Desaxuste: Table Check

Desaxuste is a specialized, privacy-focused utility designed to check whether categorical tables in a report or manuscript can describe one consistent whole-number population.

When evaluating survey findings or published research tables, each table might independently sum to 100 percent. However, when multiple cross-tabulations or marginal distributions are considered jointly, they can sometimes make mutually impossible claims about the underlying population. Desaxuste evaluates whether a valid joint distribution exists under explicitly stated assumptions.

Key Capabilities

Source-Linked Review: Review categorical tables cell by cell alongside your source document. Set and confirm variable categories, counts, percentages, and rounding thresholds.

Explicit Assumption Controls: State whether tables refer to the same cohort, the same observation time, identical category definitions, and how missing responses are handled. Nothing is linked automatically.

Exact Joint Integer Verification: Runs a bounded exact search over integer population models. If the entered tables cannot describe one whole-number population under your reviewed assumptions, Desaxuste highlights the specific conflicting cells and assumptions. If the tables are compatible, it renders an example population that satisfies all conditions.

Structured Question and Record Export: Export a readable PDF report and verifiable JSON record through Android's Storage Access Framework. The export documents reviewed values, assumptions, and mathematical findings without altering source documents.

Local Privacy and Data Protection: All data is stored locally in encrypted storage on your device. Desaxuste requires zero network permissions, features no accounts or tracking, and includes an optional dedicated app PIN or biometric lock to safeguard sensitive research drafts.

Operating Limits

Desaxuste is a mathematical consistency tool, not an empirical fact checker. A verified contradiction proves only that the entered claims cannot simultaneously hold under the selected assumptions. A compatible result shows only that at least one integer population exists that fits the reported numbers; it does not verify that the report is true or authentic. Desaxuste does not automate misconduct determinations, authenticate documents, or provide legal or statistical certification.