Let X₁, X₂, ⋯, Xₙ be a random sample from a population with p.d.f.
f(x, θ) = θx^{θ−1}; 0 < x < 1, θ > 0
Show that ∏ᵢ₌₁ⁿ Xᵢ is sufficient statistic of θ.
5 marks · directive: Show
Answer
The core of a good answer
The likelihood is L(θ) = θⁿ (∏xᵢ)^{θ−1} for 0 < xᵢ < 1, which factors as g(∏xᵢ, θ) × h(x) with h(x) = 1, so by the Neyman factorisation theorem ∏Xᵢ is sufficient for θ.
An outline written by upsc.help, not an official UPSC answer; every fact checked by a separate review. How we work
What the question asks
A direct application of the Neyman-Fisher factorisation criterion. Write the joint density, show that θ enters only through ∏xᵢ, and name the two factors. Note that the support (0, 1) does not depend on θ.
Answer outline
Introduction
Joint pdf of the sample: L(x; θ) = ∏ᵢ θxᵢ^{θ−1}, 0 < xᵢ < 1.
Step 1: Joint density
- L(x; θ) = θⁿ (∏ᵢ₌₁ⁿ xᵢ)^{θ−1}, for 0 < xᵢ < 1.
Step 2: Factorisation
- Put t = ∏xᵢ. Then L = g(t, θ) h(x) with g(t, θ) = θⁿ t^{θ−1} and h(x) = 1 on (0, 1)ⁿ.
- h does not involve θ, and the range of x does not involve θ.
Step 3: Conclusion
- By the factorisation theorem T = ∏Xᵢ is sufficient for θ.
- Equivalently, any one-to-one function such as ∑ log Xᵢ is sufficient.
Step 4: Exponential-family view
- f(x, θ) = exp[(θ − 1) log x + log θ], a one-parameter exponential family with T(x) = log x, so ∑ log Xᵢ is sufficient (and complete).
Conclusion
∏Xᵢ (equivalently ∑ log Xᵢ) carries all the information about θ in the sample. Check: the MLE θ̂ = −n/∑ log Xᵢ is a function of it, as it must be.
Examples and data to cite
- This density is Beta(θ, 1); −log X has the exponential distribution with mean 1/θ, so −∑ log Xᵢ is Gamma(n, θ).
Facts
Facts to remember
- Neyman-Fisher factorisation: T is sufficient iff L(x; θ) = g(T(x), θ) h(x).
- A one-to-one function of a sufficient statistic is sufficient.
- Beta(θ, 1) density: θx^{θ−1}, 0 < x < 1.
- MLE: θ̂ = −n/∑ log xᵢ.