Work discipline, competence, and job satisfaction predicting self-rated work achievement in Indonesian agricultural research

Abstract

We re-analyzed the respondent-level records of a census of the 58 employees of the Bakrie Agriculture Research Institute in Kisaran, North Sumatra, an Indonesian plantation company's research unit, to determine whether self-rated work discipline, competence, and job satisfaction can be ranked as predictors of self-rated work achievement. The records reproduced the original thesis output exactly. The three five-point composites jointly accounted for 86.0% of the variance in work achievement (F (3, 54) = 110.43, p <.001), and each partial coefficient was positive (discipline β = .280, competence β = .378, satisfaction β = .312). No pairwise difference between standardised coefficients was significant; competence minus discipline was 0.098 (p = .639; 95% bootstrap confidence interval [-0.328, 0.743]). Only .060 of the explained variance was unique to single predictors, against .800 shared; every heterotrait-monotrait ratio exceeded 1, and average variance extracted was below .50 for every scale, so the instruments do not discriminate the constructs. The discipline coefficient was fragile: with heteroscedasticity-consistent standard errors p = .096, and without four influential cases β = .130 (p = .358). Two criterion items, on job knowledge and time discipline, restated predictor content; removing both lowered R² to .749 and left the competence and discipline coefficients (β = .570 and .063) not significantly different (p = .074). The high R² reflects shared variance and criterion contamination, not three distinct contributions. These data cannot show which predictor matters most, and the expectation that competence outweighs discipline in knowledge-intensive work remains a proposition for comparative designs.