Keep missingness visible before summarizing or comparing a pet's history. Otherwise, a tidy result can describe evidence you never collected.
Four states that should stay distinct
| State | Example | What it means |
|---|---|---|
| Recorded zero | A measured count of 0 over a stated complete window | A numeric observation |
| Missing | No amount was recorded | No available numeric observation |
| Partial | 80 mL measured only until noon | An observation covering part of the intended window |
| Unknown completeness | 80 mL without a stated window | The amount exists; its coverage is uncertain |
The examples describe recording states. They are not recommendations or clinical interpretations.
A complete day can contain a zero. A partial day can contain a large amount. The number alone does not establish coverage.
A fictional evidence table
Fictional record: Period A has three complete-day measured drinking amounts: 100, 110 and 120 mL. Period B has two complete days: 140 and 150 mL, plus an 80 mL partial day.
The primary comparison includes three days in A and two in B. It excludes the partial entry from the complete-day mean, while retaining it in the visible record.
A's mean is 110 mL and B's is 145 mL. The difference is 35 mL, approximately 31.82% relative to A. These values describe the fictional complete-day entries only.
If the diary covers a longer intended period, the other days remain missing. Do not imply that the two included B days represent a fully observed week.
Why filling gaps changes the question
Replacing a blank with zero asks “what is the average if we assume nothing was recorded because nothing occurred?” That assumption is not supplied by a blank.
Repeating the previous day's value invents a new observation. Expanding a partial-day amount to a full day invents the unobserved part of the window. Either may make a chart look smoother while reducing its truthfulness.
For this guide's comparison method, use eligible recorded entries and report coverage. Do not fill the missing values.
Which next observation would help?
Ask what makes the existing records hard to compare:
- Is the unit absent? Add a unit to the next measured entry.
- Is the window unclear? Record the start and end.
- Was collection interrupted? Mark partial coverage.
- Did another pet share the source? Record the shared context.
- Did measurement become estimation? Keep the method change visible.
- Is the source unclear? Add a source-note reference.
- Are there too few complete observations? Acknowledge the limited basis rather than claiming a trend.
More detail is useful when it answers a specific uncertainty. A longer note is not automatically better evidence.
Missingness and a personal reference
A selected period mean is not automatically the pet's usual pattern. A maintained personal reference has a different basis and supported eligibility. Read the distinction.
Coverage describes the record. It does not score the pet's health. A day with many entries can still contain uncertain observations.
Preserve gaps in a handover
When preparing a summary, state the intended period, complete-day count, partial entries and missing windows. Preserve the original log so the reader can see the basis.
Use the blank template for future entries. If your pet needs attention, do not wait to complete a log before contacting a veterinarian.