It is not true that the higher the confidence level, the
narrower the confidence interval. Suppose you have a large population of people
and you want to know what their average height is. It’s not practical to
measure each and every person, so you take a sample of a hundred people and you
measure their height. Let’s say that the average of the sample is 5.8 feet
tall, and you want to know how confident you should be about this. So you ask
yourself, if I were to repeat this exercise 100 times, what percent of the time
would the average of the sample be in a specific range
(the range is the confidence interval).
Now let’s think about how the confidence interval is related to
the confidence level. Let’s take an example: if the confidence interval were 1 foot to 8 feet, you would know that more
than 99 times out of a 100, the sample average would fall in that range, right?
However, if you start to narrow the confidence interval to “5 feet to 6 feet”,
the confidence level might drop significantly (especially if there are a lot of
children in the population). So the confidence level would tend to go down as
the confidence interval gets narrower.