Epistemology: Is a 'Proven' Correlation Really Knowledge?
A philosophy seminar debates a local health survey: towns with more ice-cream sales report more drowning incidents. A student concludes: 'Ice cream causes drowning - the data prove it.' A second student says: 'Then knowledge is just whatever the numbers show.' The teacher introduces three distinctions: correlation vs. causation, necessary vs. sufficient condition, and the difference between doxa (opinion) and episteme (justified knowledge). The class must also connect this to a classical epistemological problem: how do we justify a belief without circular reasoning? The final essay prompt asks whether statistical regularity can ever count as scientific knowledge.
1. Using the ice-cream example, explain the difference between correlation and causation, and identify the hidden variable. Then state whether the survey gives a necessary condition, a sufficient condition, or neither.
Read the worked answer
Correlation vs. causation:
- Correlation = two variables move together (ice-cream sales ↑, drownings ↑). It is an observed association.
- Causation = one variable produces the other. To claim causation you need a mechanism, temporal order, and control of confounders - none of which the survey supplies.
Hidden variable (confounder): summer heat / season. Hot weather raises both ice-cream consumption and swimming (hence drowning). Heat is the common cause; ice cream is a symptom, not a cause.
Necessary vs. sufficient:
- Eating ice cream is not necessary for drowning (people drown in winter).
- It is not sufficient either (most ice-cream eaters never drown).
- So the relationship is neither necessary nor sufficient - it is a spurious correlation.
Epistemic lesson: a statistical regularity is a doxa-level observation until a causal mechanism is justified. The student's leap is the fallacy post hoc ergo propter hoc (after this, therefore because of this).
2. A second student claims 'knowledge is just whatever the numbers show.' Evaluate this claim using the doxa/episteme distinction and the problem of justification.
Read the worked answer
The claim is false as stated, for two reasons.
1. Doxa vs. episteme.
- Doxa = opinion: a belief held, possibly true, but not grounded.
- Episteme = justified, systematic knowledge: the belief is true and the holder can give reasons that survive scrutiny.
- 'The numbers show X' is doxa if the numbers are uninterpreted. The same data (ice-cream/drowning) 'show' a causation to one reader and a confounder to another - so data alone do not decide what is known; interpretation and method do.
2. The justification problem (the regress).
- If I justify belief A by belief B, and B by C, I face an infinite regress.
- If I justify A by A, I am circular.
- If I stop arbitrarily, I am dogmatic.
- The scientific answer is not to escape the trilemma but to make justification public and revisable: methods, controls, and replication. Knowledge is then justified true belief held to communal standards, not 'whatever the numbers show.'
Conclusion for the essay: statistical regularity can be evidence for knowledge but is not itself knowledge until a mechanism and method justify the inference. Numbers are raw material; episteme is the disciplined interpretation of them.