Facts, management claims and inferences in research notes
Label a research statement by what it actually establishes: reported fact, management claim, researcher inference or unknown. These labels prevent a plausible explanation from becoming a supposedly verified fact as a note passes between people.
Four labels with different jobs
A reported fact is an observation within a stated document and scope. “The filing reports 80 units of sales” is narrower than “the business is healthy.” A management claim identifies who supplied an explanation. An inference connects observations through reasoning. An unknown records an unresolved question; it should not disappear because a summary needs to sound confident.
MD&A presents management’s view of results. Quote or paraphrase that view with attribution. The label does not imply dishonesty; it preserves the distinction between the explanation and the underlying observation.
Worked example: an improved margin
Assume a fictional manufacturer reports margin rising from 20% to 23%. Its chief executive attributes the change to automation. The filing also mentions higher selling prices but does not quantify their contribution.
| Sentence | Label | What would strengthen it? |
|---|---|---|
| Reported margin rose three percentage points | Reported fact | Source, period and unchanged cost classification |
| Automation drove the improvement | Management claim | Quantified cost bridge or comparable plant data |
| Automation may explain part of the gain | Inference | Explicit reasoning and alternative explanations |
| Automation explains all three points | Unknown, if unsupported | Evidence separating price, mix and cost |
The difference between percentage points and percent matters. A move from 20% to 23% is three points, or a 15% relative increase. State which calculation you mean. Also check whether warranty costs moved below the gross-profit line; identical labels can hide a changed calculation.
Build an attribution trail
For each important inference, attach the observations used and one competing interpretation. In this case, pricing, product mix and capacity utilization may also change margin. These are candidate explanations, not established facts about the fictional company. The next research task is to locate a margin bridge or identify an observation that distinguishes them.
When updating the note, preserve its earlier label. A claim can become better supported, but the new evidence should be visible. Avoid replacing “management expects” with “will” while condensing a paragraph. That tiny edit can remove the uncertainty that made the original statement accurate.
This method is especially useful for collaborative notes and summaries. It is not a substitute for assessing whether a reported number is accurate, restated or selectively presented. Even a statement correctly labeled “reported fact” requires a scope and may later need correction.
A further check on the example
Read the final summary aloud and underline causal verbs such as drove, caused and proved. For each one, ask whether the source observed the connection or merely proposed it. In the automation example, a cost-saving estimate announced before installation belongs to a forecast, not an achieved result. A later realized-margin observation can update the note, but cannot retroactively convert the earlier forecast into something that was known at the time. Keep that temporal distinction in the wording.
FAQ
Is everything in an audited report a verified explanation?
No. The document contains different sections and types of statements. Audit coverage does not automatically extend to every causal claim in the narrative.
Can an inference be useful without proof?
Yes. Label its assumptions and identify the next test. The problem is presenting an inference as if its causal link had already been observed.