Null associations between experience, workload, skills, and self-rated productivity among Indonesian regional government employees

Re-estimated census survey of the 41 employees of the Communication and Informatics Office of Asahan Regency, Indonesia, with self-rated five-point composites of work experience, workload, work skills, and employee productivity (three, three, five, and five items). Because the thesis's regression output was irreproducible and internally inconsistent, the model was re-estimated from the item responses printed in the thesis. The predictors jointly accounted for 7.1% of the variance in productivity, F(3, 37) = 0.94, p = .433, and no standardized coefficient differed significantly from zero: experience β = .113 (p = .483), workload β = -.239 (p = .142), skills β = .080 (p = .621). Heteroscedasticity-consistent standard errors, the printed scale totals (R² = .042), and a criterion without its ability item gave the same result; experience reached significance only after four influential cases were excluded (β = .345, p = .035), a fragile result. Between 56.1% and 82.9% of respondents gave one identical answer to every item of a scale, within-scale alphas were .93 to .98, and between-scale correlations were near zero, the pattern that invariant responding produces; printed totals differed from item sums for 14 to 26 respondents per scale. The experience items record perceived benefits of experience rather than tenure, and the workload items record the clarity of targets and standards rather than demands or their appraisal. These data provide no evidence of associations, their intervals do not rule out moderate ones, and all estimates are conditional on the accuracy of the transcribed responses.