Sie sind auf Seite 1von 3

Confidence Interval Estimation Chapter 9 AWZ

1. Introduction
Given an observed data set we want to make INFERENCES about some larger population. For e.g. A mail-order Company has thousands of customers. The company would like to infer the average time its customers take to pay their bills, so it takes random samples of a relatively small number of customers, sees how long it takes them to pay their bills and draws inferences about the population.

In the above example the unknown population parameter that the Company wants to estimate is the mean length of time customers take to pay their bills. Therefore random samples are selected and inferences about the population are drawn from the samples. Random samples allow us to use the rules of probability to make inferences about the population as a whole. The inferences drawn are always based on an underlying probability model that is; some type of random mechanism must generate the given data. Two random mechanisms are commonly used: Sampling randomly from a larger population. Randomised experiments we select a set of subjects and randomly assign them to two treatment groups.

2. Statistical Inferences
Two types Confidence Interval Estimation for this we use the data to obtain a point estimate and a confidence interval level around this estimate. A point estimate is a single numeric value, a best guess of a population parameter. A confidence interval is an interval around the point estimate calculated from the sample data where we strongly believe the true value of the population parameter lies. Hypothesis Testing discussed in the next chapter

3. Sampling distributions
When estimating the population mean, the point estimate is the sample mean. The sampling distribution is the distribution of the sample means for all possible samples from the population. The standard error is the sample standard deviation divided by the square root of the sample size.

The Multiple is equal to 2. Typical form of confidence interval : Point Estimate +/- Multiple x Standard Error Whenever we make inferences about one or more population parameters, such as a population mean or the difference between two means, we always base this inference on the sampling distribution of a point estimate such as t sample mean. the From the central limit theorem we know: - If the size n is reasonably large then for any population distribution the distribution distribution of X bar (sample mean) is approximately normally distributed with mean (Mu) and standard deviation sigma divided by the square root of n, where Mu and Sigma , are the population mean and standard deviation. - Refer to formula on page 434. - The problem with the above is that we usually do not know the population standard not deviation therefore we replace Sigma ( the population standard deviation) by the population sample standard deviation s by doing this we introduce a new source of variability variabil and the sampling distribution is no longer normal, it is instead called the t distribution.

4. The t distribution
We want to estimate a population mean Mu with a s sample size of n We therefore will use the standardised value of X bar and the sample deviation s This standardised value has a t distribution with n-1 degrees of freedom. 1 Please refer to formula on page 434. n The degrees of freedom is a numerical parameter of the t distribution; it defines the precise shape of the distribution. The t distribution looks very much like the standard normal distribution. standard It is bell shaped and is centered at 0. It is slightly more spread out. The increase in spread is greater for small degrees of freedom. When n is large the degrees of freedom will be large, under these circumstances the t large, distribution and the standard normal distribution are practically indistinguishable. are indistingui The t-value represents the number of standard errors by which the sample mean differs value which from the population mean. Therefore t-values are perfect candidates for the multiple in expression 9.1 on page 434. values expression TDIST and TINV can be calculated using excel. REFER TO PAGE 436 and 437 for examples.

5. Confidence Interval for a mean


We use results about sampling distributions to construct confidence intervals. sampling We assume the data have been generated by some random mechanism. random We want to use the data to infer the values of one or more of the population parameters values such as the mean, the standard deviation, or a prop proportion.

For each such parameter we use the data to calculate a point estimate which can be calculate considered a best guess for the unknown parameter parameter. We then calculate a confidence interval around the point estimate to gauge its accuracy. late estimate We derive a confidence interval for a population me mean and we use X bar, the sample mean, as the point estimate of . To obtain a confidence level for we first specify a confidence level usually 90%, 95% 95 or 99%. We then use the sampling distribution of the point estimate to determine the multiple of the standard error. PLEASE READ PAGE 440 to determine how to calculate the t-multiple.

6. Confidence interval for a proportion


This section will show us how to form a confidence interval for any population proportion p. We find a point estimate. We then find the standard error of the point estima estimate. Then we find the multiple that depends on the confi confidence level Point estimate +/- multiple x standard error Please refer to page 448 and 449 for examples and f formulae.

Das könnte Ihnen auch gefallen