A plan for 59 surveys is not 59 released datasets
Statistical modernisation needs separate evidence for approval, execution and release before progress can be evaluated.
Where should modernisation progress be counted?
When an official statistical system expands, investment reviewers need to know which activities have started and which outputs can actually be used. A table that mixes planned surveys with released datasets cannot answer that question. Truthound can check the declared stages and required evidence in a programme register. It does not evaluate statistical representativeness, determine public-policy effectiveness or turn approval into completed delivery.
An approval describes work still to come
India’s Cabinet approved the VISTAR statistical-transformation scheme on 4 November 2026 for the period through financial year 2030–31. The announcement describes expanded surveys, digital statistical processes, state capacity and an economic census. Its 59 surveys are proposed activities. Recasting that figure as surveys already conducted or datasets already released would remove the programme’s actual stage from the record.
Model activities and outputs separately
A proposed register treats the programme, survey activity and published output as different entities. Record the approval reference, stage, responsible institution and period for each activity. A released dataset needs its own publication reference. One survey can produce several outputs, while an output can draw on several surveys. Do not assume a 1:1 relationship simply because a dashboard prefers a single row per activity.
Allocate attention to the next bottleneck
A planned activity may need design and staffing review; an executing survey may need collection support; a pre-release output may need quality and disclosure review. Rather than ranking every item by one progress percentage, identify the evidence required for its next stage. If an activity says released but lacks a publication reference, ask the owner to resolve it before including it in a delivery total.
Check evidence required by a stage
Truthound Core Python SDK 3.3.4 (open-source SDK, Polars backend) documents ConditionalExpressionValidator for caller-defined conditional row rules. An operator could require a particular reference for a declared stage, while InSetValidator checks the approved stage vocabulary. Programme staff define the expressions and null semantics. A populated publication reference does not automatically prove that a dataset is accessible, that its methods are sound or that its disclosure review was completed correctly.
Compare against the process already maintained
Pandera 0.35.0 (MIT open-source Python DataFrame validation library; pandas, Polars and other supported backends) declarative and custom checks provide another route to the same stage contract. If an established planning process already maintains those rules and exceptions, inspect its evidence-free completion states before replacing the tool. Choosing Truthound still leaves institutions responsible for supplying and interpreting evidence. Documented validation features do not establish that a programme delivered its intended policy benefits or that one library improves delivery outcomes.
Preserve ambition without overstating completion
Start by tracing one survey through the evidence that admits it to each next stage. Do not calculate an overall completion rate while activity identity or the meaning of released remains unresolved. Preserve plans as plans and count usable published outputs separately. That distinction lets a resource discussion consider both the programme’s intended scale and its confirmed delivery, without allowing either to conceal the other.
Conditions
Separate activities from outputs and define institution-owned evidence for each stage.
Alternatives
Retain a Pandera 0.35.0 contract that already maintains stage and evidence requirements.
Limitations
A planned survey count is not a count of completed surveys or released datasets.
Evidence
- Event checked
- Subject checked
- Modified
