EQD Ch1: Experiments and Generalized Causal Inference

1
EQD Ch1 Experiments and Generalized Causal
Inference

• Krystal Meredith

2
Experiments and Causation

• Defining Cause, Effect and Causal Relationships
• Cause a variable that produces an effect
• Reciprocal relationship two variables that
  cause each other (ex low grades and not
  studying)

3
Experiments and Causation

• Cause
• Inus condition From philosopher J.L. Mackie
  (1984), the idea that a cause is an insufficient
  but non-redundant part of an unnecessary but
  sufficient condition for bringing about an
  effect.
• Example match and the forest fire, cancer drugs
  and tumor results
• Most causes are more accurately called inus
  conditions. We rarely know all the factors for a
  specific effect.

4
Experiments and Causation

• Effect
• Effect the difference between what did happen
  and what would have happened.
• Counterfactual the state of affairs that would
  have happened in the absence of the cause.
• Example PKU and the infant diet

5
Experiments and Causation

• Causal Relationship
• A causal relationship exists if
• The cause preceded the effect
• The cause was related to the effect
• We can find no plausible alternative explanation
  for the effect other than the cause

6
Experiments and Causation

• Causation, Correlation and Confounds
• Causation cause of an effect usually causation
  and effect are correlated
• Correlation a measure of the strength of a
  relationship between two variables.
• Correlation does not prove causation.
• Example Education and income
• Confounds a relationship that may not be causal
  but due to another variable

7
Experiments and Causation

• Manipulable and Nonmanipulable Causes
• Manipulable causes variables that can be
  changed/manipulated
• Example dose of medicine, amount of sleep or
  food, number of children in class
• Nonmanipulable cannot be causes in experiments
• Example age, biological sex, height

8
Experiments and Causation

• Causal Description and Causal Explanation
• Causal Description Identifying that a causal
  relationship exists between A and B.
• Causal Explanation Explaining how A causes B
• Example light switch and illumination
• Causal explanation helps generalize causal
  relationships to other situations.

9
Experiments and Causation

• Molar and Molecular
• Molar Causation An interest in the overall
  causal relationship between a treatment package
  and its effects, in which both may consist of
  multiple parts.
• Molecular Causation An interest in knowing
  which parts of a treatment package are more or
  less responsible for which parts of the effects
  through which mediational processes.

10
Modern Descriptions of Experiments

• Randomized Experiment
• Randomized Experiment An experiment in which
  units are randomly assigned to conditions.
• Commonly called true experiment
• Independent variable deliberately manipulated
  variable control cause
• Dependent variable varies in response to
  independent variable outcome effect

11
Experiments and Causation

• Quasi-Experiment
• Quasi-Experiment An experiment in which units
  are not randomly assigned to conditions
• Quasi-experimentation is falsificationist.
• Falsification to show that data are
  inconsistent with a theory or hypothesis

12
Experiments and Causation

• Natural Experiment
• Natural Experiment investigates the effects of
  a naturally occurring event, sometimes limited to
  events that are not manipulable and sometimes
  used more generally.
• Example earthquakes and property value

13
Experiments and Causation

• Nonexperimental Design
• Nonexperimental study any study that is not an
  experiment. A cause and effect are identified
  and measured but other structural features of
  experiments are missing.
• Missing elements randomization, control,
  pretests
• Also know as correlational study and
  observational study

14
Experiments and the Generalization of Causal
Connections

• The strength of experimentation is its ability to
  illuminate causal inference. The weakness of
  experimentation is doubt about th extent to which
  that causal relationship generalizes.

15
Experiments and the Generalization of Causal
Connections

• Local Experimentation and Generalizations
• Experiments try to generalize to more people and
  settings than represented in a single experiment.
• Causal generalization how well a causal
  relationship extends across the conditions that
  were studied.

16
Experiments and the Generalization of Causal
Connections

• Construct Validity
• Construct Validity the degree to which
  inferences are warranted from the observed
  persons, settings, and cause-and-effect
  operations sampled within a study to the
  constructs that the samples represent.
• Example patient education and surgery

17
Experiments and the Generalization of Causal
Connections

• External Validity
• External Validity to infer whether a causal
  relationship holds over variations in persons,
  settings, treatments, and outcomes.
• Example Head Start Memphis, TN and Dallas, TX

18
Experiments and the Generalization of Causal
Connections

• Making Causal Generalizations
• Sampling and Causal Generalization
• Formal probability sampling
• Delineate populations and sample from within each
  sampling
• Random selection used with population samples

19
Experiments and the Generalization of Causal
Connections

• Making Causal Generalizations
• Grounded Theory
• Causal Generalizations that are presented as a
  theory grounded in the actual practice of
  science.

20
Experiments and the Generalization of Causal
Connections

• Making Causal Generalizations
• Grounded Theory
• Five principles
• Surface Similarity
• Ruling Out Irrelevancies
• Making Discriminations
• Interpolation and Extrapolation
• Causal Explanation

21
Experiments and Metascience

• Modern Social and Psychological Critiques
• Scientific knowledge is partly determined by
  social and psychological forces and partly by
  issue of economic and political power within
  science and the larger society

22
Experiments and the Generalization of Causal
Connections

• Implications for Experiments
• Experimental results partly relative to those
  assumptions and contexts and may change with new
  assumptions and contexts.