI scales. A brief description of the clusters is as follows:

Type 1 contains 47 Ss (25.9”,, of the sample). They are the only group which are intropunitive in terms of their hostility. They are also self-sufficient, sober and taciturn and, in common with Type 3. shy and timid. This taken together with them being the least expressive and outgoing of the identified groups has led to them being labelled as ‘Withdrawn’.

Personality

Table

1. Four

types

of young

types

and recidivism

delinquents:

One-way

Type 1

*P < 0.05: ***p

analysis

borstal

of variance

Type 2

(Withdrawn) HDHQ Guilt HDHQ Self-criticism HDHQ Projected Hostility HDHQ Acted out Hostility HDHQ Criticism of Others HDHQ General Hostility HDHQ Direction of Hostility 16PF Outgoing I6PF Intelligent 16PF Emotionally Stable 16PF Assertive 16PF Happy-go-lucky I6PF Conscientious 16PF Venturesome 16PF Tenderminded 16PF Suspicious 16PF Imaginative 16PF Shrewd 16PF Apprehensive 16PF Experimenting 16PF Self-sufficient 16PF Controlled 16PF Tense PSI Alienation PSI Social Non-conformity PSI Discomfort PSI Expression PSI Defensiveness

among

(Normal)

.r

SD

x

SD

3.43 4.96 1.85 4.53 5.74 20.5 I 0.53 3.53 6.98 4.28 2.70 3.96 4.40 2.30 2.21 2.66 3.55 4.00 4.81 5.19 5.23 4.08 3.60 7.60 13.17 11.17 9.72 I 1.68

1.43 2.38 1.18 1.44 2.50 5.35 6.68 1.60 0.99 1.87 1.53 1.79 1.51 1.86 1.64 1.52 1.46 1.28 1.85 1.60 1.86 1.38 2.10 2.21 2.96 5.01 3.23 I .89

2.33 3.43 1.89 4.54 6.00 18.16 3.23 4.80 7.15 5.25 3.64 6.36 4.89 5.21 1.98 3.05 3.23 3.69 3.10 4.85 2.05 5.56 2.31 7.54 13.80 6.67 14.36 11.52

1.51 1.95 1.38 1.54 2.37 5.11 5.57 1.64 1.08 1.25 1.86 1.62 1.59 1.69 1.61 1.81 1.35 1.23 1.60 1.46 1.50 1.32 1.81 2.30 2.63 3.57 3.26 2.18

trainees

167

of HDHQ,

Type 3 (Disturbed) ‘z SD 1.37 2.12 1.72 2.60 1.64 5.14 6.07 1.78 1.57 1.70 1.76 1.74 1.34 1.61 1.69 1.74 1.32 1.31 1.80 1.69 1.73 1.56 1.97 2.77 2.58 4.22 3.23 2.53

4.70 7.18 3.95 7.34 8.52 31.68 - 0.79 4.84 5.88 3.61 4.52 5.39 3.73 3.32 2.77 4.39 3.30 3.86 4.93 4.43 3.21 4.11 4.82 11.06 16.21 14.75 12.75 9.04

16PF and PSI scores Type 4 (Truculent) B SD 2.24 3.76 2.82 9.06 8.00 25.94 10.12 4.47 7.12 3.47 5.76 6.53 2.00 4.00 2.35 5.76 3.47 3.53 4.41 5.47 2.53 3.82 3.59 7.35 17.29 9.29 14.00 9.18

1.56 1.53 1.19 1.43 1.94 3.77 5.21 1.50 I .05 1.50 1.30 1.42 1.32 1.62 1.11

1.09 0.87 1.23 2.06 1.12 2.12 0.88 2.48 2.03 2.42 2.80 3.34 1.38

F 29.45*** 33.84*** 25.96*** 44.58*** 19.83*** 77.58*** 14.75*** 6.72*** 12.51*** 12.20*** 17.47*** 20.46*** 19.08*** 27.54*** 2.42 21.06*** 0.60 0.84 12.91*** 3.06* 31.30*** 16.03*** 15.25*** 27.73*** 18.53*** 37.99*** 19.35*** 19.75***

4 0.001.

Type 2

This is the largest group containing 61 Ss (33.7% of the sample). They are less hostile than any other group and relaxed, controlled and self-assured. They also have a ‘normal’ PSI profile and hence were labelled as ‘Normal’. Type 3 Type 3 contains 56 Ss (30.9% of the sample). They have the highest level of general hostility and are the most self-critical and the most critical of others of the four groups. They are tense and they score highest on the alienation-indicating similar responses to those of psychiatric patients-and discomfort scales of the PSI. Accordingly, they were labelled as ‘Disturbed’. Type 4

This is the smallest group containing only 17 Ss (9.4% of the sample). These subjects score highest on the HDHQ scales measuring extrapunitive hostility and acted-out hostility. They are also assertive, expedient, suspicious, happy-go-lucky and socially non-conformist. This group was called a ‘Truculent’ group. Of the total sample of 181 Ss, 140 (77.7%) were reconvicted within 3 years of release from the borstal. Table 2 shows the reconviction rates of the four personality types. The Withdrawn Table 2. Four

No. of cases No. of recidivists No. of non-recidivists Recidivism rate (‘I,) I’=

8.12 (#=

3); P < 0.05.

types of young

delinquents

Type 1 (Withdrawn)

(Normal)

47 30 17 64

Type 2 61 53 8 87

and recidivism

Type 3 (Disturbed) 56 44 12 79

Type 4 (Truculent) 17 13 4 76

I (,S

Table

3. Four

types of young delinquents: logistic analysis of reconviction rates Percent

No. of prcviou< convictions

reconvicted Others (Normal + Disturbed + Truculent)

Type I (Withdrawn)

OL3 4-7 X+

regression

59 59 xx

83 Xl x2

72 = 5.76 (C// = I); P < 0.02.

group have a 64”,, reconviction rate. the ‘Normal’ group 87’:/,, the ‘Disturbed’ group 79’:; and the ‘Truculent’ group a 76’!,, reconviction rate. As can be seen there is a significant difference in reconviction across the clusters. The significance of this difference, however, is entirely due to Type I, the Withdrawn group. Types 2, 3 and 4 do not differ in their recidivism rates (x’ = 1.8, NS). As it may be the case that this difference between Type 1 and the rest could be due simply to difference in age and previous convictions the influence of these variables was examined. No significant difference was found in age between any of the four types. In addition differences in age within clusters, between reconvicted and non-reconvicted Ss. were examined and no significant differences were found. A KruskaLWallis one-way analysis of variance (Siegel, 1956) failed to reveal a significant difference in number of previous convictions across clusters but the analysis of between-cluster differences with a Mann-Whitney ‘C”-test (Siegel, 1956) revealed Type 2. the Normal group (Z = 2.01. P < 0.05)and Type 4, the Truculent group (2 = 2.17, P < 0.05) to have recorded greater numbers of previous convictions than Type 1, the Withdrawn group. Given these differences the reconviction rate of Type I was compared with the other three clusters by a logistic regression analysis which permits the comparison of reconviction rates whilst controlling for the influence of previous convictions. This is shown in Table 3. Thus, having controlled for previous convictions. there remains a significant difference between the Withdrawn group and the other subtypes identified. Next. the subtypes were contrasted on their performance on Raven’s Matrices, the standardized prison literacy test and on the institutional history and offence-related variables. There were no differences between the subtypes in their scores on intelligence and literacy tests, nor were there any differences in their previous institutional experience. In the analysis of the offence-related data it was discovered that the majority of categories failed to present information suitable for between-groups comparisons. For example, in no fewer than I3 of the categories considered less than 25”,,, and often less than 10”,,, of the total number of Ss had no recorded convictions. For the four remaining categories the presence and absence of recorded off‘ences for each subtype were contrasted using Chi-square tests, but no between-groups dilyercnces were found. Finally. the results of the cluster solution were analysed by discriminant function to check the potential utility of the classificatory model employed. The results of this analysis are presented in Table 4.

Table 4. Four types

of young delinquents

and predicted function analysis PredIcted

Actual

group

membership

Type I (Withdrawn) ‘Y

Type I (Withdrawn. N = 47) Type 2 (Normal. ,Y = 61) Type 3 (Disturbed. ,Y = 56) Type 4 (Truculent. ,V = 17) Pcrccntage

of GLSC’Scorrectly

45 (95.7)

7 (3.3) 0 (0.0) 0 (0.0) classified

= 9S.M

group

group

membership

membership

Type 2 (Normal) N 2 (4.3)

58(95.1) 0 (0.0) 0 (0.0)

by discrlminant

(I’,, in parentheses)

Type 3 (DIsturbed) N

Type 4 (Truculent) .Y

0 (0.0) 0 (0.0) 53 (94.6) 0 (0.0)

0 (0.0) I (I .6) 3 (5.4) 17(100.0)

Personality

types

and

recidivism

among

borstal

trainees

169

DISCUSSION The results of the present study confirm the previously cited American work demonstrating interaction effects between personality types and recidivism among delinquents. The current study found similar personality types among borstal trainees as the previous study by McGurk rt ul. (1981) of detention-centre inmates. Three of the four types found in both studies were remarkably similar being labelled each time as Normal, Disturbed and Truculent. However the groups in each study which differed significantly from the others in terms of reconviction are less similar being Iabelled as Withdrawn in this sample and as Anxious in the detention-centre sample. Both of these groups were the only intropunitive subtypes in their respective samples and were also shy and timid but the Withdrawn group in this study did not score highly on those scales measuring anxiety. Bearing in mind the identified similarities and differences between the Withdrawn group in this study and the previously described Anxious group it is an interesting possibility that it was the most highly-anxious individuals in the latter group which did not reconvict. This interpretation would be supported by an examination of the relative proportion of the total samples constituted by the Withdrawn (25.9”/,) and Anxious (36.5”/,) subtypes in the respective studies. However this suggestion is mere speculation as is the possibility that the small Truculent group discovered (9.4%) compared with the detention-centr,: sample (33.9%) is explicable in the maturing of these individuals leading them to be represented in the Norma1 or Disturbed groups, both being subtypes with whom the Truculent subjects share common scoring patterns on certain scales. Further research is obviously required to elucidate this question of the changing relative proportion of the different subtypes in the various samples and this will be reported in the near future. However, the importance of the present results lies in the identification of the Withdrawn group who are shown to reconvict to a significantly lesser extent than the Normal, Disturbed and Truculent groups. This result underlines the importance of a typological approach in the study of recidivism and further assists in the explanation of the significantly lower reconviction rate of the Anxious group in the previous detention-centre experiment. It would appear from the present results that the impact of a specific regime did not deter the Anxious individuals found in the previous study. Alternatively it is likely that individuals possessing certain of the personality characteristics of the Withdrawn and Anxious subjects in the respective samples will reconvict at a lower rate than other subgroups irrespective of the nature of the regime experienced. A~Xno~~lrd~rmenrs~The authors would like to thank Mr R. Nash, Governor of Deerbolt Borstal at the time the study was undertaken, for his assistance and Mr J. S. Christie for help with data collection. In addition thanks are due to Sheila Spcirs and David Thornton of the Young Offender Psychology Unit, Prison Department Home Office for obtaining reconviction data and providing advice on logistic regression analysis, respectively. Special thanks are due to Jackie Dickson for typing the manuscript at very short notice.

REFERENCES BAKEK R. J. and NELDERJ. A. (1978) GLIM. Numerical Algorithm Group. Oxford. BLUU F. J. and CHAGN~N M. (1967) Extraversion and subsequent recidivism for a selected group of young offenders. Can. J. Correcl. 9, 94-98. CAINET. M., F~ULDS G. A. and HYPE K. (1967) Manual of the Hustiliiy and Direction qf’Hostility Questionnaire. Univ. of London Press, London. CATTELL R. B., EBER H. W. and TATSUOKA M. M. (1970) Handbook for the Sixreen Personality Facior Questionnaire. NFER, Windsor, Berkshire. FKAAS L. A. and PKICE R. L. (1972) The Jesness Inventory as a predictor of AWOL recidivism. PsJjchol. Rep. 31, 741-742. FKAXIS R. D. (1970) Recidivism, types of crime and extraversion. Aust. N.Z. J. Grim. 3, 92-94. GKANT J. D. and GKAN~ M. Q. (1959) A group dynamics approach to the treatment of nonconformists in the Navy. Ann. Am. Acari.

Pal. Sot.

SC. 322,

126-135.

Hook R. and SPAKKS R. (1970) &_v Issues in Criminolog_p. Weidenfield & Nicholson, London. JESNESSC. F. (1965) Thr Fricot Ranch Smdy. Consulting Psychological Press, Sacramento. LANYON R. I. (1973) Manual qf the Psychological Screening Inventory. Research Psychologists Press. New York. MA(.K J. L. (1969) The MMPI and recidivism. J. uhnorm. Psvchol. 74, 612-614. MCGUKK B. J., MCEWAN A. W. and GRAHAM F. (1981) Personality types and recidivism among young delinquents. Br. J. Grim. 21(2), 159-165.

Ii0

BAKKY

J. MCGIIKK c’f rrl.

NIL N. M.. HAOLAI-HULL C.. JF.UI