Abstract
Predictions of recidivism may be formulated solely in categories of probability. In predicting human behaviour it is impossible to take account and to control all factors that influence it. Causal relationships and the general laws that explain it are still largely unknown and generally the data available on the subject are incomplete. It is therefore necessary to expect that there may be disagreement between predicted and actual behaviour. Predictions of recidivism may be formulated solely in categories of probability. In predicting human behaviour it is impossible to take account and to control all factors that influence it. Causal relationships and the general laws that explain it are still largely unknown and generally the data available on the subject are incomplete. It is therefore necessary to expect that there may be disagreement between predicted and actual behaviour.
Nonetheless, despite these reservations, individual predictions of recidivism of juvenile delinquents are to all practical purposes a constant factor in the decisions of the law courts. The essential problem therefore is not whether it is possible to make individual predictions, where there is always a chance of error, but how to arrive at predictions a large proportion of which will be correct.
Literature in the field of criminology devoted to this subject distinguishes the statistical and clinical methodes of prediction. These two methods were studied by the Department of Criminology of the Polish Academy of Sciences. The object of the study was to investigate a number of questions that were raised by research conducted in other countries on the Department's own empirical material.Below are given the problems related to the subject of clinical predictions:
1. Since clinical predictions play an important role in present practice it was advisable to learn to what extent the predictions made in our study were correct as regards juvenile recidivism.
2. It was equally important to discover how a given prediction was justified, what factors are considered significant in predictions made in individual cases.
3. It was resolved to make a study of the subjective aspects of clinical predictions: whether persons who received the same education and professional training tend to make the same predictions regarding the same juveniles? Whether predictions made by different persons for two groups of juveniles will prove to be accurate in the same extent?
Problems of statistical predictions were related to the following questions:
1. Whether the predictive factors established in the projects carried out in other countries have any bearing in the predictions relating to juvenile delinquents in Poland?
2. It was resolved that predictive factors found in one group would be incorporated into the experimental prediction table and used in making predictions for another group. It was further resolved to check-up on the correctness of the predictions. In constructing the experimental prediction table the goal was not to construct a table designed for practical use but on the basis of our own experiments to identify the problems that arise when using a prediction table.
3. Special importance was attached to a careful analysis of cases where the predictions made with the aid of the table were incorrect.
The research planned according to these guidelines was conducted in two stages.
In the first stage clinical predictions were made and experimental prediction table was constructed for a representative sample of 15 and 16 year old recidivists of Warsaw. In the second stage data was tested on a new sample of 15 and 16 year old recidivists and instances were analyzed where the statistical predictions proved incorrect.
The initial research embraced 100 recidivists of 15 and 16 years of age out of 202 recidivists, of the entife population of juvenile recidivists who in 1954 came before the juvenile court of Warsaw on charges of larceny and who were embraced by earlier research on juvenile recidivism conducted by the Department of Criminology. The earlier research yielded data on the after-conduct of the recidivists studied that covered a span of three years. It was established that 51 per cent of them commited offences in the follow-up period.
First of all the clinical predictions on the 100 recidivists were based on the findings of environmental as well as psychological and medical examinations and without knowledge of the findings of the follow-up studies. Two psychologists who had experience in criminological studies made predictions for each of the 100 recidivists. The psychologists were not in touch with each other and did not estabiish joint criteria beforehand. Good behaviour was predicted if it was assumed that the recidivist would not commit any offences in the future, bad predictions were made if the feeling was that he could commit offences and uncertain if no definite decision was reached. If the two psychologists differed in their predictions they would discuss the subject and try to arrive at a consensus.
The predictions made in this manner shaped up as follows: 18 per cent were good, 57 per cent bad and 25 per cent uncertain. There was a significant statistical relationship between the predictions and the commission or non-commission of offences in the course of the next three years by the 100 recidivists studied that may be expressed by a level of significance of p < 0.001. The bad predictions were correct in 70 per cent of the cases, the good in 83 per cent. Thus an overwhelming proportion of the predictions was correct and the proportiorr of uncertain predictions (25 per cent) inconsiderable.
The problem arises what part do subjective factors play in the clinical predictions made by two different persons? Two separate predictions regarding the same juvenile agreed in 70 per cent of the cases. Greater agreement was found in the bad predictions (77 per cent) than in the uncertain (60 per cent) and the good (61 per cent) predictions. Moreover, there were large differences in the reasons given for the predictions issued to the same individual. The two psychologists frequently listed different factors in arriving at the same decisions.
A great many factors were listed as reasons for the predictions which, based on an analysis of data relating to the individual cases, seemed to bear significantly upon the predictions regarding the juveniles studied. Among those mentioned were envinonmental factors, personality traits, demonstration of antisocial behaviour and information about the offences committed.
The next step in the first stage of the project focused on statistical predictions. A study was made of the relationship between 23 factors and the behaviour of the 100 recidivists of 15 and 16 years of age under study over a span of three years. Account was taken of factors which were found significant in the prediction of juvenile recidivism by the research conducted in other countries and of factors which were seemed significant to the problem in the study of juvenile recidivism in Poland.
It was established that a significant statistical relationship existed between the following factors and the continued antisocial behaviour of the subjects under study: 1) early age (below 11) of the onset of symptoms of demoralization, 2) early age of onset of antisocial behaviour (below 13), 3) persistent stealing, 4) membership in a group of delinquents or keeping bad company, 5) personality disorders, 6) drinking, 7) running away from home, B) Iack of schooling or work.
The findings indicate that the early age of the onset of antisocial behaviour and the far-gone demoralization of the juvenile are important factors in predicting recidivism. However, no relationship was found, and this seemed strange and called for explanation, between recidivism and any of the factors that characterized the family environment. This contrasted with the findings of the previous study that embraced all the juvenile recidivists between the ages of 8 and. 16. The oldest of these were included in the present study.
In order to find an explanation for the disparity an additional study, one that was not initially planned, was made of the 28 factors and their relationship to recidivism that continued over a period of three years among the youngest of the recidivists studied at an earlier time in the Department of Criminology. Toward this end 68 of the youngest subjects between the ages of 8 and 13 were isolated from the whole population of recidivists ranging from 8 to 16 years of age.
It was found that the following factors had a statistically significant relationship with continued recidivism in the younger age group: 1) alcoholism in the family, 2) the home atmosphere, 3) lack of supervision by parents, 4) systematic truancy, b) early age of first symptoms of demoralization, 6) early age of first offences, 7) membership in a delinquent group, B) personaiitv disorders. Consequently, a slighty different set of factors ought to be taken into account when making predictions for younger recidivists.
Environmental factors of the home are far more significant in predictions for younger delinquents. In older delinquents it was totally immaterial whether they came from a good or a bad home environment as far as predictions were concerned. A good home which had failed to guard a child of up to 15 and 16 years of age from becoming a delinquent couId handły guard the child against recidivism. In younger delinquents a good lamily atmosphere, excellent supervision, absence of alcoholism all are positive predictive factors. Younger juveniles are still highly responsive to the influence of the home and careful supervision may guard them against further demoralization.
Our research substantiated the thesis that research on the prediction of juvenile recidivism ought to be conducted separately for narrow and strictly defined age levels. The age of the subject at the time the prediction is made is an important factor that must be kept in view.