Outsourced EmploymentPowered by StealthAgents

Research note

How often do investigation notes separate fact from opinion?

A local method for measuring how often investigation interview notes keep observations and quotations apart from conclusions, with a clear evidence standard.

Published

Research note illustration for How often do investigation notes separate fact from opinion?
Table of contents

Question and scope

The question covers interview notes written during a workplace investigation. It includes notes from the person who raised the concern, from witnesses, and from the person named in the concern. It excludes the final decision memo and any note written only for a lawyer. The review covers one employer, one window, and the cases handled by the internal team.

A case enters the sample when the investigation file contains at least one interview note and a dated scope. If the note is a summary written after the case closed, the case enters as a reconstructed note. Record the source of the scope and the interview date. A note assembled from memory should be marked as such.

The stop point is the note filed with the case, not a draft in a personal folder. The note must show what the person said and what the interviewer observed in separate passages. A note that reads as one continuous argument does not meet the test. Report the cases with no note separately.

Definitions and evidence standard

Define a separated note before reviewing any case. A quotation, a paraphrase, an observation, and a conclusion each need a visible label or a separate section. A conclusion must point to the fact that supports it. Record the definition with the case so a later reader applies the same rule.

Use the interview note, the scope, and the follow-up log as the primary evidence. A memory of the conversation can add context but cannot replace a dated entry. If only a verbal account exists, mark the case as unverified. Report the share of unverified cases with the result.

Separate a note that separates fact from a note that merely includes both. A page can hold facts and opinions without showing which is which. Measure the structure, not the presence of opinion. The two checks answer different questions about the record.

Sampling and measurement

Freeze the list of investigations before reviewing any of them. Include more than one interviewer, more than one case type, and more than one outcome. Record the export date and the filter used. Do not remove a case because the note is weak, since the gaps are part of the finding.

If the list is large, take a random sample and record the selection method. Keep the full count beside the sample so a reader can see the coverage. A replaced case should be reported as a deviation. The sample should be large enough to compare at least two interviewers.

Measure the number of notes, the share with a separated quotation, the share with a separated observation, and the share with a conclusion tied to a fact. Add the share with a dated scope. Report the counts by interviewer and case type. A single percentage hides which part of the note is weakest.

Findings and interpretation

Present the distribution before suggesting a cause. A cluster of blended notes under one interviewer may mean the template has no sections. A cluster in one case type may mean the note is written under time pressure. Read the notes behind the largest cluster before naming a reason.

Compare the result with the note template. A template with labeled sections will produce more separation than a blank page. Record the template version and the person who owns it. The comparison can show whether the gap is a form problem or a habit.

Check whether notes with separated facts still recorded the questions that remained open. A clean note that hides a gap is not a complete note. Record the follow-up where the file supports it. The quality of the open questions matters alongside the structure.

Limitations

The review measures the structure of the note, not the fairness of the investigation. A well-separated note can still support a poor decision. The result describes the record, not the outcome. Report it beside the decision memo so a reader does not overread the number.

The investigation process may change during the window, and an older case may follow a rule that was appropriate at the time. Record the version rather than applying the current rule retroactively. A case that missed an old standard may not be a current defect. Keep the version note with the case file.

Cases with no written note are invisible to this method. The count may understate the problem if some interviews are never documented. Pair the review with a count of interviews that have no note at all. The answer can reveal a gap the note record cannot show.

Practical next step

Use one note template with separate sections for quotations, observations, conclusions, and open questions. Give the template to every interviewer before the first conversation. A short form makes the structure measurable. Keep the sections short so the note is written before memory fades.

Have a second person read the note before it is filed. A second reader can catch a conclusion that has no supporting fact. Record the correction and the date. Tell the person who raised the concern when to expect an update.

Test the process for one quarter and compare the separation rate with the baseline. Keep the change if it improves the structure without slowing the case. If the template or the reviewer causes delays, fix that step. The goal is a note a stranger can read, not more pages.

Sources

References supplied with this research note.

Article source directory
SourcePublisherContext
U.S. Equal Employment Opportunity Commission, Prohibited Employment Policies and Practiceseeoc.govPrimary reference supplied with this article.
National Labor Relations Board, What We Donlrb.govPrimary reference supplied with this article.
National Institute of Standards and Technology, Privacy Frameworknist.govPrimary reference supplied with this article.