Standardise a legacy triadic comparison dataset — read

Reads a single CSV file from a legacy triplet experiment and converts it to the standard column format used by this package. Unrecognised or missing columns are derived where possible from the data that are present; see the Column mappings section below for full details. Two output files are written alongside the input: a standardised data CSV and a stimulus-level mapping CSV.

Usage

read_legacy(input_file)

Arguments

input_file: Character. Path to the input CSV file.

Value

A list returned invisibly with two elements:

data_file: Path of the written standardised data CSV (input filename with _v2025.csv suffix).
levels_file: Path of the written stimulus-level mapping CSV (input filename with _2025_levels.csv suffix).

Output columns

p_id: Character participant identifier.
Center: Character label of the target (reference) stimulus.
Left: Character label of the left-hand choice stimulus.
Right: Character label of the right-hand choice stimulus.
Answer: Character label of the chosen stimulus.
head: Integer (0-indexed) factor code for Center, ordered alphabetically across all unique stimuli in Center, Left, and Right.
winner: Integer factor code for Answer, using the same ordering as head.
loser: Integer factor code for the unchosen option, using the same ordering as head.
RT: Numeric reaction time. NA if not present in the input.
sampleAlg: Character sampling algorithm label: "random", "check", "validation", "uncertainty", or NA.
sampleSet: Character set label: "train" or "test".

Column mappings

Input column names are matched case-insensitively and ignoring punctuation.

Participant ID (p_id): Recognised input names: sessionID, session_ID, puid, Participant.ID, worker_id, sub_id, pid. If none are found, participants are assigned sequential IDs ("P1", "P2", ...).
Center: Recognised input names: Center, Target.
Left: Recognised input names: Left, Option1.
Right: Recognised input names: Right, Option2.
winner / loser: Recognised input names: winner / primary and loser / alternate. Derived from Answer, Left, and Right when absent.
sampleAlg: Recognised input names: sampleAlg, AlgSample. Values are recoded: "Random" → "random"; "Test" → "validation"; "check" → "check"; "uncertainty" → "uncertainty". Filled with NA when absent.
sampleSet: Recognised input names: sampleSet, Alg.Label, TrnTest. When absent, 10\ randomly assigned to "test" and the remainder to "train" (seed 2025).

Examples

if (FALSE) { # \dontrun{
result <- clean_triadic_comparisons("data/triplets.csv")
print(paste("Wrote:", result$data_file, "and", result$levels_file))
} # }