Confidence interval

Sometimes we cannot know a true answer. We can only make a good guess. Instead of one number, we use a range. This range is like a net.

It helps us catch the right answer. If we use the same way many times, the net works well. Most of our nets will catch the true answer. A 95 percent level means it works most of the time.

We can use this to guess many things. It helps us feel sure about our guess.

Imagine you want to know the average height of all kids in a school. You cannot measure every single person. Instead, you measure a small group. This group is called a sample. The average of this sample is just a guess. It might not be the true average for everyone.

To be more careful, scientists use a range of numbers. This range is called a confidence interval. It is like a net used to catch the true answer.

A common level is 95 percent. This does not mean there is a 95 percent chance the answer is in your specific net. Instead, it describes how well the method works. If you took 100 different samples, about 95 of your nets would catch the true answer.

Some nets might miss the true value. This happens because each sample is random. The size of the net can also change. If you measure more people, your net can become smaller and more precise.

Imagine you want to know the average height of every student in a huge school. You cannot measure every single person, so you pick a small group instead. This group is called a sample. The average height of your sample is a good guess, but it is not perfect.

To be more careful, scientists use a range of numbers called a confidence interval. Instead of saying the average is exactly five feet, they might say it is between four and six feet. This range acts like a net meant to catch the true answer.

How do scientists build these nets? They use math to decide how wide the range should be. One common way is using something called the central limit theorem. This method works well if the sample size is large enough. Another way is called bootstrapping, which uses the data you already have to find the range.

Sometimes, scientists use a special rule to make the interval wider to be safe. This is called a conservative interval. It errs on the safe side so it is more likely to catch the true value.

A common level used is 95 percent. Many people think this means there is a 95 percent chance the true answer is inside their specific net. However, that is not quite right. The true average is a fixed number that does not change. It is either in your net or it is not.

The 95 percent level actually describes how reliable the method is over a long time. It tells us about the process of making the nets, not just one single net.

To understand this, imagine you repeat your study many times. If you took 100 different samples, you would make 100 different nets. According to the math, about 95 of those nets would catch the true average.

About five of the nets would miss it entirely. This happens because every sample is random, so the endpoints of the interval change every time. This is why the interval is considered a random variable.

It is important to know that confidence intervals are different from prediction intervals. A confidence interval tries to estimate a group's average, like the mean. A prediction interval tries to guess where one single new thing will fall. For example, a confidence interval might estimate the average roll of a die. A prediction interval would try to guess what you will get on your next roll. Knowing the difference helps scientists use the right tool for the job.

{ "text": "In statistics, researchers often want to know a specific value about a large group, known as a population. For example, a scientist might want to know the average height of every person in a country. Because it is impossible to measure everyone, they take a smaller group called a random sample. The average of this sample provides a point estimate, which is a single guess at the true value. However, a single number rarely tells the whole story. To show how much uncertainty exists, scientists use a confidence interval.

A confidence interval is a range of values that is likely to contain an unknown statistical parameter, such as the population mean.\n\nTo understand how this works, we must look at the mechanism of repeated sampling. In the frequentist approach, the true population mean is considered a fixed but unknown constant. When we take a random sample, the data we collect is subject to chance. Because the sample is random, the endpoints of the confidence interval are also random variables. This means that if you were to take many different samples from the same population, each sample would produce a slightly different interval. The confidence level, such as 95%, describes the long-run reliability of this entire process.

\n\nThere are different ways to calculate these intervals depending on the data available. One common method uses the central limit theorem. This method requires a large sample size to work effectively. It involves calculating the sample mean and the sample standard deviation to find the interval. Another method is called bootstrapping, which uses the existing data to estimate the range.

In some cases, researchers use a conservative confidence interval. This is an interval that is intentionally wider than necessary to ensure the true value is caught. This approach errs on the safe side to maintain the required coverage probability.\n\nWhen a population follows a normal distribution, mathematicians use specific tools like the Student's t-distribution. If we have an independent sample with an unknown mean and variance, we can calculate a t-score. This value is known as a pivotal quantity because its distribution does not depend on the unobservable parameters of the population. By using the sample mean and the sample variance, we can rearrange mathematical terms to find the specific boundaries of the interval. For a 95% confidence interval, the math ensures that the interval covers the true mean with a 95% probability in repeated sampling.

\n\nIt is very easy to misunderstand what these numbers actually mean. A common mistake is believing there is a 95% probability that the true parameter lies within one specific, already calculated interval. In frequentist statistics, the true mean is fixed. It is either inside the interval or it is not; there is no probability involved for that single result. Another mistake is thinking that 95% of the individual data points in a sample fall within the interval. This is not true. The interval is meant to estimate the group's average, not to capture every individual measurement.\n\nTo visualize the true meaning, imagine performing the same experiment 100 times. You would generate 100 different confidence intervals based on 100 different random samples. According to the definition of a 95% confidence level, approximately 95 of those intervals would successfully contain the true population mean. About five of them would miss the true value entirely.

This highlights that the "95%" refers to the success rate of the method over many trials, rather than the certainty of one single result.\n\nConfidence intervals are also distinct from prediction intervals. While a confidence interval quantifies the uncertainty in estimating a population parameter, like a mean, a prediction interval does something else. A prediction interval provides a range where a future individual observation is expected to fall. For instance, if you roll a twenty-sided die, a prediction interval would estimate the range of a single roll. In contrast, a confidence interval would estimate the average result of many, many rolls. Understanding this distinction helps researchers choose the correct tool for forecasting or estimation.

", "media": [ "File:Confidenceinterval.svg", "File:Normal distribution 50% CI illustration.svg", "File:95% confidence interval.svg", "File:Neyman_Construction_Confidence_Intervals.png", "File:Welch and Bayes intervals.svg" ] }