What's the difference between "dummy data" and "statistically synthetic data"?

Question

Accepted Answer

Dummy data copies only the structure (column names, types, format); statistically synthetic data also preserves the real distributions and relationships.

Dummy Data
Statistically Synthetic Data

Goal
Structural similarity only
Preserve statistical relationships, distributions, patterns

Realism
Values needn't be realistic
Generated by models that learned from real data

Use
Software testing & development
Analysis, research, ML training

Example
Random strings in name fields, sequential IDs, placeholder emails
Realistic income correlated with education, age-appropriate health conditions

So dummy data is fine when structure matters more than content (e.g. testing a database schema), but useless for analysis. Statistically synthetic data is harder to make but supports real analytics.
Tip: If you just need the table to look right for code, dummy data suffices. If you need the numbers to behave right for analysis, you need statistical synthesis.

	Dummy Data	Statistically Synthetic Data
Goal	Structural similarity only	Preserve statistical relationships, distributions, patterns
Realism	Values needn't be realistic	Generated by models that learned from real data
Use	Software testing & development	Analysis, research, ML training
Example	Random strings in name fields, sequential IDs, placeholder emails	Realistic income correlated with education, age-appropriate health conditions