Experimental Research Designs, Part 2

1
Experimental Research Designs, Part 2

• Research Process and Design
• Spring 2006
• Class 7

2
Todays objectives

• To answer any questions you have
• To further explore quasi-experimental and
  single-subject designs
• To talk about writing the methods section
• To apply knowledge to your group project

3
Characteristics of Experimental Research

• Statistical equivalence of subject in different
  groups (usually by random assignment)
• Comparison of 2 or more groups
• Manipulation of at least 1 IV
• Measurement of DVs
• Inferential statistics
• Maximum control of extraneous variables

4
Experimental ResearchStrengths and Limitations

• Strengths
• The only designs capable of examining causal
  relationships
• High degree of control of extraneous variables
• Direct manipulation of an independent variable
• Weaknesses
• High levels of control needed for assuring
  internal validity mean very restrictive
  manipulated conditions which in turn imply low
  external validity
• High levels of external validity mean little
  control and thus low internal validity
• Not possible in many situations

5
Quasi-Experimental Designs

• Good designs that are frequently used in
  education
• Lack of control related to randomly assigning
  subjects to groups (i.e., the use of pre-existing
  groups)
• Non-equivalent groups terminology reflects the
  lack of random assignment and the concern about
  the selection internal validity threat

6
Quasi-Experimental Designs

• Basic designs and several variations of them
• Non-equivalent groups designs
• Non-equivalent groups pretest-posttest
  experimental-control
• Non-equivalent groups pretest-posttest
  experimental-comparison
• Non-equivalent groups pretest-posttest
  experimental-control-comparison
• Statistical analyses focus on the comparison of
  groups and the use of appropriate analyses like
  ANCOVA

7
Quasi-Experimental Designs

• Time series designs groups repeatedly measured
  before and after treatments
• Single group interrupted time series (e.g., A
  OOOOOO X OOOOOOO)
• Experimental-control group interrupted time
  series
• Variations of these designs include the use of
  comparison groups and different treatment times
• Statistical analyses examine the patterns of
  performance over time

8
Factorial Designs - Definitions

• Main effect the effect of each independent
  variable on the dependent variable
• Interaction effect the effect of the
  interaction of two or more independent variables
  on the dependent variable
• Factors independent variables (e.g., two-factor
  design means two independent variables being
  studied)
• Levels independent variables vary, often
  represented in groups (e.g., high, medium, low)

9
How many variables? How many IVs? How many groups?

• 2 x 2
• 3 x 2
• 3 x 2 x 2
• 2 x 2 x 2 x 2

10
Single-Subject Designs

• The use of one subject studied over time
• Typically one subject, but there are situations
  where two or three subjects are sampled
• Focus on populations where there is a low
  incidence of specific characteristics (e.g.,
  severe and profound special education students,
  physically-challenged students, etc.)
• Need to draw causal inferences

11
Characteristics of Single-Subject Designs

• Reliable measurement
• Repeated measurement
• Explicit descriptions of conditions
• Baseline and treatment condition duration and
  stability
• Manipulation of a single variable at one time

12
Single-Subject Designs

• Three types of designs
• A-B designs baseline and treatment
• A-B-A or A-B-A-B designs baseline, treatment,
  baseline (i.e., reversal or withdraw designs)
• Multiple baseline designs

13
Single-Subject DesignsMultiple Baseline

• Use of the A-B logic with the addition of two or
  more actions, subjects, or situations
• Used when it is impossible or undesirable to
  remove a treatment condition
• Types
• Multiple baselines across behaviors (i.e., the
  use of a single treatment across several distinct
  behaviors)
• Multiple baselines across situations (i.e.,
  observation of a single target behavior observed
  in two or more settings)
• Multiple baselines across individuals (i.e., two
  or more subjects receiving treatment where
  behavior and the situation are held constant)
• Statistical analyses examine the patterns of
  behavior

14
Group Activity

• Design a study using 2 of the research designs
  described in todays or last weeks class.
• What are the implications of each design?
• What are the pros and cons of each design?

15
The method section of a quantitative study

• Vary somewhat but generally have the following
  components
• Participants
• Variables
• Instrumentation/materials
• Procedures
• Data analysis

16
More specifically, methods section should

• Identify research design
• Describe sample, population, and participants
• Describe instruments, variables, and materials
• Review procedures in detail
• Explain plans for preliminary studies or pilot
  studies (if necessary)
• Provide details about data analysis procedures
• Identify ethical issues and your approach to
  dealing with them
• Describe limitations or threats to validity

17

• Remember, you need to give enough information so
  that the study can be replicated

18
Group project

• Begin thinking about the elements of your method
  section
• What design will you employ?
• Who do you plan to study?
• What and how will you collect data?
• What are your independent and dependent variables?

19
For next week

• Non-experimental research design Correlational
  and ex post facto studies
• Readings
• Jaeger Chapter 4 (p. 61-76)
• Wiersma, W. Jurs, S. G. Chapter 7
• Distribute take-home exam
• Reminder Outline of literature review due