week 3 Flashcards
Milton friedman en experiment causaliteit etc. (24 cards)
Problems with testing hypothesis
biased in looking for evidence for you theory.
Verification
if we find anough evidence in favor of our theory it is true
falsification
if we find evidence against a theory, it is not true
statistical falsification
if we find statistically significant evidence against a hypothesis, it is not true.
What is positive economics
It is an objective science as the physical sciences; it is independent of any ethical position. Normative economics and the Art of economics depend on it.
What where the main points of Freidsman paper’s
Positive Econommics: objective, independt to provide a system of generalization used for predicitons
2. goals: theory that yield predictions about yet observed phenomena.
3. Test quality of hypothesis = comparison of predictions with data.
4. evaluating alternative theories with simplicity and fruitfulness. Simplicity means less factors, fruitfullnes more and wider predictions
5. in social testing by predictions will work, but the evidence is often difficult to interpret (causally)
6. testing assumption, if the assumptions are realistic does it make the theory better?
7. assumptions can be widely inaccurate but still have proving power.
8. you can only test by prediction and not the assumptions.
9. theories can be unrealistic but still have proving power.
What is theory?
Theory is a complex intermixture of two elements. Language; a set of tautologies (system of definitions/ concepts used to articulate/ formulate
a body of substantive hypotheses aimed at explaining/ predicting a certain phenomenon. Hypotheses should predict, not only describe the consequences of actions.
How do you test the quality of hypothesis
by comparing its prediction with its data.
can you reduce to a singe hypothesis?
No, of one hypothesis is consistent with the evidence there are infinitely many others aswell!. New evidence may rule out some of the alternatives but can never reduce them to a single one.
what are important criteria when evaluating theories?
Simplicity: just a few factors
Fruitfulness: the more precise, the wider the area which is yields predictions and the more additional lines for further research suggests.
what does testing assumptions mean?
Instead of looking at the predictions, you can look at the quality of the assumptions the prediction is made on.
What are problems with assumptions in hypothesis?
Troly important and significant hypothesis will be found to have assumptions that are widely inaccurate descriptive representations of reality.
What are criteria for a good hypothesis?
Explains much by little, abstracts the crucial elements from the mass of complex detailed cicumstances
Is it important that assumptions are real?
it is not important but it is important that they are a sufficiently good approximations for the given purpose, and this can only be found out by their predictions.
how do you test if assumptions are good?
Testing hypotheses by assumptions is inappropriate. Only testing by predictions is possible. It tests the quality of the theories and the usefulness of assumptions at the same time. As assumptions are false, they cannot be tested by empirical data.
What are downside of experiments
Costly, ethical problems, non-response bias, a mixup of experimental and control groups, specific to a particular setting (geography), limited duration, small groups, experimental design.
Is it hard to identify causal relationships and to isolate factors?
Yes
What are issues with causation?
confouding factors, another factor dirves the association between the factors of interest. these are not accounted factors. Reverse causality, measurement errors and ofcourse selection.
what is correlation?
is association between two variables
wat is causation?
this is a direct connection between two variables
What is econometrics?
Econometrics transform theoretical claims into testable empirical propositions that can be expressed in a precise mathematical form and statistically estimates their parameter.
What is the main toolkit in econometrics?
ITs regression analysis shows the relations between 2 or more variables.
What is important when we talk about regression.
Ceteris paribus: holding other things constant, is a latin phrase that is commonly translated into English as, all else being equal.
What is the alternative in economics for an experiment?
The natural experiments are often referred to as quasi experiments. Natural experiment is a situation that is not experimental but is close enough to experimental conditions. natural experiments feger to cases that generate variation in X, which is good variation. The data collected from such cases can then be used in estimation to claim the causal relationship.
