Section 5. Experimental Design

2 Types of Errors in Evaluating ABA Research

Errors may be made when conducting ABA research. There are two main types of errors: 1. Type I error (AKA: False Positive): Assuming the IV affected the DV, when it actually did not do so. -Statistical analysis tends to lead to more Type I errors. 2. Type II error (AKA: False Negative): Assuming the IV did not affect the DV, when it actually did. -Visual analysis used in ABA studies tend to lead to more Type II errors.

Ethical Issues About Reversals

Ethical Warning: If your client is displaying severe and dangerous behaviors (e.g., SIB, elopement), then do not spend time just taking baseline data from the start. It is your ethical responsibility to get in there and immediately provide treatment for the health and safety of your client. In such circumstances, you can use a B-A-B reversal.

Changing Criterion Design

Experimental design in which an initial baseline phase is followed by a series of treatment phases consisting of successive and gradually changing criteria for reinforcement or punishment. There is only one behavior in this design. Behavior in this design has to already be in the subject's repertoire. Evaluates treatment that is applied in a graduated or step-wise fashion. Technically, it is a variation of the multiple baseline design. Ex: A researcher may use this design to assess how a person's behavior changes when the researcher provides the person with reinforcement contingent upon 10 responses per minute, then 20 responses per minute, then 30 responses per minute, and so on.

1. Prediction (Part of Baseline Logic)

In the graph, the section in the box is the "Prediction" part. -Solid data points= Actual measures of behavior -Open data points= Predicted level of responding based on the solid data points if the environment remains constant. The anticipated outcome of a presently unknown measurement. Data should be collected until stability is clear. The more data points, the better the predictive power. There is no "magic number" of data points. Main question: Are data stable enough to serve as the basis for experimental comparison?

Why does ABA have a problem with traditional psychology's group approach to research?

In traditional psychology's group comparison approach, they randomly select a pool of subjects from the relevant population, divide the subjects into experimental and control groups, pretest, apply the IV to the experimental group, and posttest. ABA has three problems with this approach: 1. Group data not representative of individual performance. -Individuals within a group could stay the same or decrease, while the improvement of others could make it appear as overall average improvement. 2. Group Data Masks Variability -Hides variability that occurs within and between subjects. -Statistical control is not a substitute for experimental control. -To control effects of any variable, must either hold it constant or manipulate it as an IV. 3. Absence of Intrasubject Replication: -Power of replicating effects with individual is lost.

Affirmation of the Consequent

Inductive Logic: -If the IV were not applied, the behavior (as indicated by baseline data) would not change. -The experimenter predicts the IV will change the behavior. -If the IV is controlling the DV (A), then the data path in the presence of the IV will show that the DV has changed (B). -When the IV is present, data show DV has changed (B is true). -Thus, the IV is controlling the DV (thus, A is true).

b. Delayed Multiple Baseline Design

Initial baseline and intervention begin and subsequent baselines are added in a delayed or staggered fashion. Effective when (1) reversal design is not possible, (2) limited resources preclude a full-scale design, and (3) when a new behavior, subject, or setting becomes unavailable. Limitations: Shorter baselines do not show interdependence of DVs.

Reversal Design

(AKA: A-B-A-B; B-A-B) Any experimental design in which the researcher reverses responding to a level obtained in a previous condition. Encompasses experimental designs in which the IV is withdrawn (A-B-A-B) or reversed in its focus (e.g., DRI/DRA). Alternation between baseline and a particular intervention. Each reversal in a reversal design strengthens experimental control. Evidence of a functional relation is strengthened with each reversal (e.g., switch from one condition to the other with a corresponding change in trend and level). For a reversal to occur the behavior must approximate the initial baseline level. Requires at least 3 consecutive phases: 1. Initial Baseline (A) 2. Intervention (B) 3. Return to Baseline (A) A-B-A-B preferred over A-B-A as stronger design. Most powerful within-subject design for demonstrating function.

Sequence Effects

(AKA: Carryover Effects, Alteration Effects) Effects on a subject's behavior in a given condition that are the result of the subject's experience with a prior condition. Ex: You are using a A-B-C-B-C design. You cannot analyze the effects of intervention C on its own, as C is always preceded by intervention B. C is affected by the sequence of B coming before it. In order to properly evaluate the effects of C, you may then decide to add on a few more phases to your design. To your A-B-C-B design, you may add on A-C-A-C. This way you can really properly assess C on its own.

2. At Least 1 Behavior (Component of Experiments in ABA)

(AKA: Dependent Variable, DV) In some studies, more than 1 DV is measured. Reasons for multiple dependent variables include the following: -Provide data patterns that can serve as controls for evaluating and replicating the effects of IV. -Assess if any collateral effects occurred. --Collateral Effect= A phenomenon in which the IV effects behaviors other than the targeted behavior. --Determine whether changes in the behavior of a person other than the subject occur during the course of an experiment and if such changes can explain changes in the subject's behavior.

Confounding Variables

(AKA: Extraneous Variables, Unrelated Variables) (Hint: E = Extraneous and Environment) Generally, these terms are used as AKAs to refer to variables that exert an uncontrolled influence on a research study. However, sometimes there is a distinction between the terms, "confounding variable" and "extraneous variable." Extraneous Variable: Any aspect of the environment that must be held constant to prevent unplanned environmental variation. Ex: Lighting, space, temperature of room. Confounding Variables: Any uncontrolled factor known or suspected to exert influence on the dependent variable. Ex: Suppose a researcher wants to analyze the effects of guided lecture notes on high school biology students' learning as measured by their scores on next-day quizzes. One potential confounding variable that the researcher would need to take into account would be the student's changing level of interest in and background knowledge about the specific curriculum content (e.g., a student's high score on a quiz following a lecture on sea life may be due to his prior knowledge about fishing and not the guided notes provided during the lecture). -The effects of all these variables should be reduced or eliminated as much as possible in order to demonstrate experimental control.

Experimental Control

(AKA: Functional Relation, Analysis, Control) When a predictable change in behavior (i.e., dependent variable or DV) can be reliably produced by the systematic manipulation of some aspect of the individual's environment (i.e., independent variable or IV). The analysis dimension of the 7 dimension of ABA.

4. At Least 1 Treatment (Component of Experiments in ABA)

(AKA: Independent variable (IV), Intervention, Experimental Variable) The particular aspect of the environment that the experimenter manipulates to find out whether it affects the subject's behavior.

Treatment Integrity

(AKA: Procedural Fidelity; Fidelity of Implementation; Program Integrity. Extent to which the IV is implemented or carried out as planned. Low treatment integrity: Very difficult to interpret experimental results. Treatment Drift: When application of the IV in later phases differs from the original application. How to Ensure a High Level of Treatment Integrity: -Precise operational definition of treatment procedures. -Simplify, standardize, and automate, as simple treatments are more likely to be consistently delivered and simple, easy-to-implement techniques are more likely to be used and socially validated. -Training and practice for individuals who will conduct the experimental sessions (e.g., providing a detailed script, verbal instructions, etc.) Assessing Treatment Integrity: -Collect treatment integrity data to measure how the actual implementation of the conditions matches the written methods. -Observation and calibration give the researcher the ongoing ability to use retraining and practice to ensure high treatment integirty. -Reduce, eliminate, or identify the influence of potential confounding variables.

Alternating Treatment Design

(AKA: Simultaneous Treatments Design; Concurrent Schedules Design; Alternative Treatments Design; Multi-Element Baseline Design; Multi-Element Design; Multiple Schedules Design) (Acronym: SCAMMM) An experimental design in which two or more conditions are presented in rapidly alternating succession independent of the level of responding and the differential effects on the target behavior are noted. Compared 2 or more IVs to one another to see which IV would be best to utilize with a client. Based on stimulus discrimination (each IV has an obvious SD signaling which IV is in effect at any given time). For each IV, data are plotted separately on the same graph. IVs may be: -Alternated across daily sessions. -Given in sessions occurring in the same day. -Implemented during each portion of the same session.

1. At Least 1 Subject (Single-Subject Design) (Component of Experiments in ABA)

(AKA: Single-Case Designs, Within-Subject Designs, Intra-Subject Designs) ABA uses single-subject designs. This does not mean there is only one subject (although sometimes there is only one individual). Called single-subject because the subject acts as his/her own control. This means repeated measures of the subject's behavior during each phase of the study provide the basis for comparing experimental variables as they are presented or withdrawn in subsequent conditions (i.e., the presence and absence of the IV). The individual is exposed to each condition several times over the course of a study. ABA studies usually involve more than one subject (4-8 is common). Each subject's data are graphed separately. ABA does not use group comparison designs that are traditionally used in psychology that have a large number of subjects. Group designs mask individual progress.

Steady State Responding

(AKA: Stable State Responding) A pattern of responding that exhibits very little variation in its measured dimensional quantities over a period of time. Provides the basic for baseline logic.

4 Patterns of Baseline Data

(Acronym: DAVS) 1. Descending Baseline 2. Ascending Baseline 3. Variable Baseline 4. Stable Baseline

3 Problems Avoided by Alternating Treatments Design

(Acronym: ISU) 1. Irreversibility 2. Sequence Effects 3. Unstable Data

5 Main Experimental Designs

(Acronym: MCRAW) 1. Multiple Baseline 2. Changing Criterion 3. Reversal 4. Alternating Treatments 5. Withdrawal

3 Parts of Baseline Logic

(Acronym: PVR) 1. Prediction 2. Verification 3. Replication

External Validity

(Hint: External Validity is Generalizable to the External World) Degree to which a study's results are generalizable to other subjects, settings, and/or behaviors. Degree to which a functional relation discovered in a study will hold under different conditions. External validity is on a spectrum ranging from a little to a lot. Replication establishes external validity. 2 Major Types of Scientific Replication Methods Used in ABA: 1. Direct Replication: -Researcher exactly duplicates a previous study. -Intrasubject direct replication = same subject used. -Intersubject direct replication = different subject used. 2. Systematic Replication -Researcher purposefully varies one or more aspects of an earlier experiment. -Demonstrates reliability and external validity by showing the same effect can occur under different conditions. -ABA research generally uses systematic replication.

Internal Validity

(Hint: IV (Internal Validity) = IV (Independent Variable) (Acronym: MISS) The extent to which an experiment shows convincingly that changes in behavior are a function of the IV and not the result of uncontrolled or unknown variable. An internally valid study involves only one IV at a time. Multiple IVs are not confounded (i.e., presented at the same time). This is the best way to see the effect of the IV on the DV. High internal validity = Designs showing strong experimental control. Four Confounding Threats to Internal Validity: 1. Measurement Controls: -Refers to the number and intricacy of the behaviors you are targeting. If you are targeting numerous complicated behaviors, your internal validity may be affected. -Measurement confounds may occur due to: --Observer Drift: When observers unknowingly alter the way they apply a measurement system. --Reactivity: This can refer to the behavior of our clients changing when observed. It can also refer to observers being affected by their data being monitored. --Observer Bias (i.e., expectations): The observer's expectations that change follow in a particular direction. How to reduce observer bias = keep observer naive to expected outcomes of a study. 2. IV Confounds -IVs are complicated and given together usually in a treatment package. Ex: When giving someone money as a reinforcer, the person giving is also providing attention in addition to the money. Thus, it is hard to say if the money or attention (or a combination) is the maintaining reinforcer. -How to reduce IV confounds: Placebo control or double-blind control procedures (in which subject is not aware if the IV is present or not). 3. Subject Confounds: -Maturation: Changes in subject over course of study. -Repeated measurement detects uncontrolled variables. 4. Setting Confounds: -Studies in natural settings are more prone to confounding variables than in controlled laboratories. -You should hold all possible aspects of the study constant until repeated measurements again reveal stable responding. -Bootleg Reinforcement (secretive reinforcement that is not part of your behavior plan) may also occur in the natural environment.

Prediction, Verification, and Replication (PVR) in the Alternating Treatments Design

(How to Demonstrate Functional Relations w.This Design) On Graphs: -Visual inspection of the differences between or among the data paths produced by each treatment. -Functional relation shown when: one data path is consistently higher than the other; no overlapping data paths. -The degree of differential effects produced by two different treatments is determined by the vertical distance between the respective data paths. Prediction, Replication, Verification: -Not identified in separate phases of the design. -Each successive data point in treatment plays all 3 roles.

Prediction, Verification, and Replication (PVR) in the Multiple Baseline Design

(How to demonstrate functional relations with this design) A functional relation requires a change in behavior with the onset of the intervention. -Apply IV to Behavior 1 when you can confidently predict that the behavior would remain the same in constant conditions. -If Behaviors 2 and 3 remain unchanged after the application of the IV to Behavior 1, this verifies the prediction. -If the IV changes Behavior 2 like it did Behavior 1, the effect of the IV has been replicated. -The more replications, the more convincing the demonstration. -Most commonly 3-5 tiers.

Prediction, Verification, and Replication (PVR) in the Reversal Design

(How to demonstrate functional relations with this design) Involves prediction, verification, and replication. The IV is responsible for behavior change if repetition of baseline and treatment phases approximate the original phases.

6 Components of Experiments in ABA

1. At least one subject 2. At least one behavior (DV) 3. At least one setting 4. At least one treatment (IV) 5. A measurement system and ongoing analysis of data 6. An experimental design

Behavior: 4 Important Elements

1. Behavior is Individual: -Behavior is defined as a person's interaction with the environment. -Groups of people do not behave. -Experimental strategy of ABA is based on single-subject methods of analysis; not large groups. 2. Behavior is Continuous: -Behavior changes over time (i.e., it is not a static event). -Thus, it requires continuous measurement over time. 3. Behavior is Determined: -The occurrence of any event is determined by the functional relations it holds to other events. -Behavior is a natural phenomenon and subject to the same natural laws as other natural phenomena. 4. Behavior variability is Extrinsic to the organism: -Variability (i.e., change in behavior) is the result of the environment, such as: the IV under investigation, some uncontrolled aspect of experiment (e.g., another child in the student's classroom elopes from the classroom), uncontrolled factor outside of experiment (e.g., weather changes).

2 Types of Validity in Experimental Designs

1. Internal Validity 2. External Validity

5 Variations of the Reversal Design

1. Repeated Reversal 2. B-A-B Reversal 3. Multiple Treatment Design 4. NCR Reversal Technique 5. DRO/DRI/DRA Reversal Technique

Guidelines for Multiple Baseline Design

1. Select independent, yet functionally similar baselines. -Behaviors are functionally independent of one another. -Behaviors share enough similarity that they will change with the application of the same IV. -Behaviors should be of different response classes (i.e., independent). 2. Select concurrent and plausibly related multiple baselines. -Behaviors must be measured concurrently. -All relevant variables that influence one behavior must have the opportunity to influence other behaviors. 3. Do not apply the IV to the next behavior too soon. 4. Vary significantly the lengths of multiple baselines. -The more baselines differ in length, the stronger the design. 5. Intervene on the most stable baseline first.

3 Variations of Alternating Treatment Design

1. Single Phase Without Baseline; Does not require an initial baseline. 2. With Baseline: Whenever possible, baseline should be conducted, as it shows the change produced by each treatment compared to the natural level of performance without an intervention. 3. With Baseline and Final Best Treatment Phase: Most widely used.

B-A-B Reversals

A 3-phase reversal design: Phase 1: IV (B) Phase 2: IV removed (A) Phase 3: IV reintroduced (B) Weaker than the A-B-A Design because it does not enable assessment of the effects of the IV during baseline. Disadvantage: Sequence effects. -Sequence effects are a disadvantage because the level of behavior in condition A may have been influenced by the IV before it. Best design when your client displays severe and dangerous behaviors, as you do not wait to start intervention with this design. Also appropriate for when an IV is already in place and you have limited time.

After I have created a treatment package, how do I analyze the individual treatments that are part of that package?

A process called Component Analysis looks at the effects of each part of the treatment package. Conduct a component analysis to determine the effective components of an intervention package. -You want to keep those effective components going while getting rid of the ineffective components.

b. Multiple Baseline Across Settings

A single behavior is targeted in two or more different settings or conditions. After steady state baseline responding, the IV is applied to the first setting while other settings are kept in baseline. When steady state responding is reached for the first setting, then the IV is applied to the next setting.

3. Multiple Treatment Reversal

A type of reversal design that compares two or more IVs to baseline and/or to one another. You can tell you are dealing with a multiple treatment reversal when letters are added, like C and/or D, etc. Ex: A-B-A-C-A-B-A-C; A-B-C-D-A-C-A-D Disadvantage: Sequence effects

Advantages and Disadvantages of Reversal Design

Advantages: -Clear demonstration of existence or absence of a functional relation between the IV and DV. -Enables us to count the amount of behavior change. -Return to baseline tells us we needed to program for maintenance. Disadvantages: -Irreversibility. -Ethical Warning: Ethical issues, as well as social and education issues, can arise when you remove an effective IV.

Advantages and Disadvantages of Changing Criterion Design

Advantages: -Does not require reversal of improved behavior. -Enables an experimental analysis within the context of a gradually improving behavior. Disadvantages: -The target behavior must already be in the person's repertoire. -Not appropriate for analyzing the effects of a shaping program. -It is not a comparison design.

Advantages and Disadvantages of Alternating Treatment Design

Advantages: -Does not require treatment withdrawal -Speedy comparison. -Minimizes irreversibility problem. -Can be used with unstable data. -Can be used to assess generalization of effects. -Intervention can begin immediately without baseline data. Disadvantages: -Multiple treatment interference: This is always a problem with this design, as multiple treatments are going on at the same time. -Unnatural nature of rapidly alternating treatments. -Limited capacity of the design (suggested maximum comparison of four conditions, although more have been reported in research). -Selection of treatments: Should be significantly different from one another.

Advantages and Disadvantages of Multiple Baseline Design

Advantages: -Successful intervention does not have to be removed. -Evaluates generalization. -Easy to implement. Disadvantages: -Functional relationship is not directly shown in this design. -Effectiveness of the IV is demonstrated, but not information regarding the function of the target behavior. -IV may be delayed for certain behaviors, settings, or subjects. -Takes resources to implement properly.

Experimental Question

All well-planned experiments begin with this. A brief but specific statement of what the researcher wants to learn from conducting the experiment. Can be in question or statement form: -Question Form: What are the effects of the IV on the DV for what population and in what setting? -Statement Form: The purpose of the study was to see the effects of the IV on the DV.

5. DRO/DRI/DRA Reversal Technique

An experimental technique for showing the effects of reinforcement by using DRO, DRA, or DRI as a control condition instead of a baseline condition in which no reinforcement is provided. DRO: Reinforcement following any behavior other than the target behavior. DRI: Reinforcement following behavior that is physically incompatible with the target behavior. DRA: Reinforcement following an alternative behavior other than the target behavior. Allows us to examine contingent reinforcement.

4. Non-contingent Reinforcement (NCR) Reversal Technique

An experimental technique for showing the effects of reinforcement by using NCR as a control condition instead of a baseline condition in which no reinforcement is provided. Allows us to examine contingent reinforcement. The reinforcer is presented on a fixed or variable time schedule independent of the subject's behavior.

a. Multiple Probe Design

Analyzes relation between the IV and acquisition of skill sequences. Instead of simultaneous baselines, probes provide the basis for determining if behavior change has occurred prior to intervention.

3. At Least 1 Setting (Component of Experiments in ABA)

Control 2 sets of environmental variables to demostrate experimental control: 1. IV (present, withdraw, or vary its value) 2. Extraneous variables (prevent unplanned environmental variation) In laboratories we can control environments better, but in applied settings like home, schools, etc., it is harder to control the environment. When unplanned variations take place you must try to wait them out or incorporate them into the design. Repeated measures of behavior tell us whether unplanned environmental changes are of concern.

Baseline Logic

Refers to the experimental reasoning inherit in single-subject experimental designs. Entails 3 elements: 1. Prediction 2. Variation 3. Replication -Each of these elements depends on an overall experimental approach called state state strategy.

Steady State Strategy

Repeated exposure of a given subject to a given condition while trying to eliminate extraneous influences on behavior and obtaining a stable pattern of responding before introducing the next condition.

3. Replication (Part of Baseline Logic)

Replication is the essence of believability. Shows reliability of behavior change, we can make it happen again! Replication is accomplished by reintroducing the IV.

Function of Baseline Data

Serves as a control condition. Does not imply the absence of intervention. It can be the absence of a specific IV.

Changing criterion designs seem to have a lot in common with shaping, but one of the disadvantages is the changing criterion design is not appropriate for analyzing effects of a shaping program. Why?

Shaping is a behavior change strategy (not an experimental design). It is used to teach novel behaviors. A novel behavior is developed by reinforcing responses that meet a gradually changing criterion (which are successive approximations) towards the terminal behavior. The changing response criterion in shaping are topographical in nature, requiring different forms of behavior at each new level. The Changing Criterion Design is an experimental design that results in behavior change. With the changing criterion design you cannot use it with a novel behavior. the behavior you choose to use in this design must already be in your client's repertoire (unlike shaping). The changing criterion design is best for evaluating the effects of instructional techniques on step-wise changes in rate, accuracy, duration, or latency of a single target behavior.

1. Descending Baseline (Pattern of Baseline Data)

Shows the behavior is already changing. Generally, one should not implement IV when baseline is descending. But you can do so if the behavior you are trying to change is something you want to increase (e.g., a functional skill) and the descending trend shows it is worsening. Ex: If the graph attached is a client's toileting skills, one should not wait until the data is stable to implement the IV because the client is losing their critical toileting skills. -If descending baseline is due to a behavior you want to decrease you should wait because the behavior is already improving.

2. Ascending Baseline (Pattern of Baseline Data)

Shows the behavior is already changing. Generally, one should not implement the IV when baseline is descending. But you can do so if the behavior you are trying to change is something you want to increase (e.g., a functional skill) and the descending trend shows it is worsening. Ex: If the graph attached is a client's toileting skills, one should not wait until the data is stable to implement the IV because the client is losing their critical toileting skills. -If descending baseline is due to a behavior you want to decrease you should wait because the behavior is already improving.

1. Repeated Reversals

Simple extension of A-B-A-B. Ex: A-B-A-B-A-B-A-B The more reversals, the stronger your evidence of control. Redundancy may be a conern.

Irreversibility

The level of behavior observed in an earlier phase cannot be reproduced even though experimental conditions are the same as they were during the earlier phase. Ex: How to ride a bike is something that, once you have learned it, you will never know how to do it. It is irreversible. -When irreversibility is a problem, use DRO/DRI/DRA conditions as control techniques or multiple baseline designs.

6. Experimental Design

The particular arrangement of conditions in a study so that meaningful comparisons of the effects of the presence, absence, or different values of the IV can be made. 2 Types of Experimental Designs: 1. Nonparametric Analysis: IV either present or absent during study. Ex: Medication is either given and taken away in the course of a study. 2. Parametric Analysis: The value of the IV is manipulated. Seeks to discover the differential effects of a range of values. Ex: Various doses of medication are given in the course of a study. Important Rules of Experimental Design: Change only one variable at a time: -If examining a treatment package, ensure that that the entire package is presented or withdrawn at the same time. A treatment package (AKA: Behavioral Package) is when multiple IVs are bundled into one program such as a token economy with praise and time-out. Do not get locked into textbook designs. -Select and combine designs that best fit the research question.

Other Larger Concerns about Single-Case Designs

The range of questions about intervention effects that can be addressed with these designs: -Single-subject designs: best for treatment package evaluation The generality of the research results: -Is the finding generalized beyond the subject in the deesign? -To assess generality, use replication of your IV across subjects, etc.

2. Verification (Part of Baseline Logic)

Verification of a previously predicted level of baseline responding by termination or withdrawal of the treatment variable.

How to Identify Practical and Ethical Considerations in Single-Case Experimental Designs to Demonstrate Treatment Effectiveness

Your #1 goal when using single-case designs is to clearly show that your IV changed the target behavior and nothing else. However, sometimes there is a lack of clarity about our treatment's effectiveness. It is your job to identify the practical and ethical issues about single-case designs showing treatment effectiveness. The 3 practical and ethical issues are described below: 1. Baseline Trends: a. Increasing or decreasing trends in your data during baseline data collection do not allow you to clearly demonstrate that your IV caused the change in behavior. b. How to Address: i. Continue observations for a longer period of time. Ii. Try to reverse the trend (e.g., using a DRO schedule of reinforcement). iii. Select designs that do not require a stable baseline. iv. Use statistical techniques that take initial trends into account. 2. Excessive Variability in Data a. Variability in your data can obscure intervention effects. b. How to Address: i. Block consecutive data points and plot blocked averages rather than day-to-day performance. ii. Search for causes of the variability or the situation (e.g., variation among the environmental stimuli). 3. Duration of Phases: a. The duration of each phase in your design can involve problems related to trends and variability in the data. b. How many data points at minimum do you need in each phase of your design. i. No rigid rules about this because the clarity and utility of a set of observations is a function of the data pattern in adjacent phases. c. Objective criteria: Use this for deciding when to shift phases. This helps reduce any subjectivity you may have about when to shift phases.

2 Variations of Multiple Baseline Design

a. Multiple Probe Design b. Delayed Multiple Baseline Design Both inherently weaker than traditional multiple baselines. Use these when extended baseline measurement is unnecessary, impractical, too costly, or unavailable.

When we see variability in our data, what should we do?

Behavior analysts should attempt to experimentally manipulate factors suspected of causing the variability in the data to look for causal factors. In practice, behavior analysts seek treatment variables robust enough to overcome variability.

a. Multiple Baseline Across Behaviors

Two or more different behaviors of the same subject. Each subject serves as his/her own control. After steady state baseline responding, the IV is applied to the first behavior while other behaviors are kept in baseline. When steady state responding is reached for the first behavior, then the IV is applied to next behavior.

Multiple Baseline Design

Most widely used design. Highly flexible. Staggered implementation of the intervention in a step-wise fashion across behaviors, settings, and subjects. Do not have to withdraw a treatment variable in this design. Ethical Warning: When it is unethical or impractical to reverse conditions or when the behavior is irreversible use this design instead of a reversal design.

3. Variable Baseline (Pattern of Baseline Data)

No clear trend. If one's data is variable, wait it out and do not introduce the IV. Variability is assumed to be due to the environmental variables that are uncontrolled. If you introduce the IV now you will not be able to tell if it changed the behavior or not. You should try to control uncontrolled sources of variability.

4. Stable Baseline (Pattern of Baseline Data)

No evidence of ascending or descending trend. All of the values of the DV fall in a small range of values. Best way to look at the effects of the IV on the DV. You can introduce the IV now.

5. Measurement System and Ongoing Analysis of Data (Component of Experiments in ABA)

Observation and recording procedures must be conducted in an standardized manner. Standardization involves every aspect of the measurement system (e.g., from the behavior definition to scheduling of observations). Behaviorists must detect changes in level, trend, and variability.

c. Multiple Baseline Across Subjects

One target behavior for two or more subjects in the same setting. After steady state baseline responding, the IV is applied to the first subject, while other subjects are kept in baseline. When steady state responding is reached for the first subject, then the IV is applied to the next subject. Most widely used multiple baseline design. The graph attached is from a research article that used multiple baseline across subjects. If you get an exam question about it, you should be able to read and recognize it as such.

Withdrawal Design

Some authors use the term "withdrawal design" to describe experiments based on A-B-A-B analysis and reserve the term "reversal design" for studies in which the behavioral focus of the treatment variable is reversed (or switched to another behavior), as in the DRO/DRI/DRA reversal techniques.

Prediction, Verification, and Replication (PVR) in the Changing Criterion Design

The criterion lines should have a large separation to show a functional relationship. Experimental control is evidenced by the extent that the level of responding changes to conform to each new criterion. If data points do not fall around the criterion lines, that shows that there is very little experimental control. The greater the vertical distance between the criterion lines, the more experimental control.

Ethical Issues Regarding Experimental Designs and Research

There are lots of ethical issues regarding how behavior analysts should conduct research designs addressed in the Professional and Ethical Compliance Code for Behavior Analysts (the Code). See Ethics Section for all of them.

Guidelines for Changing Criterion Design

Three Parts: 1. Length of Phases: -Each phase must be long enough to achieve stable responding. -Target behaviors that are slower to change require longer phases. -Validity of the design is increased when you vary the length of each phase. 2. Magnitude of Criterion Changes -The size of the changes between each criterion should vary to prove strong functional relations. -Changes in size must be large enough to be detectable, but no so large as to be unachievable. -Changes in size can be smaller if you are dealing with stable data. 3. Number of Criterion Changes -The more criterion changes the better proof of experimental control.

Benefits of Baseline Data

To use the subject's performance in the absence of the IV as an objective basis for detecting change. To obtain descriptions of ABC correlations for the planning of an effective treatment. To guide us in setting the initial criteria for reinforcement. To see if the behavior targeted for change really warrants intervention.