CHAPTER 7: EXPERIMENTAL, QUASI-EXPERIMENTAL AND EX POST
FACTO DESIGNS:
EXPERIMENTAL DESIGNS:
-Used to identify cause-and-effect relationships
-Researcher considers many possible factors that might cause or influence a particular
condition
-Researcher attempts to control for all influential factors except those whose possible
effects are the focus of investigation
-Researcher exposes one group to a treatment or intervention
-If everything is the same between the two groups expect for the intervention,
we can reasonably conclude that the treatment or intervention brought about
the change that was observed
-There are clearly identifiable independent and dependent variables present:
o Independent variable:
-Is the one the researcher studies as having a possible effect on one or more of the other
variables
-Researcher directly manipulates one or more independent variables
o Dependent variable:
-The variable that is potentially influenced by an independent variable
-Its value depends to some degree on the value of the independent variable
-Whenever we compare two or more groups that are or might be different in ways in addition
to the particular treatment or intervention we are studying, we have confounding variables.
-Presence of such variables makes It difficult to draw a conclusion about a cause and
effect relationship as it threatens the internal validity
CONTROLLING FOR CONFOUNDING VARIABLES:
-To maximize internal validity when researcher wants to identify cause-and-effect relationships
and has to rule out confounding variables as explanations for any effects observed.
Confounding variables: Outside influence that changes the effect of a dependent and
independent variable.
Strategies To Control Confounding Variables:
1. Keep some things constant (same factor…environment, age gender for everyone)
2. Include a control group (group that receives no/neutral intervention – compare to
experimental group)
, -Use of placebo ethical issues:
-Must be aware if placebo treatment as well as experimental treatment
-At end of study those who have looked for professional use after study
who was involved in placebo should have opportunity to seek more
effective treatment
-When studying treatment researcher must weigh the benefits of the
new knowledge gained against the lives that might be saved by including
all participants in the treatment group
3. Randomly assign people to groups (on average groups quite similar – representative of
population)
4. Use one/more pre-tests to assess equivalence before the treatment (determine
whether groups are similar with respect to other variables that might have an influence)
5. Expose the participants to all the experimental treatments (have participant undergo all
experimental & controlled treatments)
6. Statistically control for confounding variables (partial correlation, ANCOVA, SEM)
PRE-EXPERIMENTAL DESIGNS:
-Isn’t possible to show cause-and-effect relationships because:
a.) Independent variable doesn’t vary
b.) Experimental and control groups are not comprised of equivalent or randomly selected
individuals.
-Designs are helpful only for forming tentative hypotheses – should be followed up with more
controlled studies.
FACTO DESIGNS:
EXPERIMENTAL DESIGNS:
-Used to identify cause-and-effect relationships
-Researcher considers many possible factors that might cause or influence a particular
condition
-Researcher attempts to control for all influential factors except those whose possible
effects are the focus of investigation
-Researcher exposes one group to a treatment or intervention
-If everything is the same between the two groups expect for the intervention,
we can reasonably conclude that the treatment or intervention brought about
the change that was observed
-There are clearly identifiable independent and dependent variables present:
o Independent variable:
-Is the one the researcher studies as having a possible effect on one or more of the other
variables
-Researcher directly manipulates one or more independent variables
o Dependent variable:
-The variable that is potentially influenced by an independent variable
-Its value depends to some degree on the value of the independent variable
-Whenever we compare two or more groups that are or might be different in ways in addition
to the particular treatment or intervention we are studying, we have confounding variables.
-Presence of such variables makes It difficult to draw a conclusion about a cause and
effect relationship as it threatens the internal validity
CONTROLLING FOR CONFOUNDING VARIABLES:
-To maximize internal validity when researcher wants to identify cause-and-effect relationships
and has to rule out confounding variables as explanations for any effects observed.
Confounding variables: Outside influence that changes the effect of a dependent and
independent variable.
Strategies To Control Confounding Variables:
1. Keep some things constant (same factor…environment, age gender for everyone)
2. Include a control group (group that receives no/neutral intervention – compare to
experimental group)
, -Use of placebo ethical issues:
-Must be aware if placebo treatment as well as experimental treatment
-At end of study those who have looked for professional use after study
who was involved in placebo should have opportunity to seek more
effective treatment
-When studying treatment researcher must weigh the benefits of the
new knowledge gained against the lives that might be saved by including
all participants in the treatment group
3. Randomly assign people to groups (on average groups quite similar – representative of
population)
4. Use one/more pre-tests to assess equivalence before the treatment (determine
whether groups are similar with respect to other variables that might have an influence)
5. Expose the participants to all the experimental treatments (have participant undergo all
experimental & controlled treatments)
6. Statistically control for confounding variables (partial correlation, ANCOVA, SEM)
PRE-EXPERIMENTAL DESIGNS:
-Isn’t possible to show cause-and-effect relationships because:
a.) Independent variable doesn’t vary
b.) Experimental and control groups are not comprised of equivalent or randomly selected
individuals.
-Designs are helpful only for forming tentative hypotheses – should be followed up with more
controlled studies.
