When Is a Pretest-Posttest Design Appropriate?

When Does Measuring Before and After an Intervention Make Sense?

Suppose you introduce a new teaching strategy, training program, clinical procedure, or organizational policy. You want to know whether participants' outcomes changed after its implementation. Measuring the outcome before the intervention and again afterward seems like a sensible starting point.

But is that enough to justify a pretest-posttest design? And what conclusions can you legitimately draw from the difference between the two measurements?

The answer depends on your research question, the nature of the intervention, the timing of the measurements, and whether you need to describe change or attribute that change to the intervention.

Use a Pretest-Posttest Design When Change Is the Outcome of Interest

What Makes a Pretest-Posttest Design Appropriate?

What Exactly Is a Pretest-Posttest Design?

A pretest-posttest design involves measuring an outcome before an intervention or event and measuring that outcome again afterward. The initial measurement establishes a baseline, while the subsequent measurement provides information about the outcome following exposure to the intervention.

The simplest arrangement is:

In conventional design notation, this arrangement is written as O1 X O2. The observations represent measurements, while X represents the intervention.

Importantly, the pretest-posttest structure describes the timing of measurement. It does not automatically identify the study as experimental. Whether a design is experimental, quasi-experimental, or pre-experimental depends on additional features, particularly how participants are assigned to conditions and whether a comparison group is included.

When Is a One-Group Pretest-Posttest Design Useful?

A one-group pretest-posttest design measures the same participants before and after an intervention without including a separate comparison group.

It may be reasonable when the primary objective is to document observed change rather than establish a definitive causal effect. For example, a researcher might evaluate whether participants' digital literacy scores increased during a professional development program.

The design can also be useful during early intervention development. Researchers may need preliminary evidence about whether an outcome can be measured reliably, whether participants complete the assessment, and whether the observed pattern warrants further investigation.

Such findings can inform a subsequent controlled evaluation, but they should not be presented as definitive evidence of effectiveness.

When Should You Include a Comparison Group?

If you want to determine whether an intervention produces better outcomes than an alternative condition, a comparison group becomes substantially more valuable.

A pretest-posttest comparison-group design measures outcomes in at least two groups before and after the intervention. One group receives the intervention, while another receives usual practice, an alternative intervention, or another ethically appropriate condition.

Simply including another group does not eliminate bias. The groups must be sufficiently comparable for the intended analysis, or their differences must be addressed through an appropriate design and analytical strategy. Researchers should also consider whether a control group is necessary for the research question.

Does the Research Question Require Within-Participant Change?

A pretest-posttest design is especially informative when the research question concerns change within the same individuals, organizations, classrooms, or other units.

For instance, asking whether students' research-methods knowledge differs after a workshop is a different question from asking whether workshop participants outperform students who did not attend.

The first question concerns observed change. The second introduces a comparative effectiveness question.

When the same individuals provide both measurements, the observations are statistically dependent. Analyses should account for this pairing rather than treating pretest and posttest observations as unrelated samples.

Can the Outcome Be Measured Comparably Before and After?

The usefulness of a pretest-posttest design depends on measurement quality. Ideally, the pretest and posttest should assess the same construct using measures that support meaningful comparison.

Using identical items may improve direct comparability, but repeated exposure can also introduce practice effects. Participants may remember questions, become familiar with the response format, or learn how to perform the assessment independently of the intervention.

Alternative test forms may reduce familiarity effects, provided their scores are demonstrably comparable. Researchers should also consider whether the meaning of a construct changes after the intervention. For example, participants may interpret a self-assessment of technological competence differently after learning how much they previously did not know.

When comparing latent constructs measured by questionnaires, longitudinal measurement invariance may need to be examined. A numerical change is difficult to interpret if the measurement instrument functions differently across occasions.

How Should the Timing of the Pretest and Posttest Be Determined?

The pretest should ordinarily occur before participants experience the intervention. The posttest should occur when the intervention could plausibly have influenced the outcome.

A knowledge assessment administered immediately after a short workshop may be appropriate for measuring immediate learning. It would be less suitable for assessing whether participants subsequently apply the knowledge in their professional practice.

The timing should follow a theory of how change occurs, rather than simply the researcher's available schedule. A baseline collected too early may no longer represent the participant's condition immediately before intervention exposure. A posttest collected too soon may miss delayed outcomes.

Where the outcome is expected to develop gradually or fluctuate, additional measurements may be more informative than a single pretest and posttest. The number of longitudinal measurement occasions should therefore reflect the process being studied.

What Can You Calculate From Pretest and Posttest Data?

For a continuous outcome measured on a comparable scale, a simple individual change score is the posttest score minus the pretest score.

Researchers may summarize the distribution of individual changes and estimate their uncertainty. Depending on the outcome and research question, suitable analyses may include paired-samples tests, regression models, or repeated-measures methods.

For comparison-group studies, researchers may use baseline-adjusted regression, analysis of covariance, or models of change. These approaches answer related but not always identical questions, and their suitability depends on the design, assumptions, and estimand of interest.

Statistical significance alone does not establish that a change is educationally, clinically, or practically meaningful.

What Are the Main Threats to Interpretation?

The principal limitation of an uncontrolled pretest-posttest design is that time and intervention exposure occur together. Other processes may influence the outcome between measurements.

Participants may mature, gain experience elsewhere, encounter relevant events, or change because they have already completed the pretest. Extreme baseline observations may also move closer to typical values through regression to the mean.

These explanations matter because the difference between observing change and establishing an intervention effect is fundamental to research validity.

When Is a Different Design More Appropriate?

A pretest-posttest design may be unnecessarily complicated when the research question concerns only a single current condition, prevalence estimate, or descriptive snapshot.

Conversely, it may be insufficient when researchers need to distinguish an intervention effect from an existing trend, seasonal pattern, or other time-related process. An interrupted time-series design may be more informative when repeated observations are available before and after implementation.

When random assignment is feasible and ethically appropriate, a randomized controlled trial may provide a stronger basis for evaluating comparative intervention effects. The design should follow the inferential question, not the apparent simplicity of collecting two measurements.

Evaluating an AI Literacy Workshop for University Faculty

The result indicates that the participating faculty members scored higher after the workshop than before it. If the study's purpose is to describe observed change during the program, the design addresses that purpose.

However, the researchers cannot conclude from these measurements alone that the workshop produced the entire increase. Participants might have independently studied AI-related materials, become familiar with the assessment, or acquired relevant experience through other activities.

Suppose a comparable group of faculty members who did not attend the workshop was also assessed. If that group improved by 5 points over the same period, the difference between the two groups' mean changes would be 11 points.

That additional comparison would be informative, although causal interpretation would still depend on how the groups were formed, their comparability, and the assumptions supporting the analysis.

The practical lesson is that a one-group design may adequately document improvement while remaining insufficient to attribute that improvement confidently to the intervention.

Common Misinterpretations of Pretest-Posttest Research

How to Decide Whether a Pretest-Posttest Design Fits Your Study

Begin by identifying the claim you ultimately want to make. If you want to document whether participants changed during a defined period, a well-planned one-group pretest-posttest design may be proportionate to your objective.

If you want to establish whether the intervention caused an improvement relative to what would otherwise have happened, you need a design that addresses the missing counterfactual more convincingly.

Whichever design you select, specify the primary outcome, measurement schedule, analysis plan, and intended interpretation before collecting data. Anticipate missing follow-up measurements, since attrition can distort the observed pattern of change.

A modest design with carefully bounded conclusions is more defensible than an elaborate analysis used to support a claim the design cannot identify.

Before Choosing a Pretest-Posttest Design

Frequently Asked Questions About Pretest-Posttest Designs

Choose the Design for the Claim You Need to Make

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.