Juvenile delinquency among children in outside home care – does type of care matter?
3. EMPIRICAL METHOD
4.2 Effects on number and type of verdicts
Tables 4 and 4a-b showed the effects of placement on the likelihood of getting a verdict. However, could placement type also affect the number of verdicts and the type of verdict/sentence? This investigation brings new evidence to the question in relation to Ernjæs (2011) who only looks at the effect of placement type on the likelihood of crime and who also finds a positive (though insignificant) effect on boys’ crime of being placed in residential institutions. Lindquist and Santarvirta provide estimates of the effect of care type on both extensive (any crime) and intensive margins (sum of crimes over a time period) and severity of crime (imprisonment and length of prison term). Since few juveniles are imprisoned, our measure of severity is based instead on the nature of the crime committed. We turn to these results in Tables 5a-7b. Given the findings in Tables 4a-4b, we estimate gender separate models in each case. Table 5a shows that boys placed in residential care have significantly greater criminal intensity as measured by number of verdicts. Here too, OLS is preferred to IV despite having a strong instrument (see Hausman test result). Thus, the point estimate from the OLS model with controls show that boys in residential care have 0.11 more verdicts than children in foster care relative to an overall mean number of verdicts for boys of 4.95. The IV estimate is 0.13, though insignificant. The background variables have the same signs and significances as in Table 4a, except that days in placement and paternal crime are now significant and father basic education no longer significant.
In Table 5b similar results for girls are presented. The most striking finding is that while placement in a residential institution does not increase girls’ likelihood to commit crime, it has a large impact on the number of verdicts, in fact, even larger than for boys. Girls in residential institutions have 0.49 verdicts more than girls in foster homes, even including zero verdicts. This should be seen relative to the mean of 1.93 verdicts for girls overall. The IV estimate here is somewhat different, negative, reducing number of verdicts by 0.21 but it is insignificant. The same background factors predict girls’ crime intensity as girls’ crime likelihood, in fact mother on disability pension is now significant and positive, i.e. increasing number of verdicts and father mental disorder is no longer significant. Just as for boys, OLS cannot be rejected so column 2 estimates are the preferred ones. The basic finding is that while placement in residential care does not cause girls to become criminal, it increases their criminal intensity.
Tables 6a-b, 7a-b explore the effects of different types of verdict and sentencing. As mentioned earlier, the type of crime/sentence could be an indicator of the severity of the crime. Here, we choose to show multinomial logit models given that all the previous models failed to find
76 signs of endogeneity. The MNL estimates show a very interesting and consistent picture of more severe adolescent criminality of children who have been placed in residential care. From Table 6a we see that boys in residential care have significantly higher likelihood of verdicts for violence/sexual offences and theft and lower likelihood (though not significant) for less serious offences such as drunk driving and other offences compared to boys in foster care. For girls in residential care too, Table 6b shows a significantly higher tendency to commit violence/sexual offence and a significantly lower tendency for drunk driving and lighter offences and no significant difference in thefts compared to girls in foster care. Again, estimates of the background characteristics of the children, their parents and their municipality show a similar pattern with respect to type of crime as in crime likelihood and crime intensity – i.e. poor health, placement characteristics (age at first placement etc.), maternal age at first birth, maternal single status, maternal low education, maternal crime, paternal income and paternal crime are the important factors associated with juvenile crime behavior. Interestingly, these characteristics have stronger effects on the more serious types of crime-violence/sexual offences and theft.28
An alternate measure of crime severity is given by the type of sentence imposed. Tables 7a-b show that residentially placed children, both boys and girls, have a significant higher likelihood of obtaining an unsuspended conviction (the most serious type). In fact, for boys the marginal effects fall monotonically with the severity of the sentence so that they have the highest likelihood of getting a unsuspended sentence, a somewhat lower likelihood of obtaining a suspended conviction, an even lower one of obtaining fines and the lowest likelihood of receiving other convictions including acquittal and dismissals. For girls, the likelihood is highest for the other convictions including acquittal/dismissal although it is not significant, followed by unsuspended sentences, suspended sentences and lastly fines. In fact, residentially placed girls are less likely to be fined, while residentially placed boys are more likely to be fined compared to equivalents in foster care. The source of this gender difference can be explored further in future work, for instance, by looking at the type of verdict for which fines are imposed.
To sum up, the evidence from these tables shows that children in institutions are more likely to commit more serious types of crime compared to children in foster care and at the same
28Crimes committed by juveniles under the category Violence/Sexual offences are mainly of the first type, i.e. Violence. The
Violence category can be further split up into murder attempts/violence against government authorities/violence against private persons (simple, serious, especially serious)/threats. For this age group, the most frequent registered crime is simple violence against private persons.
77 time, they also tend to receive stiffer sentences than children from foster homes, although some gender differences exist.29