1 Evaluating Evidence

Assess source quality and relevance

Evidence should be judged by how well it supports a particular claim, not simply by deciding whether a source is “reliable.” Assess the source’s quality, provenance, corroboration, context, and limits of inference to distinguish what it directly shows from what can reasonably be concluded.

Examine how a source was made

Begin with the research question and ask whether the source is relevant. Then consider how its information was produced and what the source can contribute to the claim.

• Creator and expertise: Who made the source, and what access or knowledge did they have?

• Purpose and audience: Was it created to record, persuade, report, entertain, justify a decision, or serve another purpose?

• Method: How was the information gathered or recorded? Are there known errors, omissions, or constraints?

• Specificity and completeness: Does the source contain the details needed to assess the claim, or only a partial account?

A source’s type does not determine its quality by itself. A firsthand account may offer direct observations but still be limited by memory, perspective, or purpose. A later analysis may compare many records, but it depends on the evidence and methods it uses. Assess each source in relation to the question being asked. The Library of Congress recommends examining a source’s creator, purpose, point of view, and the evidence it provides.

Trace provenance and authenticity

Provenance helps researchers assess whether a source is authentic and understand its relationship to other records. In archival practice, keeping records connected to their creators and original groupings can preserve information about their creation and use.

Uncertain provenance does not automatically make a source useless, but it makes some claims about that source harder to support.

Compare sources and check independence

Compare sources to see where their accounts agree, differ, or leave gaps. Agreement can strengthen a claim, while disagreement may reveal different perspectives or point to questions that need further investigation. Comparing sources can therefore identify both support and productive discrepancies.

Check whether sources are genuinely independent. Look for sources created by different people or organizations, using distinct methods or evidence; several reports that repeat one original account are not necessarily separate confirmations. For example, a diary entry about a flood might be compared with a dated weather record and a municipal damage report. Agreement among them supports some conclusions about the event, but each source may document a different aspect of it.

Interpret context and perspective

Interpret evidence in the circumstances in which it was created. Relevant context may include the time and place, the creator’s role, intended audience, prevailing conventions, and the conditions under which information was collected. Without context, researchers can mistake a source’s purpose or meaning.

A public announcement and a private letter written on the same day may describe the same event differently because they had different audiences and purposes. Neither should be accepted or rejected solely on that basis. Ask what each source was positioned to observe and what it may have left out.

Separate evidence from inference

Separate what a source directly contains from the conclusions drawn from it. A photograph may show a damaged building, but by itself it may not establish when the damage occurred, who caused it, or how widespread it was.

For each claim, ask:

1. What does the evidence directly establish?

2. What additional assumptions are needed to reach the conclusion?

3. What alternative explanations fit the evidence?

4. What evidence is missing, incomplete, or uncertain?

Match the strength of the conclusion to the strength of the evidence. When the record does not justify certainty, use qualified language such as “supports,” “suggests,” or “is consistent with.” Primary sources are often incomplete or lack context, so further sources may be needed to test interpretations.

Account for preservation and ethical limits

The surviving record is not necessarily representative. Some materials were never created; others were lost, destroyed, or not selected for preservation. Archival institutions make choices about what to retain, and those choices can shape what future researchers are able to study. The National Archives’ appraisal policy recognizes that records are selected for preservation and considers their context, uniqueness, usability, and research value.

Do not treat silence in the archive as proof that an event or perspective did not exist. Ask whose records are present, whose are missing, and how recordkeeping and preservation practices may have affected the evidence available. Represent sources fairly by distinguishing documented facts from interpretation, acknowledging uncertainty and gaps, and avoiding claims that go beyond what a source can show.

• Observation

• Preservation bias