2.7

2.7 The Question of Causation

  • Examples of Pairs of Variables
    • Accepted Causation:
    • Example: The amount of fertilizer used and the yield of crops.
    • Rejected Causation:
    • Example: Declared television ownership and life expectancy.
    • Reason for Decision Making:
    • Causation acceptance or rejection is typically made based on experimental evidence or strong associative relationships.

Explaining Association

  • Causation Definition:

    • A change in variable x causes a change in variable y.

  • Common Response Definition:

    • A lurking variable z causes a change in both x and y, leading to a spurious association.

  • Confounding Definition:

    • Variables x and z both affect y, and their separate effects cannot be distinguished from one another.

  • Examples:

    • Causation Example:
    • Amount of fertilizer leads directly to increased crop yield.
    • Common Response Example:
    • Nations with more television sets tend to have higher life expectancy, with wealth as a lurking variable.

Establishing Causation

  • Experimental Approach:

    • The most reliable way to establish cause and effect is through experiments that manipulate one explanatory variable while controlling other influences that may affect the response.

  • Limitations of Experimental Approach:

    • Conducting experiments may not be feasible in studying complex environments or human conditions where ethical and practical constraints exist.

Establishing Causation without an Experiment

  • Criteria to Establish Causation Without Experimental Data:
    1. Find a Strong Association:
    • Example: There is a strong correlation between smoking and lung cancer.
    1. Consistency of Association:
    • The association is consistent across different regions and countries, indicating a stable relationship.
    1. Correlation with Dose:
    • Higher amounts of exposure (e.g., cigarette smoking) lead to more severe effects (increased prevalence of lung cancer).
    1. Temporal Precedence:
    • The alleged cause must precede the effect in time (e.g., lung cancer appearing after years of smoking).
    1. Plausibility of Cause:
    • The causal relationship must be supported by plausible biological mechanisms, as seen in animal experiments demonstrating that tobacco smoke tars can cause cancer.