# Hypotheses

## Primary questions
- H1: EMF or related environmental readings are higher in the NWI corridor than in matched control sites.
- H2: Peaks cluster by source type or operating cycle.
- H3: Exposure differs by time of day and weekday/weekend activity.
- H4: If a survey is added later, symptom reports should only be compared after exposure is characterized.
- H5: Utility-adjacent sites and energized equipment states produce distinct readings from inland controls.
- H6: Electric AMI module installs, gas AMI module installs, and AMI meter exchanges are distinct utility-operation subtypes that should be analyzed separately from generic utility-adjacent stops.
- H7: Readings fall off with increasing distance from an identifiable source and with shorter dwell time near that source.
- H8: Cleanup subarea and managed corridor context remain distinct from exposure signal once route metadata are split out.

## Make each hypothesis testable
For every hypothesis, define:
- the measurement
- the comparison group
- the sampling window
- the threshold for calling it meaningful

## Example success criteria
- Corridor average exceeds control average by a pre-set margin.
- Repeated peaks occur at the same locations or time blocks.
- Results replicate across at least two sampling sessions.
- Utility-adjacent sites are elevated only when the source is energized or under load.
- AMI subtypes separate cleanly once module-install, exchange, and access-status metadata are split out.
- Source-distance bins or source-dwell bins show a repeatable gradient rather than a one-off spike.
- Managed corridor geometry and cleanup subareas separate cleanly from cleanup phase and source-state.

## What not to do
- Do not start with conclusions.
- Do not collapse exposure and health into the same step.
- Do not treat one weird reading as a theory.