Please enable JavaScript.
Coggle requires JavaScript to display documents.
Explanation: issues and aspects (Causal explanations (Four criteria for …
Explanation:
issues and aspects
many marketing theorists think that explanation and prediction are not systematically interrelated
Does explanation for a phenomenon imply that we could have predicted it? Does being able to predict phenomenon imply that we can explain it?
Hempel: the thesis of
structural identity/symmetry
Every adequate explanation
is potentially a prediction
The E-P Argument
(explanation-implies-prediction argument)
The conclusions of the E-P argument do seem to be logically implied by the premises and the E-P argument is true: every adequate explanation is potentially a prediction, if we could not have predicted the occurrence of a phenomenon, we cannot satisfactory explain it
Dubin:
Power paradox
Many social science theories do not provide precision in prediction
Many of these same theories
contribute powerfully to understanding
a. This is false! Marketing and other social
science theories that do not predict do not
make powerful contributions to
scientific understanding
The preceding seems paradoxical
Many marketing theorists believe that their theories and models contribute to the understanding but admit the lack of predictive power of their theories
Every adequate prediction
is potentially an explanation
P-E argument
(prediction-implies-explanation argument)
Many people, who accept the E-P argument, reject the P-E argument
This is false, because accidental generalizations have predictive capacity, and because explanations must contain laws, and because accidental generalizations are not laws -> all adequate (accurate) predictions are not potential explanations
Retrodiction:
implies making
inferences about the past on the
basis of present observations
Explanation:
the phenomena that do the explaining occur in time before the phenomenon to be explained
Prediction
: the phenomena that do the predicting antecede the phenomenon to be predicted
Retrodiction
: the phenomena that accomplish the retrodicting occur after the phenomenon to be retrodicted
Although retrodiction would be a desirable characteristic of any model,
neither explanatory nor predictive adequacy implies the ability to retrodict
Causal explanations
Are all explanations causal? If not, are they adequate?
Fundamental problem of causation: What evidence can empirically or logically separate the assertion “X causes Y” from the assertion “X and Y occur regularly in the same pattern”?
How can we know that X has the power to produce Y or any of the other conceptualizations of cause? What are the kinds of evidence or criteria? What makes a relationship causal?
Four criteria for
classifying an explanation
as causal
(necessary or minimal
conditions for causality)
Associative
variation
If A is a cause of B, then changes in the level or presence of factor A must be systematically associated with changes in the level or presence of factor B correlation is evidence in support of causation
Temporal
sequentiality
If changes in A are to be used to explain causally B, then the occurrence of the changes in A must precede in time the occurrence of changes in B
Nonspurious
association
If A causes B, then there must be no factor Z that if introduced in to explanation, would make the systematic association between A and B vanish
Theoretical support
Well-conformed theories can be used to support the assertion that A causes B
In marketing
Granger conditions
-> testing for causal relationships: a variable X is causally related to Y if we are better able to predict Y by using all of the available variables, including X, than by using the same set of variables without X
Structural equation
approach to causal modeling (SEM): uses, among others, the maximum likelihood method for estimating parameters