Hypothesis Specification Type
Jump to navigation
Jump to search
A Hypothesis Specification Type is a statistical classification type that categorizes statistical hypotheses based on how completely they specify the parameter space of a probability distribution.
- AKA: Hypothesis Type, Statistical Hypothesis Type, Parameter Specification Type.
- Context:
- It can typically classify a Simple Hypothesis that specifies exact parameter values.
- It can typically classify a Composite Hypothesis that specifies parameter ranges or sets.
- It can often determine the test statistic distribution under different hypothesis types.
- It can often influence the statistical power calculation for hypothesis tests.
- It can range from being a Point Hypothesis Specification Type to being a Interval Hypothesis Specification Type, depending on its parameter constraint.
- It can range from being a One-Sided Hypothesis Specification Type to being a Two-Sided Hypothesis Specification Type, depending on its directional constraint.
- It can range from being a Parametric Hypothesis Specification Type to being a Non-Parametric Hypothesis Specification Type, depending on its distributional assumption.
- It can range from being a Univariate Hypothesis Specification Type to being a Multivariate Hypothesis Specification Type, depending on its parameter dimensionality.
- ...
- Example(s):
- Mean Hypothesis Specifications, such as:
- A Point Null Hypothesis: μ = 0.
- A One-Sided Alternative Hypothesis: μ > 0.
- A Two-Sided Alternative Hypothesis: μ ≠ 0.
- Variance Hypothesis Specifications, such as:
- A Simple Variance Hypothesis: σ² = 1.
- A Composite Variance Hypothesis: σ² > 1.
- Distribution Hypothesis Specifications, such as:
- A Normality Hypothesis: data follows N(μ, σ²).
- A Non-Parametric Hypothesis: F(x) = F₀(x).
- ...
- Mean Hypothesis Specifications, such as:
- Counter-Example(s):
- A Research Hypothesis, which is conceptual rather than statistical.
- A Scientific Theory, which is broader than a statistical hypothesis.
- A Prediction, which doesn't specify parameter values.
- See: Null Hypothesis, Alternative Hypothesis, Simple Hypothesis, Composite Hypothesis, Statistical Hypothesis Testing Task, Parameter Space, Test Statistic, Likelihood Ratio Test, Uniformly Most Powerful Test.