Pivotal function for a random sample from a normal distribution Let $y_1,...,y_n$ be a random sample from a normal distribution, $N(\theta,\theta)$, where $\theta>0$. Show that $$U=\frac{\overline{Y}_n-{\theta}}{\sqrt{\frac{\theta}{n}}}$$ is a pivotal quantity where $\overline{Y}_n$ is the sample mean. I am stuck on property 2 of the pivotal quantity. So, how do I show that the probability distribution of $U$ does not depend on $\theta$?

Since $U$ is a linear combination of Gaussian variables, it is a Gaussian variable. We only need to show that the expectation of $U$ and the variance of $U$ do not depend on $\theta$.

* The expectation of $y_i$ is $\theta$ for all $i$. Due to the linearity of the expectation, the expectation of $U$ is $E(U) = \frac{1}{\sqrt{\theta/n}} \left(\frac{1}{n}\,n\theta - \theta\right) = 0$ .
* The variance of $y_i$ is $\theta$ for all $i$. Due to the scaling and summation properties of the variance of uncorrelated variables, the variance of $U$ is $V(U) = \frac{n}{\theta} \left(\frac{1}{n^2}\,n\theta - 0\right) = 1$ .



Thus, $U\sim\mathcal{N}(0,1)$ is a pivotal quantity.