boyhood behaviour problems as precursors of criminalty: a fifteen-year follow-up study

16
J Uh,ldl\y<hut H^yihmt Vol.J'i.pp. lyioJJ Pers^mun Press I.id. l ^ L i'nn(cd in Gtrai Brii Assirf-iarion fc.r Child HsyihoiuKy anil Pnyihlalry. BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALITY: A FIFTEEN-YEAR FOLLOW-UP STUDY SHKII.A MITCHELL* and PETE;R ROSA Department of Sociology, University of Stirling INTRODUCniON THKRF. I.S A widespread beliet at both "folk" and "professional" level that "the childhood shows the man, as morning shows the day" (John Milton, Paradise Lost, IV, 220). It is nol surprising therefore that many attempts have been made to establish relationships between childhood experiences and behaviour and subsequent involvement in criminal activities. Most of these studies, however, have been of a retrospective type, i.e. they have taken adults involved in criminal activity and, with or without the use of non-criminal controls, have attempted to work backwards and establish what these individuals were like as children, relying on the recall of the subjects (and possibly their parents). Such an approach clearly involves problems of accuracy of recall and of the introduction of bias caused by later events—problems which can be avoided by adopting a prospective ap])roach, i.e. by collecting con- temporary information on a group of individuals as children and then relating this to the same individuals' experiences in later life. An increasing number of such studies have been carried out in the last thirty years (see Robins, 1966) but they are still relatively rare and in most cases relate to attempts to follow-up individuals who were already .officially labelled as "deviant" in some way when they were first investigated (usually because they had already arrived either at a psychiatric clinic or in court). Very few studies exist which relate criminality to data collected from parents before the child concerned first came to official notice. This is scarcely surprising when it is considered that such an approach must involve, firstly, the behavioural screening of a fairly large representative sample of "normal" children, most of whom will never becx)me labelled as "offenders", and, secondly, facilities to obtain reliable information about the subsequent criminal "careers" of all those involved. We are therefore fortunate in Britain in already having such information produced by two longitudinal cohort studies, the National Survey of all children born in England, Scotland and Wales in one week of March, 1946 and the smaller but more specifically crime-oriented study of working-class London boys carried out by D. J. West and his colleagues. Each of these investiga- tions (Wadsworth, 1979; West and Farrington, 1977) has now published information which relates their subjects' criminal involvements up to the age of twenty-one to information gathered when they were younger. In neither case, however, has the focus been primarily directed at exploring the n s [ s lur oil-prints should be made to: Dr. Sht-ila Mitchell, Senior Lecturer, Department of S(itii)iogy, Univt-rsity of Stirling. Stirling FK9 4LA, Scotland. Accepted manuscript received 2'^ November 1979 19

Upload: sheila-mitchell

Post on 14-Jul-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

J Uh,ldl\y<hut H^yihmt Vol.J 'i .pp. l y io J JPers^mun Press I.id. l ^ L i'nn(cd in Gtrai BriiAssirf-iarion fc.r Child HsyihoiuKy anil Pnyihlalry.

BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OFCRIMINALITY: A FIFTEEN-YEAR FOLLOW-UP STUDY

SHKII.A MITCHELL* and PETE;R ROSA

Department of Sociology, University of Stirling

INTRODUCniON

THKRF. I.S A widespread beliet at both "folk" and "professional" level that "thechildhood shows the man, as morning shows the day" (John Milton, Paradise Lost,IV, 220). It is nol surprising therefore that many attempts have been made toestablish relationships between childhood experiences and behaviour and subsequentinvolvement in criminal activities. Most of these studies, however, have been of aretrospective type, i.e. they have taken adults involved in criminal activity and, withor without the use of non-criminal controls, have attempted to work backwards andestablish what these individuals were like as children, relying on the recall of thesubjects (and possibly their parents). Such an approach clearly involves problems ofaccuracy of recall and of the introduction of bias caused by later events—problemswhich can be avoided by adopting a prospective ap])roach, i.e. by collecting con-temporary information on a group of individuals as children and then relating this tothe same individuals' experiences in later life.

An increasing number of such studies have been carried out in the last thirty years(see Robins, 1966) but they are still relatively rare and in most cases relate toattempts to follow-up individuals who were already .officially labelled as "deviant"in some way when they were first investigated (usually because they had alreadyarrived either at a psychiatric clinic or in court). Very few studies exist which relatecriminality to data collected from parents before the child concerned first came toofficial notice. This is scarcely surprising when it is considered that such an approachmust involve, firstly, the behavioural screening of a fairly large representative sampleof "normal" children, most of whom will never becx)me labelled as "offenders",and, secondly, facilities to obtain reliable information about the subsequent criminal"careers" of all those involved. We are therefore fortunate in Britain in alreadyhaving such information produced by two longitudinal cohort studies, the NationalSurvey of all children born in England, Scotland and Wales in one week of March,1946 and the smaller but more specifically crime-oriented study of working-classLondon boys carried out by D. J. West and his colleagues. Each of these investiga-tions (Wadsworth, 1979; West and Farrington, 1977) has now published informationwhich relates their subjects' criminal involvements up to the age of twenty-one toinformation gathered when they were younger.

In neither case, however, has the focus been primarily directed at exploring then s [ s lur oil-prints should be made to: Dr. Sht-ila Mitchell, Senior Lecturer, Department of

S(itii)iogy, Univt-rsity of Stirling. Stirling FK9 4LA, Scotland.

Accepted manuscript received 2'^ November 1979

19

Page 2: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

20 SHKII.AMITCHF.I.I.ANIJPKTKRROSA

extent to which parents' perceptions of their children's behaviour may ante-dateany subsequent involvement in criminal proceedings. The very interesting findingsof the National Survey pertain mainly to the relationship between criminality andsocial conditions in earlier life. The behavioural information obtained from parentswas somewhat limited, systematically covering mothers' reports during primaryschool years of "nervousness", "aggression" and "nightmares" and, at ages 6, 11and 15 years, the presence of "symptoms" (i.e. nose picking, stammering, nail biting,tics and general fidgetiness, unexplained vomiting and abdominal pain, bedwetting,thumb-sucking and "any other habit"; Wadsworth, 1979, pp. 74, 98). West, whileobtaining apparently quite comprehensive information from parents about behaviour,has presented this in his findings in a summarized form which depends not only onwhat the parents said but on how the interviewing psychiatric social worker interpretedthe information (West, 1969, p. 41).

In the present paper, we hope to supplement the findings of these previous studiesby using the material gathered in 1961 in a large-scale sample survey of all schoolchildren in Buckinghamshire, specifically designed to obtain information aboutchildren's behaviour as perceived by their parents and teachers (Shepherd, Oppcnheimand Mitchell, 1971). In particular, by examining the subsequent convictions ofthose who were seen by their parents as particularly problematic and comparingthem with a matched group of non-problematic children, we hope to explore theextent to which parental reports distinguish, in advance, between those who sub-sequently appear in court and those who do not. Because of the small number ofgirls involved in crime we have limited the study to boys only. An earlier study(Taylor and Watt, 1977) which traced the appearance of this same sample at a courtor clinic in the Buckinghamshire area up to 1969 indicated that parental reportsdid have some predictive value. The present study extends the coverage in time upto mid-1976 when the youngest of our original sample would have passed his 20thbirthday, and in space, since our criminal data refers to all indictable offencescommitted in England and Wales.

METHODS

In the original survey, questionnaires coticernint^ behaviour and health were distributed to parentsand teachers of a random sample of one in ten pupils ai ed between five and fifteen years attendinglocal authority schools in Buckinghamshire. These questionnaires were completed and returned by93 per cent of parents providing data concerning 3258 boys and 3046 girls (see Shepherd et ai, 1971,p. 34). The questionnaires covered thirty-seven categories of behaviour, and in each case ihe parentswere asked to describe their child either in terms of a three-point scale of inlensity of behaviour (forexample: has no fear of meeting new people; a little afraid or shy of new people; generally fearful olunfamiliar people) or on a seven-point scale of frequency of behaviour (e.g. frightening dreams) whichran from "never or less than once a year" to "every day".

Using these data, tables were constructed showing the frequency with which each behaviour traitoccurred among boys and girls, considered separately, at each year of age, and thus from these wecompiled a total "deviation" scoref for each child. This was based on the allocation of one point for

•fin the original study this has been referred to as the "deviance" score, but this terminology hasbeen changed in the present paper in view of the special meaning now given to "deviance" in muchcriminological/sociological literature (e.g. Becker, 1963).

Page 3: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BKHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINAL! l'Y 21

each of the traits the child manifested which were known to occur in 10 per cent or fewer of those of hisor her age and sex. Thus the child shown as conforming absolutely to the statistical norm lor his/herage group would have a "deviation" score of zero, while the child who was reported to exhibit, forexample, eight types of behaviour of Intensity/frequency found in 10 per cent or less oi his/hercontemporaries would score 8 points.

In the present study we have compared the 10 per cent of boys in our original sample who had thehighest deviation scores {i.e. a score of 4 or more points), who will be referred to as the "deviators",with a "control" group of boys, whose deviation score was 0 or l^ and each of whom was matchedto a deviator in terms of age and school. These two groups were virtually identical with those describedin the original study but include twelve fewer pairs (i.e. 321 instead of 333). This reflects (a) the,perhaps not surprising, disappearance of some questionnaires ..n the course of the intervening yearsand several changes of location, and (b) a few errors of coding/jjunching in the original survey which,when corrected, removed the individuals concerned frotn the "deviator" category.

Information of criminal convictions has been obtained from official sources and covers all appear-ances up to mid-1976 in criminal courts (including juvenile courts) in relation to indictable offences.Because of the small numbers involved in most specific categories ofcrime, offences were grouped intoseven broad categories: (a) theft; (b) damage to property; (c) violence towards persons; (d) fraud;(e) sexual oOences; (g) drug oflences.

As however, the number ol people involved in some of these criminal categories was exiremely small(see Table 1) we have limited our analysis of specitlc types ofcrime to theft, damage to property andinter-personal violence. Driving offences, which might be expected on the basis of national normsto involve a relatively high proportion of our subjects, have not been dealt with here as a separatecategory, owing to the partial coverage of our data which relate solely to indictable offences or to caseswhere non-indictable (i.e. minor) offences had been committed by someone who had also committedan indictable one. For example, the person who had stolen a car might also be prosecuted for drivingwithout a licence or without insurance or for failing to report an accident. This meant that drivingoffences were very much under-represented in our figures and our driving offenders, who were nearlyail car thieves, could not be taken as in any way representative of the general run of traffic offender.The same is probably true of other minor offences and we havj therefore, omitted from analysis anyspecific consideration of such categories. Where such offences have appeared in our records theyhave been included as part oi the total number of oflences committed by the individual concerned.

FINDINGS

Considering the two groups in toto, it was very e\ident that those in the deviatorgroup were more Hkely than the controls to have committed at least one indictableoffence leading to conviction. As Table 1 shows, out of the 321 matched pairs,approximately one in five of the deviators had been convicted of at least one suchoffence, compared with one in eleven of the controls.

This trend was also evident when we looked at conviction for specific types ofoffence. The deviator group were more likely to have been convicted of theft, damageto property, violence towards persons, and fraud. They were not more likely to havefeatured in cases concerning drugs, drink or sexual offences (but each of theseinvolved very few people).

Offenders from the original deviator group were also significantly more likelythan those from among the controls to appear in court on several occasions. Table 1shows that 72 per cent of the control group offenders had each appeared in court onone single occasion and that only one member of the group had undergone trial(and been found guilty) more than four times. In the deviator group, nearly half theoffenders (46 per cent) had appeared in court more than once, over a quarter (27

4:In 42 cases WIKTC nci suitable match wiili a zero score was available it was necessary to use a boy ofapproprialf age antl si hool with a deviation score of 1.

Page 4: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

22 . S M K I l . A M I T C H H l . i . A N D PK I K K R O S A

TABI.K l. NUMBP:R AND PKRCKNTAGK OF OFFFNDERS IN DEVIATOR AND CONTROI. GROUPS BV TYPEOFOFFENCE AND NUMBER OFCCHiRTAPPEARANC'ES

Type of offence committed

Deviator group(A = 321)No. (%)

Control group(A' = 321)

No. (%)Significance

by ;t'test

TheftDamage to properlyViolence towards personsFraudDrugolTencesSexual offencesDrinking offences

501611843I

(15.6)(5.0)(3.4)(2.5)(1.2)(0.9)(0.3)

22543332

(6.9)(1.6)(1.2)(0-9)(0.9)(0.9)(0.6)

0.0000.0150.0670.128

•*

All offenders 63 (19.6) 29 (9.0) 0.0002

Number of court appearances

12-45

341217

(10.6)(3.7)(5.3)

2171

(6.5)(2.2)(0.3)

For offendersonly

p = 0.029 (2 d.f)

•The number ol offenders is too low lor meaningful application of j ; ' test but the difference isunlikely to be significant.

pt r cent) had appeared more than four times and six boys (10 per cent) had appearedon between eight and eleven separate occasions.

It is, however, an oversimphrication to treat the ck-viators and their controls as ifeach group, while dit'tcrintf from the other in terms of childhood behaviour wasinternally homogeneous, since within the deviator group, and to a lesser extentamong the controls, there were internal difiercnces both with respcc t to the numberof deviating items underlined by parents (which varied from four to fourteen) andwith reference to the type of behaviour items underlined.

Table 2 provides an analysis of offenders (differentiated in terms of single ormultiple court appearances) against non-offenders in terms of their individualdeviation score. It tnust be remetnbered, however, that the control group has beenconstructed by selecting each of its members on the grounds that, in terms of age,school, etc. he closely resembled one of the deviator group. The control group,therefore, is in no way representative of children in our 1961 sample (and hence inthe population they represented) whose parents saw them as problem-free andcannot be used as a basis for generalizations about problem-free children. Thedeviators, on the other hand, do comprise all the worst 10 per cent of the boys in theoriginal sample, making it justifiable to consider them as representative of thistype of boy (i.e. those presenting the most atypical profiles) in that area (i.e.Buckinghamshire) at that time (1961).

Concentrating on the deviator group, then. Table 2 sht)ws that, with some iluctua-tion, the proportion of offenders increases as the number of deviating itemsunderlined by parents rises to a total score of eight. After this the proportion ofoffenders begins to decline so that the boys with the highest deviation scores areamong the least likely to commit crimes. The same thing is true of recidivismwhere the peak incidence occurred among boys with seven or eight deviating traitsunderlined while not one of the 14 boys with a deviation score greater than nine had

Page 5: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BI'.HAViOURi'RtJBLKMS AS PRF.flURSOR.S OK f;RIMINAI.rrY 23

TABLE 2. RELATIONSHIP BKTWEF.N NUMBER OF DEVIATIN<; BEHAVK.)UR TRAITS UNDERI.INEH HVPARENTS IN 1961, ANDSUBSE(.)UENTCONVIC:T1ONS

Group

Controls{n = 321)Deviators'(n = 321)

Deviationscore

0(n = 279)I {n = 42)4(fi = 132)5{n = 76)6{n = 40)7 {n = 30)8(n = 16)9 ( n = 13)

10(n = 6)11I2{n = 8)14

Non-offenders

No.

25339

1106729238

104

7

(%)

(90.7)(92.9)(83.3)(88.2)(72.5)(76.7)(50.0)(76.9)(66.7)

(87.5)

One courtappearance

No,

192

12661512

1

(%)

(6.8)(4.7)(9.1)(7.8)

(15.0)(3.3}

(31.2)(7.7)

(33.3)

(12.5;

C3ffender.sTwo or

more courtappearancesNo.

7I

1035632

(%)

(2.5)(2.4)(7.6)(4.0)

(12.5)(20.0)(18.8)(15.4)

Totaloffenders

No.

263

229

117B32

1

(%)

(9.3)(7.1)

(16.7)(11.8)(27.5)(23.3)(50.0)(23.1)(33.3)

(12.5)

•For the deviator group, non-offenders vs all offenders, with scores 9, tO and 11 + grouped becauseofsmall cell frequencies: xM-'''^./) = 14.5,^ = 0.013.

appeared in court inorc than once. On the basis ol present data we are unable tosay why this should be so. Nor, indeed, whether this apparent decrease in criminalityreflects some " r ea l " association or whether, since we are dealing with very sinallnumbers in the extreme categories, it is illusory.

Tables 1 and 2 show that, though the extent of behavioural deviation perceivedby parents in their children is significantly associated with future criminality, thepredictive value of such reports is limited. In the control group, selected on the basisof their statistical normality, nine per cent (i.e. one in every eleven) oi the boys waslater convicted of at least one indictable offence, while among the deviators, four outof every five boys had no such convictions.

So far the analysis has concentrated solely on the extent to which parents per-ceived their child's behaviour as atypical. It is, however, also possible to examinethe qualitative aspect of the data, i.e. to consider how far association may existbetween reports of specific types of behaviour and the subsequent delinquency ofthe child concerned.

Table 3 shows that out of the 38 types of behaviour reported on by parents in the1961 survey (see Shepherd et a/., 1971, App. Ill) only six showed an associationwith criininality and/or recidivism. Of these, lour iiems—stealing, destructiveness,wandering from home and lying—were positively related to criminality and reci-divism, while two behaviour traits—excessive worrying and food fads—were negativelyassociated.

Looking first at the proportion of persons who had been found guilty at least once ofan indictable offence we can clearly see that this bears a close relationship to theextent of stealing, lying, wandering and destructiveness reported by the parents in1961. Thus two out of every three boys described by their parents as having"stolen things on several occasions" had subsequently appeared in court (compared

Page 6: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

24 SHF.ll.A MirCHKLI. ANiJPKI l.R ROSA

with 19 per cent of the total deviator group, i.e. one in five) v hile more than halfwere recidivists (compared with one in eleven of the total group). As tnight beexpected nnost of these boys had been convicted of theft, but they were also five timesas likely to be convicted of damage to property and seven times as likely to be con-victed of inter-personal violence than might be expected in terms of total groupproportions. It must also be stressed that only one of the boys described by theirparents as thieves had first appeared in court before the parental questionnaire wascompleted. Even excluding this boy, however, the parents who reported stealingwere still predicting criminality in 61 per cent of cases, recidivism in 50 per cent andfuture involvement in theft in 56 per cent. Parental reports of stealing, then, appearto indicate a very unfavourable prt)gnosis for the boy concerned. Similarly, if weconsider destructiveness, we find that the likelihood of a very destructive boy fromthe deviator group becoming guilty of an indictable offence was five times as greatas that of a non-destructive boy and twice as great as that of a moderately destructiveone; the same trend is obvious with respect to lying and wandering from home.

As will be seen from Table 3, lying, wandering and destructive behaviour werealso strongly associated with subsequent conviction for theft. So far as involvementin other types ofcrime was concerned, however, there was some variation. Thus,while the very destructive were more likely than other children to steal, damageproperty and commit inter-personal violence, those who wandered from home werenot more likely than others to damage property or become involved in violence.Similarly, while the very destructive boys and persistent liars were more likely thanothers to appear in court more than once, this was less true of those who wanderedfrom home.

The other two types of behaviour covered in Table 3—food fads and worrying—show a significantly negative relationship lo officially recognized criminality. Thosemost likely to appear in court here are the apparently "problem-free", i.e. the boysdescribed by their parents as "carefree" and hearty eaters, while the worriers andthose faddy about lbod are relatively law-abiding and, where they have appeared in

T..\BI,K :i, C ") SlMa.llIC BKH.WIOLR rR.A.irS L M ) 1 - , K I , ! M : I ) < ) N

NAIRKIN 1961 (Devialor group only)

Ferti-nta^c of those in (.-aih behaviour ciHegory who had bt-cn convicted oloffences by 1976

Behaviour traits

At least one Two or Conviction Damage Convictionindictable One court naore court for to for

offence appearance appearances theft property violence

Total deviator group 19.3

Never takes auylhinir thaibelongs to someone else

(A = 165) 9.l!Has helped himself tosomeone else's things atleast once or twice

{N = 138) 2fi.lHas stolen things onseveral occasions

(A' = 18) 66.7

10.1 9.2

6.1

15.9 10.."}

11.1 55.6

15.8 4.7

19.6

61.1

5.1

22.2

3.2

0.6':'"

4.3

22.2

Page 7: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BKHAVIOUR I'KOBLKMS AS PRKCURSORS OF CRIMINALITY 25

T.\Yi\.K 3—continued

Seldom or never purposelydestroys things (A'= 99) 9 . 1 ' " ' * * 8.1 1.0<^'*" 5.l(^>**' 4 0 1 O '^ ' "About as destructive asmost children of the same

as*-- (A^= 196) 21.4 9.6 11.8 18.9 4.6 3.6Very destructive (A^ = 26) 46.2 23.1 23.1 30.8 11.5 11.5

Never wanders off fromhome without .saying wheregoing (A'= 160) 14.4">'" 8.8 5.6*^'** 1 1 . 3 ^ * ' * 4.4 2.5Loiters on way home fromschool, goes with friendswithoul telling parents

(A' = 123) 20.3 7.6 12.7 16.3 4.9 5.7Wanders of}' long distances/period.s withoul parentsknowingwhere (A'= 38) 39.5 26.3 13.2 31.6 7.9 0.0

Always tells the truth(TV = 24) 8.3^' '**' 4.2 ^//'^)»'* 4 2(:')*»- 4 2 0 O'- '

Tells an occasional fib(N = 238) 16.4 9.1 6.5 11.8 4.6 2.5

Tells deliberate untruthsquiteoften (A^ = 59) 37.3 15.3 22.0 35.6 6.8 8.5

Very carefree, doesn'tworry about anything

(A'= 92) 2 9 . 3 ( ' l " 10.9 18.5(3)*- 26.1 7.6 6.5Occasionally worries (e.g.about tests at school, ill-ness in the family)

(A'= 149) 14.8 10.1 4.7 12.1 4.7 1.3Often seems worried,worries about many things

(A' = 72) 16.7 9.7 6.9 11.! 1.4 2.8

Rather iussy about food,will eat only certainthings (A'= 85) 10.6*''** 8.2 2.4 9.4 1.2 0.0Has fairly defmlte foodpreferences but will eatmost foods if hungry

(TV' = 107) 16.8 7.5 9.3 12.1 4.7 2.8Will eat nearly anything

(A- = 123) 27.6 13.8 13.8 23.6 7.3 5.7

" ' X ' t e s t for offenders vs non-ottenders on triple choice behaviour item- 2df.- * " = 6 < 0 0 0 1 -•" = p<0.0\;* = p<0.05. ' r • •

'* X^ t e s t fo r n u m b e r o f c o u r t a p p e a r a n c e s o n t r i p l e c h o i c e b e h a v i o u r i t e m - 3 f^/: • *•• = 6 < O O 0 J -*• = / 7 < 0 . 0 1 ; • = / j < 0 . 0 5 . r • ^

'•''X^ test for specific type of offenders vs all others on a triple choice behaviour item- 2 d f.-" ' = p < 0 . 0 0 1 ; • • = p<0.0\: * = p<0.05.

Page 8: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

26 SMKIl.AMITCHKI.l. AND PETER ROSA

court, tend to do so only once. As this relationship, though significant, could havebeen a chance one in view of the large number of behaviour traits examined, i.e.38, it is also perhaps worth noting that it is supported by similar but non-signifieanttrends in other behavioural items. Thus a below-average number of offendersoccurred among those who, in 1961, were described as frequent thumb-suckers,afraid of the dark, having poor appetites, whining and complaining a lot, andsuffering often from stomach-ache.

Some items which might have been expected to relate to criminality did not, infact, do so. Children who were seen by their parents as "disliking school", forinstance, showed very little difference in the proportion of offenders they generated(20.4 per cent) from that found among those boys who "liked school very much"(17.4 per cent) and those who "liked it as much as most children" (19.1 per cent).Some other items exhibited a U-shaped distribution—for example "moods". Here26.7 per cent of the "very moody" had criminal records but so did 27.5 per centof those at the alternative extreme of that behaviour item, those said by their parentsto "run on a very even keel", while the incidence in the "moderate" categorydropped to 14.6 per cent {p = 0.05). Similarly those children perceived as dis-obedient were no more likely to appear in court (23.5 per cent) than those who"always obeyed" (25 per cent).

Information obtained from parents, though important because it reflects detailedknowledge of the child under many different circumstances, can only supply oneviewpoint on the child's behaviour. It was therefore decided to supplement theparents' reports by using the information about the child's behaviour at school,obtained from teachers during the original survey. In each case the teacher hadbeen given a check list of types of behaviour problems which might occur in schooland was asked to underline those that applied to this child. Since no item had beenunderlined for more than 10 percent of children at any age, we did not, in this case,use the concept of "deviating" behaviour. We merely scored 1 point for each"problem" underlined. The teachers' "problem score" was then added to the"deviation score" derived from the parents' questionnaires and the combined scorerelated to criminal outcotiie in terms of number of court appearance. So far as thecontrols were concerned, the additional information was of little help in distinguishingpotential offenders from non-offenders; of the fifteen boys seen as most problematicby their teachers only two (9 per cent) subsequently appeared in court, whileamong those who were seen as problem-free the proportion was 8 per cent. Examiningthe situation within the "deviator" group more closely, however (see Table 4), wefound that those deviators with high combined scores were twice as likely to becomeoffenders as those whose scores did not exceed a total of 7. Recidivists were morethan three times as likely to come from the high scoring group (16.8 per cent of allthose scoring 8 or more were recidivists compared with 4.8 per cent of the lowerscores).

Such highly significant differences cannot be discounted. It must be remembered,however, that we are still speaking of successful recognition of future criminalityin only quite a small number of cases. Thus as Table 4 shows, 71 per cent of thosewith the highest combined scores are not criminal (or at least have not been officiallylabelled as such). Bearing this comment firmly in mind, it may, however, be usefulto look at the extent to which the additional information provided by teachers has

Page 9: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD B I : H A \ ' I O I K I'KOHl.KMS AS HKliCURISORStiF CRIMINAL! I T 27

TABLE 4. COMBINED SCORE OF DEVIATING/PROBLEMATIC riEMS RECORDED BY THE PARENIS AND

TEACHERS BY SUBSEQUENT COURT AI'PFARANCES LEAOINC; TO C()NVIC:TION AMONt: BOYS [N THE

DKVLATOR GROUP

Number of court appearancesleading to conviction

NoneOneTwo or more

All offenders

Total

No. (%)

179 (85.6)20 (9.6)10 (4.8)

No. 30(%) (14.4)

No. 209(%) (100.0)

Combined score

8-11No. (%)

59 (70.2)10 (11.9)15 (17.9)

25(29.8)

a4(100.0)

12 +No. (%)

12 (70.6)3 (17.6)2 (11.8)

5(29.4)

17(100.0)

{offenders vs non-offenders) (2 rf./) = 10.279,/) 0.006.X (number of court appearances) (4 (/./) = 15.1203,/) = 0.0003.

improved the power of prediction which we have already demonstrated (in Table 1)as resting on parental perceptions alone. To do this we have looked at the latercriminal careers of those boys in the deviator group who fell into the "worst" 20per eent (approx.) in each case according to their parents alone, according to theirteachers (bearing in mind that the group has been preselected so far as the teacherswere concerned) and fmally, on the ba.sis of the combined score of both teachers andparents (see Table 5). This shows that the addition of teachers' information con-cerning extent of problematic behaviour in school appears to add nothing to thepredictive value of the parental reports in cases where the child is seen by his parents asexhibiting a fairly marked degree of deviating behaviour at home. Whether or not this relation-ship would hold for less deviating boys cannot be established from this investigation.Nor do we know if it applied to girls.

Examination of the nature of the specific items underlined by the teachers of theoriginaJ "deviators" also produced some interesting information in that, once again,there emerged a highly significant association between the teacher's perception ofthe child as a thief or a liar and his subsequent criminal career. Those boys describedby teachers as having stolen or telling lies were not only more likely to becomeoffenders but also very strikingly more likely to become persistent offenders. Thusthe proportion of teachcr-elassified thieves who appeared in court more than once(42.9 per cent) was six times as high as that of the other boys (7.1 per eent); the samewas true of those boys described as "often tells lies" (40.9 per eent) compared with6.3 percent of the rest of the deviator group).

It is also interesting that the teachers who ascribed dishonesty to their pupils didnot just pick out future thieves. Those suspected of stealing by teachers were alsoeight times as likely to be tried for offences involving damage to property (28.6 percent to 3.7 per cent for the rest of the group) and four times as likely to be convictedoi inter-personal violence. The liars were five ilnus as likely to be convicted ofdatnage to property (18.2 per cent : 3.8 per cent of the other boys); four times aslikely to be tried for violence (13.6 per cent: 2.8 per cent) and three times as likely tobecome thieves (45.5 per eent: 12.8 per cent).

Page 10: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

28 SHEILA M I T C H E L L AND PE l E R ROSA

u

o

o

T3 O O O

c >o fn (^O O CO '.Oc o o ^

C -C

Hffl

Page 11: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BEHAVIOUR PROBLEMS AS PRECUKSORS Of CRIMINAL! IV 29

Since the reports of both teachers and parents of boys in the original deviatorgroup demonstrate a similar significant association between stealing and lying andlater court appearances, it must now be asked to vv'hat extent each source supple-ments the other. Table 6 shows that the "worst" prognosis occurred where bothteachers and parents were agreed in 1961 that the boys were already stealing or lying.The number identified as thieves by both informants is small. Indeed it is noteworthythat more than half (61.1 per cent) of the boys said by their parents to have stolenwere not so labelled by their teachers while, of those seen as thieves by their teachers,half were categorized more leniently by their parents. Nevertheless, of the sevenboys labelled as thieves on both questionnaires, five (71.4 per cent) not only becameoffenders but also persistent offenders.§ Without confirmation from the teachers,parents' attribution of theft—though slightly less successful—still pointed to fairlyhigh involvement in crime later, since 63.6 per cent became offenders and most ofthese had appeared in court at least twice. This must be compared with the proportionof offenders found among those where only the teacher mentioned stealing (28.6 percent) or where neither informant did so (25.8 per cent).

In the case of lying, again we fmd the highest rate of subsequent criminalityoccurred among the boys identified by both parents and teachers where 64.3 percent became offenders and 50 per cent appeared in court more than once. Here,however, boys seen as liars either by their parents alone or by their teachers aloneappeared to have equal, and reduced, chances of becoming offenders (26 per centand 25 per cent respectively). Indeed, nearly two-thirds of boys described as tellinglies by their parents, and half those seen as liars by their teachers, appear to haveavoided all contact with the courts.

So far as later participation in specific types of criminal behaviour was concerned,the chance that a boy, seen as a thief by his parents and teacher or by his parentsalone, would later become one in the eyes of the law also was twice as high as thatof the boy accused of stealing by his teacher alone and more than four times that ofthe boy unreported by either. Those children seen as thieves at home and at school,also produced a very much higher than expected proix)rtion of those who later causeddamage to property or violence to other persons.

In the case of lying, it was again the group where the boys were identified as liarsby both parents and teachers which generated tht highest proportion of thieves,violent offenders and those convicted of damage to property. Labelling by the parentsalone or by teachers alone showed a much reduced relationship to all three types ofcrime.

DISCUSSIONIn this paper we have suggested, firstly, that boys who are reported by their parents

as displaying several types of behaviour atypical for those of their age are more likelyto become criminals and persistent criminals than are comparable boys whosebehaviour is entirely, or almost entirely, "normal". Secondly, we have establishedthat, among the most problematic group of the school-age male population (usingparental criteria) certain types of atypical behaviour are found to carry an increasedrisk of future criminal involvement. The boys described by their parents as stealing,lying, destructive or wandering from home were more likely subsequently to acquire

§ThfSf included one boy (previously rnt-niionftl) who had been convicted prior to the originalsurvey and who has re-appeared in court three lirncs since then.

Page 12: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

30 SHKILA MI'ICHELL AND PETER ROSA

inCQ

.y 3

CO "*• O — f ^

CM —. 03 C-J —

• = c ^

>f W I-H ' ^r-" flj k-* O

I .S ^ .S§ "E -a i 3 « 3

£: i i ra -=

ft. a. H z

- ^ • * c-j CO cn

c in Ol in

u"0c3

-C

o-^

j "

u

JJ

C

_j2cs.-rt

Cu

TJ'Sink-

_ ^

urt

_T I

C

" u1.1

C

c1>k.

rea,

-acOJ

rto.

-a

!_-^c3

u

u

sH

c

^

z

.s

--*Z J

1/

•her

5^^c3oH

-au

lin

s-

3s..U

-Curt^

s . .u

ch

Page 13: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS Oi- CRIMINAi.nV 31

at least one criminal conviction, to appear in court on more than one occasion and tobe involved in theft. Thirdly, we have shown that the addition of material fromteachers does not improve the accuracy of parental prediction when we look solely atthe relationship between extent of deviation (measured in terms of numbers of itemsof "problem" behaviour reported) and criminality. When, however, we look atthe cases where there is agreement by parents and teachers that the child is a thiefor a liar we find that this considerably increases the accuracy of prediction of futurecourt appearances and, in particular, of repeated court appearances, over thatfound when either parents or teachers alone are making the judgement.

These findings have both theoretical and practical implications. Firstly, at thepractical level, it is obvious that though the quantitative data may generatestatistically significant findings, they do not provide a basis for accurate prediction offuture criminality. Thus, though the "deviators" were twice as likely to becomeoffenders as the controls, there was still only a one in five chance that they woulddo so and four-fifths of the deviator group had still not been convicted of anindictable offence by 1976. The qualitative data relating to reports of stealing andlying carry a much higher chance of successful prediction, particularly where bothparent and teacher are in agreement about the boy's behaviour. Our numbers arevery small and consequently open to distortion by chance elements. Nevertheless ourfindings on the adverse prognosis associated with lying and stealing and, to a lesserextent, on wandering and destructiveness, accord well with Robins' (1966)follow-up study of child guidance patients in which she reported that 44 per cent ofboys originally referred for "anti-social" behaviour (which would have included ourfour types of behaviour among many others) had subsequently appeared in court inconnection with "major crime", compared with 12 per cent of other male clinicoffenders and 3 per cent of non-clinic controls (Robins, 1966, p. 47). This wouldsuggest that, from a practical point of view, parental rejxjrts of anti-social behaviour,particularly if supported by school information, sliould be treated as importantadvance warnings of the likelihood of increased later involvement in delinquency.This point has already been made by Robins (1966, p. 158) who says: "Anti-socialbehaviour in childhood seems to be an excellent predictor both of one particularpsychiatric syndrome, sociopathic personality, and also of the advent of adult anti-social behaviour that cuts across diagnostic lines". Robins' study, however, relatedsolely to children who had attended child guidance clinic and who, in the case of halfthe boys, had been referred to the clinic by a juvenile court (though in some casesas an alternative to court proceedings). These, then, were boys who had beenofficially "labelled" as "problematic" and in many cases as "delinquent" beforeinformation was gathered from the parents. Furthei'more, the information relatingto childhood state was obtained by the clinic staff in an "official" setting and in thelight of what parents and staff saw as the most important issues for discussion sothat the coverage of behaviour traits was not necessarily the same in all cases. Ourstudy, though in some ways less comprehensive, has the advantage of referring to a"free" population of boys who, at the time of the ;iurvey had not yet (apart fromthe one exception mentioned on page 24 been officially labelled as delinquent.Both the parents' and teachers' reports of anti-social behaviour are. therefore,spontaneous and do not reflect thejudgements of a court.

Page 14: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

32 SHEILA MITCHKLL AND PE TER ROSA

This has also important theoretical connotations in light of the controversy incriminology in recent years between those who look for the causes of criminalbehaviour in the personal circumstances, constitutional or environmental, of thoseconcerned and those who see criminality as a "label" fixed on the individual byothers and, in particular, by official social control agencies such as the police, courts,etc., which causes him to adopt a criminal "identity" (Becker, 1963; Lemert, 1972).Robins' study would appear to provide evidence for either viewpoint and our ownwork certainly does not resolve the dilemma. It does, however, suggest that iflabelling is instrumental in the development of a criminal career the sources of effectivelabelling may be in home or school and not necessarily in the actions of officialagencies.

One result of our findings, then, may be to indicate to future researchers theimportance of parental reports. While we have not been able to test the predictivepowers of parents as against teachers in this study, we have shown that, dealing witha group pre-selected on the basis of parental information, teachers' data havesupplemented rather than superseded that of the parents. We hope, therefore, thatfuture researchers will use both resources and avoid placing sole reliance on theviews of teachers (as, for example, in Stott and Wilson, 1977).

This raises another theoretical issue. In our study—as indeed in most othersconcerning children's behaviour—we do not possess accurate information aboutwhat the child actually does. What we have is a picture of how our informants(parents and teachers) see the child's behaviour (or, indeed, how they are willingto say they see it). For example if a boy is said to be "very destructive" this maymean that he leaves a trail of ruin in his wake or it may mean that his mother per-ceives as "very destructive", behaviour that another mother might see as quite"normal". We cannot say that the description is objectively "true", we can onlysay that this is how the situation appears to the informant and "a situation definedas real is real in its consequence", in the often quoted words of W. I. Thomas (1928).

Again this leads us to another issue, ln our study we have necessarily relatedchildhood behaviour to subsequent convictions for indictable offences. This mustresult in gross underestimation of criminal activity among those we have beenconsidering. Even at the official level we have not been able to consider those foundguilty of minor offences, but perhaps even more important is the fact that we knownothing of those who have taken part (perhaps quite regularly) in criminal activitybut who have never been officially processed. Our findings, therefore, relate not tothose who A^com^criminals but to those who are officially recogniseda^ such. West andFarrington (1973, 1977) who use self-reports on criminal activity to supplement theofficial data suggest that those who break the law but are not caught do not differsubstantially from those who are. Unfortunately, however, they do not relate this tobehavioural criteria apart from "troublesomeness". We cannot therefore make anyassessment of the extent to which our findings might have been affected by ourreliance on "official" data.

Clearly there is much need for further research, based on random samples of"normal" (i.e. not yet officially labelled) children of sufficient size to allow fbr theidentification of particular "at risk" groups—for example those seen by parentsand/or teachers as thieves, liars, destroyers, wanderers or in other ways as anti-social—whose present state and subsequent progress could be studied in more detail.

Page 15: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY

BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALITY 33

SUMMARY

Material gathered in 1961 as part of a large scale sample survey of the behaviourof school age children has been used to select a group of boys whose behaviourdeviated most from that typical of other boys of their age. Their subsequentconvictions for indictable (major) offences have been ascertained over a fifteen yearperiod together with those for a matched control group originally exhibiting"normal" behaviour. F"indings show those in the deviator group to be significantlymore likely both to become offenders and to become recidivists. Qualitatively,parental reports of anti-social behaviour (stealing, lying, destructiveness andwandering from home) were shown to carry the worst prognosis for subsequentconviction(s) particularly where teachers' reports supported those ot the parents.The practical and theoretical implications of the study are discussed.

Acknowledgements—The authors wish to thank the staff of the Home Office Research Unit and theCriminal Records Office for their invaluable assistance with this project.

REFERENCES

BECKER, H . S. (1963) Outsiders. Free Press, New York.BI.AUX:K, H . M . (1972) Social Statistics. 2nd edition. McGraw-Hill, London.LEMERT, E . (1972) Human Deviance, Social Problems and Social Control, 2nd edition. Prentice Hall,

Englewood Cliffs.ROBINS, L . N . {\9^€))Deviant Children Grown Up. Willianns &Wilkins, Baltimore.SHEPHERD, M . , OPPKNHEIM, B. and MITCHELL, S. (1971) Childhood Behaviour and Mental Health.

University of London Press, London.STOTT, D . H . and WlLSON, D. M. (1977) The adult criminal as juvenile. Br. J. Criminal. 17, 47-57.TAYLOR, T . and W.^rr, D. C. (1977) The relation of deviant symptom.s and behaviour in a normal

population to subsequent delinquency and maladjustment. Psychol. Med. 7, 163-169.THOMAS, W. I. (1928) The Child in America. Knopf, New York.WADSWORTH , J. W. B. {\97 9) Roots of Delinquency. Martin Robertson, Oxford.WEST, D . J . {\9b9) Present conduct and Future Delinquency. Heinemann, London.WEST, D.J. and FARRINGTON, D . P. (1973) Who Becomes Df/m^t^fr'Heinemann, London.WEST, D.J . and FARRINCTON, D . P. (1977) The Delinquent Way of Life. Heinemann, London.

Page 16: BOYHOOD BEHAVIOUR PROBLEMS AS PRECURSORS OF CRIMINALTY: A FIFTEEN-YEAR FOLLOW-UP STUDY