How Can History Effects Threaten Internal Validity in Before-and-After Studies?

What If Something Else Happened Between Your Pretest and Posttest?

You introduce a new teaching strategy at the beginning of a semester. Students complete an assessment before the intervention and another afterward. Their average scores improve substantially, suggesting that the strategy may have worked.

During the same semester, however, the institution introduced an academic support program, students gained access to additional learning resources, and the examination schedule changed. Could these developments have contributed to the improvement?

This is the problem of history effects. When events outside the intervention occur between observations, researchers may incorrectly attribute their influence to the intervention being evaluated.

The concern is particularly consequential in before-and-after studies because the intervention and the passage of time occur together. Without an appropriate counterfactual comparison, separating their effects can be difficult.

How Do History Effects Threaten Internal Validity?

Why Events Outside an Intervention Can Distort Before-and-After Findings

What Counts as a History Effect in Research?

In research methodology, history refers to events occurring during the observation period that may influence the outcome independently of the intervention being studied.

The term does not necessarily refer to historical events in the distant past. A history effect might arise from an institutional policy introduced last week, an unexpected disruption during data collection, or a public information campaign launched between baseline and follow-up.

Consider an educational intervention intended to improve students' digital literacy. Between the pretest and posttest, the university requires all students to complete an additional digital skills workshop.

If scores improve, the workshop provides an alternative explanation for at least part of the observed change.

Other examples include changes in curriculum, staffing, funding, examination conditions, institutional services, public policy, economic circumstances, or access to technology.

An external event does not have to be dramatic to matter. Routine organizational changes may be more consequential than highly visible events when they directly affect the outcome.

Why History Is Especially Problematic in Single-Group Before-and-After Studies

A single-group pretest-posttest study observes participants before an intervention and again afterward.

The difference between the two measurements is often interpreted as evidence of improvement or deterioration. Yet the comparison does not directly establish what would have happened to those participants without the intervention.

Three developments occur together: time passes, the intervention is introduced, and other circumstances may change.

Without additional evidence, the observed outcome difference cannot generally be separated into an intervention effect and changes attributable to other causes.

This is a central limitation of the design, not necessarily an error in statistical calculation.

A paired-samples t-test, for example, may establish that the mean posttest score differs from the mean pretest score under its statistical assumptions. It cannot establish that the intervention caused the difference.

The broader problem belongs to the credibility of the study's main causal comparison.

How Can History Effects Be Confused With Intervention Effects?

Suppose a hospital introduces a patient education program and observes improved medication adherence over the following three months.

During that period, a national public health campaign also encourages patients to follow prescribed treatments.

The hospital's improvement may reflect its intervention, the national campaign, or both.

In a simple before-and-after comparison, the intervention and campaign are temporally associated with the observed improvement. Their separate contributions are not identified merely by knowing that the intervention preceded the outcome.

The same reasoning applies when external events worsen outcomes. An effective intervention might appear ineffective if a concurrent disruption offsets its benefits.

History effects can therefore exaggerate, diminish, conceal, or potentially reverse an apparent intervention effect.

What Is the Difference Between History and Maturation?

History and maturation are often discussed together because both concern changes occurring over time. Their underlying mechanisms differ.

For example, students may improve their reading skills through ordinary development and accumulated learning experience. That possibility concerns maturation as a threat to internal validity.

If the same students improve partly because a separate reading campaign was introduced during the semester, that is a potential history effect.

Both mechanisms may operate simultaneously. Researchers should distinguish them conceptually while recognizing that the available data may not identify their separate contributions.

Can History Effects Occur in Studies With Comparison Groups?

Yes. A comparison group can substantially improve a study's ability to address history effects, but its presence does not automatically eliminate the threat.

Suppose intervention and comparison groups experience the same external event, and the event has comparable effects on both groups. A suitable concurrent comparison may help distinguish the intervention's effect from the common change.

However, external events may affect the groups differently.

For instance, a technology subsidy introduced during an educational intervention might benefit students at one institution but not students at the comparison institution.

Even when both groups experience the same event, its consequences may vary because of differences in baseline resources, exposure, or institutional conditions.

These are sometimes described as differential history effects or selection-history interactions.

They are particularly important when comparison groups differ in characteristics that influence their response to external events.

Why Concurrent Comparison Groups Are Usually More Informative Than Historical Controls

A historical control group consists of participants observed during an earlier period rather than concurrently with the intervention group.

Historical controls can be useful when concurrent allocation is infeasible, but comparisons across periods may confound intervention effects with changes in circumstances.

For example, comparing students taught with a new curriculum this year against students taught with an older curriculum last year may involve differences in institutional policy, instructors, student composition, or examination conditions.

A concurrent comparison group is observed during the same general period as the intervention group, potentially reducing exposure to different calendar-time events.

Nevertheless, concurrent observation alone does not establish exchangeability. The groups must also be sufficiently comparable for the intended causal inference.

How Do Interrupted Time-Series Designs Help Address History Effects?

An interrupted time-series design examines outcomes at multiple points before and after an intervention.

Rather than relying on one pretest and one posttest, researchers estimate the preintervention pattern and assess whether the outcome's level or trend changes following implementation.

This can provide stronger evidence than a simple before-and-after comparison because the analysis considers the outcome trajectory that existed before the intervention.

For example, a university might record monthly course completion rates for two years before introducing a student support initiative and continue measuring them afterward.

If completion rates were stable before implementation and increased abruptly afterward, the pattern may be consistent with an intervention effect.

However, an interrupted time series remains vulnerable when another event occurs at approximately the same time as the intervention.

Multiple observations do not automatically distinguish two simultaneous causes of a change.

Adding a suitable comparison series, where feasible, may help evaluate whether a similar change occurred in an unaffected population.

Researchers must also consider seasonality, autocorrelation, underlying trends, changes in measurement, and the plausibility of the chosen statistical model.

Can Researchers Identify History Effects by Reviewing an Event Timeline?

Constructing a timeline of relevant events is a practical way to identify plausible competing explanations.

The timeline should include intervention implementation, baseline and follow-up assessments, institutional changes, external programs, disruptions, and other developments that could influence the outcome.

Researchers should consider the event's timing, affected population, plausible mechanism, and expected direction of influence.

An event that occurs after outcome measurement cannot explain an earlier change through ordinary forward causation. An event affecting only one comparison group may be more concerning than one experienced similarly by both groups.

Documentation may come from institutional records, implementation logs, administrative data, public announcements, or contemporaneous field notes.

Retrospective interviews can help identify events, but recollections may be incomplete or influenced by knowledge of the study results.

Importantly, an event timeline helps evaluate plausibility; it does not automatically quantify the causal effect of each event.

Which Research Designs Offer Protection Against History Effects?

The appropriate design depends on the intervention, setting, available observations, ethical constraints, and causal question.

Researchers should prioritize design-based safeguards against internal validity threats rather than expecting a statistical test to resolve all competing explanations afterward.

Can Statistical Adjustment Eliminate History Effects?

Statistical models can incorporate measured external events, time trends, seasonality, and other relevant variables when the design and data support such adjustment.

However, an external event that occurs at exactly the same time as an intervention and affects the same population may be difficult or impossible to distinguish statistically from the intervention effect.

For example, if every participant receives a new educational intervention on the same day that a separate mandatory training program begins, there may be no independent variation allowing their effects to be identified separately.

Adding both variables to a regression model does not create information that the study design never collected.

Sensitivity analyses may help assess how alternative assumptions affect the findings. They should not be presented as proof that unmeasured history effects have been eliminated.

When Is a History Effect a Serious Threat Rather Than a Minor Concern?

The seriousness of a history threat depends on whether the external event could plausibly affect the outcome and distort the intervention comparison.

Researchers should consider the event's timing, intensity, reach, and relationship to the outcome.

A national policy affecting the outcome immediately after intervention implementation may present a substantial concern. An unrelated event with no plausible pathway to the outcome may not.

It is also useful to distinguish the existence of an event from evidence that the event materially affected results. Researchers should avoid claiming bias merely because an external development occurred.

The appropriate conclusion reflects the available evidence and remaining uncertainty.

How a Concurrent University Initiative Can Distort an Intervention Evaluation

The initial single-group results appear encouraging.

From these data alone, the researchers cannot determine how much of the fifteen-point improvement resulted from the online intervention rather than the workshops or other changes over time.

Suppose a comparable group that did not receive the online intervention nevertheless participated in the same workshops.

The example illustrates why the entire fifteen-point improvement should not automatically be interpreted as the intervention effect.

It does not establish that the workshops caused exactly ten points of improvement or that the online intervention caused exactly five points. Other influences, group differences, and sampling uncertainty remain possible.

The researchers should examine the timing and implementation of both programs, evaluate group comparability, and report the assumptions required for their causal interpretation.

Common Misconceptions About History Effects

How to Anticipate and Address History Effects in Your Study

When planning an intervention evaluation, identify external developments that could influence the outcome during the observation period.

For educational research, these may include examination preparation programs, changes in academic calendars, institutional technology initiatives, curriculum revisions, or unexpected disruptions to instruction.

Where feasible, choose a design that permits intervention-related changes to be compared with changes occurring under an appropriate alternative condition.

Document relevant events prospectively whenever possible. This reduces reliance on retrospective explanations developed after the results are known.

When external events cannot be ruled out, the appropriate response is not necessarily to discard the findings. Instead, distinguish the observed change from the causal effect that the study can credibly estimate.

If several alternative explanations remain plausible, prioritize the most consequential internal validity threats and report how they affect interpretation.

History Effect Checks for Before-and-After Research

Frequently Asked Questions About History Effects

Observed Change Is Not Automatically an Intervention Effect

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.