Suppose that $X_1, X_2, ..., X_n$ and $Y_1, Y_2, ..., Y_n$ are independent random samples from populations with means $\mu_1$ and $\mu_2$ and variances $\sigma_1^2$ and $\sigma_2^2$, respectively. Show that the random variable

$U_n = \frac{(\bar{X} - \bar{Y}) - (\mu_1 - \mu_2)}{\sqrt{\frac{(\sigma_1^2+\sigma_2^2)}{n}} }$

satisfies the following conditions of the theorem below and thus that the distribution function of $U_n$ converges to a standard normal distribution function as $n -> \infty$.

Theorem:enter image description here

I'm not exactly sure how to exactly approach this but I was told to consider $W_i = X_i - Y_i$, for $i = 1, 2, ..., n$.]

  • $\begingroup$ A first step would be to determine the mean and variance of $W_i$. $\endgroup$ – user133281 Mar 5 '14 at 20:30

Let $W_i=X_i-Y_i$, then $\mu_W=E[W_i]=E[X_i-Y_i]=E[X_i]-E[Y_i]=\mu_1-\mu_2$,

Also, you can show that the variance of $W_i$ is $\sigma^2_W=\sigma^2_1+\sigma^2_2$.

Now we know that $W_i$ is drawn from some distribution (difference of two normals) with limited variance ($\sigma^2_1+\sigma^2_2$), therefore you can apply the weak law of large numbers:

Let $U_n=\frac{\frac{1}{n}\sum_{i=1}^nW_i-\mu_W}{\sqrt{\frac{\sigma^2_W}{n}}}$

By substituting $\sigma_W^2$ and $\mu_W$ the proof is complete:

$\frac{1}{n}\sum_iW_i = \frac{1}{n}\sum_i (X_i - Y_i)=\frac{1}{n}\sum_i X_i-\frac{1}{n}\sum_i Y_i=\bar{X} - \bar{Y}$

$U_n=\frac{(\bar{X} - \bar{Y})-(\mu_1-\mu_2)}{\sqrt{\frac{\sigma^2_1+\sigma^2_2}{n}}}$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.