Skip to content
Back to search results

Alternating Treatments Design

Introduced by
Single-case methodology tradition; design standards by Kratochwill et al.
Year
2013
Type
Within-subject experimental pipeline comparing two or more conditions by rapid alternation
Family
Single-case / single-subject experimental methodology
Font size

What does it do?

The alternating treatments design (ATD), also called the multielement design, is a single-case experimental method that rapidly alternates two or more conditions or treatments within a single individual to compare their effects directly.

Rather than running each treatment in a long block, the analyst switches between conditions in close succession — often session by session — in a counterbalanced or randomized order so that time, sequence, and setting do not favor one condition. The signature of an effect is a consistent vertical separation between the data paths of the conditions: when one treatment's outcomes reliably sit above another's across the alternation, the design attributes the difference to the treatment rather than to extraneous variables. Because comparison is built into the rapid alternation, the ATD can demonstrate differential effects quickly and without withdrawing an effective treatment to baseline. It fits squarely within the single-case design standards formalized by Kratochwill and colleagues in 2013, which treat systematic manipulation and replicated demonstrations as the basis for experimental control.

The intuition behind the method

Suppose a teacher wants to know which of two prompting strategies helps a student with a disability learn faster. Running strategy A for three weeks and then strategy B for three weeks would confound the comparison with everything else that changes over six weeks. The alternating treatments design instead interleaves them: strategy A in one session, strategy B in the next, flipping back and forth in a balanced schedule. If you plot A's sessions as one line and B's as another, and A's line consistently rides above B's no matter the order or day, the gap between the lines is hard to explain by anything except the strategies themselves. The rapid switching means the two conditions share the same time period, the same general context, and the same person, so a stable separation between their data paths is convincing evidence that one treatment genuinely outperforms the other for that individual.

History

The alternating treatments design emerged from the experimental analysis of behavior, where multiple-schedule and concurrent-schedule arrangements showed that behavior could be brought under the control of rapidly changing stimulus conditions. Applied researchers adapted this into the multielement or alternating treatments design to compare interventions efficiently within a single subject, and methodologists clarified its logic of differential effect via separation of data paths and its vulnerability to multitreatment interference. The 2013 What Works Clearinghouse single-case intervention research design standards, authored by Kratochwill and colleagues, placed the ATD within a unified framework alongside reversal, multiple-baseline, and changing-criterion designs, articulating the systematic-manipulation and three-demonstrations requirements that any single-case design — including the ATD — must satisfy to provide credible experimental evidence.

When to use it

Use the alternating treatments design when the question is comparative — which of two or more interventions works better for an individual — rather than whether a single intervention works at all. It is well suited to behaviors that respond quickly and can be measured repeatedly, to settings where conditions can be implemented in rapid succession, and to situations where withdrawing an effective treatment to baseline would be undesirable, since the ATD compares treatments directly without a return-to-baseline reversal. The design assumes that the treatments do not strongly interact or carry over between alternations; it is therefore a poor fit when one condition durably changes the learner so that subsequent sessions are contaminated, or when the behavior changes too slowly to show condition-specific differences within a rapid schedule. It also requires that conditions be discriminable enough that the participant and measurement system can keep them distinct.

Limitations and considerations

  • Vulnerable to multitreatment interference and carryover, where exposure to one condition contaminates performance in the other.
  • Requires behaviors that respond quickly; slowly changing outcomes will not separate within a rapid alternation schedule.
  • Frequent switching can confuse participants or be difficult to implement faithfully in applied settings.
  • Limited to comparing a small number of conditions, since many alternated treatments produce cluttered, hard-to-read data paths.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Directly compares two or more treatments within a single individual, answering 'which works better' rather than just 'does it work'.
  • Avoids withdrawing an effective treatment to baseline, since comparison comes from alternation rather than reversal.
  • Rapid alternation holds maturation, history, and context roughly constant across conditions, sharpening the contrast.
  • Counterbalanced or randomized scheduling controls sequence and time effects and supports randomization-test analysis.

Applications

  • Comparing two instructional or prompting strategies for a student with a disability to identify the more effective one.
  • Contrasting reinforcement schedules or antecedent interventions for reducing challenging behavior in an individual.
  • Evaluating which of two assistive technologies or communication systems yields better performance for a single user.
  • Testing treatment versus a control or attention condition within one person without a return-to-baseline reversal.
  • Quickly screening candidate interventions in clinical or educational settings before committing to a longer course.

How it works

How it works

1

The design starts by selecting two or more conditions or treatments that are meaningful to compare and that act on the same operationally defined target behavior measured the same way. The conditions might be two instructional methods, two reinforcement schedules, or a treatment versus a control condition. Because the entire logic rests on contrasting outcomes between conditions, the analyst ensures the treatments are genuinely distinct yet comparable on a shared dependent variable, and that each can be implemented cleanly without bleeding into the others. Limiting the number of conditions matters too, since alternating too many makes the schedule unwieldy and the data paths hard to discriminate.

Formula
\{C_1, C_2, \dots, C_k\},\ k \ge 2,\quad \text{conditions matched on a common target behavior } y_t
2

Next the analyst builds an alternation schedule that counterbalances or randomizes which condition occurs when, controlling for sequence and time-of-day effects. If treatment A always preceded treatment B, any difference could reflect order or fatigue rather than the treatments; counterbalancing across sessions, therapists, settings, and times of day removes these confounds by ensuring each condition is equally exposed to them. Randomizing the order also supports later randomization-test analyses. This scheduling step is where the ATD earns its internal validity, because the comparison is only credible if no extraneous variable is systematically aligned with one of the conditions.

Formula
\sigma: t \mapsto C_{\sigma(t)} \text{ randomized/counterbalanced so } \Pr(C_j \mid \text{time, order, setting}) \text{ is balanced}
3

The conditions are then implemented in rapid succession according to the schedule — often alternating each session, or even within a single session — so that all conditions are active during the same overall time window. This rapid alternation is what distinguishes the ATD from a phase-change design: instead of a long block of one condition followed by a long block of another, the treatments are intermixed closely in time. The closeness keeps maturation, history, and general context essentially constant across conditions, so that differences in the behavior can be attributed to which treatment was in effect on a given occasion rather than to slow background changes.

Formula
\dots C_1, C_2, C_2, C_1, C_2, C_1 \dots \text{ in close succession (within or across sessions)}
4

Each condition's measurements are extracted and plotted as a separate data path on the same graph, so that the outcomes for treatment A form one series and treatment B another, interleaved in time. Visualizing the conditions as distinct paths is essential because the analytic question is whether the paths separate, not whether a single series shifts between phases. Clear plotting — distinct markers and connected series for each condition — lets the analyst and reviewers judge the magnitude and consistency of any gap between treatments across the alternation, and reveals whether one condition is uniformly superior or whether the difference emerges only later.

Formula
\{y_t : \sigma(t)=C_j\} \to \text{path}_j,\quad j = 1,\dots,k \text{ plotted as distinct series}
5

Finally, the analyst infers a differential effect from a consistent vertical separation between the data paths: when one condition's path stays reliably above (or below) the other's across repeated alternations with little or no overlap, the design demonstrates that the treatments produce different outcomes for the individual. The degree of separation, its consistency, and the number of times it is replicated within the alternation correspond to the demonstrations-of-effect logic in the single-case design standards. Overlapping or crossing paths indicate no clear differential effect, while a stable, replicated gap supports a causal comparison without ever having to withdraw an effective treatment to baseline.

Formula
\text{effect} \iff \min_t \text{path}_a(t) > \max_t \text{path}_b(t)\ \text{(consistent, non-overlapping vertical separation)}

Common pitfalls

  • Failing to counterbalance order, so a difference between treatments is confounded with sequence or time of day.
  • Ignoring carryover effects and attributing a path difference to the treatments when one condition is contaminating the other.
  • Plotting the conditions as a single series instead of separate data paths, hiding the very separation the design tests.
  • Declaring an effect from one or two non-overlapping points rather than a consistent, replicated separation across the alternation.

Frequently asked questions

How does an alternating treatments design show that a treatment works?

It shows differential effect through separation of data paths rather than a phase shift. By rapidly alternating two or more conditions in a counterbalanced order and plotting each as its own series, the design reveals whether one condition's outcomes consistently sit above the other's. A stable, replicated vertical gap between the paths — with little overlap — indicates that the treatments produce different results for that individual, because the rapid alternation holds time, context, and the person constant across conditions.

What is multitreatment interference and why does it threaten this design?

Multitreatment interference is carryover, where exposure to one condition alters how the individual responds in another, so the alternated treatments no longer reflect their isolated effects. Because the ATD interleaves conditions closely in time, an effect of treatment A may persist into a treatment B session and shrink or distort the difference between paths. Researchers guard against it by choosing treatments unlikely to carry over, counterbalancing thoroughly, and, when interference is suspected, interpreting the comparison cautiously or isolating conditions to verify their separate effects.

When should I choose an alternating treatments design over a reversal (ABAB) design?

Choose the ATD when your question is comparative — which of several treatments is better — and when you do not want to withdraw an effective treatment to baseline, since the ATD compares conditions directly through alternation. Choose a reversal design when you want to demonstrate that a single intervention controls the behavior and when the behavior can ethically and practically return toward baseline. The ATD is faster for head-to-head comparisons but assumes minimal carryover, whereas the reversal design isolates one treatment's effect through repeated on-off demonstrations.

Related methods

Sources

  1. Kratochwill, T. R., Hitchcock, J. H., Horner, R. H., Levin, J. R., Odom, S. L., Rindskopf, D. M., & Shadish, W. R. (2013). Single-case intervention research design standards. Remedial and Special Education, 34(1), 26-38.

    Open DOI reference
    Copied.