Abnormal Psychology

(やまだぃちぅ) #1

160 CHAPTER 5


Correlation
The relationship between the measurements
made of two variables in which a change in
the value of one variable is associated with a
change in the value of the other variable.

Correlation coeffi cient
A number that quantifi es the strength of
the correlation between two variables; the
correlation coeffi cient is most typically
symbolized by r.

Statistically signifi cant
The condition in which the probability of
obtaining the value of a statistical test is
greater than what would be expected by
chance alone.

To be a true experiment, you would need to have assigned participants randomly to
the three groups (loss with helplessness, loss without helplessness, and no loss) during
childhood. Obviously, this is undesirable and impossible. But in a quasi- experimental
design, you can sort participants into groups—those who had a childhood loss and
experienced helplessness, those who had a childhood loss but didn’t experience help-
lessness, and those without a childhood loss. Then you would show people in all three
groups a fi lm that involves relationships breaking up. After viewing the fi lm, partici-
pants in all three groups would rate their mood. Your hypothesis is that participants
who experienced early loss and helplessness (like Carlos) will report greater sadness
after seeing the fi lm than will those in the other two groups. With a quasi-experimental
design, you still try to control as many variables—such as age, health, education, and
economic level—as you can in order to make the groups as similar as possible.

Correlational Research
Experiments and quasi-experiments allow researchers to zero in on which variables
cause which effects. In some cases, however, manipulating the variables, even in a quasi-
experiment, can be unethical or diffi cult. This is an issue for any study that involves
participants’ histories—such as medical histories, family histories, or experiences they
had when younger. When independent variables can’t or shouldn’t be manipulated, re-
searchers can study the relations among variables by looking for a correlation, a rela-
tionship between the measurements made of two variables in which a change in the
value of one variable is associated with a change in the value of the other variable.
A correlation compares two measurements and records the amount of similarity in
their variations; the stronger the correlation, the more closely related the variables are.
There are no independent and dependent variables incorrelational research: Nothing is
manipulated; instead, naturally occurring variations among measurements of different
variables are compared. These comparisons can involve measures from different indi-
viduals or groups or measures from the same participants at different times. For exam-
ple, if your study were correlational, your two variables of interest might be the extent
to which a child experienced helplessness during a loss (perhaps rated by relatives who
were present at the time or by the person’s memory of how severe the sense of feeling
helpless was) and the number of symptoms of depression experienced as an adult, after
a breakup. Twin studies (discussed in Chapter 2) often involve correlational research.

Correlation Does Not Imply Causation
A major disadvantage of correlational methods is that they only indicate that two
variables are related. A correlation does not demonstrate that either variable causes
the other to change. In an experiment or in a quasi-experiment, the point is to show
that changes in the independent variable cause changes
in the dependent variable. In contrast, a correlational
research study can only show that the values of two
variables are related. For example, although the de-
gree of helplessness felt during childhood loss and
the amount of depression after an adult relationship
breaks up may be correlated, the loss experienced after
the breakup may not be the cause of the depression.
As discussed earlier, it could be that depression comes
fi rst, and it causes the breakup! Simple correlations do
not control for possible confounding variables. Per-
haps experiencing helplessness as a result of an early
loss leads people to be passive, and that in turn leads
them to have fewer friends or family members who
are supportive. It may be the relatively low amount
of support received after a relationship breaks up that
makes people vulnerable to depression. Such a factor
could have contributed to Carlos’s depression after his
relationship with Liana ended.

A correlational study found that suicide is
more likely to be the cause of death in
high-altitude states, such as Colorado, than
in lower altitude states (Cheng, 2002). This
relationship may arise for any number of reasons,
such as decreased levels of oxygen from the
higher altitude (McCook, 2002), the challenges
of mountain living, differences in the age and
education levels of the populations, or perhaps the
fact that depressed people seek a type of solitude
more often found in high-altitude states. One pos-
sible confound the study did not control for was
the amount of time those who committed suicide
had lived at the higher altitude: Those who were
born in high-altitude states would have adapted to
the lower oxygen level. Researchers could have in-
vestigated one possible explanation by examining
this variable (McCook, 2002).

Paul Chesley/Getty Images

Free download pdf