|
Activity Number:
|
384
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistical Education
|
| Abstract - #305799 |
|
Title:
|
Post-Hoc Analysis for a Class of Chi-Square Tests
|
|
Author(s):
|
Edward Markowski*+ and Carol A. Markowski
|
|
Companies:
|
Old Dominion University and Old Dominion University
|
|
Address:
|
IT Decision Sciences Department, Norfolk, VA, 23529,
|
|
Keywords:
|
introductory statistics ; chi-square tests ; post-hoc analysis
|
|
Abstract:
|
Chi-square tests such as the test of independence are standard methods in introductory statistics texts. When the null hypothesis is rejected, the conclusions of such tests are too generic to fully identify the sources of relationships identified. Authors of introductory texts address this issue in different ways and with varying sophistication. In this paper we point out several weaknesses of these common approaches and propose an alternative approach to further analysis of significant chi-square results. This analysis uses the components of the chi-square statistic in identifying critical cells for interpretation. In addition, it uses subsequent chi-square tests to explicitly determine the appropriate degree of investigation of cell comparisons.
|
