ONE SAMPLE T-TEST

[Link Map]
begin comments text sound
line

The t-test is used to validate the hypothesis that a sample is consistent with a specified mean, or to test for the equality of the means of two sample data sets.

The one sample t-test assesses a set of data against a particular value. For example, if a model predicts the average milk yield of a breed of cow to be 18.5kg/day, the t-test can assess whether the difference between 18.5 and the mean of all of the data could have happened by chance.

No formal one-sample t-test exists in Excel but you can find a confidence interval which is a range of values that would not be rejected by a t-test at a particular level of significance (often 5%).
A confidence interval can be easily obtained either by using the Descriptive Statistics tool in the Analysis ToolPak, or by selecting CONFIDENCE from the FunctionWizard's list of statistical functions.

Using this method, select Tools > Data Analysis > Descriptive Statistics from the menubar, enter the input range of the sample, and tick the confidence level option. Insert '95' as the confidence level for the mean if you are accepting the hypothesised value when it lies within the 95% confidence interval. Click OK.

For the milk example the following output appears:


The sample mean is 20.04, thus the corresponding 95% confidence interval for the mean is (19.176,20.904). Since 18.5 doesn't lie in this interval, the null hypothesis is rejected. The higher the confidence level, the wider the interval. For example, a 99% confidence interval outputs the value 1.1451 so for the mean would be (18.895,21.1851).
line
Basic statistics in Excel   23.2.99   Page: 18 of 25