# Homework Help: Finding the confidence interval

1. Mar 27, 2010

### Pietair

1. The problem statement, all variables and given/known data

What formula do I need to find the confidence interval, when I have got:

- Number of samples
- Level of Confidence
- The assumed (1st guess) accuracy

2. Relevant equations

I found the following equation online: µ = z * [p * (1 - p) / n] ^ (-1/2)

3. The attempt at a solution

When I fill in this formula, I get µ = 125.5, while I think the confidence interval should be around 3 percent.

2. Mar 27, 2010

### rock.freak667

If you want a 95% CI, then you want P(-a<Z<a)=0.95 where

$$Z=\frac{\bar{x}-\mu}{\frac{\sigma}{\sqrt{n}}}$$.

So $\bar{x} \pm a \frac{\sigma}{\sqrt{n}}$ will be a 95% CI for μ

3. Mar 27, 2010

### Staff: Mentor

This may or may not be relevant to your problem. This formula looks vaguely related to a binomial distribution. You haven't said what the distribution is, so it's hard to say if this is something you need to use.
Again, you have provided enough information for me to tell if this is a reasonable value for µ. What you said about the confidence interval makes no sense at all. A confidence interval is an interval, with a left endpoint and a right endpoint. It is not given as a percentage.

4. Apr 3, 2010

### Pietair

All the information I have got considering this practice situation:

Information written down on a form will be put in a database. The information in the database can be correct (match the information written on the form) or can be incorrect (do not match the information written on the form). A mismatch occurs when the database administrator enters the wrong information (for example: putting "b" in the database when "a" is written on the form).

Now I would like to execute a sample to judge whether the data found in the database is reliable (ie consistent with the source information) or not. The database contains a total of 4000 entries. I would like to execute a sample because it is quite time consuming to check if all 4000 entries are correct or not. With this sample I would like to state something about the reliability of the entire database (4000 entries).

So, suppose I have 100 entries checked, and 2 of them do not match. Then I find that 98% of the database entries of the corresponding sample is consistent with the source information. But what can I say about the confidence level and interval of this 98% considering the entire database (4000 entries).