# Week 9 reform evaluation proposal

**Student / date:**  
**Working title:**  
**Status:** Proposed study; supplied lab data are original synthetic teaching constructions, not empirical replication.

Write two pages plus this component/outcome ledger. Choose a feasible reform. State actual data access and leave unavailable records explicitly unavailable.

## Decision, population, and comparison

Name the decision-maker, initial eligible establishments and workers, job/workflow, intervention version and control regime. Define the primary contrast: bundle, conditional component, factorial marginal, interaction or rollout policy. Which treatment combinations are feasible and observed? What would a two-arm design leave unidentified?

Record announcement, assignment, activation, competence assessment, first substantive use, support withdrawal, maintenance, changes and control crossover. Give the assignment/dependence unit, number of independent assignments, outcome horizon, inference procedure and exact null. Explain why actual use cannot simply replace assignment.

## Component and implementation ledger

| Component / version | Who assigns and implements it? | Assignment and eligibility | Delivery / use / maintenance evidence | Timing and control exposure | Setup, recurring and worker costs |
|---|---|---|---|---|---|
| Tool or practice | | | | | |
| Workflow, review or authority change | | | | | |
| Training, support or compensation | | | | | |

## Outcomes and incidence

| Outcome family | Unit, denominator and initial population | Source and independent validation | Horizon / observation rate | Observed, elicited, modeled or missing? | Firm / worker incidence and uncertainty |
|---|---|---|---|---|---|
| Verified output and independent quality | | | | | |
| Total inputs, paid/unpaid time and resource costs | | | | | |
| Worker conditions, control and burdens | | | | | |
| Retention, closure, composition and destinations | | | | | |
| Implementation and sustained routines | | | | | |

Keep numerators and denominators separate. Distinguish output from rates, time saved from changed logging, and worker conditions from firm profit. Transfers cancel in joint-resource accounting; real burdens remain. State discounting, running costs and terminal assumptions before reporting value.

## Competing mechanisms and decision criteria

| Mechanism | Distinguishing prediction | Required endpoint or variation | Remaining ambiguity |
|---|---|---|---|
| Explanation 1 | | | |
| Explanation 2 | | | |

Name a result that would change the organization's decision. If equivalence is relevant, prespecify its meaningful margin and a suitable procedure; nonsignificance is insufficient. Explain what the design cannot identify.

## Spillovers, attrition, and feasibility

Draw a small exposure map for collaborators, supervisors, shared review capacity or infrastructure. Specify contamination and how a pilot differs from full rollout. Explain the independence assumption and any change in assignment level it requires.

Preserve initial cohorts. Describe follow-up for exits, closures, nonresponse and destinations, with observation rates by assignment. Do not replace missing destination earnings or conditions by zero. A closure may give zero establishment production for a defined endpoint without implying zero former-worker welfare.

State permitted sources, linkage authority, privacy protections and actual access. Do not contact providers or assume confidential logs are available. End with the most informative feasible next measurement and its cost.