Pitch Accuracy by Scale Degree, Interval, and Direction in Two-Note Sight-Singing: Data and Code

Data and reproducible code for “Target Scale Degree and Accuracy in Cued Two-Note Sight-Singing” (RECIEM). The main dataset contains 6,656 observed binary scores from 104 music students completing 64 staff-notated two-note items in C major (C4–C5), without notated rhythm or a time limit. Each item followed a C4–E4–G4 tonic chord and a supplied starting pitch; item order was fixed. Correct = 1 only when both starting and target pitches were correct on their final renditions; spontaneous corrections were allowed. Scoring was auditory. A separate preliminary sample (n = 57) is not included. Forty randomly selected main-study recordings were independently rescored: 99.29% agreement over 2,240 paired decisions (Cohen’s κ = .986, 95% participant-bootstrap CI [.977, .993]). Original main-study scores were retained. Primary logistic mixed-effects models use 56 non-unison items (5,824 responses) to assess categorical target degree, interval category, and melodic direction. A post hoc 49-item comparison (5,096 responses) examines tonic-chord targets versus other diatonic targets. In this single-key, one-octave design, degree also identifies exact pitch/register. Sensitivity checks cover C4 exclusion, item order, interval quality, expanded random effects, and starting degree. A further post hoc comparison uses 11 cells containing both directions (22 items; 2,288 responses); starting pitches are not matched. Analyses are exploratory where indicated and were not preregistered. The ZIP contains 17 files: three unchanged observed-data inputs, twelve R/Python scripts, a README, and checksums. Six analysis workflows, agreement statistics, and figures can be regenerated; outputs are produced under rerun/. No responses were imputed or altered. Findings are task-specific associations, not evidence of a causal direction effect or an isolated tonal-hierarchy mechanism.