An imaginary editorial team read a time series in spring and calculated a change from it. In autumn it opens the source again. The figure for the same earlier period is different. Before the headline “Market suddenly collapses” emerges, two things must be separated: which period was observed and what publication version is in place?
Official statistics can revise data retrospectively. Destatis explains that provisional results in certain areas are based on still incomplete foundations and are later revised when the data situation is completed. [DE-REVISION] The article uses invented figures to show how an editorial team can classify such changes in a traceable way. It does not report an actual current real-estate trend.
The two timelines at a glance
In the working paper reporting period and publication version are kept separate. “Year 1” in the following model is the period under consideration; “Version A” and “Version B” are two different publication versions. A retrieval date is added as documentation, but it does not become a new observation period.
- Year 1 – Fictitious Version A: 100; Fictitious Version B: 102
- Year 2 – Fictitious Version A: 110; Fictitious Version B: 108
The numbers are our own teaching values. They were not taken from any market report. In Version A the change from Year 1 to Year 2 is ten per cent. In Version B it is 108 divided by 102 minus one, that is roughly 5,88 per cent. Both calculations refer to the same two model periods, but to different data versions.
The difference between the results is not automatically an additional market movement between spring and autumn. Appropriate data for this new period would be required for that. Here the earlier observation has been revised.
Do not mix a new figure with an old base
The editorial team could by accident divide 108 from Version B by 100 from Version A and report eight per cent. That would be a self‑mixed calculation that does not correspond to either of the two fully presented versions.
Before publication it is therefore checked which values and definitions belong together for the intended comparison. For a current analysis the appropriate up‑to‑date series version is read in full. For a historical question about what was known at an earlier time, the earlier version with its contemporaneous information remains relevant.
These two tasks may coexist. An older report is not quietly rewritten as if the new figure had already been available then. Conversely, a current explanation does not have to continue the earlier value as an unchanged present‑day figure.
Read the reason from the specific source
Not every change arises in the same way. Destatis distinguishes in its revision presentation, among other things, routine revisions and method‑change induced revisions. The specific cycles depend on the respective statistic. [DE-REVISION]
For Austria Statistik Austria describes a documented distinction between scheduled and event‑related revisions as well as supplementary information in standard documentations. [AT-REVISION] The German procedure is therefore not presented as a blanket Austrian calendar.
The editorial team looks for the exact revision note for the series it actually used. If the cause is not clarified, it does not write its own explanation. A new figure alone proves neither manipulation nor a harmless routine case. Both would be more than the available observation.
What conclusion remains after the revision?
In the model the increase was positive in both versions, but different in magnitude. The editorial team therefore also checks the previous interpretation. Did the old headline make a claim that is no longer supported by the revised version? Does only the number need to be changed or also the interpretation?
A published decimal place is not more important than the core message. If the direction or the decisive comparison changes, this is explained to readers in an understandable way. With a limited change the explanation remains correspondingly factual, without turning it into a dramatic event.
All follow‑on content that actually depends on the changed figure receives an appropriate review note. The editorial team does not have to touch every unrelated page. But it must know where it has reused the affected statement.
A small note with great benefit
One’s own source note contains the series, reporting periods, publication version, values used and calculation method. If two versions were compared, the difference remains traceable. The original source is linked; foreign tables or images are not simply presented as one’s own depiction.
The workshop result is a more precise message: “The source has revised the values for the earlier periods; under the stated conditions the current comparison turns out differently.” Whether the market has moved again in the meantime remains a question to be evidenced separately. This very separation prevents a data revision being told as new market activity.