When to use it
Two reports appear to show a change over time.
Property decision guide
Comparing Property Data Without Misleading Conclusions
Two figures can look comparable while describing different populations or measurement rules. A useful comparison begins with a small definition record and ends with a conclusion that preserves those limits.
Align population and geographic coverage
Name the property, buildings, accounts or residents covered by each measure. For regional statistics, retain the issuing agency’s geography. The site’s Hillsborough-plus-Pinellas totals and BLS Tampa-St. Petersburg-Clearwater metropolitan measures describe different geographic scopes.
Counts can be added when populations and definitions are compatible and do not overlap. Medians cannot be combined by averaging two published median values. A pooled rate needs the corresponding numerators and denominators, not simply the mean of two county percentages.
Match time, unit and collection method
Distinguish an observation date, a reference interval, a publication date and an editorial review date. A review performed today does not make a historical observation current. Compare equivalent reporting periods and preserve preliminary or revised status.
For property metrics, document whether elapsed time uses calendar or business days and whether a cost includes tax, credits, wind coverage or another component. A changed definition should be flagged in the series. Recalculate the earlier period on the same basis only when the underlying records support it.
Reconcile the denominator before explaining a trend
State the population eligible for the metric, the number observed and the number excluded. A closure rate may improve because fewer difficult actions remained in the denominator. An inspection or study submission count does not become a compliance rate without a validated count of the entities required to submit.
Keep unavailable data distinct from a real zero. For a percentage, define what happens when no eligible items exist. Report the measure as not applicable or unavailable as appropriate rather than returning a healthy-looking zero. Investigate changes in coverage before attributing a result to improved management.
Write an interpretation that can be checked
Record the source, reference period, calculation, rounding and conclusion beside the comparison. Separate the observed result from a hypothesis about its cause. Identify the next evidence needed when a causal explanation is not yet supported.
Hypothetical example: one provider closes 18 of 20 due actions and another closes 27 of 30. Both have a 90% closure rate, but the actions may differ in difficulty, age and service impact. Use the shared definition, show overdue critical items and examine the supporting record before selecting a provider based on the rate.
Hypothetical worked example
Two buildings use different eligibility rules
Two fictional buildings report a completion percentage but use different periods, asset boundaries and waiting-job exclusions. Ranking the numbers would not answer a common question.
Original fictional example · no completed client assignment or property inspection is represented.
| Record to obtain | Review question or calculation |
|---|---|
| Formula and denominator for each report | Align definitions where feasible. Rebuild each numerator and denominator from records using one stated eligibility rule before comparing percentages. |
| Coverage and source cutoff | Explain remaining population differences. Where exclusions cannot be aligned, present separate results and the specific limitation rather than a league table. |
| Status and exclusion rules | Decline an unsupported ranking when alignment is not possible. Check whether a changed percentage reflects completed work or merely a changed population; document both movements separately. |
Useful working output: A comparison note stating which conclusions the data can and cannot support.
What would change the decision?
Open each challenge to test the reasoning against a different finding.
Two counties are compared using different estimate vintages
Check: Keep geography, reference date, estimate vintage and unit together. A new Census vintage can revise earlier years as well as add a year.
Decision effect: Use a consistent series for the change calculation and label any unavoidable difference in coverage.
An average is mistaken for a median
Check: Read the source definition and population base. A median describes the middle observation; a mean reflects the sum divided by the number of observations.
Decision effect: Correct the comparison label and method before interpreting an apparent property-cost gap.
Adapt the record request to authorized access, governing documents and the actual property. The example does not verify any unknown input or replace a professional conclusion.
Resident Survey Data Quality: Denominators, Coverage and Privacy · More property data and metrics examples
Before acting on this guide
Set the review boundary and the acceptance test.
What the evidence does not prove
Do not compute a trend before confirming a common population and method.
What to verify before acceptance
Record geography, unit, cutoff and methodology differences; separate a comparable subset from unsupported comparisons.
Original proposed coordination checks · adapt to authorized access, the actual obligation and the appropriate professional review. These are not findings about your property.
Follow the related handoff workflow · Read the review definitions
A number, its evidence and the decision
Keep the denominator attached to a result
A percentage can change because the event count changed, the population changed or the definition changed. Show the numerator, denominator and period before describing a trend.
Definition
State the event and eligible population.
Source
Identify the authoritative records and period.
Calculation
Apply the same unit and denominator.
Interpretation
Check selection effects and missing records.
A worked example with visible assumptions
A fictional survey receives 47 unique valid household responses from 120 eligible households invited. The response rate is 39.2% after rounding. It describes participation in this survey, not the opinion of every resident. Multiple responses from one household require a defined reconciliation rule.
Original hypothetical arithmetic example. The amounts describe a fictional scenario, not observed property performance or a recommended target.
Continue with Comparing Property Data Without Misleading ConclusionsExplore the calculation
Enter non-negative values using the same scope and period. This calculation stays in your browser tab.
Counts can add; rates need weights
Separate, non-overlapping counts for the same period can be added. Rates need their denominators; medians cannot be pooled by averaging them.
Label the geography
The Tampa metropolitan area and Hillsborough–Pinellas county pair are different populations. Keep their measures separate.
Preserve the observation period
An editorial update does not make an old observation new. Keep provisional status, survey period and publication date visible. Run the same definition on both populations before ranking results. If one source counts tickets and the other counts buildings, keep the figures separate until a documented mapping exists; a common percentage format does not make the units comparable.
Frequently asked questions
Can two county median rents be averaged into a regional median?
No. A valid regional median requires the relevant distribution or an agency-published measure for that geography. Keep the two county medians separate.
Is a newly reviewed statistic the latest observation?
The review date and observation period are separate. Check the issuing source for the current release and retain the reference period.
What should a zero-denominator percentage show?
Define its treatment explicitly. If no eligible items exist, report not applicable; if the denominator is missing, report unavailable. Neither condition establishes a 0% performance result.