Formula: x_i = w_i (y_i - theta_FE)^2; y_i = (theta_FE - theta_FE(-i))^2 / var(theta_FE(-i))
Value
List with x (contribution to Q), y (influence on the pooled estimate),
theta_fe, theta_loo, k.
References
Baujat, Mahe, Pignon and Hill (2002), A graphical method for exploring heterogeneity in meta-analyses, Statistics in Medicine 21:2641-2652. Paywalled; the two axes are as documented by metafor::baujat, the reference implementation – the x-axis is each study's contribution to the Q statistic and the y-axis the standardised squared difference between the overall estimate with and without that study.
Examples
V <- c(1, 2, 3, 4, 5, 6, 7, 8)
Baujat(V, V)
#> $x
#> [1] 3.777174373 0.445091786 0.001064257 0.279050492 0.845842233 1.557036728
#> [7] 2.350747080 3.196029845
#>
#> $y
#> [1] 2.1987709447 0.1003427536 0.0001487728 0.0282685014 0.0671874682
#> [6] 0.1017196182 0.1304159268 0.1540785738
#>
#> $theta_fe
#> [1] 2.943495
#>
#> $theta_loo
#> [1] 4.074844 3.156200 2.935597 2.836469 2.780142 2.743817 2.718447 2.699725
#>
#> $k
#> [1] 8
#>
#> $method
#> [1] "Baujat plot coordinates"
#>
#> attr(,"class")
#> [1] "morie_rich_result" "list"