Direct answer: a worked example of revisions
A “worked example of revisions” is a step-by-step scenario that shows how a previously published value changes after an update, and how that change affects calculations. Here, “revisions” means the act of updating earlier reported information—typically because new or corrected inputs become available.
Worked examples focus on transparency: you state every assumption, you keep the scenario’s other conditions fixed, and you show how the revision alters results.
Mechanism: how revisions work in a calculation
To understand revisions, separate two parts:
-
The revised input: an original figure is later replaced with an updated figure (the revision). For example, a “first estimate” might be updated after additional data collection.
-
Your calculation method: the rule that converts inputs into outputs (for example, computing a change from one period to another, or applying a percentage to a revised base).
A worked example typically compares:
- Outcome A using the original number
- Outcome B using the revised number
Key assumption: the method stays the same. Only the input changes.
Worked scenario example (with every assumption stated)
Assume a simple index update process used by an analyst to track a metric over two months.
Assumptions (fixed for the entire example)
- You want to compute a two-month change as a percentage.
- The formula is: Percent change = (New − Old) / Old × 100.
- “Old” is taken from an earlier publication for Month 1.
- “New” is first published for Month 2, but later revised.
- No other data changes (timing, definition of the metric, and unit scale are identical).
- Roundings: keep results to one decimal place at the final step.
Step 1: Original publication
- Month 1 (Old) = 100.0
- Month 2 (New, original) = 105.0
Compute percent change:
- (105.0 − 100.0) / 100.0 × 100 = 5.0%
Step 2: Revised publication
Later, Month 2 is revised:
- Month 2 (New, revised) = 103.0
Recompute percent change:
- (103.0 − 100.0) / 100.0 × 100 = 3.0%
Step 3: Compare outcomes
- Original computed change: +5.0%
- Revised computed change: +3.0%
- Difference due to the revision: −2.0 percentage points
This is the core value of a worked example: it shows exactly what changed and quantifies the impact under controlled assumptions.
Limitations and risks (material failure modes)
Even a clear worked example has important limitations:
-
Change in definition, not just the number: A “revision” might include methodological changes. If the meaning of the metric shifts, redoing arithmetic alone may not be enough.
-
You may be comparing incompatible versions: If “Old” and “New” come from different publication cycles or different vintages, your comparison can mix assumptions unintentionally.
-
Second-order effects: Many real analyses use revised outputs downstream (for example, forecasts or models). A revision can cascade through multiple dependent computations.
-
Omitted costs and frictions: In real environments, timing, access, and transaction-related frictions can make the practical outcome differ from a purely numerical recalculation.
-
Historical relationships are not guarantees: Even if revisions correlate with later outcomes in one context, that does not establish what will happen next.
Verification: how to independently check a revision
To verify a revision-driven calculation, you can:
- Identify the exact metric and the relevant time periods.
- Locate both the original and updated publication values.
- Recompute your formula with the updated input while keeping your stated assumptions constant.
- Record whether any definition or methodology changed, not only the numeric value.
If you can’t find the original and revised values (or if definitions changed), then the worked calculation becomes less verifiable and you should treat the output as conditional on what can be confirmed.