Chapter 9 (Sam) Flashcards
Fundamental problem of causal inference
the challenge that we face making causal inferences due to the fact that we cannot observe given cases in both their treated and untreated states.
Random Assignment
the challenge that we face making causal inferences due to the fact that we cannot observe given cases in both their treated and untreated states.
Experimental (treatment) group
a group of subjects exposed to the intervention of interest in an experiment; identical to the control group is all respects except that the control does not receive the treatment.
control group
a group of subjects randomly assigned not to receive the treatment in an experiment; identical to the treatment group in all other respects.
outliers
a case that differs significantly from the others in the distribution.
single-blind design
an experimental design in which subjects remain unaware of various types of information—such as whether they are part of the treatment, placebo, or control group—in order to reduce the possible bias that this information could induce.
double-blind design
a research design in which both the subjects and the research team are unaware of who receives the treatment and who receives a placebo; intended to reduce the risk of the researchers providing subjects with cues about how they should react and to control for bias in the data collection.
Experimental Context
The experimental context is where the experiment occurs: in a laboratory, over the telephone, online, or in the real world. Contemplation of the experimental context is crucial. For instance, it might be unwise to test whether subjects are responsive to the subtleties of a TV political advertisement if they are recruited in a mall while doing last-minute Christmas shopping.
placebo group
a form of stimulus that does not contain the precise treatment the researcher is testing; often takes a form similar to the experimental treatment without exposing subjects to the specific ingredient that is hypothesized to have a causal effect (e.g., a sugar pill); a type of control group in that it offers the experimenter a point of reference to estimate the treatment’s effect.
between-subjects design
an experimental design in which different subjects are randomly assigned to various treatment and control groups; causality is inferred based on post-treatment differences observed between these groups.
within-subjects design
an experimental design in which researchers evaluate subjects before and after exposure to a given treatment; may also involve comparisons made to the before and after observations of the control group; causality is inferred based on any differences observed between comparisons.
stimulus (stimmy uh)
the treatment(s)—often involving exposure to different forms of information—to which subjects in an experimental group are exposed.
factors (the parameters manipulated by the experimental design)
the parameters manipulated by the experimental design.
levels
the different possible settings for a factor.
factorial design
an experimental design where the researcher manipulates two or more factors simultaneously.
Laboratory experiments
an experiment in which subjects are recruited to a common location where the researcher exerts a relatively large degree of control over the experimental setting.
Survey experiments
an experiment implemented in the context of a survey involving the random manipulation of a part (or parts) of the survey instrument.
split-ballot designs
a survey experiment that compares responses to various versions of a single survey.
field experiment
an experiment that takes place in the subjects’ natural environment.
natural experiment
a course of naturally occurring events in which different conditions appear to be randomly assigned to subjects without the planned intervention of the researcher (e.g., a lottery); to qualify as a natural experiment, assignment to experimental and control groups must be allocated as if by random assignment.
quasi-experiments
studies that do not meet the full conditions of experimental design and that are not naturally occurring; a course of events in a subject’s environment that implies neither “as if” random assignment nor any planned intervention by the researcher.
Case study
a method of analysis that involves an in-depth investigation of a single individual, group, or event; in political science, generally used with the intent of identifying general causal principles.
Comparative research
: a research design that seeks to compare phenomena across different political systems or cultures.
Counter-intuitive
: a condition that occurs when a situation, event, or outcome differs from dominant theoretical expectations or common sense.
Descriptive case study
: a study that aims to chronicle or describe some aspect of reality; involves questions of who, what, where, and when.
Equivalent measures
: indicators that measure the same concept across different countries, political systems, or cultures.
Failed most-likely case
: a case that is expected to confirm the operating assumptions of a particular theory but in practice refutes them.
Galton’s problem
: a problem in comparative research that emerges when the units under observation are not independent of one another.
Most-different-systems design
: a comparative research design in which the researcher compares very different systems in an attempt to explain similarities between them.
Most-similar-systems design
: a comparative research design in which the researcher compares very similar systems in an attempt to explain differences between them.
Process tracing
: a research method that generates causal pathways between the independent and dependent variables of a case by connecting a series of observations.
Scope conditions
: the limits to which particular research can make valid claims.
Successful least-likely case.
: a case that is expected to refute the operating assumptions of a particular theory but in practice confirms them
Theory testing case study
: research that seeks to modify an existing theory or generate a new theory.