Practice questions Flashcards
What is methodology? How does “methodology” relate to “methods”?
Methodology is the systematic assessment and justification of method choice. A method is an empirical practice within science, such as observational studies, field experiments or model studies.
Why can the conventional view of methodology be problematic for a scientist?
It reduces the critical potential within the group and can cause difficulty collaborating with other disciplines.
Name some typical goals of science and explain what they mean. Come up with three scientific examples of research aimed at each of the typical goals of science.
Typical goals for science are Predicting, Explaining and Designing.
What is the classical definition of knowledge?
“Knowledge is true, justified belief.” - Plato
Give some examples of different types of inductive inferences. Come up with an example of each.
Inductive inference:
- Direct inference
- Projection
- Generalisation
Deductive inference
- Modus ponens
- Modus tollens
Why is the relation between falsification and confirmation asymmetric? Is it asymmetric in scientific practice?
You need an infinite amount of confirmation but only one case of falsification. Popper argues yes, but others argue no.
What was Karl Popper’s view on the demarcation of science, and what is his view of how science should progress?
Popper insists on falsification in science. Asymmetry between falsification and confirmation: we need indefinite confirmation to confirm a hypothesis but we only need one case of falsification to conclude that the hypothesis is false. For confirmation we need to use inductive inference, but for falsification we can use Modus tollens, and reject the hypothesis with deductive inference. Popper argues that this means we don’t fall into Hume’s problem of inductive inference, because we can just use deductive inference (to reject hypotheses, but you can never confirm them). BUT in order to deduce consequences, Duhem-Quine thesis argues that you need to do inductive inference because you use auxiliary hypotheses.
What is an auxiliary hypothesis? Try to come up with an auxiliary hypothesis you might need to include to test “If someone is home, then there is a light in the window”.
Ex. auxiliary hypothesis: That our measurement equipment works properly
Explain the Duhem/Quine problem, make sure you use the term auxiliary hypothesis.
The Duhem-Quine thesis argues that there is no asymmetry between falsification and confirmation because of the auxiliary hypotheses you need to make in order to test your primary hypothesis.
When a hypothesis should be falsified because of our test, we might instead modify one of the assumptions instead. However, it is important that this modification is not ad hoc. Why?
If the modification is ad hoc, it reduces the falsifiability of the hypothesis.
What is empiricism? Do you think there are disciplines whose underlying view on knowledge is different from empiricism? (Hint: you have, most probably, taken courses in a discipline where this question at least can be discussed.)
Empiricism: Sensory experience is considered to be the ultimate basis for knowledge.
The feeling of heartburn is compatible with the hypothesis “I have stomach cancer”. However, this hypothesis is under-determined. Explain what this means and come up with an alternative hypothesis.
Under-determination: An inference is underdetermined if multiple conclusions would be supported by the premises.
There are multiple reasons you could have heartburn, several which are more likely than stomach cancer.
A proposed solution for this problem is to introduce some criterion of relevance for observable consequences. Only the truth of relevant implications in the stomach cancer case, such as various effects of the cancer, would confirm the hypothesis. Specifying
such relevant conditions is tricky, however, and requires a lot of background knowledge.
Another solution for this problem could be considering alternative hypotheses. Preferably, have two hypotheses that are distinguishable, one of them would not be true if the other one is.
Distinguish, in your own words, direct, aided and indirect observation from each other. Can indirect observations be made without direct observations?
Direct observation: Sense experience of the event itself.
Aided observation: Sense experience of the event itself with the help of a tool.
Indirect observation: Sense experience of the effect of an event, but not the event
itself.
Explain the terms manipulation for control, manipulation for intervention and observation. Use these to distinguish an experiment from an observational study.
Observation: Registering an empirical phenomenon.
Manipulation: Changing a variable in a study.
Intervention: Changing an independent variable.
Control: Reducing the influence of disturbing factors/variables from influencing the outcome.
Intervention is what distinguishes an experiment from an observational study. Once you intervene/manipulate the variable of interest in a study, it is no longer an observational study.
Astrophysics is a non-experimental science, as well as macro economics.
Advantage of experiments over observational study: easier to study casual relationships
Advantage of observational study: external validity
Most of the time observational studies are performed out of a need, i.e. there is an inability to perform an intervention (because such intervention would be illegal, unethical, or such).
In some cases of high homogeneity between control and treatment group (without intervention) you can apply Mill’s method of difference, and draw a causal conclusion from an observational study (see Dr. Snow in Soho) but this is very rare.
The observation “Adam has not called me today” is not a severe test of “Adam does not want to be my friend anymore”. Give an example you think would be a severe test and explain why.
Severe test: A hypothesis test is a severe test if the probability to observe a consequence would be low if the hypothesis were false.
A severe test of Adam not wanting to be my friend anymore could be Adam telling me “I don’t want to be your friend anymore”. Or Adam murdering my dog.
Severe testing concerns confirmation and not falsification.
Severe testing is not a single method, rather it is a way of thinking.
Are severe testing and frequentism incompatible, since severe testing implies assigning probabilities and frequentism claims those probabilities have no meaning?
No, they are not incompatible. Because severe testing concerns the probability of the observation given the hypothesis and not the probability of the hypothesis itself.
