Usually, we cannot manipulate things like: gender; race; age; ethnicity etc.
These are subject variables:
Natural treatment: independent variable where exposures to events, situations, or
settings that emanate from the ‘real world’ define how participants are selected
Exposure and nonexposure would be the levels of this variable
Researchers can only provide ex post facto (‘after-the-fact’) analysis of the
effects on a dependent variable
Quasiexperiment: resembles a true experiment except for the degree to which an
experimenter can directly control and manipulate one or more of the independent
Whether the experimenter can randomly assign participants to experimental and
control conditions (in this case, they can’t)
Nonequivalent-control-group designs: have experimental and comparison groups that
are designated before the treatment occurs and are not created by random assignment
Random assignment cannot be used to create groups
Confounds related to equivalency of groups (control vs. experimental) cannot be
o Often high in external validity:
o Particularly ecological validity
Def.: method of selection of a control group for a quasiexperiment
Individual matching: individual cases in the treatment group are matched with similar
Aggregate matching: identifying a comparison group that matches the treatment group
in the aggregate rather than trying to match individual cases
Matching can lead to a problem known as regression artifact: threat to internal validity.
This occurs whenever a pretest measure is used for matching.
People who originally had extreme scores on a test will later, after interacting
with a group, have scores that are closer to the groups’ scores regression to
o This can be a confounding factor for interpreting results of studies that
match groups based on pretest measures Regression to the Mean (THIS IS EXTRA FROM THE LECTURE)
Regression to the mean: a statistical phenomenon that can make natural variation in
repeated data look like real change. It happens when unusually large or small
measurements tend to be followed by measurements that are closer to the mean.
Graphical example of true mean and variation, and of regression to the mean using a
How to Reduce the Effects of Regression the Mean
1. Random allocation to the comparison groups
a. If subjects are randomly assigned, the responses from all the groups
should be equally affected by RTM.
b. The difference between the mean change in the control group and the
mean change of the comparison group is then the estimate of the
treatment effect after adjusting for RTM
2. Selection of subjects based on multiple characteristics
a. The effect of RTM increases with larger measurement variability. To
reduce variability, you can select subjects based on two or more baseline
b. Study selection criterion (cutoff) is then applied to either the mean of
multiple measurements or the second, later measurement (as long as
RTM has taken place between the first and later measurements. This can
method can be thought of as an attempt to get a better estimate of each
subject's true mean before the intervention.
Before-and-after designs: have a pretest and posttest but no comparison group.
The participants act as their own control group which means no comparison
Simplest type of B-&-A is the fixed-sample panel design that has one pretest and one
Interrupted-time-series design: examines observations before and after a naturally
occurring treatment. Experimental group(s) where multiple observations have been obtained before
and after a naturally occurring treatment.
John Gibbons and colleagues (2007)
Wanted to know whether the policies that discouraged the use of antidepressants in
treating children and adolescents had inadvertently led to untreated depression that
later led to a jump in suicide rates.
Turns out it was right.
The results had ecological validity as they suggested that antidepressants may
help reduce suicide rates of children, adolescents, and adults.
Viewed as quasiexperiment of the effects of a natural treatment (public health
warnings) on a dependent variable or outcome measure (suicide rates)
Multiple group before-and-after design: several before-and-after comparisons are made
involving the same independent and dependent variables but with different groups
David P. Phillips (1982)
Study of the effect of TV soap-opera suicides on the number of actual suicides in the
In 12 of 13 comparisons, deaths due to suicide increased from the week before
each soap-opera suicide to the week after
Repeated-measures panel design: includes several pretest and posttest observations of
the same group
Stronger than simple B-&-A designs because they allow the researcher to study
the process by which an intervention or treatment has impact over time
Time series design: compare the trend in the dependent variable up to the date of the
intervention or event whose effect is being studied and the trend in the dependent
variable after the intervention
Disparity between the p