Spring 2001

REHB 509B

Behavior Analysis Research Designs:

Group Experimental Designs

The syllabus is subject to error correction and minor change in content during the

course.

Instructor: Anthony J. Cuvo, Ph.D.

Rehn 311A

Phone 536-7704; FAX (618) 453-8271

Syllabus On-line:

Time: 12:00-1:15 PM, Tuesday and Thursday

Classroom: Rehn 326

COURSE DESCRIPTION & GOALS:

The purpose of this course is to provide a foundation in applied research

methods pertinent to program evaluation, group experimental design, and related data

analysis. After completing the course you should be able to do the following:

a) Be a knowledgeable consumer of group design and related statistical analysis

literature (i.e., understand and critically evaluate research in journal articles and other

research presentations).

b) Have intermediate level skill generating group design studies, knowing which

data analysis techniques are appropriate, using a computer for basic statistical

analyses, and drawing appropriate conclusions.

Principal Text (available at local bookstores)

Christensen, L. B. (2001). Experimental Methodology (8th ed.). Boston: Allyn &

Bacon.

Huck, S. W., (2000). Reading statistics and research (3rd ed.). New York:

Addison Wesley Longman.

REHB 509B Syllabus

2

Additional Required Readings

Additional readings are available from the Printing Plant, 606 S. Illinois Avenue.

These readings, indicated by asterisks in the syllabus, supplement and are equally

important to those in the textbooks. Page through the entire reading packet as soon as

you get it and compare it to the syllabus. If you find missing pages or pages that are

not legible go to the Printing Plant and ask them to rectify the situation. You are

responsible for all assigned readings on the due date.

Requirements and Grading

1. There will be 4 tests @ 100 points each on: February 13, March 20, April 12, May 9.

Tests will emphasize the material since the previous test; however, the content is

cumulative and you should be able to relate earlier concepts to the current material on the

tests. Students must remain in the classroom until finished with the test. Take care of any

personal needs before coming to the classroom.

Total possible: 400 points

2. There are 3 conceptual projects that require you to apply the material in this

course. The forms for the first two are available on-line at

The form in is Microsoft Word format

and can be downloaded on disk or to your computer. You will need to use Word or a

program that will open Word. The form is the same one used in Rehab. 509A. Although

projects 1 and 2 could be on the same general topic (e.g., child abuse, biofeedback,

mental retardation), each must be on a different specific topic. Projects should include a

new literature review and independent variable. Projects should not be just minor

variations of each other. About 90% of the points lost in past years have been due to

not following APA style and not answering all components of the questions. Put

Projects in instructor's mailbox in Rehn 317 by 4:00 PM on the due date. Note that

Rehn 317 will be locked promptly by 4:30 PM. The office also will be closed between

12:00-1:00 PM. There will be a 33% per day reduction in the maximum point total for

late assignments, including weekend days.

Project due dates and point values:

Project 1 (3/5/01) 20 points

Project 2 (4/2/01) 25 points

Project 3 (4/30/01) 10 points

Total possible: 55 points

REHB 509B Syllabus

3

3. A 15 minute quiz will be given at the beginning of 22 classes indicated on the

syllabus. Each quiz will be worth 10 points. If you come to class while the quiz is being

administered, you will have until time expires on the quiz to finish. If you come to class

after the quiz has been completed, you will not have the opportunity to take it and you

will receive a grade of 0 for that quiz. If you plan to be absent from class, it is your

responsibility to arrange to take the scheduled quiz or test in advance of the class you

will not attend.

Total possible: 220 points

Point to letter grade conversion:

A = 675-607 points

B = 606-540 points

C = 539-472 points

Lower grades are available on the same proportional scale.

If you have earned 90% of the points up to and including the first 18 quizzes, first

three tests, and the three projects (i.e., 481 points exactly, no rounding) and made a

minimum score (not average) of 9 on each quiz in the last course unit, you will be

exempt from taking the final exam and receive an “A” in the course. The quiz points for

the final unit are not included in the 90% criterion.

• If you are having difficulty with this material, see the professor as soon as

possible.

• If you wish to drop this course for any reason, the GraduateSchool has a final

date that you can do this. It is your responsibility to drop by the date designated by

the GraduateSchool.

• A grade of Incomplete will be given only under the conditions specified in the

GraduateSchool Catalog.

The reading list includes a number of journal articles that present experimental

research. These articles serve as models for the integration of conceptual issues,

research questions, measurement procedures, experimental design, data analysis, and

inferences from the data to past research and conceptual issues. You should read

these articles and try to understand the integration of the various components of the

research process; however, an article has been assigned for a particular class because

it illustrates the topic for that class. Focus, in particular, on the aspect of the article that

is related to the topic for that class.

REHB 509B Syllabus

4

You will have three statistical projects to do in this course using a computer

and software of your choice. It is recommended that you identify the computer and

software that you will use and gain familiarity with both before you start your first

project.

The readings and exercises from the Internet are no different from other

assigned readings with respect to their importance and availability for material for

quizzes and tests.

It is recommended that you take your readings to class, especially the ones

for that day.

A number of computer data analysis examples have been included in your

reading packet. They present a research question, design, computer data analysis, and

interpretation. Various statistical packages have been used for these examples. Try to

understand the research and data analysis, and don’t get bogged down in the

mechanics of how to use the software. You should try to understand the printouts and

interpretation of results.

Try to do the readings in the sequence indicated on the syllabus.

REHB 509B Syllabus

5

***January 16, 2001

Introduction to course

Video: Facilitated Communication

***January 18-23, 2001 (Science & Theory)

Readings:

1/18/01

"Much like the law of gravity, the laws of learning are always in effect.

Thus, the question is not whether to use the laws of learning, but rather

how to use them effectively." From "Learning Principles" (Spreat & Spreat)

The above quote is related to the goal of science, to discover the orderliness or

lawfulness in nature. Those lawful relations about human behavior always have existed,

and they are there waiting for us to discover them. We discover them using scientific

methods, and that discovery can lead to useful applications in human services. This

course focuses on the role of certain aspects of scientific methodology as a tool for

understanding variables that relate to human services.

These initial readings are a varied collection that have a common theme- the big

picture pertaining to research and program evaluation. They address philosophical

issues, definition and characteristics of science, research questions, and conceptual

issues. They address the larger issues that surround the more focused tactics of

research methodology.

Christensen Chap. 1

* Shermer, M. (1992). A skeptical manifesto. Skeptic, 1, 15-21. (This article

advocates a basic approach, a “mind-set” as some might say, to examine claims about

causal events in the universe. Do you believe everything that people tell you? How do

you decide what to believe? What is your criterion for truth?)

* Cuvo, A.J. Rational Skepticism and Research Methodology

* Green, G. (1996). Evaluating claims about treatments for autism. In C. Maurice,

G. Green. & S. C. Luce. (Eds.). Behavioral intervention for young children with autism

(pp. 15-28). Austin: Pro-Ed.

REHB 509B Syllabus

6

1/23/01

Christensen, Chaps. 3 & 4

* Cuvo, A. J. How to Ask Research Questions or Words Mean Something

* Exploring psychology scientifically-Asking testable questions

(Download from Internet.

Test yourself with this exercise.)

* Riggin, L.J.C. (1990). Linking program theory and social science research. In

L. Bickman (Ed.), New directions for program evaluation: Advances in program theory,

no. 47 (pp. 109-120). San-Francisco: Jossey-Bass. (This article makes a point similar to

that made by Cuvo above. It shows not only that human service programs should have

a theory on which the program delivered to clients is based, but also the program theory

should be evaluated in light of the manner in which the program actually is delivered. Is

the program as conceptualized on paper, congruent with the program in operation?

Does the program theory have to be revised to reflect reality?)

QUIZ 1 1/23 only, based on readings for both dates. No quiz on 1/18

***January 25-30, 2001 (Variables Used in Experimentation; Measurement

Principles and Applications)

Readings:

1/25/01

You might want to read pp. 101-105 from Meltzoff and then the pages from

Christensen on the independent variable first, and then the remaining of the Meltzoff

and Christensen chapters on the dependent variable.

Christensen Chap. 6

* Cuvo, A. J. Independent Variables and Conceptual Models

*Meltzoff, J. (1998). Chap. 7 Criteria and criteria measures. In Critical Thinking

About Research. APA: Washington.

QUIZ 2

REHB 509B Syllabus

7

1/30/01

* Cuvo, A. J. Translating Conceptual Variables to Measurable Variables

* Cuvo, A. J. Documenting Client Progress

Huck Chap. 4

* Anastasi, A. & Urbina, S. (1997). Psychological testing (7th ed.).Upper Saddle

River, NJ: Prentice Hall Chaps. 4-5. (You should understand each type of reliability and

validity. What is its purpose? How does one establish the specific type of reliability and

validity procedurally? Which ones would you use in a particular situation?)

QUIZ 3

***February 1, 2001 (Pseudo- or Pre-Experimental Designs)

Readings:

* HCB (1st. Ed.), Chap. 11 (focus on experimental design and not statistical

analysis)

Christensen, pp.232-238.

* Holden, P. & Neff J. A. (2000). Intensive outpatient treatment of persons with

mental retardation and psychiatric disorder: A preliminary study. Mental Retardation,

38, 27-32. (Focus on design and not data analysis. What pre-experimental design was

used? What research questions can it and can it not answer?)

QUIZ 4

***February 6, 2001 (Internal-Validity)

Readings:

Christensen, Chap.7

Re-read the HCB Chap. 11 from last class and focus on internal validity of

designs.

* Cuvo, A. J. Threats to Internal Validity in Experimental Research

* Cuvo, A. J. A Note on Testing as a Threat to Internal Validity and Pretest and

Posttest Sensitization as Threats to External Validity. (Read first paragraph)

REHB 509B Syllabus

8

* Kruger, J., Savitsky, K., & Gilovich, T. (1999). Superstition and the regression

effect. Skeptical Inquirer, 23(2), 24-29.

QUIZ 5

***February 8, 2001 (Quasi-Experimental Designs)

Readings:

* HCB (1st. Ed.), Chap. 14 (up to p.323, focus on experimental design and not

statistical analysis)

Christensen, Chap. 10

The following articles involve quasi-experimental designs. Understand which

specific design was employed, what research question(s) it can and cannot answer,

and how well this design controls for threats to internal validity.

* Slate, J. R., & Jones, C. H. (1989). Can teaching of the WISC-R be improved?

Quasi-experimental exploration. Professional Psychology: Research and Practice, 20,

408-410.

* Schnelle, J.F. & Lee, J.F. (1974). A Quasi-experimental retrospective

evaluation of a prison policy change. JABA, 7, 483-496.

* Wilderman, R. (1981). Psychotherapy in a community mental health facility.

Evaluation and the Health Professions, 4,189-205.

QUIZ 6

***February 13, 2001

TEST 1

***February 15, 2001 (True Experimental Designs & External Validity)

Readings:

Note: True experimental designs control for most threats to internal validity by

random assignment of subjects to conditions, and do not require an individual analysis

of the plausibility of each threat as was the case for pre-and quasi- experimental

designs.

REHB 509B Syllabus

9

* HCB (1st ed.), Chap. 12 (Focus on design issues and not data analysis. The

use of gain scores and the independent t test are not a recommended approach to

statistical analysis for the pretest-posttest control group design as suggested in this

chapter. When you use a gain score, you lose the reliability of the underlying

measurement that was the basis for calculating the gain score. The pretest-posttest

control group design is also called a Two Factor Mixed Design with One Repeated

Measure. You will learn about an ANOVA for this design in a future class).

* Cuvo, A. J. A Note on Testing as a Threat to Internal Validity and Pretest and

Posttest Sensitization as Threats to External Validity. (Read all. See previous class)

Christensen Chaps. 8 (Except sections on counterbalancing) & 14

* Designing research studies in psychology

(Download from Internet)

* Class Exercise-matching and randomization-Take these to class

QUIZ 7

***February 20-27, 2001 (Basic Statistical Issues)

Readings:

2/20/01

Christensen, Chap.12

Huck, Chap. 2

* Normal Curve

QUIZ 8

2/22/01

Huck, Chaps. 5-6

* Cuvo, A. J. & Hewes, R. L. Population & Sampling

QUIZ 9

2/27/01

Huck, Chaps. 7-9

REHB 509B Syllabus

10

QUIZ 10

REHB 509B Syllabus

11

***March 1, 2001 (T-Tests)

Readings:

You need to understand the logic of the t-test as represented in the formula.

What does the numerator mean? What does the denominator mean? What does the t

test do and how does it do it? What does the calculated value of t really mean?

Re-read comments above about not using t test for pretest-posttest control

group design under introduction to True Experimental Design.

Huck, Chap.11 (Skip sections on “Inferences Concerning a Single Mean”,p.285-

290.

Christensen, pp. 326-335.

* Independent t Test Example (This analysis was done on the Excel

software and serves as a model for Project 1. This provides you an applied research

example where the independent t test was used. )

* Dependent t Test Example (This analysis was done on the Excel

software and serves as a model for Project 1. This provides you an applied research

example where the dependent t test was used.)

* Cuvo, A. J. & Hewes, R. L. Using t Tables (Includes “Critical Value of

‘Student’s’ t statistic”)

* Holden, P. & Neff, J. A. (2000). Intensive outpatient treatment of persons with

mental retardation and psychiatric disorder: A preliminary study. Mental Retardation,

38, 27-32. (See pseudo-experimental design class. Focus on data analysis and

interpretation)

* Cuvo, A. J. Effect of Within Groups Variability on the t Test

QUIZ 11

***March 5, 2001

Applied Exercise 1 - Conduct a literature review, derive a research question, and design

an experiment whose data would be analyzed by either an independent or dependent t

test. Generate hypothetical data for this exercise. Use the Experimental Research

Design Form to do this and attach a table of the raw data for each experimental

condition, and a computer printout of results. Include in your data analysis the M and

S.D. for each experimental condition. Do statistical analysis using any computer and

software of your choice. State at the top of your report the name of the computer and

software. The t test examples on the reading list provide models for this project.

REHB 509B Syllabus

12

***March 6-8, 2001 (One Way ANOVA; Post hocs.)

Readings:

3/6/01

Huck, Chap. 12

* Cuvo, A. J. The logic of ANOVA (You need to know the logic of the ANOVA as

represented in the formula and discussed in this paper. How are the components of

ANOVA calculated?)

* Cuvo, A. J. & Hewes, R. L. Using F tables (Includes “Percent Points in the F

Distribution”).

* Read the relevant section of Cuvo, A. J. Relationship Between Experimental

Design and ANOVA (See Mixed Factorial Design Class below)

QUIZ 12

3/8/01

Huck, Chap. 13

* The Tukey test and table (Note: A post-hoc test, such as the Tukey test, will be

needed for any significant F that is based on 3 or more means. The n per group is

based on the number of subjects that went into the calculation of the mean per group.)

* Kregel, J., Wehman, P., & Banks, P. D. (1989). The effects of consumer

characteristics and type of employment model on individual outcomes in supported

employment. JABA, 22, 407-415. (Focus on ANOVA and post-hoc analyses and not

chi-square analyses)

* ANOVA For One factor CRD (For this and subsequent computer analyses,

focus on the example of applied research and interpretation of results rather than the

mechanics of the computer data analysis.)

QUIZ 13

REHB 509B Syllabus

13

***March 20, 2001

TEST 2

***March 22-27, 2001 (Between Subjects & Randomized Blocks Designs)

Readings:

Huck, Chap. 14

* HCB (1st ed.) pp. 281-284.

Christensen, pp. 246-252, 346-357.

Note: If an ANOVA has both significant main effects and interactions, you will

have to compute more than one Tukey test because the q value and the n per group,

which is in the denominator under the square root, will change. The n per group is

based on the number of subjects that went into the calculation of each mean for rows,

columns, or cells. The Tukey critical difference will not be the same for main effects and

interactions when they differ with respect to number of means and n per group. Assume

a 2 Factor Completely Randomized Design with 2 levels on Factor A and 3 levels on

Factor B. Further assume that there are 10 subjects per cell or a total of 60 subjects

(2X3X10=60). If the F for Factor A is significant, no post-hoc test would be needed

because there are only 2 levels and 2 means involved. If the F for Factor B is

significant, you will have to calculate a Tukey based on a q involving the 3 means for

the three levels of the independent variable, and an n of 20 subjects per row group. In

this case, the group is one entire row or column that contributes to the mean. For the

interaction, there are 6 means for the 6 cells (2x3) in the design and an n of 10 per

group. The MS within group or error will be the same for all Tukey tests based on the

same ANOVA because there is only one MS error in the ANOVA.