Although the correlation between child abuse and substance abuse in adulthood has been the subject of multiple studies, there remain methodological, measurement, and conceptual issues in conducting research. By using empirically collected data and statistical analyses to analyse this data, researchers have come to agreement that even though there may be a range of confounding variables or external factors, the correlation between child abuse and substance abuse later in life is an established fact.

Lo & Cheng (2007) investigated the relationship between individuals’ experiences of abuse in childhood and their abuse of substances later in life. Having designed their quantitative descriptive research as a longitudinal study, Lo & Cheng (2007) used a large sample of to test their three hypotheses. The hypotheses were: 1) the experience of child abuse leads to various substances abuse later in life; 2) the more years the abuse is experienced in childhood the greater the impact of a person’s substance abuse later in his of her life; 3) the symptoms of depression act as mediators of childhood victimization on substance abuse later in life. Whereas the study is based on data from National Youth Survey from the first five waves and the seventh wave, the final sample was 762 responses based on the quality of responses.

The variables were distinguished as follows: current alcohol abuse, marijuana abuse, and other drug abuse were dummy variables; the child-abuse variables were physical abuse, chronic physical abuse, sexual abuse and chronic sexual abuse; in addition, depression was designated as a continuous variable; five control variables were gender, race, age, neighbourhood problems, and receipt of public welfare. Two-step logistic regression analysis was used to test the hypotheses. For a start, the substance abuse likelihood in adulthood was regressed on all variables of child abuse and all control variables. Next, Lo & Cheng (2007) entered the depression variable in every model. Using the chi-square statistic, it was found that all models were significant, and each of them predicted the values for measures of current substance abuse.

The results of logistic regression to a certain extent supported the hypothesis of positive association between child abuse and presence of every kind of substance abuse in young adults. At the same time, no statistically significant results were shown by sexual abuse measure while physical abuse measures turned out related to current alcohol, drug, and marijuana abuse. Moreover, after the variable of depression was introduced to the equations, the coefficient size for every child-abuse measure dwindled. Nevertheless, no support for depression’s mediating role was found between physical abuse measures and current abuse of marijuana.

Nomura, Hurd, & Pilowsky (2012) also measured the long-term effects of child abuse on substance abuse during the period of transition from the period of adolescence to that of adulthood. Using the random regionally representative sample of 1,748, the study took its data from National Survey of Drug Abuse. For the purposes of statistical analysis, potential confounders were determined as follows: age during the last interview, race, gender, poverty level at birth and at the age of seven, mother’s mental health status, number of siblings, and mother’s education at the time when the child was born; childhood behavioural characteristics. The statistical analysis occurred in several steps: first, preliminary analyses of groups differences were done using t-tests, and then categorical outcomes were compared by X2-test.

Then, with the aim to control the aforementioned potential confounding effects, the researchers used logistic regression. ANCOVA of linear analysis of co-variance was employed to analyse continuous outcomes and child outcomes, such as number of various substances employed as dependent variables and with child abuse the predictor variable. As a result, it was found that those participants who suffered from physical abuse in childhood were subject to considerably higher rates of substance abuse, including heroin, cocaine, and marijuana. Further, it was found that cumulative risks were from two- to threefold higher among abused participants; earlier initiation to substance abuse was found in young adults who were abused in childhood; it was found the potential confounders do not impact the nature of substance abuse; it was also found that child abuse was associated with increased risks for both functional and medical impairment later in life (Nomura, Hurd, & Pilowsky, 2012).

Min, Farkas, Minnes, & Singer (2007) used structural equation modeling to research the relationships between two long-term sequences of childhood trauma, adults’ substance abuse, and psychological distress. The self-reported data was collected from 285 respondents using the Childhood Trauma Questionnaire; psychological distress was measured with help of Brief Symptom Inventory; substance abuse was measured using the Addiction Severity Index; educational attainment was measured using self-reported data and number of years of education, avoiding coping was measured using three subscales of the COPE questionnaire. The results showed that childhood trauma was related both directly and indirectly to psychological distress and substance abuse, yet they also showed that the psychological distress was not correlated with substance abuse if trauma, avoidant coping, and education were controlled for. Childhood trauma was also considerably correlated with lower education levels. Girls were more likely to employ avoidant coping strategies (Min, Farkas, Minnes, & Singer, 2007).

All three studies provide empirically based evidence that child abuse correlates with substance abuse in adulthood, including alcohol abuse, marijuana abuse, and other drug abuse. Two studies (Lo & Cheng, 2007; Min, Farkas, Minnes, & Singer, 2007) also found correlation between child abuse and psychological problems (depression; psychological distress) in adults. Both found that depression/psychological distress does not correlate with substance abuse while child abuse does. Nomura, Hurd, & Pilowsky (2012) complemented their findings with established correlation between child abuse and medical or functional impairment later in life, findings of earlier initiation to drugs among abused children, and much higher cumulative risks of substance abuse (two to threefold).

Still, the studies have used different methods of data collection and analysis and different approaches to defining variables, which means that their results should be viewed in the context. In particular, Lo & Cheng (2007) did not provide a clear definition of sexual abuse and incorporated in their research only individuals aged 21-24 back in 1987, who suffered abuse before they turned 18; Nomura, Hurd, & Pilowsky (2012) used self-assessment data without any external validation, included only maternal (but not paternal) data on alcoholism and mental health, and did not include mild forms of abuse; likewise, Min, Farkas, Minnes, & Singer (2007) relied on adults’ retrospective self-reported measurement of their childhood traumas; they failed to provide the criteria for trauma specification; the sample only included participants from low socio-economic background. In this way, even though Lo & Cheng (2007) present more rigorously obtained findings due to validating data obtained as a result of sets of interviews and establish the causal relationship between the variables of child abuse and substance abuse in later life, the remaining two studies provide sufficient and reliable evidence that abuse in childhood correlates with and hence may predict substance abuse in adulthood.