# Week 6 proposal: career transitions and opportunity sets

**Length:** two pages plus a denominator table.  
**Name / setting / date:** …  
**Status:** research proposal; label any illustrative numbers synthetic.

## 1. Population, boundary and opportunity set

Define baseline workers, organizational boundary, grades/job families and eligible openings. Record which positions are created, visible, frozen, filled or cancelled. Explain eligibility before treatment and whether appointments without applications use a separate route. Define current risk-set membership without dropping people from the baseline policy cohort after exit.

## 2. Intervention, comparison and clocks

Specify one intervention or exposure: offer, course receipt, disclosure, confidential application, release rule, developmental assignment or opportunity supply. Name the assignment/comparison mechanism and exact unit. Separate nomination, offer, enrollment, completion, assessment, certificate disclosure, application, promotion offer, release and actual start dates. Address anticipation and repeated exposure. Explain why the comparison would support the stated target, and the strongest remaining threat.

## 3. Competing mechanisms

| Mechanism | Distinct observable prediction | Counterprediction / potential falsifier | Additional data needed |
|---|---|---|---|
| Mechanism 1: … | … | … | … |
| Mechanism 2: … | … | … | … |

Distinguish skills, beliefs, effort, assignment, access and opportunity supply. State whether a common assessment might miss skills specific to a new task. Do not interpret persistence alone as learning.

## 4. Denominators and outcomes

| Outcome / gate | Numerator | Denominator / risk set | Clock and horizon | Post-exit observability | Causal target or descriptive diagnostic? |
|---|---|---|---|---|---|
| Training offered / received / completed | … | … | … | … | … |
| Application | … | … | … | … | … |
| Promotion offer / release / actual start | … | … | … | … | … |
| First promotion before exit | … | … | … | … | … |
| Exit / retention | … | … | … | … | … |
| Independent common-task or quality measure | … | … | … | … | … |
| Worker welfare | … | … | … | … | … |

Report baseline-cohort rates alongside conditional gate rates. Preserve missing outcomes after exit as missing unless a substantive zero is justified. List total slots and competitors; distinguish own advancement from displacement and total production. A completion flag is not an assessment, and a routine rating is not automatically quality.

## 5. Inference, access and claim limits

Define the assignment/support and dependence unit, primary outcome family, planned diagnostic and uncertainty method. Explain direct effects, peers and fixed-slot competition. Distinguish sharp-null inference from heterogeneous average-effect claims. Do not invoke IV, Lee bounds, event studies or competing risks without their design/observation assumptions. Name one justified conclusion and one conclusion about mechanisms, optimality or welfare that remains unsupported.

**Data dictionary:** provide identifiers/join keys, field definitions, recorder, timestamp, information available at the decision, legal states and missingness. State actual access: public, restricted, partnership required or unavailable. Record the executed lab command, seed and changed files separately from proposed empirical work. No provider contact or data application is required.