Tensions in Practice: Population coverage in tension with a selected group’s relevance¶
A toy eligibility register
Two qualifications, X and Y, are independent fair yes/no attributes in a toy population. Entry requires X or Y. The full population has four equally likely combinations. Among entrants, the neither-qualified combination disappears: someone without X must have Y. Studying entrants is useful for entrant questions, but their association cannot be read back as the population relationship.
Describe the population
Retain people outside the eligibility gate.
Describe entrants
Condition on the group the service actually handles.
Why these aims pull against each other
The gate makes the selected universe more relevant to entrant questions while changing the relation between the two attributes.
Choose an arrangement to see what changes and what remains difficult.
The same X/Y combinations remain in place. Entry removes only the neither-qualified cell and rescales the remaining probabilities.
What this choice protects
What it costs
When it fits
Compare the arrangements
Describe everyone
Use the four equally likely population combinations.
| Y = 0 | Y = 1 | |
|---|---|---|
| X = 0 | 1/4Probability | 1/4Probability |
| X = 1 | 1/4Probability | 1/4Probability |
- What it protects
- The independent population relationship is visible.
- What it costs
- It includes people the service will not handle.
- When it fits
- Fits questions about the whole stipulated population.
Illustration note: X and Y each have probability 1/2 and each joint combination probability 1/4.
Describe entrants
Remove X0,Y0 and renormalize the remaining three combinations.
| Y = 0 | Y = 1 | |
|---|---|---|
| X = 0 | ExcludedNo entry | 1/3Probability |
| X = 1 | 1/3Probability | 1/3Probability |
- What it protects
- The distribution now answers questions specifically about entrants.
- What it costs
- The selected association does not describe the population or establish one qualification causes the other.
- When it fits
- Fits when entry is explicitly part of the target question.
Illustration note: Among entrants without X, Y is certain; among entrants with X, Y has probability 1/2.
What this illustration does—and does not—establish
The source supplies the tension. The invented setting, alternatives and any numbers illustrate a limited comparison; each arrangement retains its stated costs and conditions.
- The OR gate and the probability law are declared toy assumptions.
- Conditioning changes the description of a population; it does not change anyone’s attributes.
- No general rule says always condition or never condition. The question and selection mechanism matter.
Source entries
Conditional Probability
This source passage supplies the contextual tension. The concrete arrangements and schematic examples are editorial illustrations, not measured findings.
Conditioning Set as Given versus as Choice (scopal)
The framework treats the conditioning event as supplied, but what one conditions on is the most consequential modelling decision, and it is selectable. The failure mode is conditioning on a post-selection or collider variable — slicing the universe in a way that manufactures a spurious dependence (Berkson's paradox, selection bias) — so the re-normalization itself injects the correlation later "discovered."
The source operation
Conditional probability is the probability of one event $A$ relative to the assumption that another event $B$ is known to have occurred — formally $P(A \mid B) = P(A \cap B) / P(B)$, defined when $P(B) > 0$.