Consistency Between Cases
Why comparable cases get handled differently, how to find out what your own pattern is, and what to do when an earlier case was decided wrongly.
Four comparable matters at one employer, read together
The group was large and the cause was a queue.
Same conduct, different site, different manager.
Explanation accepted; recorded.
Nothing in the file distinguishes it from the row above.
Different conduct; properly distinguished.
Two pairs describe the same conduct and were handled differently, and the files record no reason for the difference beyond a different manager. This is one employer's own set, read together for the first time during a later case.
Nobody compares cases until somebody else does. Each matter is handled by whoever was available, in its own month, without reference to the last one — and the pattern across a year is only ever assembled by somebody challenging an outcome.
The practical lesson in “Consistency Between Cases” is that visibility is not certainty. For teams researching employee time tracking, review the product approach can add time and project context to the operational record, provided the purpose is explained, access is restricted and any material inference is checked through conversation and proportionate human review.
That is avoidable, and the work is small: a list of comparable matters, what was found, what followed, and the reasons given.
A broader reference for the question in “Consistency Between Cases” is the Alberta employment standards. Read it alongside the local facts so that an external framework informs the assessment without replacing case-specific judgement.
Why outcomes diverge
Different managers with different tolerances. Different sites with different norms. Different HR advisers. The order the cases arrived in, and whether the organisation had just had a bad experience.
None of those is a reason that can be written in a letter, which is the test.
Building the comparison
- List matters of this kind from the last two or three years.
- Record the conduct, not the label it was given.
- Record what followed.
- Record what distinguished the cases, if anything.
- Look at the distribution by site, by manager, and by length of service.
- Write down what the pattern is.
Step two is the one that makes it work. Cases are labelled inconsistently — the same conduct appears as misconduct, as a procedural breach and as gross misconduct — and comparing labels compares the labelling rather than the handling.
Distinctions that hold
Different conduct: falsifying a record is not the same as a missed punch. Different scale, different duration, different concealment. Different record: a live warning changes the position. Different explanation: one accepted, one not.
Each of those is a reason that can be stated. If the file does not state it, it did not operate.
Distinctions that do not
Who the manager was. Which site. How busy the organisation was. Whether the person is liked. Whether somebody had recently been annoyed by something unrelated.
Those are the real drivers in a surprising number of cases and none of them appears in any file, which is precisely why the aggregate is the only way to see them.
When an earlier case was handled badly
The awkward finding. Somebody did the same thing two years ago and kept their job, and the file shows no reason.
The honest options are to treat the earlier case as the standard, or to announce that the approach is changing and apply the new standard from a date. What is not available is applying a stricter outcome to the current person while the earlier file sits there unexplained.
Announcing a change of approach
Where an organisation decides to take something more seriously than it has, saying so is both fair and effective.
A notice that from a stated date this conduct will be treated as a disciplinary matter removes the tolerated-practice problem and gives everybody the same starting point. It also disposes of the strongest argument against the first case brought under the new approach.
Keeping the comparison current
A short register: date, conduct, finding, outcome, distinguishing reasons. One line per case.
Maintained as cases arise, it takes a minute each and it is available when the next decision-maker needs to know what the organisation has done before — which is the question they should be asking and currently cannot.
Across sites as well as across time
The comparison that most often shows something is geographic. Two sites with the same operation, the same systems and the same policy produce different outcomes for the same conduct, and the difference is the managers.
That is not misconduct by anybody. It is what happens when judgement is exercised without reference to how it has been exercised elsewhere, and the only remedy is making the comparison available.
Who holds the register
Somebody central, because the whole point is the view across managers and sites. A register held by each site is four registers and no comparison.
It also has to be consulted rather than merely maintained. Adding a line to the investigation template — what comparable matters have been handled before, and how — is what turns a record into a practice.
What to do with an outlier
Where the register shows one outcome well outside the pattern, the question is whether it was wrong or whether something distinguished it.
Either answer is useful. A distinction that was never written down can be written down now; an outcome that was simply harsher than comparable ones is a fact the organisation needs to know about itself before anybody else establishes it.
What the register is really for
Not to constrain decisions mechanically. Cases differ and judgement is required.
It is to make sure the judgement is informed, and to make sure that when somebody asks why this person was dismissed and that one was not, there is an answer written down rather than assembled under pressure.