Skip to main content
Through 3.3 · A practical walkthrough

Reconciling two counts

Two reports. Two totals. Before changing a number, find out what each report is actually counting.

A fictional enrollment investigation · About 8 minutes
Our setting: Sierra Vista Middle School

SDLA’s fictional urban middle school serves 800 students in grades 6–8. Its connected systems still depend on staff to reconcile transfers, schedule changes, and overlapping enrollments.

The school context comes from the SDLA school profiles. All report dates, records, staff actions, and findings below are invented for this learning exercise. These are local enrollment reports, not CALPADS certification results.

01 / Unify

Put both reports on the table

Keep the totals attached to their definitions, dates, and sources.

Report A · Current SIS roster
800

Unique enrolled students

As of
November 9, 2026 · end of day
Generated
November 10 · 8:00 a.m.
Scope
Sierra Vista · grades 6–8
Count rule
One student per local ID, actively enrolled on the as-of date
Excludes
Future starts and students no longer enrolled on that date
Source / owner
SIS enrollment records / school registrar
Report B · Saved weekly snapshot
797

Unique enrolled students

As of
November 6, 2026 · end of day
Generated
November 6 · 6:00 p.m.
Scope
Sierra Vista · grades 6–8
Count rule
One student per local ID, actively enrolled on the as-of date
Excludes
Future starts and students no longer enrolled on that date
Source / owner
Saved SIS extract / district data analyst

The difference is 3 students. What would you check before calling it an error?

1. Reporting date and refresh time

“As of” tells you which day the records describe. “Generated” tells you when the report was produced. Check both.

2. Population and exclusions

Confirm school, grades, enrollment status, and filters. Similar titles can hide different populations.

3. Unit of record and identifiers

Count unique students, not course enrollments or file rows. Match on approved identifiers, not names alone.

4. Source, transformation, and owner

Trace the source extract, filters, joins, and feed status. Name the person who can explain or correct each step.

02 / Simplify

Follow an explanation to the evidence

Select a possible cause to see what it would take to establish it.

Supported in this fictional example

The reports describe different days

Three students began enrollment on November 9. They belong in the November 9 roster and do not belong in the November 6 snapshot. The exercise assumes no exits or other membership changes between those dates.

797 + 3 new enrollments − 0 exits = 800
Invented evidence reviewed for the exercise
RecordEffective startNov. 6 snapshotNov. 9 roster
Demo student ANov. 9ExcludedIncluded
Demo student BNov. 9ExcludedIncluded
Demo student CNov. 9ExcludedIncluded

Verification: The registrar checks the effective start dates. The analyst reruns the SIS report as of November 6: 797. Both November 6 student-ID sets match, with no unmatched IDs or duplicates.

Resolution: Keep both original totals with their dates. No source correction is needed. Choose the dated population that answers the question being asked.

Matching totals are a starting point. Two reports can both show 797 and still contain different students. Compare the record sets as well as the totals.

How to carry out the record comparison

Preserve both original reports and their parameters. Produce comparable extracts for the same as-of date, school, grades, and enrollment rule. Check that the local student ID is unique at the intended grain before matching records.

List records found only in A, only in B, and in both. Review enrollment dates and status for every unmatched record. Equal counts can hide offsetting differences, so resolve both lists. Have the registrar verify source facts and the analyst verify report logic; rerun affected outputs after an authorized correction.

03 / Multiply

Leave an explanation someone else can use

Share the finding, the evidence, and the next action together.

The note below starts with the resolved different-dates example. Exploring other explanations does not change your note. Edit it for the exercise, then print or download it. Changes stay in this open page and are lost on reload; download before closing. Keep student-level details in your approved records location.

School Data Leadership Association · Through the Data
Fictional learning exercise · Sierra Vista Middle School

Reconciliation resolution note

School registrar and district data analyst
November 10, 2026 (exercise date)
Why does Sierra Vista’s current SIS roster show 800 students while its saved weekly snapshot shows 797? Report A: active unique students as of Nov. 9, 2026, generated Nov. 10 at 8:00 a.m. Report B: same school, grades 6–8, and count rule as of Nov. 6, generated that day at 6:00 p.m.
The reports describe different dates. In this fictional exercise, three students started Nov. 9; there were no exits or other membership changes between the two dates. The registrar verified the start dates. A Nov. 6 as-of rerun returns 797, and the analyst confirms that both Nov. 6 student-ID sets match with no unmatched or duplicate IDs.
No enrollment correction is needed for this example. Use 800 for the Nov. 9 population and 797 for the Nov. 6 population. The district data analyst will label both as-of dates and preserve the report parameters, originals, comparison results, and this note. The school registrar will review any new discrepancy before a source change is made.
Exercise only — no actual evidence files attached
Illustrative completed review; not a real approval

Use the same method locally: preserve the evidence, confirm the explanation, and record who verified it. A selected explanation alone is not proof.

In the print dialog, select your printer or Save as PDF. The note prints your current entries with room for wrapped text.