Since taking over the Business Data Analysis (Statistics) module a few years ago at the College I have been endeavouring to improve the module content and cover more tests. Many statistical tests assume that populations are normally distributed and that variances are equal. Student's t-test for comparing the means of two samples uses a slightly different formula, and a completely different way of determining degrees of freedom, depending on if the variances of the two samples are considered equal or unequal.
Fortunately the test for Equality of Variance is an easy one - it is called the F test. Simply divide the higher sample variance by the lower sample variance - this gives an F statistic which can be compared to a critical value for F in an F Distribution statistical table. From this you can determine the appropriate t-test formula to use.
Here is my latest YouTube video showing how to carry out the F test, both manually and using Excel's Data Analysis Toolpak:
No comments:
Post a Comment