The AB Design “interrupted time-series design.”
Measures on a sample or a series of samples from the same population are obtained several times before and after a manipulated event or a naturally occurring event (considered a type of quasi-experiment)
Its main disadvantage its inability to distinguish the experimental effect from possible confounds that might occur coincident with the change of condition.
The BA design
A treatment is withdrawn to determine its effectiveness.
The ABA or Reversal Design (The BAB design)
The effect of the experimental variable is studied by repeatedly introducing and withdrawing the experimental variable.
The main disadvantage of the ABA design is its inability to be used with variables with irreversible effects or when it is undesirable to return to a baseline condition for practical or ethical reasons.
A-BC-A-B-A-C-AThe Multiple Sequential Withdrawal Design is a variation on the reversal design.
It is used to separate the effects of each component of a compound independent variable. The components are presented individually and in combinations to determine their effects. Any number of components can be examined.
The main disadvantages of the multiple sequential withdrawal design is the inability of the design to deal with variables with irreversible effects, the length of time required to examine several components, and difficulty interpreting order and interaction effects.
Multiple Baseline Designs
The Multiple Baseline Design
is used when a return to baseline is undesirable. Experimental control is demonstrated by the repeated changes in the dependent variable with each successive introduction of the independent variable. It protects against the inability of AB – type designs to demonstrate unequivocal control by the independent variable by showing the effects at different times. It is unlikely that a confound could repeatedly coincide with the introduction of the experimental variable. The effects of a confounding variable would be seen as a change in the behavior in all three baselines at the same time. The multiple baseline can be used across different subjects, settings, or responses. A standard variation of the design is the Staggered-Start Design in which a series of AB designs with equally long baselines are started at successive, overlapping times. If the target behavior is not irreversible, combining the multiple-baseline and reversal designs produces a highly effective demonstration of experimental control with multiple subjects, settings, or responses. The main disadvantage of the multiple baseline design is that a high degree of planning is required to produce a successful implementation.
Multiple baseline designs actually refer to the lagged introduction of the treatment condition across subjects, conditions (settings), or behaviors. The reason for the name is that each of these levels of manipulation has a baseline with typically at least three different replications across subjects, conditions, or behaviors (see Figure 6). This design is particularly useful when it is not possible to remove a treatment (either because it resulted in a skill, which could not return to baseline conditions to show the relative effects of the treatment, or because of ethical reasons). In lagging the treatment and determining whether the levels (or rates) do not change until the treatment is introduced, this design provides partial validation of the treatment as the cause of any behavior changes.
Figure 6. Graphic Depiction of Multiple Baseline Design
In addition to the requirement of both comparability and independence across the subjects, conditions (settings), or behaviors, this design assumes that delayed access to treatments in later baselines is not a problem. For some subjects or with some conditions and behaviors, an extended baseline needs to be conducted in such a manner that other factors do not enter into the outcomes (like frustration, fatigue, or vigilance on the part of the subject). In addition, an extended baseline also implies comparability in the conditions in which no changes occur in the presence of the treatment or the collection of the outcome data (instrumentation and testing) during the entire phase; otherwise, any of the previously noted threats to validity may be present.
A multiple baseline design across settings may be functional in sorting out various scheduling accommodations in different subject areas having a common response demand. For example, in many discipline-specific tests (math, science, and social sciences), students may be given a problem to solve in which they write their answers. For students with individual needs in writing who cannot write as proficiently or fluently as other students, it may be important to break the test sessions up into smaller time periods (e.g., three 15-minute periods instead of one 45-minute period). To test this accommodation, a student would take the test in two areas (e.g., geography and economics) initially using the standard time twice each week for two weeks, then take the test daily in 10-minute periods in geography for two weeks while the economics problem-solving task is completed using the standard time. After two weeks, this test in economics is then administered in five 10-minute periods for two weeks. As can be seen from this example, the problem with this design is the extended baseline (in economics) for four weeks. Another problem is the sheer length of time overall in this example, a total of six weeks.
Multiple Probe Designs
In multiple probe designs, the baseline condition is prolonged and only sample probes are taken to ascertain the levels (rates) of behavior (see Figure 7). The major reason to use this design is that an extended baseline may be completely unnecessary once it is documented that the performance levels are low. It is important to both establish this low performance level initially and then again just before the intervention is implemented. If multiple probes are taken just before the intervention only (when both baseline and treatment conditions are being implemented across the subjects, conditions, or behaviors), the data display is less convincing in documenting that the changes are concurrent with the introduction of the treatment. Levels of performance are never available for comparing subjects, conditions, or behaviors under a common baseline condition.
Figure 7. Graphic Depiction of a Multiple Probe Design
The Changing Criterion Design
is used to determine the effects of an independent variable when the final version of the target behavior cannot be emitted initially. Experimental control is demonstrated by the repeated changes in the dependent measure as the criterion is changed. The steps in the changing criterion design must be large enough to clearly show the effects of the independent variable, but not so large that the subject cannot meet the changed criterion.
The critical element of changing criterion designs is the systematic introduction of a criterion level of performance over successive phases so that the behavior is essentially shaped into a final level, with each change in behavior occurring concurrent with the change in criterion (see Figure 8). Experimental control is established by the simultaneous co-occurrence of both. In this design, successive levels of the criterion are changed only upon attainment of previous levels.
Figure 8. Graphic Depiction of a Changing Criterion Design: Increase Only
The following example illustrates the use of a changing criteria design. For a student with an attention deficit, poor performance may be a function of not attending to the problems and working only in brief periods. If the student is trained to remain attentive to a read aloud condition using a specific reinforcement schedule, then a test may be more appropriately used to assess academic skill (possibly math or other content areas in which reading should not limit performance). In this example, note that the accommodation includes both a behavioral skill (attending) as well an access skill (e.g., reading). With a changing criterion design, this accommodation may be investigated by successively increasing the length of time in which eye contact is made with the person reading the test. To exhibit experimental control, the researcher would systematically manipulate the reinforcement after different amounts of time. Concurrently, performance would be tracked on an outcome measure (e.g., math test) to determine if attending has an influence.
In comparative designs, different treatments are considered using any one of several strategies.
The multi-treatment Design
is used to directly compare the effects of two or more different experimental variables across the same span of time in the same subject. It is also known as the multiple schedule design.It is also highly effective in controlling for systematic changes in the subject or setting across time. Its main disadvantages are (1) its inability to deal with irreversible effects, (2) potential generalization from one condition to the other, and (3) interpretation problems due a variety of interaction, carryover, and order effects.
For example in a multi-treatment design, successive phases highlight the distinctions between a variety of treatments and baseline (e.g., A-B-A-C-A-B-C or A-B-A-C). When making comparisons across successive phases, it is important to consider the sequence so that each phase is preceded and followed by every other phase in a balanced manner. In the example noted parenthetically, the alternate phases of B, C, and BC all follow a baseline, however, C and BC are confounded by being sequenced in that particular order. In some instances, this building of treatments in a sequence is unavoidable. In Figure 9, the sequence is depicted with two different treatments sequenced after each baseline in a multi-treatment design.
Figure 9. Graphic Depiction of a Multi-treatment Design
Alternating Treatment Design
In contrast, in the alternating treatment or multi-element design, the treatment and control conditions are presented in a random order (counterbalanced within a session) so that successive days (or sessions within days) contain an unordered sequence (see Figure 10). This design relies on stimulus discrimination, allowing the subject to identify the conditions, and depends on a treatment being readily implemented and removed in a more quick fashion than a reversal-withdrawal design.
Figure 10. Graphic Depiction of an Alternating Treatment Design
An accommodation study may involve the use of an assistive device (e.g., a calculator or Franklin Speller) during the first accommodation and then a prompt to use the device in the second accommodation intervention; or in many instances, both a social skill and an administration assist may be needed and in that order. For example, a student may need to first be taught to work independently for 30 minutes using a reinforcement system and then a response accommodation might be implemented with various kinds of assistive devices (e.g., special computer keyboard).
Using another example, a student may be assisted in a reading test by having two forms of the test administered: (a) one form includes the passage written with one sentence per line, multiple choice items listed below, and a bubble sheet; and (b) one form has the passage presented in a standard form but allows a student to mark the multiple choice items in the booklet instead of on the bubble sheet. To use an alternating treatment design, a student would receive passages randomly with either of these two accommodations and directed to respond accordingly. The advantages of this design include the efficiency in comparing several treatments concurrently, the capacity to dismantle essential components of a treatment quickly, as well as the lack of reversal needed to understand the relative effects of a treatment.