You are given an interview segment (interviewer question(s) + interviewee answer) and a specific sub-question extracted from it.

Task:
Classify the answer into one of the evasion types listed in the definitions below (9 for Step 2 / 6 for Step 3, as specified by the definitions block).

{prior_analysis_block}

{definitions_block}

{confusion_guidelines_block}

{fewshots_block}

### Part of the interview ###
{interview_question}

{interview_answer}

### Sub-question ###
{sub_question}

Output format (must follow exactly):
<analysis>
Step 1 - Identify what the sub-question is asking for (key information required).
Step 2 - Check if the answer provides that key information (fully, partially, or not at all).
Step 3 - If key info is provided: determine Explicit vs Implicit vs Partial.
Step 4 - If key info is missing: determine which evasion type best fits.
</analysis>
<confidence>High/Medium/Low</confidence>
<label>LABEL</label>

Confidence Level Guide:
- **High**: The label clearly fits; no ambiguity.
- **Medium**: The label is reasonable, but there's some overlap with another category (e.g., General vs Deflection, Implicit vs Partial).
- **Low**: Uncertain; could arguably be multiple labels.
