The big idea
Every experiment asks the same question: when I change this, does that change too? The thing changed on purpose is the independent variable (IV). The thing measured to see whether it changes is the dependent variable (DV). In applied behavior analysis the IV is usually an intervention or one feature of it, and the DV is a measure of the target behavior.
The names describe the logic. The DV depends on the IV: if the experiment works, the DV changes as a function of what the experimenter does. The IV is independent because the experimenter, not the participant’s behavior, decides its value.
This matters well beyond research. A practitioner who cannot say what was changed and what was measured cannot say whether the plan worked.
Breaking it down
The independent variable
The IV is whatever the experimenter manipulates: a procedure, a condition, a stimulus, or a particular value of one (for example, how often reinforcement is delivered). On a graph, the IV appears as the condition labels and phase-change lines, such as Baseline and Token economy.
Single-case designs present the IV in conditions that are introduced, withdrawn or alternated (see D.7). A sound study changes only one variable at a time when moving between conditions, so any change in behavior can be pinned on that variable.
The dependent variable
The DV is the measured dimension of the target behavior: its rate, duration, latency, percentage correct or another measure (see C.3 and C.4). It is plotted on the vertical axis. The DV is a measure, not a person and not a goal. Ava is the participant; Ava’s words read correctly per minute is the DV.
Extraneous and confounding variables
Anything other than the IV that might affect the DV is an extraneous variable: time of day, a new classmate, a change in sleep. Researchers try to hold these constant. When one changes along with the IV, so that its effect cannot be separated from the IV’s, it becomes a confounding variable. Confounds are the main threat to a study’s conclusions (see D.2 and D.3).
Functional relation
When the DV changes reliably when, and only when, the IV is manipulated, the experimenter has shown a functional relation: the behavior is a function of the IV (see A.1).
Checking the IV: treatment integrity
A conclusion about the IV is only as good as the IV’s delivery. Behavior analysts therefore measure how accurately the procedure was carried out, which is called treatment or procedural integrity (see C.12). Poor integrity means the IV actually tested was not the one described.
Seeing it in practice
| Situation | Independent variable | Dependent variable |
|---|---|---|
| A teacher introduces a token economy and records assignment completion. | Token economy (in place or not) | Percentage of assignments completed |
| A clinic compares 1-minute and 5-minute breaks between work periods. | Break length | Minutes on task per session |
| A supervisor gives written feedback to technicians and counts errors in their session notes. | Written feedback | Errors per session note |
| A functional analysis alternates attention, escape, play and alone conditions. | The test condition (which consequence follows the behavior) | Rate of problem behavior |
| A behavior analyst teaches a parent to use descriptive praise and records the child’s compliance. | The parent’s descriptive praise | Percentage of instructions the child follows |
The same behavior can be either variable, depending on the question. In the last row, the parent’s praise is the IV. If the question were whether parent training worked, the parent’s praise would be the DV and the training would be the IV.
Common mix-ups
- The DV is a measure of behavior, not the person. The participant is not a variable.
- Participant characteristics are not IVs in single-case research. Age, diagnosis or gender cannot be manipulated. Group studies sometimes compare people on such characteristics, but those comparisons are correlational, not experimental.
- The IV is not only “the intervention.” Any manipulated variable counts, including a value of a variable (such as reinforcer size) or the presence of a stimulus.
- Staff behavior can be the DV. In supervision and training studies, the trainee’s performance is often the DV and the training procedure is the IV.
- Changing two things at once confounds the IV. If a new reinforcer and a new teacher arrive together, the change cannot be credited to either one alone.
Exam tip: Ask two questions of any scenario. What did someone deliberately change? That is the IV. What did someone count, time or score to see the effect? That is the DV. If something else changed at the same moment as the IV, look for a confound.
Quick recap
- The IV is manipulated by the experimenter; the DV is measured.
- The DV is a dimension of behavior plotted on the vertical axis. The IV appears as the conditions or phases.
- Extraneous variables that change along with the IV become confounds.
- A reliable change in the DV produced by manipulating the IV is a functional relation.
- Treatment integrity checks that the IV was delivered as planned.
Further reading
- Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.), chapter 7.
Key terms
- Independent variable (IV)
- The variable that the experimenter deliberately manipulates to see whether it affects behavior. In applied behavior analysis it is usually an intervention, a condition or a particular value of a procedure.
- Dependent variable (DV)
- The variable that is measured to see whether it changes when the independent variable is manipulated. In applied behavior analysis it is a measured dimension of the target behavior, such as its rate, duration or latency, plotted on the vertical axis of a graph.
- Confounding variable (Confound)
- An uncontrolled variable that changes along with the independent variable and may affect the dependent variable, so that its effect cannot be separated from the effect of the independent variable.
- Functional relation
- An experimentally demonstrated relation in which a specific change in the dependent variable (behavior) can be reliably produced by specific manipulations of the independent variable, and the change is unlikely to be the result of confounding variables.
Check yourself
5 questions on D.1. Choose an answer to see whether you're right and why.
Show the answer
Answer: B. The dependent variable is the measured dimension of behavior, here the latency from cue to starting. The visual schedule is what the teacher manipulates, so it is the independent variable. Kai is the participant, not a variable.
Show the answer
Answer: A. Video modeling is what the supervisor introduces and controls. The technicians' accuracy is what the supervisor measures, so it is the dependent variable. In training studies, staff behavior is often the dependent variable.
Show the answer
Answer: B. The medication changed at the same time as the procedure, so its effect cannot be separated from the effect of DRA. That makes it a confounding variable. It is not the independent variable, because the analyst did not manipulate it as part of the study.
Show the answer
Answer: B. An independent variable must be something the experimenter can deliberately change. A diagnosis is a characteristic of the person, and it cannot be introduced and withdrawn. Comparisons based on such characteristics are correlational, not experimental.
Show the answer
Answer: B. The vertical axis shows the measure of behavior, which is the dependent variable. The condition labels and phase-change lines show the independent variable, and the horizontal axis shows time, usually sessions or days.
Adds a step to your study record. Saved in this browser only.