A missing subtotal is not a reported zero

When reading a Taiwan university student census history, distinguish the original cell from the number available for comparison. CollegeVerse retains printed source values. A sex subtotal printed as a space followed by a dash remains a source marker and has no numeric value in the corresponding subtotal field.

The catalogue text checked for this dataset does not explain that marker's meaning. It should therefore not be relabelled as zero, confidential, suppressed or not applicable without further official evidence.

Can the missing subtotal be calculated from the total?

Suppose a table gives a total and one available sex subtotal, while the other subtotal is a dash. Subtracting the available subtotal from the total produces an arithmetic remainder. It does not establish the value the source intended to report in the missing cell.

CollegeVerse does not reconstruct that missing subtotal. Keeping the original marker makes the limit visible to the reader and preserves the evidence if the source later clarifies it.

Older files may not contain newer columns

The retained student census spans multiple file layouts. A column absent from an older layout differs from a column present with a printed zero. If a newer file adds preparatory or year-level fields, older records do not acquire those fields retroactively.

Check the qualification and class labels as well as the year. The official catalogue describes a preparatory-population coverage change from year 113. Comparisons that ignore layout and coverage changes can attribute a difference to student growth when the available evidence does not support that conclusion.

A practical reading sequence

Open a record from the Taiwan student census and read the exact population labels. Then inspect the annual total, original male/female cells and year-level cells. Where the page identifies a source marker, preserve it in your notes rather than filling it with an estimate.

If you calculate a difference between two available totals, label it as your calculation and keep both original observations. A difference between historical census totals is not an applicant count, graduation count or admission forecast.

When another directory shows a different number

The department overview and the named student census do not use an automatic code or name crosswalk. Different populations can produce different counts without either value being a transcription error. Inspect the file, year, qualification, day label and named class before combining observations.

If an exact source cell appears to have been transcribed incorrectly, retain its source URL, original row identity and year when reporting it. Those details make a correction traceable and avoid replacing one uncertainty with another.