Augment Data with Swapped Profile Order — augment_profile

Doubles the dataset by swapping left/right profiles and inverting the outcome. This satisfies the profile order constraint required for valid CRT hypothesis testing with HierNet. Apply this BEFORE calling cj_fit() with method = "crt".

Usage

augment_profile_order(data, outcome, left, right)

Arguments

data: A data frame containing conjoint data.
outcome: Character name of the binary outcome column (0/1).
left: Character vector of column names for the left profile attributes.
right: Character vector of column names for the right profile attributes. Must be the same length as left.

Value

A data frame with 2 * nrow(data) rows: the original data followed by the swapped copy (left/right exchanged, outcome inverted).

Examples

# \donttest{
df <- data.frame(
  y = sample(0:1, 100, TRUE),
  left_a = factor(sample(c("x","y"), 100, TRUE)),
  left_b = factor(sample(c("p","q"), 100, TRUE)),
  right_a = factor(sample(c("x","y"), 100, TRUE)),
  right_b = factor(sample(c("p","q"), 100, TRUE))
)
augmented <- augment_profile_order(df, "y",
  left = c("left_a", "left_b"), right = c("right_a", "right_b"))
nrow(augmented)  # doubled
#> [1] 200
# }