Resource Library

This page discusses the procedures and applications of the two sample t test and the paired t test.

This case study explores statistics on divorce rates using Markov chains. Two closely related statistics are presented: the chance of divorcing in a given year and the chance of divorcing over the lifetime of a marriage. Accompanying teacher instructions are found at http://ublib.buffalo.edu/libraries/projects/cases/markov/markov_notes.html

This tutorial on the One Sample t test includes its definition, assumptions, hypotheses, and results. An example using output from the WINKS software is given, but those without the software can still use the tutorial. An exercise is given at the end that can be done with any statistical software package.

Explains how to set up the Kruskal-Wallis test and gives the formula for the test statistic.

This site offers an involved definition of the likelihood ratio test with examples and formulas.

This section on Common Statistical Tests uses an example on faculty publications to show users how to perform a one-sample t test. The discussion includes one-tailed and two-tailed tests.

This exercise will help the user understand the logic and procedures of hypothesis testing. To make best use of this exercise, the user should know how to use a z table to find probabilities on a normal distribution, and how to calculate the standard error of a mean. Relevant review materials are available from the links provided. The user will need a copy of the hypothesis testing exercise (link is provided), a table for the standardized normal distribution (z), and a calculator. The user will be asked several questions and will be given feedback regarding their answers. Detailed solutions are provided, but users should try to answer the questions on their own before consulting the detailed solutions. The end of the tutorial contains some "thought" questions.

This site explains the relationship between hypothesis testing and confidence intervals.

This applet simulates finding confidence intervals for the mean of a normal random variable. A sample of size 20 is generated from a standard normal random variable. The blue marks represent the sample data. The sample mean X and sample standard deviation s are found and used to calculate the confidence interval. The black intervals are the confidence intervals which include the true mean 0, and the red intervals are those which exclude 0. This applet needs to be resized for optimal viewing.

This simulation applet shows groups of confidence intervals for a given alpha based on a standard normal distribution. It shows how changes in alpha affect the proportion of confidence intervals that contain the mean. An article and an alternative source for this applet can be found at http://www.amstat.org/publications/jse/v6n3/applets/confidenceinterval.html.