Write Pay Equity Analysis Narratives

Before opening Claude, compile the key numbers from your analysis. You'll need:

• The unadjusted (raw) pay gap percentage by protected group
• The adjusted gap (after controlling for legitimate factors) by protected group
• The p-values (statistical significance) for each gap
• Your R² (how much of pay variation the model explains)
• The control variables you included (job level, tenure, performance rating, location, etc.)

Write these in a plain text list. You'll paste them directly into your Claude prompt.

Go to {{tool:Claude.url}}. Click New Chat. Pay equity narratives contain sensitive data. Use your corporate Claude account if your organization has one, and confirm your data handling policy first.

Start with a short framing message explaining what you need:

I'm a compensation analyst writing a pay equity analysis summary for HR leadership review before it goes to legal. 
I need you to help me write a clear, factual, legally neutral executive summary of our findings. 
Use precise statistical language but also explain what the numbers mean in plain terms. 
Avoid any language that could be read as an admission of liability. 
The audience is the CHRO and HR leadership team.

In your next message, paste your findings:

Here are our pay equity analysis results:

Scope: [N] employees analyzed across [company description]
Model: OLS regression
Control variables: [list your variables]
R² = [value]

Unadjusted gaps (before controls):
- Gender: Women earn [X]% less than men
- Race: [Group] earns [X]% less than white employees

Adjusted gaps (after controls):
- Gender: [X]% gap (p=[value])
- Race: [X]% gap (p=[value])

Statistically significant gaps: [yes/no by group]
[Any other findings]

Based on these findings, please draft:
1. A 3-paragraph executive summary (250–350 words) suitable for CHRO review
2. A bullet-point recommendations section (3–5 bullets)
3. A one-sentence "bottom line" finding

Use objective language throughout. Do not speculate about causes of any gaps. 
For statistically insignificant gaps, note that they are within statistical noise.

What you should see: A structured draft with an introduction, findings, and recommendations, ready for your edits and legal review.

Review the draft for accuracy. Check that every number matches your actual output. Adjust any language that feels too strong or too weak for your company's situation. Then send to your legal team before distributing to HR leadership.

Troubleshooting: If Claude's language feels too definitive about cause (e.g., "this demonstrates pay discrimination"), add: "Rewrite to remove any language suggesting cause. Note only that the gap exists and remains unexplained by the model's control variables."