Carryover effects occur when the influence of one treatment, task, or condition persists into a later phase of a study and alters the results of whatever comes next. They are one of the most common threats to validity in any experiment where the same participants are tested more than once. If you’ve encountered this term in a research methods class, a clinical trial report, or a psychology textbook, understanding carryover effects is essential for interpreting whether study results can be trusted.
How Carryover Effects Work
In many experiments, researchers test the same group of people under multiple conditions rather than assigning separate groups to each condition. This is called a within-subjects or crossover design, and it’s popular because it requires fewer participants and controls for individual differences. The tradeoff is that whatever happens in the first condition can bleed into the second.
A carryover effect is specifically the lingering influence of a previous treatment or condition that changes how a participant responds to the next one. In a drug trial, for instance, if the first medication hasn’t fully cleared the body before the second medication is given, the participant’s response to the second drug is contaminated by the residual effects of the first. The same logic applies outside of pharmacology. In a psychology experiment measuring reaction times under different lighting conditions, if the first lighting condition causes eye strain, that strain carries into the next condition and skews the data. The key feature is that the effect travels forward in time from one condition to the next.
Carryover vs. Order vs. Sequence Effects
These three terms often get confused because they all involve doing things in a particular order. They describe different problems.
- Order effects depend on position in the sequence, not on what came before. A participant might perform worse on the fourth task simply because they’re tired from doing four tasks in a row. It doesn’t matter which task was third.
- Sequence effects depend on what specific condition came immediately before. Judging the weight of an object feels different depending on whether you just held something light or something heavy.
- Carryover effects are longer-lasting changes that persist across sessions or conditions. If repeatedly lifting heavy objects over several experimental sessions makes participants physically stronger, their improved strength carries over and affects all subsequent measurements.
In practice, researchers and statisticians sometimes treat carryover and sequence effects as interchangeable because they’re difficult to separate statistically. In a simple two-period crossover design, the carryover effect is embedded within the sequence, meaning the statistical test for one is effectively the test for the other.
Physical and Psychological Sources
Carryover effects come from two broad categories: physiological and psychological.
On the physiological side, the clearest example is a drug that hasn’t left the body. Every medication has a half-life, the time it takes for half the drug to be eliminated. Even after you stop taking a medication, measurable amounts remain in your system for many half-lives. Some drugs clear quickly, while others linger for weeks or months. Certain cancer immunotherapy drugs, for example, require washout periods of seven to eight months before a patient can start a new treatment, because the drugs have half-lives of roughly 20 days and need about 10 half-lives to fully clear.
On the psychological side, the sources are more varied. Both physical and mental exertion can produce carryover effects on later performance. Prolonged cognitive effort leads to perceptions of mental fatigue that reduce not only subsequent cognitive performance but also physical performance, including strength, endurance, and motor control. Prior exertion can also negatively affect self-efficacy, mood, and motivation, all of which shape how someone performs on the next task. Practice effects work in the opposite direction: participants may improve on a task simply because they’ve done it before, not because the experimental condition helped.
Why They Threaten Study Validity
When carryover effects go undetected, they create a specific kind of bias. If a participant’s response to Treatment B is partly caused by the lingering influence of Treatment A, the researcher can’t cleanly attribute the results to Treatment B alone. This undermines the entire logic of the experiment, which depends on isolating the effect of each condition.
The problem is especially damaging for subjective measurements. Pain ratings, mood scales, and self-reported symptoms are more vulnerable to psychological carryover than objective measures like blood pressure or reaction time. A participant who felt relief during the first treatment phase may rate the second treatment more favorably simply because their expectations shifted. Researchers studying this bias have recommended validating subjective measures against objective ones whenever possible to catch these distortions.
In crossover clinical trials, where patients receive two or more treatments in sequence, undetected carryover can lead to incorrect conclusions about which treatment is more effective. If Treatment A has a strong residual effect, it can make Treatment B look better or worse than it actually is, depending on whether the residual effect enhances or interferes with the next treatment’s action.
How Researchers Prevent Carryover
The most direct strategy is building a washout period between conditions. This is a gap long enough for the effects of the previous condition to dissipate before the next one begins. In drug studies, the washout is calculated based on the medication’s half-life, typically requiring enough time for the drug to drop to negligible levels. For behavioral studies, the washout might be a rest period, a distraction task, or simply enough calendar days between sessions for fatigue or practice effects to fade.
The second major strategy is counterbalancing, which means varying the order in which participants experience conditions. In a simple two-condition experiment, half the participants get Condition A first and Condition B second, while the other half get the reverse order. This doesn’t eliminate carryover effects, but it distributes them equally across conditions so they don’t systematically favor one treatment over another. A common formal method for this is the Latin Square design, which assigns conditions across positions so that each condition appears in each ordinal position equally often. This approach works well when the number of conditions makes it impractical to test every possible order.
Counterbalancing has an important limitation. It only controls for carryover if the carryover effect doesn’t interact with the treatment itself. If Treatment A creates a unique physiological state that specifically amplifies the effect of Treatment B but not vice versa, simply reversing the order for half the participants won’t solve the problem. In those cases, the carryover is asymmetric, and more complex designs or statistical models are needed.
Detecting Carryover in Data
Researchers have developed formal statistical approaches to check whether carryover effects are present in their data. The most classic method, introduced by Grizzle in 1965, is a two-stage procedure. In the first stage, the analysis tests whether a carryover effect exists. If it does, the researcher drops the contaminated data (typically from the second period) and analyzes only the first-period data, essentially converting the crossover design into a simpler parallel-group comparison. If no carryover is detected, the full dataset from both periods is used.
A second approach involves building the carryover effect directly into the statistical model as a parameter. This is more flexible but comes with a risk: if the model doesn’t correctly specify how the carryover operates, the results can be misleading. Because carryover effects are genuinely difficult to isolate statistically in simple designs, many researchers emphasize prevention over detection. Planning the study so that carryover effects are unlikely to occur, through adequate washout periods and appropriate counterbalancing, is generally more reliable than trying to fix the problem after the data is collected.
Carryover Effects Beyond the Lab
While the term comes from experimental research, carryover effects show up in everyday contexts. A grueling morning workout that leaves you mentally drained for afternoon meetings is a carryover effect. Studying one foreign language and then mixing up vocabulary when you switch to studying another is a carryover effect. The concept captures any situation where a previous experience lingers and shapes your response to the next one.
In applied settings like workplace safety and military operations, the interaction between physical and cognitive carryover is particularly relevant. Research has shown that excessive physical and cognitive exertion can independently reduce performance on subsequent tasks, and when both types of fatigue are present, the combined effect can be worse than either alone. Understanding this helps explain why scheduling, rest protocols, and task sequencing matter in high-stakes environments where performance degradation has real consequences.