Table Creation & Cross-tabulation

For Researchers: Create frequency tables and cross-tabulations with automatic cell suppression for small counts.

What ACRO Supports:

• crosstab() - Cross-tabulation of two or more variables with frequency counting
• pivot_table() - Spreadsheet-style pivot tables with aggregation functions
• table() - Simple frequency tables for categorical data (R interface only)

Technical Details:

• ACRO suppresses, and reports the reason why, the value of an aggregation statistic (mean, median, variance, etc.) for any cell is deemed to be sensitive.
• The current version of ACRO supports the three most common tests for sensitivity: ensuring the number of contributors is above a frequency threshold, and testing for dominance via N-K rules.
  • N-K Rule: A dominance test where if the top N contributors account for more than K% of the total, the cell is considered disclosive.
  • Frequency Threshold: Cells with fewer than a specified number of contributors are suppressed.
• All thresholds are configurable via YAML configuration files.
• For detailed methodology, see our research paper.
• Automatic flagging of negative or missing values for human review.

Example Use Cases: - Survey response analysis by demographics - Clinical trial outcome tables - Market research cross-tabulations - Educational assessment reporting

Statistical Modeling

For Researchers: Run regression analyses with automated checks on model outputs and residual degrees of freedom.

What ACRO Supports:

• ols() - Ordinary Least Squares linear regression
• logit() - Logistic regression for binary outcomes
• probit() - Probit regression for binary outcomes

Technical Details: - For regressions such as linear, probit, and logit, the tests verify that the number of residual degrees of freedom exceeds a threshold. - The functionality acts as a wrapper around standard statistical packages.

Example Use Cases: - Economic modeling and policy analysis - Medical research and clinical studies - Social science research - Business analytics and forecasting

Automated Sensitivity Testing

What ACRO Checks:

For Tables: - Minimum cell counts (frequency thresholds) - Dominance rules (N-K rules for concentration) - Presence of negative or missing values

For Statistical Models: - Residual degrees of freedom thresholds - Model fit diagnostics - Parameter significance testing

For Non-Technical Users: ACRO automatically identifies when research outputs might reveal sensitive information about individuals or organizations, applying industry-standard privacy protection rules without requiring manual review of every result.

Output Management

What ACRO Provides:

• Suppression Masks - Clear indication of which results are hidden and why
• Summary Reports - Detailed explanation of all disclosure checks performed
• Audit Trails - Complete record of all analysis steps and decisions
• Exception Handling - Process for requesting release of flagged outputs

Workflow Integration: The finalise() function will: 1. Check that each output with “fail” or “review” status has an exception (if not you will be asked to enter one). 2. Write the outputs to a directory. This directory contains everything that the output checkers need to make a decision.

Research Environments

Where ACRO Works: - Trusted Research Environments (TREs) - Data safe havens - Secure data centers - Academic research computing facilities - Government statistical offices - Healthcare research environments