Required sample size to reduce the interval size by half

1.1k Views Asked by At

I have a small doubt regarding a statistics problem and would like a confirmation.

If we want to reduce the length by half of the confidence interval (at $95\%$), we should: Multiply the size of the sample by $2$ ? $4$ ? or by $8$ ?

As for me, I think that the sample size should be multiplied by $4$.

Why do I think so? We are looking for $\frac{1}{4}E$, $E$ being one side of the interval:

$$\frac{1}{4}E= 1.96 * \frac{\sigma}{\sqrt{n}} $$

So, $4n=(1.96)^2 * \frac{\sigma^2}{E^2} $

So it should be multiplied by $4$, as said before.

Am I right ?

Thanks in advance.

Regards

1

There are 1 best solutions below

0
On

Because this is a homework tag, I won't give you the explicit confirmation/explanation, but a simple way to check yourself is to simulate this with known $\sigma$ and known confidence intervals.

Simulate using a variety of sizes for $n$ and see if you can deduce the relationship explicitly. The main things to think about are

1- How the value of $\sqrt{n}$ changes as $n$ changes 2- How the symmetry of the confidence interval effects the answer, i.e, accounting for the $+/-$