Transcript VARIABLES

VARIABLES
Concepts that are operationalized
(things that vary)
Because things vary, research is
conducted
 How or why things vary
 How differences in one variable are related
to another
Types of Variables
 INDEPENDENT/DEPENDENT
 CRITERION/OUTCOME
 DICHOTOMOUS
 ATTRIBUTE/ACTIVE
 EXTRANEOUS
 INTERVENING
 CONFOUNDING
 UNCONTROLLED
Independent Variable
 The presumed cause
 The manipulated variable in an experiment
 The treatment
 The “active” variable
The Dependent Variable
 The variable that the researcher is
interested in understanding, explaining or
predicting
 The outcome variable
 The “critereon variable”
Independent /Dependent
Variables
 Variables are not inherently one kind or the
other, you must look at them in the context
of the study.
 Causality is not necessarily implied, rather
the directionality of the influence
independent variable changes in the
dependent variable
Criterion variable
 The dependent variable--the outcome
Dichotomous variable
 A variable having two (and only two) values
e.g., male, female
smoker, nonsmoker
alive, dead
Attribute and Active variables
 Attribute variable--preexisting
characteristic which researcher simply
observes and measures, e.g. bloodtype,
medical diagnosis, etc.
 Active Variable--researcher creates or
manipulates this, e.g. metrology class,
experimental drug, etc.
Variables to be controlled for:
Extraneous Variables which have an
unwanted or irrelevant effect on
 Intervening
the
dependent
variable
under
 Confounding
investigation. The kind of
 Uncontrolled
independent variable which may
unexpectedly alter the results of
the study

CONTROLLING
EXTRANEOUS VARIABLES
 1.
HOMOGENEITY/ELIMINATION OF
VARIABLE
 2. INCLUDE IN DESIGN
 3. MATCHING
 4. STATISTICAL CONTROL
 5. REPEATED MEASURES
 6. RANDOMIZATION
Homogenity, Elimination of Var.
 Use subjects who are homogeneous for the
variable which is suspected may have a
confounding effect on the results of the
study.
eg. If gender or age is thought to be a
confounding factor use all one gender or
all one age cohort as subjects, thus
eliminating this as a factor
Include in the design
 Include the presumed intervening variable
in the design of the study.
 For example: If age is a factor include
several age groups in the design and
analyze the results separately for each
group.
Matching
 Match subject in two groups on relevant
characteristics such as age, gender,
diagnosis.
Statistical Control
 After study is completed perform an
analysis of covariance to determine if the
presumed extraneous variable had an
effecrt.
Repeated Measures
 If the variable is subject to fluxuation,
measure several times under several
different circumstances.
Randomization
 Overall, the best way to control extraneous
variables.
 Random assignment to groups means that
all possible extraneous variables should
be spread evenly between groups.