Method of Writing
Research Paper 3
Trương Văn Ánh
HUFLIT

CAUSAL-COMPARATIVE STUDIES
An experiment is a study in which treatments are given in
order to observe their effects. When researchers conduct
experiments, they ask, “Does the treatment given by the
researcher cause changes in participants' behavior?” When
researchers
want
to
investigate
cause-and-effect
relationships, they usually prefer experimental over
nonexperimental studies. However, for physical, ethical, legal,
or financial reasons, it may not be possible to conduct an
experiment. For example, comparing a smoking group to a
nonsmoking control group, whose members are forbidden to
smoke for 15 years. Clearly, the harm done to those forced to
smoke
would
create
ethical
problems
with such a research design. For this research problem,
researchers cannot conduct an experiment.

If it is impossible or impractical to conduct an experiment to
answer a causal question, a researcher must get the
information from nonexperimental studies. The researcher
then recruits a control group of individuals who do not have
lung cancer but who are as similar to the experimental group
as possible in demographics (basic statistical characteristics of
humans that are used as identity markers, such as
socioeconomic status, gender, or age) and compare them on
prior cigarette smoking as well as other key characteristics,
such as diet, exercise, alcohol use, prescription drug use, and
other measures that may help determine exposure to
environmental toxins. By choosing groups that are as similar
to one another as possible except for having lung cancer,
researchers can feel more confident in suggesting that the
differences they do find between the groups may help to
explain the difference in lung cancer status.

For instance, if the researchers find that smoking
differentiates the two groups who are otherwise
demographically similar, it suggests that smoking is a
possible cause of lung cancer. However, there are
several dangers in making a causal interpretation from
a nonexperimental study. First, smoking and cancer
might have a common cause. For example, perhaps
stress causes cancer and causes individuals to smoke
excessively. If this is the case, banning smoking will
not prevent cancer, only reducing stress will. Another
danger is that the researcher may have failed to
match the correct demographics or other
characteristics of the two groups.

For instance, perhaps the researcher compared the
group with lung cancer to a group without lung cancer
based on age, race, gender, alcohol use, and activity
level, but did not take into account whether the
people in the study lived in urban or rural areas. If the
group with lung cancer mainly resided in urban areas
with heavy pollution, while those without lung cancer
were all residents of rural areas, this weakens the
argument that smoking was the key factor.

These types of problems would not arise in an experiment in
which participants are divided by random assignment to form
two groups. They would not exist because the random
assignment would produce two groups that are equally likely
to experience stress and equally likely to live in either rural or
urban areas and, in fact, be roughly equal in terms of all other
potential causes of cancer.
The above example of smoking and lung cancer illustrates a
specific type of nonexperimental study known as a causalcomparative study (sometimes called an ex post facto
study).3 The essential characteristics of this type of
nonexperimental study are (1) researchers observe and
describe some current condition (such as lung cancer) and (2)
researchers look to the past to try to identify the possible
cause(s) of the condition.

Notice that researchers do not give treatments in causalcomparative studies. Instead, they only describe observations.
Hence, they are conducting nonexperimental studies.
Although the causal-comparative method has more potential
pitfalls than the experimental method, it is often the best
researchers can do when attempting to explore causality
among humans. Note that when it is used properly, and the
comparison groups are selected carefully, the causalcomparative method is a powerful scientific tool that provides
data on many important issues in all the sciences.

TOPIC REVIEW
1. According to the topic, do “experimental” or
“causal-comparative” studies have more potential
pitfalls when one is trying to identify cause-and-effect
relationships?
2. Researchers look to the past for a cause in which
type of study?
3. Is causal-comparative research a type of
experiment?
4. Are treatments given by researchers in causalcomparative studies?
5. Random assignment to treatments is used in which
type of study?

6. How is the term demographics defined in this
topic?
7. A researcher compared the health of low-income
adolescents who had received free lunches during
their elementary school years with the health of a
comparable group of low-income adolescents who
had not received free lunches. The purpose was to
determine the effects of free lunches on health. Did
the researcher conduct an “experimental” or a
“causal-comparative” study?

8. A researcher divided diabetes patients who were being
released from the hospital into two groups. Upon their
release, the researcher provided brief counseling
for individuals with diabetes to one group while providing the
other group with extended counseling. The purpose was to
determine the effects of the two types of counseling on
patients' compliance with physicians' directions during the
first month after hospitalization. Did the researcher conduct
an “experimental” or a “causal-comparative” study?
9. What is another name for a causal-comparative study?

DISCUSSION QUESTIONS
1. If you wanted to investigate the causes of child abuse,
would you use the “experimental” or the “causalcomparative” method? Explain.
2. Suppose you read a causal-comparative study indicating
that those who take vitamins A and E tend to be less
overweight than the general population. What possible
dangers are there in the interpretation that the vitamins
cause individuals to maintain a healthy weight?

RESEARCH PLANNING
If you will be conducting a causal-comparative study,
briefly explain why you chose this method. What are
you comparing? What are the relevant characteristics
that you would want to match in your two groups?
What are some potential pitfalls that readers may
want you to account for? How can you address these
in your research plan?

NOTES
1. Because of the laws of probability, the larger the sample,
the more likely that two groups formed at random will be
equal
in
terms
of
their
characteristics.
2. The relationship between smoking and health has been
examined in many hundreds of causal-comparative studies.
On this basis, almost all experts agree that alternative
interpretations are without merit. However, the results of
some of the early studies on the health effects of smoking
were disputed because the studies were not true experiments
(i.e., they did not have random assignment to groups).
Remember that most findings are based on multiple,
sometimes hundreds of studies that support similar
conclusions.
3. Other types of nonexperimental studies are covered in the
next topic.

Good luck!