Analysis of Data through Non-Parametric Statistical Techniques
CECS 6800.030

Instructor: Gerald Knezek

Online Student Information Sheet

Room: Matthews 308
Dates and Times: May 12,13,15 (optional), 20, 22, and 27, in Matthews 308 from 6:30 to 9:30 pm. Final project presentations will be held May 27 from 6:30 until 9:30 pm. This is a web extended course.

Contact Information:
Dr. Gerald Knezek
Voice Mail: 940-565-4195
FAX 940-565-2185

Dr. Dana Arrowood

Mailing Address:
Technology and Cognition/UNT
P.O. Box 311337
Denton, TX  76203

This course will introduce practical distribution free (nonparametric) techniques for analysis of social science data sets. The emphasis will be on drawing conclusions from data not well suited to techniques that assume a normal distribution.  Participants are encouraged to use their own data sets for final project analysis. It is expected that students will apply at least three data analysis techniques to homework (practice) sets of data.

Gibbons, Jean D. (1976). Nonparametric methods for quantitative analysis. New York: Holt, Rinehart and Wnston.
(on reserve in Willis Library)

Kendall, M., & Gibbons, J. D. (1990). Rank correlation methods. Oxford University Press.
(on reserve in Willis Library)

Dunn-Rankin, P., Knezek, G., Wallace, S., & Zhang, S. Scaling Methods (2nd Edition - prepublication).This book will be given to all participants in a bound format.

Other Class Materials
CD-Rom with class materials will be distributed in class.

Students are responsible for making arrangements for the following:

     E-mail access capable of posting attachments
     Internet Browser Access
     Realplayer G2 capability (for playing encoded audio and video files)

Note: Each participant is responsible for seeking technical support via the computing center help desk (940.565.2324) or other means to resolve connectivity problems.


 Data Analysis of Student Chosen Dataset








90 - 100


Project Presentations on May 27, 2003








80 - 89


Home Exercises 








70 - 79


Participation (In class & online) 








60 - 69










below 60




May 12

6:30 - 9:30


Matt 308

Distribution Free Stat, Probability, Binomial Tests


May 13

6:30 - 9:30


Matt 308

Strict/Partial Order, Ranking,. Rank Corr., Single Subject Consistency


May 15 (optional)

6:30 - 9:30


Matt 308

Variance-stable Rank-Sum Tests


May 20

6:30 - 9:30


Matt 308

Fisher Exact Test, Non-parametric alternatives to t, ANOVA, etc.


May 22

6:30 - 9:30


Matt 308

Chi-Square Goodness of Fit


May 27

6:30 - 9:30

final project presentations

5 pages, 5 ref., 5 minutes Matt 308


Description of Homework Exercises (submit at least 3 for 10 pt. each credit)


  1. Binomial Test.  Locate some real word data that has dichotomous categories (M/F, etc.) and carry out a sign test to determine if the distribution of one group or the other is unlikely to have occurred by chance.  Compute the exact p level of the event you defined as a “success.”  Turn in a brief summary of your findings.


  1. Rank Order Correlation. Use the graphical method introduced by Wilcoxon and Wilcox to compute rank-order correlations (Kendall’s Tau) of two data sets.  Visually examine the crossovers and write a brief summary of how the graphical representation adds value to the numerical index.  Include at least 10 entries in each set of data.


  1. Single Subject Internal Consistency. Construct a paired-comparisons experiment in which pairwise-selections of at least 5 items are presented to at least 10 people.  Run the data through the TRIAD program and examine the differing preferences of each person in conjunction with his/her consistency.


  1. Rank Sum Differences.  Locate at least 6 objects to be rank-ordered by 10 or more judges.  Produce the rank sums by hand and/or by running the program RANKO on the CD provided.  Briefly describe which objects are significantly different.


  1. Fisher’s Exact Test.  Find 2 x 2 contingency table data relevant to your interests and carry out Fisher’s Exact test on the data.  Run and compare a Chi-square test on the same data, if it is appropriate, or describe why it is not appropriate otherwise. What are the advantages and disadvantages of each technique? Which outcome do you trust more?


  1. Chi Square Goodness of Fit.  Find a set of data that you believe should conform to (be drawn from) some theoretical a population with some theoretical distribution such as the normal curve.  Use a Chi-square test to determine the goodness of fit of your data. What does it mean if the data fails to deviate from the hypothesized distribution?


Final Project.  Choose one technique and set of data to explore in detail.  Write 5 pages, 5 references, and present 5 minutes during the final class. This is not counted as one of the three Homework Exercises listed above.



Cheating and disciplinary action is defined by the UNT Policy Manual Codeof Student Conduct and Discipline.
Cheating is an act of academic dishonesty. It is defined and is to be handled as follows: "Plagiarism and
cheating refer to the use of unauthorized books, notes, or otherwise securing help in a test; copying tests,
assignments, reports, or term papers; representing the work of another as one's own; collaborating, without
authority, with another student during an examination or in preparing academic work; or otherwise practicing
scholastic dishonesty."

Statement on Discrimination:
The University of North Texas provides academic adjustments and auxiliary aids to individuals with disabilities as defined under the law, who are otherwise qualified to meet the institution's academic and employment requirements. Please see the instructor outside of class to make any arrangements involving special accommodations. ADA/EEO/AA