Method of Writing
Research Papers 5
Trương Văn Ánh
HUFLIT

VARIABLES IN NONEXPERIMENTAL STUDIES
A variable is a trait or characteristic that can vary within a
population. It is a variable, then, if it has at least two
categories or values that will apply to participants or units in
the study. The opposite of a variable is a constant—a value
or category that stays the same for everyone or for all units
included in a study.
Research includes different types of variables that are
generally grouped into categorical variables and continuous
variables. Categorical variables have mutually exclusive
categories. There are two types. When variables have at
least two categories, but the categories have no inherent
order, they are called nominal variables or name variables.
Nominal variables do not have a numerical value, but consist
of words or “names” instead. Example 1 below has two
nominal variables. Can you identify them?

Example 1
A sample of registered voters was surveyed. Each voter was
asked first to name his or her gender (male or female), then
to name the candidate for whom he or she planned to vote
(Doe, Jones, or Smith). The purpose was to explore gender
differences in voting preferences. The two variables in
Example 1 are (1) gender, with two categories (male and
female); and (2) preferred candidate, with three categories
(the three candidates: Doe, Jones, and Smith). Gender is an
example of a nominal variable. Each respondent to the survey
in Example 1 belongs to one—and only one—category, and
there is no intrinsic value that determines the order of the
categories. The terminology is important. Be careful not to
confuse a variable with its categories. For instance, “male” is
not a variable, gender is. “Male” is one of the two categories
of the variable called “gender.” Here is how to visualize it:

Variable
Gender
Nominal Categories
Male Female
Researchers define variables in such a way that the
categories within them are exhaustive, meaning all
possible categories are accounted for. Another
nominal category is “state in which you were born.”
States are a nominal category because there is no
built-in order for which state is first, and which is last.
Also, the choices are mutually exclusive because
each person can only be born in one state. There are
50 options to make the choices exhaustive; however,
a 51st option of “not born in the United States”
might need to be offered to make the selections
complete.

Nominal variables are not the only type of categorical
variable. Ordinal variables are categories that have a rank
order. Ordinal variables can be placed in an order, but
the categories do not have a specific, measurable distance
between them. An example would be asking someone to
report on their overall health, and offer the choices:
excellent, very good, good, fair, and poor. It is not possible
to say that “excellent” has a specific and definite distance
from “very good.” It is also not possible to say that the
gap between “excellent” and “very good” is the same gap as
the distance between “very good” and “good.”
Variable

Ordinal Categories

Overall Health

Excellent Very Good Good Fair Poor

By contrast, continuous variables can be quantified. They
have a numerical value, allowing them to be placed in order
and the distance between each item to be exactly
described. There are two types of continuous variables. The
first type is called an interval variable. Interval variables
have a numeric value, so they can be measured along a
continuum with clearly defined intervals between values.
Intervals are also sometimes referred to as scaled variables.
A classic example is temperature. It is a good example
because in addition to showing that it is numerical,
temperature also distinguishes the unique qualities of an
interval variable: it does not have a zero value that indicates
the complete lack of something. Because these values are
scaled to intervals, they do not necessarily compound
in the same way.

For instance, we cannot say that 80 degrees feels
exactly twice as hot as 40 degrees, but we can say
how far apart they are on the scale. In research on
people, we are unlikely to use temperature. Interval
variables are most commonly used when a test
or measure is scaled to evaluate participants. We are
familiar with a few scales of this type, such as SAT
exams or IQ tests. These tests are scaled. Technically,
there may be a zero value, but it is unlikely to suggest
that the person lacks any intelligence. The scale has
definitive intervals, but they do not necessarily
indicate specific quantities so that someone with an
IQ of 240 is twice as smart as someone with an IQ of
120.

The second type of continuous variable is called a
ratio variable. Just like interval variables, ratio
variables are numerical and can be measured along
a continuous number line. The difference is that the
number line for a ratio variable has a zero value,
and zero means that none of that variable is
present. Grade point average (GPA) is tied to a
numerical measurement that has a zero value
indicating a lack of any grade points, for instance.
GPA can be considered a ratio variable. Height,
mass, and distance are other examples of
ratio variables, where a zero value means that it has
no height, no mass, or no distance.

One method for remembering the four types of variables is
the acronym NOIR, which may help you remember both the
names and the order. With each variable type, the
criteria have greater specificity. Continuous variables can
always be reduced to categorical variables. A common
instance in the research of reducing a continuous variable to
a categorical one is age. Age is a ratio variable. It is
numerical, and it is tied to a timeline in which the zero value
means none of the age variable is present. However, it can be
treated as a categorical variable if the researcher groups age
into categories. For research on adults, this might look
something like: 18 to 24, 25 to 34, 35 to 44, 45 to 54, 55 to
64, and 65 or over. Note that the categories must include all
possible values, and the categories must not overlap.
NOIR: Nominal, Ordinal, Interval, Ratio

Another way that variables are classified is as either
independent or dependent. When researchers conduct a
causal-comparative study (see Unit 2 to review), the
presumed cause of an effect (such as smoking) is called the
independent variable, and the response or outcome (such as
lung cancer) is called the dependent variable. One way to
remember the difference is to keep in mind that variables
vary. Those that are dependent are being examined for how
much their variation depends upon the influence of another
variable that is being measured. Independent means that you
are not trying to determine what influences that variable—it
varies, but the factors are independent of those being
measured in this study. Figure 21.1 provides a visual
representation of the relationship between these two
variable types.

In nonexperimental studies, some researchers refer to any
variable that comes first (whether or not it is presumed to be
a cause) as independent and to the one that comes later as
dependent. For instance, SAT scores (the predictor variable)
are usually determined before students earn their college
GPAs. Thus, some researchers would call the SAT the
independent variable and the GPA the dependent variable. It
is also common to call the independent variable the predictor
and the outcome variable (such as GPA) the criterion. The
term criterion means standard. Hence, GPA is the standard by
which the predictive accuracy of the SAT is often judged. If
the SAT works as intended, high scores on the SAT should be
associated with high GPAs, while low SAT scores should be
associated with low GPAs.

To the extent that this is true, the SAT is judged to be
valid. To the extent that it is not true, the SAT is
judged to be invalid. Procedures for determining the
validity of a test are described in Part 5 of this book.
FIGURE 21.1 The independent variable (stimulus or
input) causes changes in the dependent variable
(response or output).
Independent variable exerts influence that affects
Dependent variable.

TOPIC REVIEW
1. What is the minimum number of categories that any variable
must have?
2. Adults who were taking a course to learn English as a second
language were asked to name their country of birth and their
number of years of formal education. In this
example, how many variables were being studied?
3. In Question 2, which variable is a categorical variable? Which
type of categorical variable is it?
4. In Question 2, which variable is a continuous variable? Which
type of continuous variable is it?
5. A sample of adults was asked their level of agreement with
the statement, “The President of the United States is doing a
good job of handling foreign relations.” They were permitted to
respond with either “strongly agree,” “agree,” “disagree,” or
“strongly disagree.” How many variables were being studied?

6. What is a good way to remember the four types of
variables?
7. Studies try to explain the variation in which type of
variable—the dependent variable or the independent
variable?
8. A researcher looked for the causes of social unrest by
examining economic variables including poverty and
income. Is social unrest an independent or a dependent
variable?
9. If a researcher administers a basic math test to middle
school children to see if its scores predict grades in high
school algebra, what is the criterion variable?

DISCUSSION QUESTIONS
1. Suppose you want to measure income on a self-report
questionnaire that asks each participant to check off his or
her income category. Name the categories you would
use. Are they exhaustive and mutually exclusive? Explain.
2. Name a quantitative variable of interest to you, and name
its categories. Which type of variable is it? Are the
categories mutually exclusive and exhaustive? Explain.

RESEARCH PLANNING
Name the major variables you will be studying. Consider the
ways you can measure each variable in terms of the
categories used, the number of categories, and any options
on which type of variable it is. Keep in mind that variables
can often be measured in different ways. Review the
literature that uses variables that are similar to yours. How
have they treated each variable? Are there advantages or
disadvantages to adopting a similar way to measure the
variable?

Good luck!