Alternating (2) / Multi-Element (3+) Treatments Designs

ARE USED FOR...

• Comparing effectiveness of two or more treatments

• Comparing one intervention to no intervention

• Performing a functional analysis (Iwata style)

Alternating Treatments (2) / Multi-Elements (3) Treatment Design

STRENGTHS ARE...

• no treatment withdrawal

• interventions can begin immediately

• rapid identification of effective intervention

Alternating (2) / Multi-Element (3+) Treatments Design

LIMITATIONS ARE...

• Multiple treatment interference

• limited capacity

• Need three phases for most experimental control, only in baseline + alt treatments + final phase. 

IN Alternating (2) / Multi-Element (3+) Treatments Design

EXPERIMENTAL CONTROL is demonstrated by...

• the vertical distance between data paths (greater distance = greater control)

• Need three phases for most experimental control, only in baseline + alt treatments + final phase. 

Alternating (2) / Multi-Element (3) Treatments Design

VARIATIONS ARE...

• Single Phase: primary tool for functional analysis

• Two Phase: Baseline and Comparison: stronger control

• Baseline + alternating treatments + final best treatment: helps show control and effectiveness of the best

• Baseline, comparison, and participant's choice

Changing Criterion Design

USED FOR...

• Systematic behavior change
• desensitization training
• task analysis
• Changing a measured dimensions of a single target behavior (duration, frequency, amplitude, accuracy, latency...)
• Evaluating irreversible behaviors (like learning)
• Increasing fluency

Changing Criterion Design

STRENGTHS ARE...

• No removal of treatment

• Experimental analysis while gradually improving behavior

• Stakeholders agree with

Changing Criterion Design

LIMITATIONS ARE...

• The target behavior must already be in subject's repertoire

• Can't teach new skills

• Not for shaping (because with shaping each phase is a new behavior)

• Difficult balance to choose levels that show control but don't impede learning

• Be cautious not to impose an artificial ceiling

IN Changing Criterions Design

EXPERIMENTAL CONTROL is demonstrated when...

• the behavior only changes when the criterion does. The behavior must quickly adapt to each new level.

Also -

• Control can be strengthened if in a return to a previous criterion, the behavior returns to that level.
• Stable responding evidences prediction.
• Varied length of phases increases validity and verification.
• Varied magnitude of changes shows more control.

Multiple Baseline Designs

USE FOR...

• Target behaviors that are irreversible or dangerous

• Interventions that are unethical to remove

• Shaping (multiple probe variation)

• All variations controlling for maturation 

• Multiple probe controls for history effects.

• Nonconcurrent does not control for history effects, because not in baseline together. Delayed baseline could be the same, depends on how far delayed.

It's the most widely used in ABA.

Multiple Baseline Designs

STRENGTHS ARE...

• can demonstrate control without withdrawing treatment (so stakeholders like)

• Easy to teach to implementors 

• Shows effectiveness of an intervention (less about the function of the intervention though)

Multiple Baseline Design

LIMITATIONS ARE...

• Ethics of having subsequent subjects wait for treatment 

• costly in time and money

• generalization conflict (ultimately, you want subjects to generalize, but if too soon then you lose experimental control)

Multiple Baseline Design

CONTROL IS demonstrated when...

• the behavior changes only when the IV is applied

• Need three (settings, participants, behaviors)

• data under the verification windows remains at baseline as the IV is applied to upper tier

Also - 

• verification is only if the baseline extends before and after previous tier
• Nonconcurrent MBL variation is without verification (due to absence of concurrent baseline)
• demonstration of control can be increased by varying lengths of phases or a short reversal (if ethical)

Multiple Baseline Design

VARIATIONS ARE...

• Multiple probe

• Delayed / extended baseline

• Nonconcurrent (across participants / settings / behaviors)

Multiple Baseline Design

VARIATION--

MULTIPLE PROBE IS good for...

• shaping
• when baseline is zero
• saving time and money.

Probes are only in baseline, not every day.

Right before start intervention, do a few in a row.

When IV starts, data will be daily.

Multiple Baseline Designs

VARIATION--

DELAYED / EXTENDED BASELINE IS good for...

• when a planned reversal is no longer possible
• limited resources
• new behaviors, participants, or settings are to be added to an in-progress experiment

Note that it's not common.

Multiple Baseline Design

VARIATION--

NONCONCURRENT IS good for...

• logistical constraints, inability to start all at the same time

• control for maturation, but not history 

• beginning even if all subjects or settings are not yet available 

Note that can be across participants (weakest), settings, or behaviors (strongest). Is usually just across participants.

Reversal Designs

USED FOR...

• Demonstrating functional relations

Reversal Designs

STRENGTHS ARE...

• Strong demonstration of functional relation

• Can measure behavior before, with, and with removal of intervention

Reversal Designs

LIMITATIONS ARE...

• Not appropriate for:
  • irreversible behaviors
  • irreversible interventions
  • interventions that would be unethical to remove (like for SIB / dangerous behaviors)

• Sequence effects

• Lost instructional time

Reversal Design

CONTROL IS evidenced by...

• Prediction, verification, and replication

Reversal Design

VARIATIONS--

A-B-A consists of...

• baseline, treatment, baseline (ends without treatment)

Reversal Design

VARIATIONS--

A-B-A-B / -A-B is best for...

• most powerful functional demonstration.
• (The third A-B replication could add more control evidence but also may not provide new information.)

Reversal Design

VARIATIONS--

B-A-B is useful for when...

• an intervention is to begin immediately

• an intervention is already in place

• learning a new skill

This graph is from a(n)  __________ design.

[image]

Alternating / Multi-Elements Treatments

with baseline + alternating treatments + final best

This graph is from a(n) _________ design:

[image]

Changing Criterion

This graph is from a(n) ___________ design:

 [image]

Multiple Baseline

This graph is from a(n) _________ design.

[image]

Reversal

Which designs are good for irreversible behaviors (due to learning, for example) and interventions that would be unethical to remove?

Multiple Baseline

Changing Criterion

Alternating / Multi-Element

Which design is NOT recommended for behaviors that are irreversible or treatments that are unethical to remove?

Reversal

Changing Criterion

VARIATIONS are...

• Range bound: with an upper and lower criterion

• Lag schedule: where reinforcement is contingent on the behavior varying from the previous response(s)

Which design is good for desensitization training?

Changing Criterion

Which design is good for task analysis?

Changing Criterion

Which design is best for changing a measured dimension of one behavior?

Changing Criterion

Which design is best for evaluating instruction on stepwise changes / evaluating irreversible behaviors (like learning)?

Changing Criterion

Which design is good for shaping behavior?

(Multiple Baseline)

Multiple Probe Variation

Which design is good for adding in new participants / settings / or behaviors to an in-progress experiment?

Multiple Baseline Designs: 

• Delayed / Extended Baseline

• Nonconcurrent MBL

What is the most commonly used experimental design in ABA?

Multiple Baseline

Which design controls for maturation effects?

Multiple Baseline

Nonconcurrent Across Participants and Multiple Probe

Which design can start before all participants are available?

Nonconcurrent Across Participants and Delayed Baseline

Which design is good for when the baseline is likely zero (like before learning a new skill)?

• Multiple Baseline: Multiple Probe

• Alternating Treatment

Which designs do stakeholders tend to prefer and why?

Alternating / Multi-element treatments

Changing Criterion

Because these don't require the removal of the (effective) intervention.

Which design is best for demonstrating functional relation?

Reversal

A-B-A-B or A-B-A-B-A-B is strongest.

Which design is used for functional analysis?

Multi-Element Treatment

Which designs are good for when an intervention needs to start immediately?

Reversal - B-A-B

Alternating / Multi-Element

Which designs are best for learning / implementing new skills?

Multiple Baseline, Multiple Probe

Alternating Treatments

Reversal IF the skill is reversible 

Which designs control for history effects?

Multiple Probe and Multiple Baseline

Alternating treatments IF you have a baseline

***Reversal A-B-A-B is strongest, the BEST***

Changing Criterion

ABA not too concerned with history effects because usually working within participants. 

Which design is this the definition for?

"Three A-B designs

evaluating the same DV and IV

conducted sequentially

with different lengths of baseline

stacked on top of one another"

Nonconcurrent Multiple Baseline

What are the general baseline logic steps (in order) to evaluate experimental control?

Important Note: these elements have variations in different designs.

Evaluating the LEVELS, VARIABILITY, TRENDS, overlapping data points, and step...

1. Prediction

• Once a stable responding baseline is established, it should continue that way if nothing changed.
• You have prediction if you extend the estimated data path from baseline into the first IV phase and there is a distance between the predicted and actual data paths.

2. Verification:

• When the IV is removed, the DV should return to baseline levels.
• You have verification if the DV data in the second baseline phase (after first intervention) are at comparable levels to the data in the first baseline phase

3. Replication: 

• The IV is reintroduced, and DV data should replicate the data level from the first IV phase. 

What is the difference between conditions and phases?

Conditions are treatments.

Phases may be baseline versus intervention or new medication or things going on outside the study.

In a delayed multiple baseline design, if the baseline doesn't span before and after the intervention in the previous tier, then you can have replication but not verification.

True or False?

TRUE

Which design is most subject to multiple treatment interference?

Alternating Treatments is most common for

Also possible in A-B-C design

Which design is most subject to generalization conflict?

Multiple Baseline

Which design is best to quickly find an effective intervention?

Alternating (2) / Multi-Element (3) Treatments

Which design is most subject to sequence effects?

The A-B-C type Reversal 

To strengthen the experimental control of a Multiple Baseline Design...

• Can perform a short reversal if ethical, but overall very solid design "workhorse" of ABA.

To strengthen the experimental control of a Changing Criterion Design...

• Vary length of phases

• vary magnitude of changes

• return to a previous criterion level (the behavior should follow it)