Meta-analysis of 380 empirical studies (438 independent samples) synthesizing the age–job performance relationship across ten distinct performance dimensions: core task performance, creativity, training performance, organizational citizenship behaviors (OCBs), safety performance, general counterproductive work behaviors (CWBs), workplace aggression, on-the-job substance use, tardiness, and absenteeism. Published in Journal of Applied Psychology, 2008, Vol. 93(2): 392–423. The study finds that although age is largely unrelated to core task performance and creativity, it demonstrates stronger and practically significant relationships with citizenship behaviors, safety, and counterproductive behaviors — challenging the narrow "productivity-loss" stereotype of older workers. Curvilinear effects are documented for core task performance (inverted-U, peaking at ages 31–35 with supervisor ratings) and counterproductive work behaviors (steepening negative slope with age), with evidence that the inverted-U in supervisor ratings partly reflects age bias rather than actual ability decline.
"Although age was largely unrelated to core task performance, creativity, and performance in training programs, it demonstrated stronger relationships with the other 7 performance dimensions."
"These observations suggest either that older workers are as motivated as younger workers to contribute to their organizations or that they more consciously engage in discretionary behaviors to compensate for any losses in technical core performance."
"Given the variance among self-evaluations, peer evaluations, and supervisor evaluations, older workers may especially benefit from the use of 360° feedback systems."
"The stereotype of older workers as difficult colleagues, then, seems largely unfounded."
The paper's key contribution is disaggregating "job performance" into ten dimensions that have different age trajectories. The productivity-loss framing that dominates popular and policy discussion is built almost entirely on the core task performance dimension, where the age effect is near-zero. Meanwhile, dimensions with clear positive age gradients — OCBs, safety compliance, reduced counterproductive behaviors, punctuality, lower voluntary absence — are systematically ignored in that framing. The curvilinear finding for supervisor-rated task performance is important and underappreciated: the negative slope for workers 40+ likely reflects supervisor age bias, not actual decline. This has direct implications for how employer accommodation offers and performance reviews interact with disability insurance (DI) application decisions in the disability literature.
The sickness-vs.-voluntary absence distinction is especially relevant for DI research: sickness absence (the disability-relevant category) shows essentially no age gradient, while voluntary absence falls sharply with age. Studies that aggregate all absence together will confound these mechanisms.
Limitation worth noting: the training studies are overrepresented by computer/technology programs in which older workers are least likely to excel. The training performance finding (−.04) is probably inflated in the negative direction relative to domain-general training contexts.