Derive mean and difference for Bland-Altman analysis

A helper function that computes the mean and difference (or ratio) of two variables after applying a transformation. Supported transformations are "identity" (no transformation), "log" (natural logarithm, ratio back on the original scale), and "logit" (logit transformation).

Usage

ba_mean_diff(
  data = stop("data must be specified"),
  var1 = stop("variable must be specified"),
  var2 = stop("variable must be specified"),
  transform = c("identity", "log", "logit")
)

Arguments

data

A data frame.

var1

1st variable to compare (unquoted).

var2

2nd variable to compare (unquoted).

transform

Transformation to apply before computing mean and difference. One of "identity" (default), "log", or "logit".

Value

The input data frame with two additional columns:

avg

Mean of the (transformed) paired observations.

dfce

Difference of the (transformed) paired observations. For "log" this equals log(var1 / var2).

See also

ba_stat, ba_plot

Examples

library(tidyr)
tbl <- temperature %>% pivot_wider(names_from = method, values_from = temperature)

# identity (default) - same as used inside ba_stat / ba_plot
ba_mean_diff(tbl, var1 = infrared, var2 = rectal)
#> # A tibble: 450 × 7
#>    animalID treatment visit            rectal infrared   avg   dfce
#>       <int> <fct>     <fct>             <dbl>    <dbl> <dbl>  <dbl>
#>  1        1 healthy   baseline           33.9     34.8  34.4  0.932
#>  2        1 healthy   visit 1            34.6     36.5  35.5  1.92 
#>  3        1 healthy   visit 2            35.4     36.2  35.8  0.780
#>  4        1 healthy   visit 3            35.4     35.7  35.6  0.310
#>  5        1 healthy   end of treatment   35.7     35.1  35.4 -0.616
#>  6        2 vehicle   baseline           37.6     38.6  38.1  0.969
#>  7        2 vehicle   visit 1            35.1     38.5  36.8  3.39 
#>  8        2 vehicle   visit 2            37.4     36.0  36.7 -1.42 
#>  9        2 vehicle   visit 3            37.9     36.1  37.0 -1.87 
#> 10        2 vehicle   end of treatment   38.7     36.4  37.6 -2.32 
#> # ℹ 440 more rows

# log transformation
ba_mean_diff(tbl, var1 = infrared, var2 = rectal, transform = "log")
#> # A tibble: 450 × 7
#>    animalID treatment visit            rectal infrared   avg     dfce
#>       <int> <fct>     <fct>             <dbl>    <dbl> <dbl>    <dbl>
#>  1        1 healthy   baseline           33.9     34.8  3.54  0.0271 
#>  2        1 healthy   visit 1            34.6     36.5  3.57  0.0540 
#>  3        1 healthy   visit 2            35.4     36.2  3.58  0.0218 
#>  4        1 healthy   visit 3            35.4     35.7  3.57  0.00873
#>  5        1 healthy   end of treatment   35.7     35.1  3.57 -0.0174 
#>  6        2 vehicle   baseline           37.6     38.6  3.64  0.0254 
#>  7        2 vehicle   visit 1            35.1     38.5  3.60  0.0922 
#>  8        2 vehicle   visit 2            37.4     36.0  3.60 -0.0387 
#>  9        2 vehicle   visit 3            37.9     36.1  3.61 -0.0505 
#> 10        2 vehicle   end of treatment   38.7     36.4  3.63 -0.0618 
#> # ℹ 440 more rows