Before discussing the studies which deal specifically with resilience in children of parents affected by addiction, we will examine some elements of the concept of resilience from three separate perspectives.
The term "resilience" is borrowed from the physical sciences, where it refers to the ability of a substance to regain its initial state after being subjected to stress. When transposed into the field of human development, resilience refers to a process of adaptation whereby individuals learn to overcome destabilizing effects resulting from traumatic experiences of greater or lesser severity (Masten, 2001). The concept of resilience arises primarily out of the great variability in personal and social adaptation processes observed in situations which pose a seemingly equal degree of risk.
The majority of studies dealing with children of alcohol- or drug-dependent parents have focused on adaptation problems and mechanisms of intergenerational transmission. In the 1970s, however, some researchers began to examine successful adaptation in groups of individuals who had been exposed to a variety of risk factors, which generated interest in what was then referred to as "invulnerability" (Anthony, 1974; Garmezy, 1974; Rutter, 1979; Werner and Smith, 1982). Since no one is completely invulnerable to stress or immune to experiencing difficulties in life and in view of the fact that the studied phenomenon encompassed both subjects who overcame harmful effects of stress and others who had suffered no harmful effects, Garmezy (1991) determined that it was more appropriate to speak of "resilience".
Even if children of alcohol- and drug-dependent parents are more likely to experience a variety of adjustment problems, a significant proportion of them, sometimes even the majority - depending on the definitions used and the age of the child in question - seem to develop normally. Such children, said to be resilient, correspond to false positive cases in epidemiological studies. These cases argue against the predictive links which are frequently established between problems of parental addiction and concomitant or subsequent problems of their children. Deemed detrimental to predictive studies, these subjects are ignored by those attempting to identify the risk factors associated with maladaptive outcomes. Some researchers, however, have grasped the importance of these children who are said to be resilient, recognizing that such children have much to teach us about possible protective mechanisms as a form of prevention.
Some would argue that resilience in these children result from personal attributes which mitigate the harmful effects of parental addiction on children's adaptive processes, an idea which is consistent with the notion of moderating effects. The term "protective factors", however, is preferred to describe environmental characteristics which also exert a moderating influence (Sher, 1993). Resilience and protective factors correspond as well to moderating factors in that they mitigate the cause-effect relationship between a risk factor - parental addiction in this instance - and certain harmful effects (such as adjustment problems in children) (Baron and Kenny, 1986). It should be noted that a moderating factor can also exacerbate this causal relationship, although it would then be more accurate to describe it as a vulnerability factor. Likewise, a resilience factor or protective factor could conceivably completely block the harmful effect of a risk factor; serving as a kind of buffer. Protective factors or resilience factors are frequently confused with compensatory factors, which exert a dominant effect opposite to that of
a) Illustration of the effect of a compensatory factor (C) on the relationship between a risk factor (R) and an adaptation problem (P).
R + (-C) --------------- P
b) Illustration of the effect of a moderating factor (M) on the relationship between a risk factor (R) and an adaptation problem (P).
R ---------------- P
risk factors. Figure 1 illustrates the roles played, respectively by moderating factors and compensatory factors with respect to the relationship between risk factors and harmful outcomes.
Based on the preceding figure, a compensatory factor1, such as a personal attribute or social experience, can exert a dominant influence which is contrary to that of a risk factor. Through addition (or, as in the present case, through subtraction), such a compensatory factor mitigates the risk factor, thereby attenuating the foreseeable adaptation problem "P". A compensatory factor contributes to an increased well-being in individuals exposed to risk factors, as well as in those individuals who are not exposed to any risk. For their part, moderating factors are more closely tied to multiplication and manifest themselves through a (statistically) significant interaction between a given risk factor and a moderating factor (for example R x M). "M" need not exert an effect which is uniquely its own: it needs only to modulate (through multiplication) the link between risk factor "R" and the potential problem. Consequently, it would be inaccurate to speak of resilience factors or protective factors in the absence of risk factors, since there would be nothing to moderate. The following analogy may be useful in clarifying the difference between a compensatory factor and a protective or resilience factor.
Let us imagine that a canoeist facing a head wind is attempting to cross a large body of water. If he has strong muscles, he will be able to "compensate" for the effect of the wind and travel across the water. It should be noted that the canoeist's muscles may be all the more developed for having faced such wind on many prior occasions, but his strength may also be the result of his genetic inheritance. Moreover, strong muscles will enable any canoeist to travel more quickly regardless of climatic conditions. Therefore, they represent a compensatory factor in the sense that they exert a dominant effect which is contrary to the unfavourable factor at hand, namely the wind. This dominant effect has a beneficial overall impact in that it enables anyone who benefits from it to advance more quickly under any circumstance. Obviously, muscular canoeists facing a strong head wind will not travel as fast as those not exposed to wind, but they will nonetheless succeed in crossing the body of water. If their muscles are exceptionally well-developed, they may even keep pace with those not exposed to the wind. Let us imagine for a moment that a mountain is blocking the wind; canoeists moving opposite to the wind will obviously travel more quickly on windy days due to the mountain's protective effect. This protective effect, however, only serves canoeists who travel against the wind. Unlike well-developed muscles, the mountain will in no way favour canoeists traveling with the wind at their back or in the absence of wind. Therefore, it can be said that the mountain exerts protective effect (in the form of an obstacle to the wind) by moderating or partially attenuating the effect of head wind or by entirely canceling it out, thus creating an advantage for - and solely for - the canoeists who are traveling against the wind.
In short, resilience represents a dynamic process which is inherent to development and manifests itself as successful adaptation at the individual level, despite harmful circumstances or life events normally considered to be risk factors from the standpoint of adaptation (Egeland et al., 1993; Luthar et al., 2000; Masten, 2001). Resilience does not depend solely on specific personal attributes, it is also dependent upon the interaction between such attributes and various life experiences and circumstances (Luthar, 1999; Rutter, 1987). This definition gives rise to a number of conceptual and methodological implications.
First, it requires the identification of the entire range of problems associated with the risk factors at issue in the present context of alcoholism, drug addiction or compulsive gambling in parents. If one were to focus too narrowly on the manifestation of equivalent problems: (alcoholism, drug addiction or compulsive gambling affecting children), one would run the risk of ascribing resilience to children who do not possess this trait since, in fact, some may present problems in other areas, such as depression or learning disabilities. The risks for children of alcohol- or drug-dependent parents extend well-beyond the risk of developing substance abuse problems of their own, just as the potential risks for children of compulsive gamblers go beyond the risk of becoming addicted to gambling.
Second, the cited definition requires that resilient children be compared with other children who are not in an equivalent risk category. Therefore, it is essential to begin by evaluating the level of addiction of all parents targeted by the study. It is equally important to ascertain the presence or absence of problems commonly associated with addiction in parents (for example, an anti-social personality or socio-familial dysfunction). In other words, when comparing resilient and non-resilient children of alcohol- or drug-dependent parents, it is crucial that other risk factors, to which the parents or the socio-familial context are exposed, remain constant. Otherwise, the apparent resilience of some children of alcohol- or drug-dependent parents may in fact result from the absence of certain risk factors associated with parental alcoholism or drug-addiction, rather than the presence of personal or socio-familial characteristics which play a protective role or promote resilience. It is also important to ensure that resilient and non-resilient children present the same characteristics of vulnerability in the areas of genetics, neuroendocrinology, psychophysiology, neuropsychology, not to mention temperament and personality, since all of these factors can influence adaptation skills from one generation to the next. Otherwise, apparent resilience may, in fact, result from the absence of personal vulnerability rather than from specific characteristics which can moderate the relationship between the vulnerability which typically results from parental addiction and problems experienced by the next generation that are commonly associated with such addiction. In other words, in the absence of vulnerability, protective or resilience factors, and even compensatory factors, become superfluous, since it is unlikely that the problems normally associated with parental addiction will arise.
Finally, in determining the extent to which individuals exposed to a given risk factor are authentically resilient, it is necessary to compare them to similar individuals who have not been exposed to the risk factor in question, nor to any other risk factor. This suggests that it is not possible to accurately determine the presence of resilience, nor that of protective or resilience factors, without comparing the subjects being examined to peers who have not been exposed to the risk factor but who otherwise provide a match in terms of the standard socio-demographic variables. Ideally, one should also compare children of parents with alcohol, drug addiction or compulsive gambling problems with children exposed to parental risk factors other than alcoholism, drug-addiction or compulsive gambling, in order to determine the extent to which resilience factors are specific.
1It should be noted that a compensatory factor can stem from a risk factor. For example, cerebral damage sustained at a young age in an accident can be "compensated" for by the formation of new neuronal connections which would not normally be present. These compensatory elements, like resilience elements, should exist solely, or especially, in at-risk or in vulnerable individuals who adapt despite their challenging situations. However, most authors use the expression "compensatory factors" to refer to beneficial factors which are available to all individuals and which, in at-risk or in vulnerable individuals, serve to counterbalance or neutralize (through addition or subtraction) risk or vulnerability factors; leading to successful adaptation despite the presence of these risk or vulnerability factors.
The preceding considerations provide an incentive for a close examination of other methodological aspects which are specific to the study of resilience.
Since the introduction of the concept of resilience, a variety of methodological approaches have been employed. Some of these present clear limitations which cast doubt on their internal validity, such as retrospective data collection methods (raising the possibility of oversight or difficulty in reconstituting events), and the inability to control for third variables due to the small number of participants. Retrospective case studies fall into this category, being analogous to anecdotal evidence rather than scientific research. Other approaches - such as cross-sectional studies - are of little use in determining the precedence of resilience factors in relation to the personal, social and academic adaptation of participants. Due to their concomitant nature, the data collected do not provide a means of determining whether characteristics of resilience precede or follow adaptation in resilient children. Consequently, without entirely discounting studies of resilience which adopt a cross-sectional methodology, it must be said that the findings and conclusions of such studies are equivocal at best. Prospective longitudinal studies are far more preferable since, unlike cross-sectional or retrospective studies, they make it possible to determine the direction of the relationship between presumed resilience factors and successful adaptation in children of addicted parents.
There are two types of longitudinal studies: "follow-up" studies and "follow-back" studies. A follow-up study tracks children of parents with an alcohol- or drug-dependency or who are compulsive gamblers and records the experiences and attributes associated with their subsequent adaptation, according to a prospective line. The follow-back method consists of identifying apparently well-adjusted children of parents with an alcohol- or drug-dependency or who are compulsive gamblers and comparing these children with maladjusted children of parents with similar problems, as well as with children of non-dependent parents, according to a series of personal attributes and life experiences which are collected prospectively. These are in fact two variants on the same research approach. The difference between the "follow-up" and "follow-back" methods emerges at the analysis stage. It is important to note that with longitudinal studies one cannot establish a causal relationship between resilience variables and successful adaptation in children. Prevention and intervention studies conducted within an experimental framework in which participants are randomly distributed into an experimental and a control condition might clarify this aspect, but studies of this type are rare. Finally, some authors have studied twins to clarify the contribution of genetic and environmental factors in resilience (Rutter et al., 1997). As far as we know, however, there are no completed studies of twins born of addicted parents, although one such study is currently underway by PÚrusse, Brendgen, Robaey and Vitaro, (2003).
In addition to the selection of a cross-sectional, longitudinal or experimental research design, two empirical approaches have been used to date, to identify the characteristics and life experiences of resilient children. The first approach, which is person-centred, consists of comparing well-adjusted children of addicted parents with children in the same circumstance who present adjustment problems. Such comparisons may be limited to concomitant variables of a personal, familial or social nature, or they may encompass variables and mechanisms (processes) which are compared longitudinally. Another option is to establish a control or normative group composed of children not exposed to the same risks but matched according to a number of control variables. As stated earlier, the presence of such a group is essential in determining (a) the extent to which resilient children resemble children not exposed to the same risks, and (b) the extent to which the resilience factors in question are specific to these children.
The other approach, which is centred on variables, consists of identifying personal or environmental variables which serve to attenuate or - in strict technical terms - to "moderate" the relationship between parental problems and the problems subsequently experienced by children, after first having determined that this relationship is not spurious, which is to say attributable to third variables. The identification of the moderating variables of resilience or protection rests on analytical strategies such as multiple regression, logistic regression, and multi-group structural equation modelling. In the case of multiple or logistic regression, a moderating effect is revealed when a significant interaction is found to occur between a risk factor (i.e., parental addiction in this instance) and a presumed protective or resilience variable (Baron and Kenny, 1986). These two methodological approaches, the person-centred and the variable-centred, are perfectly complementary and equally valuable.