4 RESULTADOS Y DISCUSIONES
4.2 Análisis de ingesta
substantial field. Rather, I draw on the literature to identify broad patterns of continuity and discontinuity, consider major theories developed to account for these patterns, and discuss some of the unanswered questions and controversies arising from the research.
2. Continuity and discontinuity in aggression from childhood to
adulthood
Before reviewing the empirical research on the developmental patterns of aggression from childhood to adulthood, it is essential to consider what is meant by the terms stability, continuity, and discontinuity. Stability refers to the observed level, amount or frequency of the same behaviour or characteristic over time. Examples include weight gain over a six-month period, or the number of fights that a child gets into per week across grades 5, 6 and 7 (Lerner, 2002). Although correlation coefficients are usually interpreted as indicators of stability, this refers to relative, rather than absolute, stability. Absolute stability would be observed if there were no weight changes for any study member, or each child reported the exactly the same number of weekly fights at each measurement. Relative stability is observed if study members retain their ranking relative to other study members over time, even if changes occur at the population level. For example, if a boy got into five fights a week in grade 5 and three fights a week in grade six, yet remained in the top 10% of the sample, his behaviour would be stable relative to the other children in the study. Longitudinal studies of aggression usually demonstrate relative stability (Huesmann & Moise, 1998).
Continuity is a more general and confusing term. It can be used to describe relative stability (for example, Huesmann & Moise, 1998), but it may also refer to the relationship between a particular type of behaviour at one point and an apparently different behaviour at a later point. For example, temper tantrums in childhood might predict fighting in adolescence. This is often referred to as heterotypic continuity. It is regarded as continuity because although manifest behaviours vary over time, it is assumed they continue to arise from some specific underlying disorder, characteristic, or process (Rutter, Kim-Cohen, & Maughan, 2006; Schulenberg, Maggs, & O'Malley, 2003). There are many examples of heterotypic continuity in the longitudinal aggression literature. In the present study, any link between school bullying and adult aggression would be a form of heterotypic continuity.
Discontinuity can refer to both the absence of relative stability and the absence of heterotypic continuity. In the first case, nonsignificant correlation coefficients would show that the relative ranking of study members differs between measurement points. For example, children who report bullying in grade 5 might be no more or less likely to report bullying in grade 6 than children who did not bully in grade 5. Abrupt decreases or increases in behaviour, such as desistance from crime or late-onset offending, are also examples of this sort of discontinuity. In the second case, behaviour at one point would fail to predict a different manifestation at a later point. For example, there would be a nonsignificant relationship between teacher-rated aggression in high school and violent delinquency in early adulthood (Lerner, 2002).
Stability in aggression and bullying
One of the most consistent findings from longitudinal studies of antisocial behaviour is that the best predictor of later antisocial behaviour is current antisocial behaviour. Most longitudinal studies of aggression find that measures of the same behaviour are correlated between time points, indicating that study members generally
retain their ranking relative to aggressive behaviour in the rest of the cohort (Farrington, 2007b; Huesmann & Moise, 1998; Loeber, 1982; Loeber & Hay, 1997; Olweus, 1979; Tremblay, 2000).
For example, Cairns and colleagues (1989) assessed 220 Carolina children annually from age 10 (grade 4) to early adolescence (grade 9). Teacher-ratings and self- reports of getting into trouble at school, fighting, and arguing in grade 4 were found to account for the majority of predicted variance in the same measures taken in grade 9 for girls and boys. Stability coefficients over time were strongest for measures of fighting. In Finland, the Jyväskylä Longitudinal Study has tracked the development of 369 individuals from 8–9 years old until the age of 42 with teacher and peer reports, interviews, and adult criminal records. Findings from early waves showed that peer ratings of aggression at age 8 were significantly associated with the same measure at age 14 for both girls and boys, and teacher ratings of aggression between 8 and 14 were related for boys, but not for girls (Pulkkinen & Pitkänen, 1993). A multi-cohort study of delinquency conducted in the Netherlands collected Child Behaviour Checklist (CBCL) reports of competencies and behavioural problems from parents of 1412 boys and girls aged 4 to 14 in 1983, and 6 to 16 two years later. There was significant stability for items assessing aggression over the two-year interval for both boys and girls, with correlation coefficients averaging .66 (Verhulst & Althaus, 1988). A four-year follow- up also found significant stability for teacher reports of aggressive and externalising behaviours (Verhulst & Van der Ende, 1991). Findings from other longitudinal studies conducted in the United States, Europe, and Australia confirm the relative stability of the same measures of aggression from early childhood to adolescence (for example, Cairns & Cairns, 1994; Eron & Huesmann, 1987; Kingston & Prior, 1995; Loeber & Stouthamer-Loeber, 1998; Maughan, Pickles, Rowe, Costello, & Angold, 2000; Stanger, Achenbach, & Verhulst, 1997).
Longitudinal studies of bullying generally report stability patterns that are consistent with other measures of aggressive behaviour in childhood and early adolescence, although findings vary depending upon whether stability/continuity is assessed using continuous measures or by grouping children according to their bullying status at each time point. Correlations between measures of bullying over time are generally very high, ranging from .30 to .60 over time (Camodeca, et al., 2002; Pellegrini & Long, 2002). However, there is evidence that stability is greater for boys rather than girls. For instance, Salmivalli and colleagues (1998) reported coefficients of .34 and .52 for peer-nominated and self-reported bullying respectively amongst Finnish boys between grades 6 and 8. For the girls, in contrast, peer-nominated bullying
measures were correlated at .28, and self-report measures over the two-year period were not significantly related. Studies using classification techniques consistently indicate that children who report bullying at one time point are more likely than other children in the sample to also report bullying at the next time point. Generally, these findings show that between 30% and 40% of children classified as bullies at one assessment are also classified as bullies at the next assessment (Kumpulainen, et al., 1999; Schäfer, et al., 2005; Scholte, et al., 2007; Sourander, Helstelä, Helenius, & Piha, 2000). Thus, bullying, like other measures of aggression, demonstrates relative stability during childhood and adolescence.
Heterotypic continuity in aggression and bullying
Longitudinal findings also show that childhood aggression demonstrates heterotypic continuity to a range of later problematic outcomes, including various manifestations of violence, teenage delinquency and adult offending. For instance, David Farrington and colleagues have examined the development of aggressive and criminal behaviour in the Cambridge Study in Delinquent Development, a prospective survey of 411 boys from working-class London assessed nine times over a period of 40
years from age 8 to age 48. Results indicate substantial continuity from childhood aggression to adolescent and adult aggression and offending. For example, teacher-rated aggression at ages 8, 10, 12 and 14 predicted self-reported violence at age 18, as well as officially recorded violent offending (Farrington, 1991; Farrington & West, 1981). Other major longitudinal studies report similar findings. Later results from the
Jyväskylä study mentioned above showed that peer and teacher ratings of aggression at age 8 and 14 predicted violent offences up to age 20 for males (Pulkkinen, 1987), and self- and other-rated aggression at age 14 were significantly associated with self-reports of aggression at age 26 for both males and females (Pulkkinen & Pitkänen, 1993). Another Scandinavian study examined the development of aggression and crime from age 10 to 35 for 1,027 girls and boys in Obero, Sweden. Teacher-rated aggression at age 13 predicted officially recorded crime up to age 26 for both boys and girls, with highly aggressive boys particularly likely to commit later violent offences (Stattin &
Magnusson, 1989). North American studies paint a similar picture (Loeber, et al., 2005; McCord, 1992; McCord & Ensminger, 1997; Tremblay, 2002). In the Seattle Social Development project, for instance, in which over 800 children were assessed annually from age 10 to 16 and followed-up two–three times to age 27, self-reports of fighting at age 10 predicted engaging in at least some violence between the ages of 13 and 21 (Kosterman, Graham, Hawkins, Catalano, & Herrenkohl, 2001).
Continuity vs. discontinuity
Overall, the longitudinal research provides strong evidence for continuity of aggression across time and situations. No matter what the measure of aggression, individuals who are relatively more aggressive than others at earlier ages are also likely to be relatively more aggressive at later ages. The evidence for discontinuity, however, is also strong: the majority of aggressive adults were aggressive children, but most aggressive children do not become aggressive adults (Robins, 1978; Tremblay, 2000).
In other words, not everybody who is aggressive in childhood is equally aggressive in adulthood, no matter how the behaviour is assessed and no matter whether stability or continuity is the focus. For example, about half of the most aggressive boys at ages 8– 10 in the Cambridge Study were still among the most aggressive at age 32, compared with about one third of the rest of the sample. Although this difference was statistically significant, Farrington (1991) notes that ‘the difference between half of one group and one third of another does not indicate a very accurate prediction over time.’ (p. 12). This effect is mirrored in the longitudinal bullying literature. Although studies generally show that children assessed as bullies at the first measurement point have the greatest likelihood of falling into the bullying group at the second measurement point, most also find that up to half the children who were bullies at the first time point ‘desist’ from bullying at the second point, while there are often substantial numbers of children who report bullying at later times who would not have been categorised as bullies at the initial assessment (Espelage, et al., 2001; Schäfer, et al., 2005).
There is thus a seeming contradiction where people are unlikely to
spontaneously become highly antisocial adults, yet most of those who were antisocial as children don’t persist into adulthood (Robins, 1978). Longitudinal studies, and
developmental theory, are devoted to disentangling these two seemingly contradictory facts. The process of disentanglement reveals several general patterns or principles that are frequently observed in the longitudinal literature, and, despite the presence of unresolved controversies, appear to receive broad assent amongst researchers: (a) the prevalence of aggression declines with age, (b) manifestations of aggression change with age, (c) patterns of continuity and change are not the same for everyone, (d) it is often possible to identify subgroups of individuals who exhibit distinct developmental trajectories of aggression, and finally (e) there are important contemporaneous and situational influences on acts of aggression.
a) The prevalence of aggression declines with age
Aggression is relatively stable from childhood to adulthood, but the absolute prevalence of aggression and violence at the population level declines over this period. Although studies vary in the timing of the change and the rate of the decline,
longitudinal evidence clearly shows that, on average, the frequency of engaging in overtly aggressive behaviours is highest in childhood and subsequently declines
throughout adolescence and into early adulthood. Findings from Richard Tremblay and colleagues’ research on the origins of physical aggression suggests that it peaks at a very young age, possibly as young as two years old. These researchers have traced the development of 1,037 kindergarten boys living in poor areas of Montreal from
childhood until early adulthood (Tremblay, Vitaro, Nagin, Pagani, & Seguin, 2003). Aggressive behaviour (eg. hitting, kicking, and fighting) was first rated by teachers when boys were six years old, then again at age ten and annually to 15 years old. There was a steady decline in aggression over this period. For example, only one in every eight boys who were judged to be very aggressive in kindergarten continued to exhibit the same high levels of aggressive behaviour in later adolescence (Nagin & Tremblay, 1999). Based on these results, the authors challenged the view that physically
aggressive responding is acquired via social learning during childhood. Instead, they argued that children ‘take up’ physically aggressive acts as soon as they are physically able to do so, and the subsequent decrease in frequency of such behaviour occurs because children learn not to aggress over the course of development (Tremblay, 2000). Although other researchers disagree with this interpretation (for example, Loeber & Stouthamer-Loeber, 1998), most findings concur with a pattern of decreasing aggression throughout the primary school years.
The population-level decline continues throughout adolescence and into early adulthood. Loeber and colleagues, for example, followed the development of three
cohorts of boys (N=1517) from middle childhood to early adulthood in Pittsburgh, Pennsylvania (Loeber, Farrington, Stouthamer-Loeber, & White, 2008; Loeber & Hay, 1997). Annual follow-ups showed that the prevalence of physical fighting started decreasing at 15 and continued through age 17. Further analyses from the multi-cohort Dutch study mentioned above revealed steadily declining trajectories of aggression as measured with the CBCL (e.g., argues, bullies, gets into fights, attacks people,
disobedient, irritable) from age 4 to 17 for both boys and girls (Stanger, et al., 1997). By early adulthood rates of violence and aggression decrease even amongst the most
antisocial individuals. In the Seattle Social Development project, for example, 54% of females and 55% of males engaged in at least one act of violence after the age of 13, but reported no violence at 21. Even those who persisted in violence at 21 showed a
relatively modest decline after the age of 18. A similar early adulthood pattern emerged when the men in the Cambridge study (Farrington & West, 1981) were followed up at the age of 21. Although those men who had been delinquent at the age of 18 continued to report more aggressive behaviour and offending than those who were not delinquent at 18, the overall level of antisocial behaviour, including self-reported involvement in fights, had declined for both groups.
It is possible that the prevailing pattern of desistance from antisocial behaviour over time may conceal late-onset aggression in adolescence and early adulthood amongst individuals with little prior history of such behaviour. Although researchers have not yet reached agreement on this issue, the available evidence for late-onset or increasing aggression during adolescence and adulthood is limited (Loeber & Hay, 1997; Moffitt, 2007). For example, Broidy and colleagues (2003) examined the developmental course of aggression from around ages 6 to 14 and its relationship with violence in later adolescence by comparing six datasets from New Zealand, Canada, and the United States. Similar patterns emerged across countries and studies. First,
desistance from problem behaviour continued from childhood to adolescence in all but the most chronically physically aggressive, and in some of the studies even these individuals exhibited decreasing prevalence by later adolescence. Second, none of the datasets provided evidence of the sudden appearance or abrupt increases in aggression after the age of 6. Although some longitudinal studies do identify individuals who begin to aggress in later adolescence, this subgroup generally represents less than 5% of the total sample under study (for example, Kosterman, et al., 2001; Nagin & Tremblay, 2005).
b) Manifestations of aggression change with age
The evidence presented for developmental declines in aggression appears to be contradicted by the existence of the age-crime curve. Across the developed world, officially recorded violent offending consistently peaks during the teenage years (Farrington, 2007b). For example, statistics from the US, the UK and Australia show that the peaks for crimes such as robbery, assault, and rape are around 17–18 years old (Federal Bureau of Investigation, 2004; Home Office, 2004). Self-report surveys
generally find similar patterns, although the peak ages tend to fall a little earlier, around 16–17 for males and 14–15 for females (Budd, Sharp, & Mayhew, 2005; Elliott, 1994).
There are two main explanations that can reconcile the contradiction between declining prevalence of aggression on the one hand, and the observed adolescent spike in the violent age-crime curve on the other. First, although aggression as a whole declines from childhood onwards, serious violence increases with age, peaking in adolescence and declining in early adulthood (Farrington, Loeber, & Jolliffe, 2008). In other words, there are qualitative changes in manifestations of aggression at different ages. Via heterotypic continuity, one form of aggression in childhood (for example, getting into fights), predicts another more serious form of aggression in adolescence (for example, carrying a weapon). Loeber and colleagues (2008; Loeber, Keenan, &
Zhang, 1997) illustrate this pattern. Prospective and retrospective parent-and self-
reports from boys in the Pittsburgh Youth Study were used to create age-of-onset curves for different forms of aggression from age 6 to 17. The different forms were minor aggression, fighting, and serious violence, such as forced sex or aggravated assault. It was shown that the age of onset was lowest for minor aggression, followed by fighting, which increased from age 10, and violence, which became more likely after the age of 12. The authors argued that this pattern represented an ordered developmental sequence, in which minor forms of aggression at younger ages ‘escalated’ to more serious forms at later ages. These findings suggest that although only a minority of children persist in serious aggression beyond childhood, those who do so are most likely to engage in violence during late adolescence.
The second explanation for the aggression decline vs. age crime curve paradox is that the societal response to violent acts changes after childhood. For instance, it is far more likely that school authorities will contact police if a 16 year-old boy punches another student in the face than if a 6 year-old boy does the same. As Cairns and Cairns observe, ‘the stakes of aggression get higher with age’ (2000, p. 421). Thus, violence committed by adolescents is much more likely to be captured by official statistics than violence committed by children.It could also be speculated that the decline in the age- crime curve after late adolescence may simply mask further developmental changes in the manifestation of aggression, as the targets of individuals’ violence shift to partners and children, such incidents being less likely to be reported.
Manifestations of bullying change with age
Consistent with trends from the wider aggression literature, longitudinal studies of bullying usually find that it declines at the population level as children grow older. Studies vary in the timing and rate of decline, but a decrease in prevalence of self- reported bullying and victimisation by late high school is a consistent finding (for a
review refer to Smith, Madsen, & Moodey, 1999).There are, however, some
differences between the longitudinal patterns reported in the bullying field and those reported for aggression and antisocial behaviour more generally. First, some studies show that the prevalence in reported bullying ‘spikes’ around late childhood and early adolescence, coinciding with the transition to high school (Pellegrini & Long, 2002; Rigby, 1996, 2002b). Pellegrini and Long suggest that this occurs because children use bullying to establish positions of dominance in new social networks, and that once power relationships are in place, reports of bullying decline.