Chapters 10

1
Chapters 10 11 (Richards text)CHANGING
CRITERIONDesigns in Single-Subject Research

• Ps534
• Dr. Ken Reeve
• Caldwell College
• Graduate Programs in ABA

2
Review Single-Case Experimental Designs Issues

• For each design we will ask the following
• Does the design allow us to see a change in DV
  (without regard to whether it was caused by the
  IV)? YES
• Does the design allow us to infer a functional
  relationship between IV and DV? YES
• Why does it allows this? WELL FIND OUT
• What threats to internal validity (confounds)
  does the design control for?
• What ethical issues are important to know about
  using a particular design?

3
CHANGING CRITERION Designs

• Main Purpose We examine the effects of one IV
  (treatments) on a SINGLE DV (target behavior)
• Specifics
• Usually used with a target behavior the person
  already has in their repertoire
• And we want to dramatically increase or decrease
  the target behavior (rate, duration, frequency,
  correct, magnitude)
• And it is a behavior that is usually very
  difficult to change a great deal all at once
  (studying, smoking, cursing, over-spending)
• Frequently used in self-management interventions

4
How To Do CC Designs

• Begin with single participant (each participant
  is a SEPARATE experiment and serves as his or her
  own control like in a withdrawal design)
• MUST take baseline to demonstrate experimental
  control (phase A)
• Prior to initial treatment step (phase B1), you
  must determine two different criteria
• a) criterion to move to next treatment step and
• b) final goal criterion
• Learner must reach step criterion for at least
  2/3 occasions before next treatment step (phase
  B2) is implemented
• Continue on with treatment phases B3BX
• Stop when final goal criterion is reached

5
Sample CC design
Phase A B1 B2 B3
B4 B5
6
Issues Using CC Designs

• You should vary each treatment phase to control
  for confound of person simply changing behavior
  when a certain number of days goes by (as opposed
  to the persons behavior actually being
  controlled by the changes in response
  requirements
• Longer phases are needed for slow-to-change
  behaviors
• Shorter phases are okay for quick-to-change
  behaviors
• More

7
Issues Using CC Designs

• How much of a step increase in response
  requirements is desirable?
• Clinically, the quicker you get to the goal the
  better BUT if you step up too fast then it is too
  difficult to succeed so better to be conservative
• There are formulas to determine the steps but
  there are no studies demonstrating what step size
  is most desirable (thesis or dissertation
  anyone?)
• Usually the first step in treatment is the
  average of the baseline responding (or slightly
  higher than the average)

8
Reversals in CC designs (step down)

• To increase internal validity (demonstration of
  functional relationship), you can step down the
  requirements to see if behavior drops
• This is a GREAT control for maturation practice
  effects, in particular

9
Reversals in CC designs (back to baseline)
Not shown in this graph but imagine that the
reversals are return to baseline

• Experimentally, it is even better if you reverse
  back to baseline instead of just stepping down
  response requirement (assuming behavior will drop
  a lot)
• Clinically, though, this may not be as ethical

10
Design a study using CC Design

• Question You wish to decrease the frequency of
  cigarette smoking in an adult group home resident
  who currently smokes 60 cigarettes a day and has
  been doing it for 20 years.
• The resident already uses a token motivational
  system to earn access to leisure activities
• How would you do it?

11
End