The Human Relations Movement of the 1930s popularized the belief that happy workers were on average more productive than less happy or unhappy workers (Cropanzano & Wright, 2001; Wright, Cropanzano, & Bonnett, 2007). This view is usually known as the happy-productive worker hypothesis (HPWH), and both researchers and practitioners have become persuaded that it is essentially correct. Happy workers have been frequently conceptualized as the individuals scoring higher in job satisfaction scales and several metaanalytic studies examined the prediction that there was a positive correlation between job satisfaction and job performance. The first meta-analysis, by Iaffaldano and Muchinsky (1985), found an average observed correlation of .17 between job satisfaction and individual performance. A second meta-analysis, by Judge, Thorensen, Bono, and Patton (2001), found an average observed correlation of .18 (ρ = .30 when corrected for measurement error in both variables), although the magnitude of the relationship between job satisfaction and job performance was moderated by the study design (cross-sectional vs. longitudinal). In the case of longitudinal designs, the observed correlation was .14 (ρ = .23), and it was .18 (ρ = .31) for the crosssectional designs. Other studies have used measures of job burnout, job involvement, engagement, positive affect, neuroticism, and climate to characterize happy workers (

This paper presents a meta-analytic study of the relationship between overall subjective well-being (SWB), cognitive SWB, affective SWB, and job performance ratings. The study examined the moderator effect of the source of job performance measure (self-report vs. supervisory ratings). The database consists of 34 independent samples (n = 5,352) using supervisory performance ratings and 38 independent samples (n = 12,086) using self-reported of job performance. These samples were located through electronic and manual searches. The results indicated that, on average, the correlation for SWB-supervisory ratings (ρ = .35) was slightly larger than for SWB-self-reported performance (ρ = .33). The correlation of affective SWB was much higher with supervisory ratings (ρ = .49) than with self-reported performance (ρ = .30). A suppressor effect of cognitive SWB was found for the prediction of supervisory ratings. Finally, we discuss the implications for the theory and the practice of SWB at work and suggest new research avenues.