PSYCH-INTRO TO RESEARCH METHODS
CHAPTER 8 NONRANDOMIZED RESEARCH AND CAUSAL REASONING
HOW IS CAUSAL REASONING ATTEMPTED IN THE ABSENCE OF RANDOMIZATION
Prospective data is collecting data by following one’s reaction forward in
time. Commonly used in longitudinal research, the defining characteristic
of such research being that individuals are observed and measured
repeatedly through time.
Retrospective data is data collected back in time.
Campbell and Stanley coined the expression quasi-experimental
research to describe all the different types on nonrandomized research
for generalized causal inference. Quasi merely means, “resembling.” This
research often resembles randomized experimental research because
quasi experiments have outcome measures, sampling units, and
something comparable to the experimental treatment (i.e.,an
intervention), but what identifies them as quasi-experimental is that they
do not randomly assign individual units to treatment conditions.
WHAT IS THE “THIRD-VARIABLE” PROBLEM?
Third-variable problem is correlated with both X and Y is the reason X and
Y covary. A third variable that is correlated with both X and Y can also
account for the correlation between X and Y.
HOW CAN CAUSAL EFFECTS BE STUDIED IN NONEQUIVALENT GROUPS?
Nonequivalent-groups design are traditionally between-subjects designs
in which the sampling units (subjects, groups, etc.) are allocated to
experimental and control groups by means other than randomization and
are observed or tested before and after the experimental treatment.
Subclassification on propensity scores reduces all of the variables on
which the “treated” and “untreated” subjects differ into a single
composite variable. This composite variable, called a propensity score, is
a summary statistic of all the differences on all variables on which the
“treated” and “untreated” subjects differ. Actual procedure requires a
WAIT-LIST CONTROL GROUPS
Wait-list control group gives information about the immediate and
delayed effect of the treatment as well as a replication of the immediate
WHAT ARE TIME-SERIES DESIGNS AND “FOUND EXPERIMENTS”?
In time series designs, the defining characteristic is the study of variation
across some dimension over time.
When the effects of some intervention or “treatment” are inferred from a
comparison of the outcome measures obtained at different time intervals
before and after the intervention, the data structure is called an
interrupted time-series design.
The term time series means there is a data point for each point in time,
and an interrupted time series means there is a dividing line at the
beginning of the intervention (a line analogous to the start of the
David P. Phillips referred to his studies as “found experiments” because
they are essentially found (or discovered) in naturally occurring situations. PSYCH-INTRO TO RESEARCH METHODS
WHAT WITHIN-SUBJECTS DESIGNS ARE USED IN SINGLE-CASE EXPERIMENTS?
A family of nonrandomized designs that is also a mainstay of behavior
modification research is single-case experimental research (also called
small-N experimental research and N-of-1 experimental research).
These designs are conceptualized as a subcategory of interrupted time-
series designs. They incorporate “treatments” (interventions) that are
manipulated and controlled for within a repeated-measures design.
In single-case experiments, [a] only one sampling unit is studied, or only a
few units are studied; [b] repeated measurements are taken of the unit (a
within-subjects design); [c] random assignment is rarely used.
Behavioral baseline—pretest—is the observations