Given the following set of values:


How do I calculate a 99% confidence interval for the sample mean? I am assuming that the variance is 10

Well, the idea I have is to assume that the distribution is normal, but after that i'm not completely sure what to do next. In particular, I am unable to find the z-score that corresponds to the z-score in the formula for the CI, i'm not sure what to input in R to find the CI. The formula for the confidence interval is:

$x-z_{\frac{a}{2}} \frac{\sigma^2}{sqrt(n)}$ and $x+z_{\frac{a}{2}} \frac{\sigma^2}{sqrt(n)}$ where x denotes the mean. Here the significance level (a) is 0.01.

  • $\begingroup$ What are your thoughts? Are you familiar with any formulas that concern confidence intervals? What computations can you do? $\endgroup$ – Brian Mar 21 at 17:56
  • $\begingroup$ Well, I know the formula for the confidence interval, but i'm not sure how to find the z-score, I know that I can use R, but i'm not sure what to write in R. $\endgroup$ – topologicalmagician Mar 21 at 17:57
  • $\begingroup$ Do you know what the "99%" in the confidence interval means? Can you relate this to the Normal distribution in some way? Do you know how z-scores relate to the Normal distribution? $\endgroup$ – Brian Mar 21 at 18:00
  • $\begingroup$ @Brian Yes, a CI is the 100(1-a)% interval in which the true parameter lies in, a is the significance level. The formula for the CI is : $x+z_{\frac{a}{2}} \frac{\sigma^2}{sqrt(n)}$ and : $x-z_{\frac{a}{2}} \frac{\sigma^2}{sqrt(n)}$ where x denotes the mean $\endgroup$ – topologicalmagician Mar 21 at 18:04

It is likely that you are asked to find a confidence interval for the (unknown) population mean, not the sample mean. The sample mean is not a parameter of interest, it can be calculated from the sample.

A conventional setup for the problem is that you have a sample $\mathbf X=(X_1,X_2,\ldots,X_n)$ of size $n=11$ from a $N(\mu,\sigma_0^2)$ population (by assumption) with $\sigma_0^2=10$. You have to find a confidence interval for the mean $\mu$.

A suitable pivotal quantity here is $$Q(\mathbf X,\mu)=\frac{\sqrt{n}(\overline X-\mu)}{\sigma_0}\sim N(0,1)$$

, where $\overline X=\frac{1}{n}\sum\limits_{i=1}^n X_i$ is the sample mean.

So if $z_{\alpha/2}$ be such that $P(Z>z_{\alpha/2})=\alpha/2$ where $Z\sim N(0,1)$, you have $$P_{\mu}(-z_{\alpha/2}\le Q\le z_{\alpha/2})=1-\alpha\quad\,\forall\,\mu$$

You have to use the above to arrive at the form $$P_{\mu}(c_1\,\le \mu\le\,\, c_2)=1-\alpha\quad\,\forall\,\mu$$

A $100(1-\alpha)\%$ confidence interval for $\mu$ is then $[c_1,c_2]$.


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.