Data Quality Diagnostics for Propensity Score Matching — diagnose_matching

Provides comprehensive diagnostics for data quality issues that can affect propensity score matching, including missing value patterns, non-finite values, and categorical variable issues.

Usage

diagnose_matching_data(
  data,
  treatment_var = "is_treated",
  matching_variables,
  exact_match_vars = NULL,
  factor_level_threshold = 5
)

Arguments

data: A data.table containing the data to be assessed
treatment_var: Character. Name of treatment indicator variable
matching_variables: Character vector. Variables to include in assessment
exact_match_vars: Character vector. Variables requiring exact matches. Default: NULL
factor_level_threshold: Numeric. Minimum observations per factor level. Default: 5

Value

A list containing:

overall_summary: Overall data quality summary
variable_analysis: Detailed analysis by variable
missing_patterns: Missing data patterns across observations
recommendations: Specific recommendations for data cleaning

Examples

if (FALSE) { # \dontrun{
# Assess data quality before matching
quality_report <- diagnose_matching_data(
  data = my_data,
  treatment_var = "is_treated",
  matching_variables = c("age", "education", "sector"),
  exact_match_vars = c("gender")
)
print(quality_report$overall_summary)
} # }