Changes in the seriousness of offending over the delinquent career

View the most recent version.

The research evidence on changes in the seriousness of crime over the course of the delinquent or criminal career is mixed. Some research has found evidence of an increase in seriousness, or escalation; other studies have found evidence of a decrease in seriousness, or de-escalation; but the majority of studies have concluded that the seriousness of offending tends to be "relatively stable" over the career (Piquero et al. 2003: 387-388, 452; Kyvsgaard, 2003: 173-174). The trend in the seriousness of offending over the career is not necessarily monotonic: evidence has been found by at least one study of an inverted U-curve in longer careers, with the seriousness of offending increasing in the earlier part of the career, then decreasing later in the career (Kyvsgaard 2003: 177). Previous Canadian research using court data found "no pronounced tendency to escalation, stability, or de-escalation", with 41% of court careers being characterized by de-escalation, 31% by escalation, and 28% by stability (Carrington et al. 2005: 33-34).

One simple method of studying changes in seriousness over the career is to compare the most serious offence allegedly committed in the first and last incidents in the career. In Table 8, incidents are classified into 6 levels of seriousness, ranging from the least serious - those whose most serious offence was a summary or hybrid "other" offence – to the most serious – those whose most serious offence was an indictable offence against the person. Only substantive incidents¹ are included in the analysis, and only those offenders with at least two substantive incidents in their careers are included. Careers are cross-classified in Table 8 by the most serious offence involved in the first substantive incident in the career (in the rows of the table), and whether the most serious offence in the last incident in the career was in a more serious category than the first (escalation), the same category of seriousness as the first (stability), or a less serious category (de-escalation). The last ("Total") row for each birth cohort shows that in the overall distribution of careers, there are substantial numbers of careers with each of the three patterns. Overall, stability is the most common pattern in both cohorts, followed by de-escalation and then escalation. The numbers of careers exhibiting escalation and de-escalation are very similar. Therefore, there is definitely no overall tendency towards either escalation or de-escalation.

The analysis in Table 8 suffers from two limitations. One is the relatively crude measurement of seriousness, in only 6 levels, with only an intuitive rather than empirical basis for their order. The second limitation is the inclusion of only the first and last (substantive) incidents in the career, which precludes detection of more complex career patterns than simply escalation, de-escalation or stability. Both limitations are overcome in Figure 39, where the x-axis is the order of the incident in the delinquent career: the first substantive incident in the career, second substantive incident, etc., for all incidents in the career. The y-axis shows the mean seriousness of the most serious recorded substantive offence in all incidents of a given order: first, second, etc.² Here, seriousness is measured by the seriousness scale developed by Canadian Centre for Justice Statistics to rank the seriousness of criminal offences.³

There is no evidence of escalation in Figure 39. The mean seriousness of substantive incidents fluctuates around 62 for both cohorts, with no evident upward or downward trend as the incident number in the career increases. When the seriousness score of the incident is regressed on the incident number, the slope for offenders born in 1987 is small and statistically non-significant (b = 0.035, std. err. = 0.042, p = .40), i.e. neither significantly positive (indicating escalation) nor significantly negative (indicating de-escalation). For offenders born in 1990, the regression slope is negative and statistically significant (b = -0.295, std. err. = 0.054, p < .001), i.e. the overall pattern is one of de-escalation.

Visual examination of the plot for the 1987 cohort suggests a U-shaped curve, with de-escalation for the earlier incidents in the career (up to the 14^th incident), followed by escalation. This is confirmed by performing separate regressions on the first 14 points and the 14^th to 25^th points. For the first 14 points for the 1987 cohort, the slope is negative and statistically significant (b = -0.25, std. err. = 0.09, p = .017); for the 14^th to 25^th points, the slope is positive and statistically significant (b = 0.30, std. err. = 0.05, p < .001). The absolute values of the two slopes are approximately equal, and approximately equal to the absolute value of the regression slope for the first 18 points for the 1990 cohort. It may well be that the finding of overall de-escalation for offenders in the 1990 cohort is due to the cut-off at the 18^th incident, and that extension of the plot to the 25^th incident would also yield a U-shaped curve. The finding for offenders born in 1987 of de-escalation followed by escalation in seriousness is partly consistent with Kyvsgaard's findings (2003: 177), which are based on an analysis that examined careers of different lengths separately: she also found de-escalation in careers with few incidents, but she found an inverted U-curve for careers with many incidents.

---

Note

1. "Substantive offences" refers to all offences except administrative offences (offences against the administration of justice). The most common administrative offences are violations of bail or probation conditions, or failure to appear for court. Other, much less common administrative offences include prison breach, escaping from custody, and other very infrequent offences. See Table A.1 in the Appendix for a detailed breakdown of offences allegedly committed by members of the study population.
2. Because of small numbers, only the first 18 substantive incidents in the career are shown for offenders born in 1990. For the same reason, the 26^th and all subsequent substantive incidents are combined for offenders born in 1987.
3. It is based on the average length of prison sentence imposed on convicted charges between 1994/95 and 2000/01 in criminal court (Robinson, 2004: 10). Its value is inversely related to the seriousness of the offence, and ranges from 1 for "First degree murder" to 112 for "Other federal statute offences". This is an ordinal (ranked) scale, and strictly speaking should not be used in analyses which are based on arithmetic operations such as averaging, However, use of such ordinal seriousness scales in analyses using averaging are common in research on career escalation (e.g. Kyvsgaard 2003: Chapter 13).