Interpreting Research and Functional Relations

1
Interpreting Research andFunctional Relations

• EDEX 5150
• Session 4

2
Agenda

• Discuss functional relations and concept of
  experimental control.
• Discuss threats to validity.
• Discuss purpose of baselines.

3
Functional Relations

• The purpose of single subject research is to
  suggest functional relations.
• Manipulation or change in independent variable
  (intervention) produces consistent change in
  dependent variable (measure of target behavior).
• Manipulation typically involves addition or
  subtraction of whole or part of independent
  variable.

4
(No Transcript)
5
Types of Dependent Variables

• Number of correct responses
• Percentage of correct responses
• Duration of head-banging
• Number of words read per minute

6
Types of Independent Variables

• Direct Instruction
• Self-reinforcement
• Time out
• Token economy
• Verbal praise

7
Experimental Control

• A functional relationship is demonstrated when
  alternative explanations of the reported behavior
  change can be ruled out--minimize alternative
  explanations (extraneous variables) of behavior
  change.
• Other events or variables that may be responsible
  in whole or part for the behavior change need to
  be identified and manipulated (extrinsic
  variability).
• These extraneous variables need to be held
  constant.

8

• Experimental control--demonstrating a functional
  relation by minimizing alternative explanations
  (extraneous variables) of behavior change.

9
Extraneous Variables

• Change in students health physical,
  psychological, or emotional development or
  social status.
• Parent/sibling intervention.
• Change in school schedule, setting, personnel,
  grading system, support available.

10
Types of Validity

• Statistical (significance)
• Content
• Predictive
• Concurrent
• Construct
• Internal
• External
• Participants
• Settings
• Responses
• Social

11
Threats to Validity

• History effects (what happens outside of research
  setting)
• Maturation effects (physical, emotional,
  psychological development)
• Testing effects (reactivity)
• Instrumentation effects (errors in observation)
• Regression to the mean (no outlier behaviors)

12

• Participant selection bias
• Attrition
• Multiple treatment interference
• Instability (cyclic variation)

13
Questions Regarding Validity

• Internal validity--Is the intervention and only
  the intervention responsible for the change in
  behavior?
• External validity--(1) Given that the
  intervention produced a change with this
  participant(s), will it similarly produce a
  change for other participants, in other settings,
  or with other responses and (2) is it possible
  to make statements about generality?

14
Social Validity

• Was the experiment/research socially acceptable?
• Applied behavior analysis Social Validity
  Quality Research
• Relevant stakeholders, including student, are
  asked to provide input on goals (dependent
  variables), procedures (independent variables),
  and effects (results).

15

• Social Validation
• To gather information about participants
  perceptions about what they had learned through
  the problem-solving process, each student
  answered the following three questions at the
  completion of his/her training about the goals he
  or she had set (1) What has changed?, (2) Did I
  do what I said I would do?, and (3) What do I
  like about it (the change)? Additionally,
  anecdotal information was obtained from the
  students regarding their general perceptions
  about the utility of the model.

16
Baselines

• Baselines provide information about the pattern
  of a behavior--how often it occurs over a
  specified period of time--prior to the
  introduction of an intervention.
• A baseline provides a representative sample of
  the behavior, or an estimate of the frequency of
  responding.

17
(No Transcript)
18
Baseline Problems

• Baselines may be too short (need a minimum of 3-5
  data points but stability steady state is key
  issue).
• Baselines may be too long.
• There may be problems with missing data points.
• Data may be cyclical.

19
(No Transcript)