Interpret RATE Estimates

Produces a concise summary of which outcomes show positive, negative, or inconclusive heterogeneous treatment effects based on RATE (Rank Average Treatment Effect) analysis.

Produce a compact Markdown summary describing which outcomes show positive, negative, or inconclusive heterogeneous treatment effects.

Usage

margot_interpret_rate(
  rate_df,
  flipped_outcomes = NULL,
  target = "AUTOC",
  adjust_positives_only = FALSE
)

margot_interpret_rate(
  rate_df,
  flipped_outcomes = NULL,
  target = "AUTOC",
  adjust_positives_only = FALSE
)

Arguments

rate_df

A data frame from margot_rate() or a list containing rate_autoc and rate_qini.

flipped_outcomes

Character vector of outcomes inverted during preprocessing.

target

Character; either "AUTOC" or "QINI" (ignored when rate_df is a list).

adjust_positives_only

Logical; if TRUE, apply multiple testing correction only to positive RATEs in comparison output. Default FALSE.

Value

For a single table: a markdown-formatted character string summarizing the RATE results. For a list input: a list containing comparison results and individual interpretations for both AUTOC and QINI.

If rate_df is a data frame, a Markdown string. If rate_df is a list, returns a list produced by margot_interpret_rate_comparison().

Details

RATE quantifies how much better CATE-based targeting performs compared to uniform treatment (ATE). The interpretation differs by weighting scheme:

• AUTOC: Uses logarithmic weighting, emphasizing top responders

• QINI: Uses linear weighting, balancing effect size and prevalence

Positive RATE values indicate heterogeneity can be exploited for better targeting. Negative values suggest CATE targeting would underperform uniform treatment.

Examples

if (FALSE) { # \dontrun{
# Compute RATE for causal forest results
rate_results <- margot_rate(causal_forest_results)

# Interpret the results (handles both AUTOC and QINI)
interpretation <- margot_interpret_rate(rate_results)
cat(interpretation$comparison)

# Or interpret just AUTOC results
autoc_interpretation <- margot_interpret_rate(
  rate_results$rate_autoc,
  target = "AUTOC"
)
} # }