Chapter 11: QuasiExperimental and Single Case Experimental Designs

1
Chapter 11Quasi-Experimental and Single
CaseExperimental Designs
2

• Quasi-experimental designs
• QUASI sort of
• Emerged from the need to conduct applied research
  in settings in which the control features of a
  true experiment cannot be achieved
• - Causal inferences can be much more difficult
  as a result but are still do-able if done
  correctly

3
Program Evaluation

• Research on programs that are proposed and
  implemented to achieve some positive effect on a
  group of individuals
• Done with schools, businesses, psychiatric
  settings, etc.
• - Programs honestly evaluated to determine
  whether they are effective

4
Program Evaluation (cont)

• Five types of program evaluations
• Needs assessment asks whether there are
  problems that need to be addressed in a target
  population (e.g., do kids need to learn something
  they arent now getting?)
• Assessment of program theory asks if program is
  based on valid assumptions and whether program
  addresses needs in appropriate ways (e.g.,
  education theory?)

5
Program Evaluation (cont)

• Process evaluation (program monitoring)
  monitors the implementation of the program to
  determine if it is reaching the target
  population, attracting enough clients, etc.
• Outcome evaluation (impact assessment) asks
  whether the intended goals are being realized
• Efficiency assessment asks whether the
  resources used to implement the program was
  worth it

6
Quasi-Experimental Designs

• One-group posttest-only design (called a
  one-shot case study

Lacks a crucial element of a true experiment a
control or comparison group. DONT USE IT!
7
Quasi-Experimental Designs (cont)

• One-group pretest-posttest design

- May be used to determine an index of change
from the pretest to the posttest but
SUSCEPTIBLE TO ORDER EFFECTS (practice, fatigue)
and maturation, historical confounds, etc. (see
more next)
8
Quasi-Experimental Designs (cont)

• Alternative (confounding) explanations to studies
  using the one-group pretest-posttest design
• 1. History event is confounded with the
  manipulation
• 2. Maturation changes that occur
  systematically over time (fatigue, hunger,
  intelligence, etc.)
• 3. Testing pretest sensitizes people to the
  purpose of experiment or make them more adept at
  a skill being tested (practice effect)

9
Quasi-Experimental Designs (cont)

• Morealternative (confounding) explanations to
  studies using the one-group pretest-posttest
  design (cont)
• Instrument decay basic characteristics of the
  measuring instrument change over time
• Regression toward the mean when participants
  are initially selected because they score
  extremely high or low on some variable scores
  later tend to change in the direction of the mean
  when tested again

10
Quasi-Experimental Designs (cont)

• Nonequivalent control group design

Better design BUT potential problem is that the
groups are not equivalent. An alternative
(confounding) explanation is selection
differences.
11
Quasi-Experimental Designs (cont)

• Nonequivalent control group pretest-posttest
  design

• EVEN BETTER but problem is participants are not
  randomly assigned to condition so differences in
  outcome may be due to self-selection or already
  existing differences in groups.

12
Quasi-Experimental Designs (cont)

• Interrupted time series design
• Examines the dependent variable over an extended
  period of time, both before and after the IV is
  implemented
• Researcher keeps measuring MANY times before
  intervention (baseline) and them MANY times
  after intervention (treatment)
• May be some interpretation problems (possible
  regression to the mean)

13
Single-Case Experimental Designs

• Traditionally called single-subject designs
• Now referred to as single-case and
  single-participant designs
• Historical tradition Behavior Analysis
• Techniques and logic applied to other research
  areas
• If done well, is a perfectly valid research
  design

14
Single Case Experimental Designs
Simple Reversal design A-B-A design Participant
s behavior is measured over time during a
baseline control period

15
Single Case Experimental Designs
A-B-A design First baseline condition data MUST
be stable before intervention! Why?

16
Single Case Experimental Designs (cont)

• Improving the A-B-A design using REPEATED
  REVERSALS
• ABAB design (treatment is tested twice)
• ABABAB design (treatment is tested three times)
• Why is it necessary to extend the basic ABA
  design?
• Single reversal is not powerful evidence for the
  effectiveness of the treatment. Repeating the
  reversal adds REPLICATION!
• - Ethics sequence ends with the treatment
  rather than the withdrawal of the treatment

17
Single Case Experimental Designs (cont)

• What if reversal of some behaviors is impossible
  or unethical?
• Some treatments might produce a long-lasting
  change that is not reversible
• this especially occurs if researcher is
  investigating a teaching technique (e.g., does a
  technique to teach language work?)
• What do we do then? (see next)

18
Single Case Experimental Designs (cont)

• Multiple baseline designs
• Effectiveness of the treatment demonstrated when
  the behavior changes ONLY after the manipulation
  is introduced so no need to do reversal

19
Single Case Experimental Designs (cont)

• Multiple baselines can be done in 3 different
  ways
• Multiple baseline across subjects
• Multiple baseline across behaviors
• Multiple baseline across situations (settings)

20
Single Case Experimental Designs (cont)

• Multiple baseline design

21
Single Case Experimental Designs (cont)

• Notice that IV (treatment) is implemented at
  different times. Why? To control for historical,
  practice, maturation confounds.

22
Single Case Experimental Designs (cont)

• Important to do replications in single case
  designs over time.
• Also researcher present results from each
  participant individually rather than group data.
  Why? Because each participant was his or her own
  group (many scores obtained for each
  individual).
• If you can show the same effect for each
  individual again and again across many people,
  then you can be very certain that the IV has an
  effect

23
The End