Research Evidence Integration

How do I combine survey data and interview findings in one report?

Keep them on their own terms rather than flattening them into one format. Survey figures carry question reference, base description, base size, weighting and filter. Themes carry a participant count out of the total, never a percentage, and their counter-evidence. Then label the relationship explicitly: converging (both measure the same construct and agree), complementary (they answer different parts of the question and are not a check on each other), or diverging (they disagree, which is reported and never averaged). Most claimed convergence is actually complementarity.

Two files give different numbers for the same thing. Which one do I use?

Neither, until you know why they differ. Investigate to origin. The usual causes are a different base, a different filter, an included or excluded "don't know", a different rounding point, a different weighting state, a reworded question, or a different time period, and each of those is itself worth reporting. Notice if one of the two numbers makes the story better: that is a reason to check harder, not less. If the origin genuinely cannot be established, report the range and name both sources.

How should I label client-supplied data in a research report?

As client-supplied, at every appearance rather than only the first, with the extraction date and the definition where it is documented. It has not been through your analysis and its provenance is not yours to vouch for. It can support a finding, and it frequently outranks your survey on what customers actually did because it measures behaviour rather than recall. It should not carry a headline or lead a chapter, because a headline resting on someone else's number puts your name on it.

Can I compare this wave to the last one?

Only after checking three things: identical question wording, identical base definition, and equivalent sample and method. Comparability is established, not assumed. Where any of the three differs, the comparison is not made and the reason is reported. The cost of checking is twenty minutes. The cost of a retracted trend claim is the client's confidence in everything else in the report.

What do I do with a quote that has no participant ID?

Recover the ID from source before the quote enters the report. A quote that cannot be checked is functionally the same as a fabricated one to everybody downstream, because there is no way for a reader to tell the difference. Identifiers are preserved end to end, and the ID is checked against the segment the quote is attributed to, since misattributing a quote to a segment is a common and damaging error.

How do I use photographs or video as evidence in a report?

Each item carries a capture reference, a date, a context, and a confirmed consent status covering this specific use. The caption states what is shown, never what it proves. A photograph of an empty waiting room shows an empty waiting room at one moment; the claim that the service is under-used is a separate assertion needing separate evidence. Images without capture references can illustrate a setting generally and cannot support a claim about a specific place or time.

What if a claim I want to make is not supported by the right kind of evidence?

Restate the claim at the level the available evidence supports, mark it as a hypothesis with its validation named, or drop it. What you must not do is substitute adjacent evidence, which usually means supporting a behavioural claim with an attitudinal measure because the behavioural measure does not exist. That is more damaging than an absent claim, because it looks like evidence. Ask what evidence the claim requires before asking what evidence you have; reversing that order is the commonest failure in integration.

What is an evidence map and when should I build it?

A table with one row per significant claim: its location in the report, the claim, its level (finding, interpretation, insight, implication, recommendation), the source codes it rests on, the base and question reference, its verification status and its confidence. Build it while you integrate, not afterwards. Built during the work it records the link between claim and source while that link is in front of you; built afterwards it records what someone remembers, which is wrong in exactly the places that matter. A row with no source is a defect, not a gap to fill later.

What do I do with a strong finding that contradicts the story?

Report it. There are four honest placements and none of them is omission: in the argument as a qualifier, which usually makes the report better and more credible; as a separate finding outside the argument; in the limitations section, where it constrains how the main finding should be read; or in the appendix, only where it is genuinely secondary to the decision. The test for that last option is whether you would be comfortable with the client finding it there. The evidence you most want to leave out is usually the evidence the reader most needs, because it is the thing that would have changed their mind.

What can I do with an input whose origin nobody can establish?

Verify it, use it with the claim explicitly marked as resting on an unverified input, or drop it. There is no fourth option, and the common failure is the input that gets used because it happened to be in the folder. Unverified material never carries a headline. If a client insists on a figure whose derivation is not documented and it must carry weight, the honest output is a report saying the conclusion depends on an undocumented figure.