Quick answer
Find the manual's age rule before you calculate. Where it is silent, follow a documented institutional rule and record which date you used.
Four dates that are not the same
A single assessment generates several candidate dates: referral, consent, administration, scoring, and report completion. They can be weeks apart. Only one of them governs the age calculation, and for most norm-referenced instruments it is the administration date.
The reason the wrong one gets used is almost never confusion about which is correct. It is that the file already contains the referral date, the scoring session is when the calculation actually happens, and a date field pre-filled with today requires an active decision to change.
Use this guide with the right tool: Open the Assessment Age Calculator for age on a specific testing or evaluation date. If the question shifts, compare it with the Test Age Calculator or the Age on a Date Calculator.
Read the age rule first, not last
The sequence that prevents most errors is to locate the manual's age instruction before opening the calculator. It takes a minute and it determines both the date and the format. Calculating first and checking afterwards means the number is already in the file, and numbers in files are rarely re-derived.
Where the manual genuinely does not specify, that gap should be filled by a written institutional rule rather than case-by-case judgement. A team that decides individually will produce inconsistent records, and the inconsistency will only surface when two reports on the same child disagree.

Multi-session and interrupted testing
Testing that spans days, or that is interrupted and resumed weeks later, needs an explicit decision. Common rules include the first session, the session containing the subtest being scored, or the date most of the battery was completed. All three exist in practice.
Whatever rule applies, document the circumstance in the report rather than only the resulting date. A reader who sees an age that does not match the cover-sheet date needs to be able to find the explanation without contacting you.
Retrospective calculation
Sometimes an age has to be reconstructed long after the fact — for a re-score, an appeal, a records request or a research dataset. The rule is the same: the age is the age on the administration date, however long ago that was.
This is where a default reference date does the most damage, because the resulting figure will be years out and will still look like a plausible age for a child. Set the reference date explicitly on every retrospective calculation, and keep the shareable result link with the record so the inputs are visible.
Worked example
Run your own birth and session dates, then let the test manual choose between them.
When the sessions straddle a band edge
The worked example above has a three-day gap between sessions, and the two ages it produces differ only in the days field. Lengthen the gap and the choice of session stops being a formality. Take the same child, born 3 May 2019, with a battery split by an interruption: the first session on 28 October 2026 and the second on 5 November 2026. The first-session age is 7;05;25. The second-session age is 7;06;02. Between the two sessions, on 3 November 2026, the child passes 7;06;00 exactly.
That crossing is what makes the manual's rule consequential rather than clerical. Where a manual bands its norm tables in six-month intervals — a block running 7;00 to 7;05 and the next running 7;06 to 7;11, a common arrangement — the two session ages do not merely differ, they select different tables. The first session is scored against the younger band, the second against the older one, and the same raw scores can convert to different scaled results depending on which session date governs. A three-day gap like the worked example's almost never crosses a boundary; an eight-day gap in late October, for this child, does. Whether a split battery matters is a question about where the sessions fall relative to the child's half-year lines, not about the gap in the abstract.
The retrospective version of the same error is quieter. Suppose this child was tested earlier, on 22 April 2026, when the age was 6;11;19. Months later the battery is re-scored, and whoever runs the calculation leaves the reference date at the scoring date, 2 June 2026. The calculator obligingly returns 7;00;30 — the age the child would have been on the day of the re-score, 41 days after the testing actually happened. The figure is internally consistent, plausibly formatted, and wrong: it describes a child six weeks older than the one who sat the test, and in a banded table it can sit in a different band from the correct 6;11;19. Nothing about the number looks like an error, which is precisely why the reference date has to be set deliberately and the administration date recorded beside every stored age.
The habit that covers all three cases is the same. Record both session dates whenever testing is split, name the rule that chose between them, and store the age that rule produced. On any later re-score, set the reference date back to the administration date before calculating — never accept whatever date the tool opens with. The ages in this article — 7;04;11 and 7;04;14, 7;05;25 and 7;06;02, 6;11;19 against a phantom 7;00;30 — are all correct answers to slightly different questions. The assessment record's job is to make clear which question was asked.
Common mistakes to avoid
- Accepting a pre-filled today. A default is not a decision. Set the reference date deliberately on every calculation, including retrospective ones.
- Using the scoring or report date. Scoring often happens days later. For most norm-referenced instruments the administration date governs.
- Inventing a midpoint for multi-session testing. A midpoint is an undocumented method. Follow the manual, or a written institutional rule, and record which.
Use the right calculator
Use the Assessment Age Calculator for age on a testing date, the Test Age Calculator for age at a test, and the Age on a Date Calculator for past or future dates.
Frequently asked questions
What if the manual really says nothing?
Use the administration date and record that you did. Then raise it so an institutional rule can be written and applied consistently.
Does a rescheduled or interrupted session change the date?
It can. Whichever rule applies, document the interruption alongside the age so the record is self-explaining.
Should the reference date appear in the report?
Yes. An age without its reference date cannot be verified, and it silently becomes wrong as time passes.


