The Scope of Inference
Statistical inference is concerned with making statements about the characteristics of the true underlying probability measure. It uses observed data to narrow down which specific distribution (or family of distributions) produced the variation we see. Whether we are estimating a parameter $s$ or predicting a future value $X$, we are trying to resolve the ambiguity of the source.
The Descriptive-Inference Link
While often viewed as simple summaries, methods like calculating the sample mean $\bar{x}$ are actually the first steps in inferring the location of the true population density.
Example: Stanford Heart Transplant Study (5.1.1)
In the foundational study by Turnbull, Brown, and Hu (1974), researchers investigated whether a heart transplant program at Stanford was "producing the intended outcome" (increased survivorship). Simply looking at raw survival times ($X$) of one or two patients was insufficient.
- Control Group: Patients receiving standard care.
- Treatment Group: Patients receiving transplants.
The researchers needed inference to decide if the survival differences were statistically significant or merely the result of the stochastic variation inherent in individual patient health.
The Dual Nature of Uncertainty
We must acknowledge a critical pitfall in analysis—uncertainty is not a monolithic "noise." It arises from two distinct sources:
- Inherent Variation: Modeled via probability (e.g., the randomness of a coin toss or biological diversity).
- Structural Ignorance: The reality that we cannot collect enough observations to know the correct probability models with absolute precision.