Express a metric as a ratio to the separate analysis

Divides each row's metric by the value the separate (no borrowing) method takes in the same scenario. The scenario key is the full set of coordinates that identify a simulated configuration, so the ratio never collapses across drift or sample size.

A zero or missing denominator makes the ratio undefined. Those rows are dropped with a warning rather than reported as Inf or as some finite stand-in.

Usage

scale_by_separate(df, metric_columns)

Arguments

df

A frequentist results frame containing separate method rows.

metric_columns

Character vector of columns to divide - typically the metric and its two confidence bounds.

Value

df with metric_columns divided by the separate analysis's value, in the original column order.