Doing Research in Behavior Analysis
4 Phases of a program Screening Baseline Treatment How do we know if the treatment actually caused the change in behavior? Screening Baseline Treatment Follow-up
ANALYZING DATA
FREQUENCY GRAPH
BAR GRAPH (FREQUENCY)
CUMMULATIVE GRAPH
CUMMULATIVE GRAPH
INTERVENTION DESIGNS TO RULE OUT ALTERNATIVE ________________ FOR CHANGE BASELINE ( MEASURE BEFORE INTERVENTION VERIFY BEHAVIOR __________ VERIFY YOUR ______________ COMPARRISON FOR ______________ BEHAVIOR MAY CHANGE EXPLANATIONS STATE MEASURE INTERVENTION
Training/Treatment-Designs TO RULE OUT ALTERNATIVE EXPLANATIONS Reversal design Multiple Baseline Across behavior, settings, or people Alternating treatments(multi-element) Changing criterion
Reversal-replication(ABAB) Dependent variable The measure of the ____________ to change Independent variable The treatment “A” is the baseline; “B” is the treatment Conditions are “reversed” to baseline Treatment is “_______________ ” Also called the “withdrawal” design behavior replicated
ABAB REVERSAL DESIGN
Some questions Assuming response definition,observer accuracy,& data recording are OK… How long a baseline? Stability, Trends How representative of the actual behavior
VISUAL ANALYSIS Immediacy of effects ____________ OF VALUES FROM BASELINE TO INTERVENTION ____________ OF DATA IN BASELINE TREND IN BASELINE IN THE EXPECTED ______________ OF INTERVENTION OVERLAP STABILITY DIRECTION
EFFECTIVE TRAINING
LESS EFFECTIVE
WEAK EFFECT
INCREASED VARIATION
CLEAR EFFECT
TREND IN BASELINE
CLEAR EFFECT
CLEAR EFFECT
EFFECTIVE TREATMENT NUMBER
NOT EFFECTIVE NUMBER
TREND IN BASELINE NUMBER
Why?
WHY? DRO PRAISE
WHAT HAPPENED?
WHY?
Other considerations Scientific—is this phenomenon new? longer Yes—perhaps a ___________ baseline Practical -how much time do we have? What resources are available? Ethical – How _____________ is the behavior? longer dangerous
Other questions How many reversals & replications? If the phenomenon is well researched & the treatment effect is large perhaps one is OK. If the effect is small and / or the phenomenon is new perhaps several
Other questions ethical Is reversing effective treatment _______ ? What if the behavior change is _________________ ? Behavior traps ( e.g., social interaction, physical restrictions[ golf driving])) irreversible
MULTIPLE BASELINE two Baseline _____ behaviors concurrently Baseline one behavior in two __________ concurrently Baseline a treatment across ________ Baseline different behaviors across different ____________ Assumes behaviors are independent situations people situations
MULTIPLE BASELINE
Potential problems If behavior changes on both situations ( due to stimulus generalization across situations for example) no conclusion can be drawn regarding the independent variable The behavior may not occur in more than one situation One treated person may influence the next to be treated person Not enough observers
ALTERNATING Rx (multi-element) How to compare two different treatments for the same behavior? A multiple baseline is not well suited to do this So…we rapidly alter RX back and forth and watch for effects under each condition
ALTERNATING Rx
Multi-element used in Functional Analyses
Potential Problems interact Two conditions may _________ (extinction burst in play condition after attention condition) or due to stimulus generalization across conditions
Changing Criterion values Specific _______ of the independent variable are introduced in order E.G. number of steps that are required for reinforcement Different values of the reinforcement schedule
Issues Scientific effectiveness > # of replications = > effect < overlapping points > effect The sooner the effect The larger the effect The more precisely defined procedures The more reliable the measures The more consistent with the literature
Issues Clinical effectiveness Social Validity How important was the problem to the person? How acceptable were the procedures How satisfied is the person with the outcome
Issues (Larry’s) If we don’t do research on clinical effectiveness and only provide “clinical service” where does better treatment come from? Just because we provide only “clinical” service and don’t do research can we afford not to take reliability measures?