How to interpret confidence intervals?

811 views

How do you interpret confidence intervals? For example does a 90/10 CI mean that we are 90% confident that the estimate is within 10% of the true population value? Does the 90 represent accuracy and the 10 precision?

In: Mathematics

2 Answers

Anonymous 0 Comments

When you form a 90% confidence interval, you are saying “There is some true value of the parameter I am trying to estimate. I do not know the value of this parameter, but I know that if I used my procedure to generate 100 confidence interval of the kind you see here, 90% of them would contain the true value.”

In general, you cannot guarantee that this interval will deviate no more than 10% from the true value. If you don’t have much data or your data are very noisy, a 90% confidence interval might be much larger than that. The “90/10” idea is therefore about experimental design. It’s always good to have more data, but gathering data is often expensive. Researchers using a 90/10 interval stop gathering data when they know that their 90% confidence intervals will have endpoints within 10% of the midpoint

Anonymous 0 Comments

90/10 would indicate that we are 90% certain that the actual true value is between 10 below and 10 above the reported/measured value.