usually quantify the magnitude of the association between two variables using a correlation coefficient. The Pearson product–moment correlation coefficient (r) is the most commonly used statistic. It reflects the linear relationship between two variables. Correlation coefficients range from 1.0 to –1.0.
The strength of association is determined by the absolute value of the number. Coefficients near 1.0 or –1.0 indicate strong covariation; if we know the value of one variable, we can predict the other with accuracy. Coefficients near 0 indicate weak or absent covariation; the value of one variable does not tell us much about the value of the other.
The direction of the association is determined by the sign of the coefficient. Positive values indicate a direct association between variables; as one variable increases, the other increases. Negative values indicate an inverse association; as one variable increases, the other decreases.
Humphreys and colleagues (2013) used correlation coefficients to investigate the association between children’s ADHD symptoms and their social functioning. They asked parents to rate the severity of their children’s ADHD symptoms, the degree of stress that they experienced caring for their children at home, and the quality of their children’s relationships with peers. The researchers found a positive correlation (r = .32) between children’s ADHD symptoms and parenting stress at home: the greater children’s symptoms, the higher their parents’ stress. In contrast, the researchers found a negative correlation (r = –.31) between children’s ADHD symptoms and the quality of their peer interactions: the greater children’s symptoms, the fewer friends they had at school.
Correlations ≠ Causality
Correlational studies allow us to identify associations between variables, but they do not allow us to say that one variable caused the other. Correlations do not imply causality for two reasons.
First, a correlation between two variables does not tell us the direction of the relationship between the two variables. It is tempting to conclude that children’s behavior problems increase their parents’ levels of stress. However, it is also possible that the opposite could be true. Parents who are experiencing high levels of stress might be more likely to lose their temper toward their children and increase their children’s disruptive behavior (Figure 3.2).
Second, correlational studies do not rule out alternative explanations for covariation. A third variable might explain both the severity of children’s behavior problems and their parents’ levels of stress. For example, parental divorce, unemployment, or illness might adversely affect both children’s behavior and their parents’ stress levels.
Cross-Sectional and Longitudinal Studies
There are two types of correlational designs that are especially relevant to researchers who study childhood disorders: (1) cross-sectional and (2) longitudinal (Kazdin, 2017).
In a cross-sectional study, researchers examine the association between variables at the same point in time. For example, Humphreys and colleagues (2013) assessed the relationship between children’s ADHD symptoms and mood. They found a significant, positive correlation between these variables: children with more ADHD symptoms also experienced more symptoms of depression. However, the researchers could not determine the direction of this relationship because the variables were assessed at the same point in time.
In a longitudinal study, researchers specify the direction of the relationship between variables by measuring them at different times. In a prospective longitudinal study, researchers measure a hypothesized predictor variable at Time 1 and measure its expected outcome at Time 2. For example, Humphreys and colleagues (2013) conducted a second, prospective longitudinal study examining children’s ADHD symptoms in early childhood and depressive symptoms in early adulthood. They found a significant, positive correlation between early attention problems and later depression. Because children’s ADHD and depressive symptoms were assessed at different points in time, the researchers concluded that children’s attention problems emerged before their depressive symptoms (Markon & Markon, 2018).
Figure 3.2 ■ Correlations Do Not Imply Causality
Image courtesy of Pixabay Creative Commons
Note: (A) Children’s behavior problems can increase parenting stress; (B) parenting stress can increase behavior problems; or (C) other factors, such as divorce, can increase both behavior problems and parenting stress.
Prospective longitudinal studies are difficult to conduct because researchers must wait a long time to test their hypotheses and participants often drop out of studies before completion. Consequently, some researchers use other methods. In a retrospective longitudinal study, researchers recruit individuals with a known disorder and ask them (or their parents) to recall events in the past that might have predicted its emergence. For example, researchers might recruit a large sample of young adults with depression. Then, they might ask their parents to recall symptoms of ADHD that these adults showed as children. The chief limitation of retrospective longitudinal studies is that people may not accurately recall past events.
In a follow-back longitudinal study, researchers recruit individuals with a known disorder and examine their medical records, school reports, or other objective data for events in the past that might have predicted its emergence. For example, researchers might ask young adults with depression for permission to review their childhood medical records. The researchers could then determine if participants were ever diagnosed with ADHD or prescribed medication to treat attention problems when they were children. Follow-back studies do not rely on people’s memories of past events. However, obtaining high-quality records is often difficult (Wright & Markon, 2018).
Mediators and Moderators
Considerable research has established a link between ADHD symptoms in childhood and depression later in life. Although this correlation exists, it does not tell us how the two variables are related or why some children with ADHD develop depression and others do not. To answer more complex and interesting questions like these, researchers look for mediators and moderators (Baron & Kenny, 1986).
A mediator is a variable that can help explain how two variables are related. Mediator variables explain the mechanism by which one variable predicts another variable. Mediators tend to be continuous variables—that is, they range from low to high and everywhere in between (Figure 3.3).
Figure 3.3 ■ Mediation and Moderation
Note: Mediators explain how two variables are related. Parenting a child with ADHD can cause stress and conflict in the home; the more stress and conflict, the higher children’s likelihood of depression (Humphreys et al., 2017).
Note: Moderators affect the direction or strength of the relationship between two variables. Children’s likelihood of depression depends on whether they are rejected by peers (Humphreys et al., 2017).
For example, parenting stress may mediate the relationship between ADHD in childhood and depression later in life. Children’s ADHD symptoms can increase parenting stress, prompting
