Evidence Interpretation: Source and Activity Questions

• Distinguish classification, source and activity questions.
• Formulate relevant alternatives at the same level.
• Explain what a likelihood ratio does and does not express.

Classification asks what kind of material or feature is present. A source comparison asks whether material could share an origin with a specified reference. Activity interpretation considers mechanisms and circumstances such as contact, transfer and later disturbance. These are related questions, not interchangeable conclusions.

Terminology varies by discipline. DNA interpretation may also distinguish sub-source questions about a DNA contributor from source questions involving the biological material. Do not infer the identity of a body fluid solely from a DNA profile.

Answering one level can help frame the next without answering it. Identifying a fibre as a particular material class does not establish a specific garment. A supported association with a garment does not, by itself, establish who wore it, when contact occurred or why the material arrived.

Keep the proposed conclusion at the level supported by the examination. If the analysis only classifies material, report that useful result. If a source comparison is possible, explain its significance and limitations. Do not imply that an unanswered activity question invalidates the analytical work or that analytical success resolves the event.

A useful comparison addresses relevant, explicit alternatives. For a source question, these might concern the proposed item versus a different relevant source. For an activity question, they might concern a specified direct contact versus an alternative transfer history under the case circumstances.

Do not compare a source proposition with an activity proposition as though they answer the same question. Nor should an examiner invent the defence account and report it as something the person actually said. Identify any assumed alternative, its basis and limitations.

Identify exactly what is being evaluated. The evidence might include the observed properties of a recovered trace, its location, a distribution across samples or a defined combination of findings. Changing that evidence description can change the question and the assessment.

Case information provides the framework in which propositions are evaluated. Distinguish information treated as agreed from information assumed for a particular calculation and information still disputed. If the interpretation depends on prior access being absent, that assumption needs to be visible.

Keep alternatives relevant and mutually exclusive for the stated comparison. A source proposition and its alternative should specify the appropriate source question; activity propositions should describe competing activities under the relevant circumstances. If the supplied question is unclear, seek clarification rather than produce an impressive number for a different issue.

A comparison needs attention to variation within a proposed source and between possible sources. Different areas of one garment may not have identical features. Similar properties may also occur in unrelated garments. A limited reference can therefore affect both the comparison and the significance assigned to similarity.

Visual resemblance is not a substitute for an appropriate examination. The methods selected should address relevant properties at their demonstrated resolution. The interpretation must then explain what those findings support rather than convert the number of techniques used into an automatic measure of source strength.

A physical fit, a chemical classification and an observed pattern similarity are different types of findings with different reasoning requirements. Do not transfer language appropriate to one method to every other discipline merely because both results are called a “match.”

An activity interpretation may require information on how material transfers, how long it can persist under relevant conditions, how it was recovered and how often comparable background material occurs. Those factors affect the observations expected under each proposed activity.

A strong analytical association cannot supply missing information about the event. If prior access or an intermediate carrier is a relevant alternative, consider its expected findings rather than dismiss it merely because a comparison result is strong.

Equally, the mere possibility of an indirect route does not make every observation equally expected under every explanation. The task is to assess the relevant propositions using appropriate evidence and clearly identified assumptions, not to list unlimited speculative stories.

A likelihood ratio compares how probable the observations are under one proposition with how probable they are under another, given the stated information and assumptions. It is not the probability that either proposition is true.

In symbols, LR = P(E | H1, I) / P(E | H2, I). Here E is the observed evidence, H1 and H2 are the propositions, and I represents the information assumed in the assessment. The choice of propositions and assumptions is part of the result’s meaning.

For a purely hypothetical arithmetic illustration, suppose a defined finding has probability 0.6 under H1 and 0.2 under H2. The ratio is 3. This describes a relative expectation of that finding; it does not say H1 has a probability of 75%, or that a person has a particular probability of guilt. These numbers are invented for arithmetic, not empirical trace-transfer data.

In the hypothetical example, P(E|H1,I)=0.6 and P(E|H2,I)=0.2, so the finding is three times as probable under H1 as under H2 within the specified model. The ratio describes relative support from E. It does not state how probable H1 was before the finding or account automatically for the rest of the case.

If prior odds are supplied in a purely mathematical example, posterior odds equal prior odds multiplied by the likelihood ratio. With prior odds 1:9 and LR 3, posterior odds are 1:3, corresponding to 25% for H1 in that two-proposition toy model. Different prior odds give a different posterior probability even though the LR is unchanged.

These invented numbers demonstrate why the LR alone cannot be read as a probability of a proposition. They are not a recommendation for an examiner to choose odds of guilt. A real legal decision also involves the applicable legal framework and information beyond the scope of a particular forensic examination.

For the specified propositions and assumptions, a ratio greater than one means the observations are more probable under H1 than H2. A positive ratio below one means the reverse. A value of one means the observations are equally probable under those propositions; it does not make the propositions equally probable.

A number’s precision should reflect its basis. Sparse or poorly representative data can make an assessment sensitive to assumptions. A model output with many decimal places is not automatically a highly precise scientific conclusion. Explain the information, method and limitations supporting the result.

Do not manufacture a number where the available method and data cannot support one. A qualified description of what was examined can be more useful than an unsupported quantitative conclusion. Where a validated reporting framework uses verbal expressions, explain the framework and keep the statement within its scope.

“The finding is unlikely if the item came from another source” is different from “it is unlikely that the item came from another source.” Reversing those statements changes the question. Conclusions about propositions depend on other information as well as the comparison findings.

Do not transfer a source-level likelihood ratio to an activity question without a suitable activity assessment. A ratio for a DNA contributor or a material comparison does not automatically quantify whether the person performed a disputed action.

Not every method supports a numerical likelihood ratio. Do not invent probabilities or attach a numerical value to an unsupported verbal impression. State the observations and the limit of any justified interpretation.

Several observations may come from the same specimen, process or source of variation. For example, morphology and composition can provide different information about one fibre, but the relationship between those properties still matters when evaluating their combined significance.

Multiplying separate likelihood ratios as though every finding were independent requires a justified model. Repeating a measurement, using closely related tests or splitting one observation into several descriptions does not automatically create independent support. The evaluation should account for relationships relevant under each proposition.

Conversely, related findings are not necessarily redundant. One test may identify a limitation or distinguish materials that another cannot. The scientific task is to model and explain their joint meaning, not to assume that more tests always multiply the evidential strength.

A clear report states the material examined, the propositions or comparison question, the findings, assumptions and limits. Explain whether the conclusion concerns a material class, a source or an activity. Identify an unanswerable question directly rather than hide it behind a confident phrase.

The scientific distinctions apply across countries. The legal treatment of an expert opinion depends on the applicable jurisdiction and proceeding; the national reference lessons address those rules separately.

Trace Evidence Transfer, Persistence and Contamination. Cognitive Bias in Forensic Investigation.

Read the conclusion alongside the propositions. Are they at the same level? Is the reported result about the observations, or has it become an unsupported probability that an event occurred? Are material assumptions and reference limitations explained in terms a non-specialist can follow?

Check the scope of exclusions as carefully as associations. A result excluding a submitted reference as the source under the method’s conditions does not identify the true source. A failure to recover expected material may affect an activity explanation without establishing a different event.

Finally, distinguish an exploratory suggestion from an evaluation of stated propositions. An investigative lead can be useful while requiring further examination. Labelling its purpose prevents an early possible explanation from later being presented as an established forensic conclusion.

• Identify whether the issue is classification, source or activity.
• Compare relevant propositions at the same level.
• A likelihood ratio concerns observations given propositions, not guilt.
• State missing information and avoid carrying source strength into an unsupported activity conclusion.