Learn why natural change over time, regression to the mean, and participant dropout can blur the true effects of an intervention. These research design concepts help art therapy professionals judge whether a study’s findings reflect its treatment or other factors—and why that distinction matters.

Multiple Choice

Maturation, regression and experimental mortality are examples of threats to which aspect of a study?

Internal validity focuses on whether observed changes truly come from the experimental manipulation rather than other factors. Maturation is a natural change in participants over time that can alter outcomes independently of the treatment, so improvements or declines might occur simply because people develop or age during the study. Regression to the mean happens when participants are chosen for extreme scores at the start; on subsequent testing their scores tend to move toward the average, which can masquerade as an effect of the intervention. Experimental mortality, or attrition, occurs when participants drop out in a way that biases the remaining sample; if those who leave differ systematically from those who stay, the observed effect may reflect who remained rather than the treatment itself. These issues threaten internal validity because they offer alternate explanations for any observed effects. In contrast, external validity concerns generalizability to other populations or settings, construct validity concerns whether the measures truly reflect the intended constructs, and reliability concerns the consistency of measurements across time or raters. Therefore, these examples are best understood as threats to internal validity.

Research on art therapy asks a simple-sounding question: did the intervention make a difference? The answer can be harder to pin down. A client’s mood may shift over time, a participant may leave a study, or an unusually high score may move closer to the average on a later measurement. These changes can look like effects of art therapy even when something else caused them. That’s why researchers pay close attention to internal validity: how well a study supports the claim that its intervention, rather than another factor, produced the observed result.

The heart of internal validity

Imagine a study in which adults attend a series of art therapy sessions and report less anxiety afterward. The change matters, but it does not automatically show that the sessions caused it. Perhaps participants’ anxiety eased naturally over the same period. Perhaps people with the highest initial scores were more likely to improve, no matter what they did. Or perhaps those who found the sessions least helpful stopped taking part, leaving a group whose results look unusually positive.

Internal validity is about sorting through these alternative explanations. A study has stronger internal validity when its design gives researchers a sound basis for linking an observed change to the intervention. It does not mean the study is flawless, or that every other influence has vanished. Rather, it means likely competing explanations have been considered and controlled as well as possible.

Three familiar threats illustrate the idea: maturation, regression to the mean, and experimental mortality, also called attrition. Each can muddy the story a study tells.

Maturation: change happens, even without an intervention

Maturation refers to changes that occur naturally as time passes. People grow, adapt, recover, and respond to events around them. In a study, these shifts can affect outcomes independently of the intervention.

For example, a group of young people might show improved emotional regulation over several months of art therapy. The sessions could contribute to that change. But development, family support, a change in school routine, or simply having more time to adjust to a stressful event could also play a role. If the study has no useful way to account for those factors, it becomes difficult to tell how much of the improvement came from art therapy.

Maturation isn’t limited to children or long studies. Adults may experience natural changes too. A person dealing with grief may feel differently several weeks later than during the first appointment. Someone coping with a medical condition may improve or decline as the condition changes. Time is not an empty space around an intervention; it can be an influence in its own right.

Researchers can address this threat through design choices. A comparison group, when appropriate and ethically sound, can help show whether change also occurred among people who did not receive the intervention during the same period. Researchers may also take several measurements over time, rather than relying only on a single before-and-after comparison. These steps don’t make life perfectly predictable, but they help separate the intervention’s possible effect from ordinary change.

Regression to the mean: extreme scores often move

Regression to the mean is a statistical pattern that can seem like an intervention effect. When people are selected because they have unusually high or low scores, their next scores often tend to be closer to the group average. This may happen even if nothing meaningful has changed.

Say a study enrolls clients who report exceptionally high distress during an initial assessment. Some may have had a particularly difficult week, slept poorly, or experienced a recent crisis. At a later assessment, their scores may be less extreme—not necessarily because of art therapy, but because the first measurement captured an unusual moment.

This can be easy to miss. The improvement feels real to the participant, and the later score really is lower. Still, the study needs to consider whether the original score was an outlier. The same issue arises when people are chosen for very low scores, which may later rise toward the average.

Repeated measurements before an intervention can help researchers understand how stable scores are. A comparison group selected in a similar way can also show whether scores shift over time without the intervention. Clear reporting matters, too: readers need to know how participants were chosen and whether selection relied on unusually high or low scores.

Experimental mortality: who leaves, and who remains?

Experimental mortality is the loss of participants during a study. The term sounds dramatic, but it usually refers to people withdrawing, becoming unreachable, or otherwise not completing the study. “Attrition” is another common name.

Some participant loss is hard to avoid. People move, face new demands, or decide they no longer wish to take part. The concern is not simply that the group gets smaller. It’s that those who leave may differ in meaningful ways from those who stay.

Suppose participants who feel uncomfortable with a group format are more likely to withdraw from an art therapy study. The final results then describe a group that may be especially comfortable with that format. If the remaining participants report positive experiences, researchers should be careful about applying that finding to everyone who began the study.

Attrition can also affect comparisons between groups. If one group loses more participants—or loses participants for different reasons—then the groups may no longer be comparable. Researchers should track how many people leave, when they leave, and, when possible, why. They should also describe how missing information was handled. A smaller study is not automatically a weak study, but unexplained gaps can make its findings harder to trust.

Related ideas, different questions

Internal validity is one part of a wider picture of research quality. It helps to distinguish it from three nearby concepts.

External validity concerns whether findings may apply beyond the study itself—to other people, settings, or forms of care. A study could have strong internal validity within one clinic but offer limited evidence about a different age group or community.

Construct validity asks whether a measure captures the idea it is meant to capture. If a study aims to assess emotional expression, for instance, does its chosen measure actually reflect that concept? A sound design cannot rescue a poor measure.

Reliability concerns consistency. Would a measure produce similar results under similar conditions, or would scores shift depending on the rater or the day? Consistency matters, though a measure can be reliable without measuring the intended concept well.

These concepts overlap in real research, but they answer different questions. Internal validity asks, in effect, “What caused the change?” External validity asks, “Where else might this finding hold?” Construct validity asks, “Did we measure the right thing?” Reliability asks, “Can we measure it consistently?”

A careful study tells a careful story

Art therapy research often deals with human experiences that are personal, layered, and shaped by context. That makes careful study design especially important. A change in a drawing, a self-report, or a clinician’s observation may carry real meaning, but researchers still need to ask what else could have shaped it.

Maturation, regression to the mean, and participant loss all offer alternative explanations for change. They do not prove that an intervention had no effect. They remind researchers that a before-and-after difference, on its own, cannot settle the question.

When studies account for these threats, their findings become easier to interpret. Readers can see not only what changed, but how confidently that change can be linked to the intervention. That’s the point of internal validity: not to drain human experience of its nuance, but to make claims about it more trustworthy.