Related Web Sites

Confidence Interval
Using colorful graphs, this page demonstrates how confidence intervals work.

New Confidence Interval Applet
Visitors to this site can manipulate a Java applet in order to see various levels of confidence intervals.

Statistics Glossary: Confidence Intervals
Presented here is a discussion of confidence intervals and how they are calculated.

Can We Be Confident?

Pollsters make reference to the results of a poll in terms of confidence. They say they are 95%, 90%, 75% confident of their results. No one, at any time or under any circumstances, can be 100% sure of a poll's accuracy.

It's actually possible to determine the level of confidence and margin of error prior to carrying out a poll. A higher level of confidence always requires a larger sample size.

What's a Confidence Interval?
The range of numbers that fall within the two ends of the margin of error is referred to as the confidence interval. If Fletcher is in the lead by 20% with a margin of error of +/- 5%, the confidence interval is from 15% to 25%. The interval reflects the calculated average and those numbers that might fall outside that average.

When a pollster is 95% confident, that means this: If the poll was repeated 100 times, then in 95 of the polls, the percent of people who said that they supported a particular candidate would be within the margin of error of the percent who said that they supported that candidate the first time. Of course, that means that in 5 polls, the pollster could expect results that were nothing like the original results.

What Can Go Wrong