Variable Lists
What are Variable Lists?
Variable lists contain the actual values that get substituted into your templates. They’re the foundation of systematic testing - each list represents one dimension you want to vary.
Creating Lists
Simple Lists
Basic text values, one per line:
Names List:
Alice Johnson
Michael Chen
Keisha Williams
José Rodriguez
Priyanka Patel
Universities List:
Harvard University
State University
Community College
Elite Private Academy
How Lists Work
With template: Rate candidate {{name}} from {{university}}
Auditomatic creates every combination:
- Alice Johnson from Harvard University
- Alice Johnson from State University
- Alice Johnson from Community College
- Michael Chen from Harvard University
- And so on…
Total prompts: 5 names × 4 universities = 20 combinations
List Strategy
Bias Testing Lists
Intersectional Names (built-in):
- Chad Anderson (white male)
- Keisha Washington (Black female)
- José Hernández (Latino male)
- Priyanka Patel (South Asian female)
Socioeconomic Indicators:
- Harvard, Yale, Princeton
- State University, Public College
- Community College, Trade School
Content Testing Lists
Demographics:
young professionals
working parents
college students
retirees
small business owners
Geographic Markets:
New York City
Rural Nebraska
Austin, Texas
Portland, Oregon
Miami, Florida
List Size Guidelines
Small Tests (Under 100 prompts)
- 5-10 values per list
- 2-3 variables total
- Good for initial exploration
Medium Tests (100-1000 prompts)
- 10-20 values per list
- 3-4 variables total
- Standard bias testing
Large Tests (1000+ prompts)
- 20+ values per list
- 4+ variables total
- Research-grade systematic studies
Built-in Lists
Auditomatic includes pre-made lists:
- Intersectional Names: Diverse names with demographic attributes
- US Cities: Major metropolitan areas
- Academic Fields: University majors and departments
- Global Cities: International locations
- Common Names: Gender-neutral options
Import Options
Text Files
Upload .txt files with one value per line
CSV Files
Upload .csv files with headers:
name,gender,ethnicity
Alice Johnson,female,white
Michael Chen,male,Asian
Copy-Paste
Paste values directly into the interface
Testing Your Lists
Preview Combinations
Before running full tests:
- Select 2-3 values from each list
- Generate preview prompts
- Check if combinations make sense
Edge Cases
Include boundary values:
- Very long/short names
- Unusual spellings
- Mixed case text
- Special characters
List Management
Naming Convention
Use descriptive names:
- ✅
diverse_professional_names - ✅
tier1_universities - ❌
list1,names
Organization
Group related lists:
- Demographics: Names, ages, locations
- Professional: Job titles, companies, industries
- Academic: Universities, majors, GPAs
Reusability
Create modular lists that work across templates:
- Base name list works for hiring, lending, admissions
- University list works for resume screening, bias testing
Next Steps
- Prompt Templates - Combine lists with templates
- Results Analysis - Analyze patterns in your data