Sign-flip or rescale an estimates_vcov object

Multiply each estimate by a per-row factor and update the variance-covariance matrix to match, keeping the object internally consistent. Use it to align the sign of estimates across studies (by of +1 / -1) or to change units (e.g. by = 100 for percentage points).

This is the correct way to transform estimate values. The dplyr methods keep the vcov row-aligned (subsetting, reordering) but never transform it, so mutate(estimate = -estimate) would flip the estimates while leaving the vcov (and its cross-study covariances) inconsistent. rescale_estimates_vcov() applies \(V \mapsto \mathrm{diag}(s)\, V\, \mathrm{diag}(s)\), so the covariances stay valid, including the sign of cross-covariances under a partial sign flip. std.error, statistic, and the confidence bounds are updated to match when present.

Usage

rescale_estimates_vcov(ev, by)

Arguments

ev

An estimates_vcov object.

by

A per-estimate multiplier: a bare column name, an expression evaluated in the estimates, or a numeric vector of length 1 (recycled) or nrow(estimates). Use +1 / -1 to flip signs, positive values to rescale.

Value

An estimates_vcov object with estimate (and std.error, statistic, conf.low, conf.high when present) and the vcov rescaled.

See also

dplyr-methods, which keep the vcov row-aligned but never transform it, and estimates_vcov for what the object guarantees.

Other estimates_vcov objects: as_estimates_vcov(), bind_estimates_vcov(), dplyr-methods, estimates_vcov, make_estimates_vcov()

Examples

library(dplyr)
library(randomizr)
library(estimatr)

set.seed(123)
dat <- data.frame(Z = complete_ra(120, num_arms = 3), Y = rnorm(120))
ev <- as_estimates_vcov(bind_rows(
  study_1 = prep_fit(lm_robust(Y ~ Z, dat), term = c("ZT2", "ZT3")),
  .id = "study"
))

# Flip the sign of the first arm only; the cross-covariance sign updates too
ev |> rescale_estimates_vcov(by = if_else(term == "ZT2", -1, 1))
#> <estimates_vcov>
#> # 2 estimates with 2x2 vcov matrix
#> 
#> # A tibble: 2 × 11
#>   id    study   term  estimate std.error statistic p.value conf.low conf.high
#>   <chr> <chr>   <chr>    <dbl>     <dbl>     <dbl>   <dbl>    <dbl>     <dbl>
#> 1 1     study_1 ZT2     0.0878     0.242     0.363   0.717   -0.391     0.567
#> 2 2     study_1 ZT3    -0.265      0.219    -1.21    0.228   -0.699     0.168
#> # ℹ 2 more variables: df <dbl>, outcome <chr>

# Rescale to percentage points
ev |> rescale_estimates_vcov(by = 100)
#> <estimates_vcov>
#> # 2 estimates with 2x2 vcov matrix
#> 
#> # A tibble: 2 × 11
#>   id    study   term  estimate std.error statistic p.value conf.low conf.high
#>   <chr> <chr>   <chr>    <dbl>     <dbl>     <dbl>   <dbl>    <dbl>     <dbl>
#> 1 1     study_1 ZT2      -8.78      24.2    -0.363   0.717    -56.7      39.1
#> 2 2     study_1 ZT3     -26.5       21.9    -1.21    0.228    -69.9      16.8
#> # ℹ 2 more variables: df <dbl>, outcome <chr>