Under the stated assumptions, work-hour shifts affect recovery and perceived physical health more than broad well-being, with increasing work hours most clearly raising fatigue and reducing sleep while decreasing work hours most clearly lowers fatigue.
Key Findings
Results
Increasing work hours by 10 per week most clearly raises fatigue and reduces sleep.
The study used three annual waves of the New Zealand Attitudes and Values Study (NZAVS) with N = 24,579 pre-retirement adults from 2020–2023.
The intervention was defined as working 10 more or 10 fewer hours per week than actually observed.
Fatigue effects were identified as the most robust findings under the upward work-hour shift.
Sensitivity analyses using E-values indicated that the fatigue effects are robust to moderately strong residual confounding.
Machine-learning methods were used to adjust for baseline differences and account for dropout.
Results
Decreasing work hours by 10 per week most clearly lowers fatigue, and the downward shift is better supported by the data than the upward shift.
The downward shift in work hours was described as 'better supported by the data' compared to the upward shift.
Fatigue reduction was the clearest outcome under the 10-hour-per-week decrease scenario.
The study compared outcomes under hypothetical shifts to what actually occurred, using a target trial emulation framework.
E-values from sensitivity analyses confirmed robustness of the fatigue-reduction finding to moderately strong residual confounding.
Results
BMI and perceived physical health shift adversely when work hours increase and favourably when work hours decrease, but these outcomes are more sensitive to residual confounding than fatigue.
BMI and perceived physical health showed adverse shifts under the +10 hours per week scenario.
BMI and perceived physical health showed favourable shifts under the −10 hours per week scenario.
Both outcomes were described as 'more sensitive to residual confounding' compared to fatigue and sleep findings.
The authors flagged these results as less certain than fatigue outcomes due to this confounding sensitivity.
Results
Perceived support increases slightly with increased work hours but remains sensitive to residual confounding.
Under the +10 hours per week scenario, perceived support showed a slight increase.
This finding was explicitly flagged as confounding-sensitive, meaning it may not reflect a true causal effect.
This was among the few outcomes showing any positive change under an increase in work hours.
Results
Most of the 28 well-being outcomes examined showed little movement under either a 10-hour increase or 10-hour decrease in weekly work hours.
The study estimated effects across 28 well-being outcomes spanning multiple dimensions.
The majority of these outcomes showed 'little movement under either policy.'
This indicates that broad well-being dimensions are less responsive to work-hour changes than fatigue and physical health.
The authors summarize this as work-hour shifts affecting 'recovery and perceived physical health more than broad well-being.'
Results
Naive baseline (observational) associations between work hours and well-being outcomes were broader, larger, and sometimes reversed in sign compared to the target trial emulation estimates.
The authors directly compared naïve correlational associations with their causal estimates from target trial emulation.
Naïve associations were described as 'broader, larger, and sometimes reversed in sign' relative to the emulation results.
This finding illustrates the problem that simple correlations cannot be used to predict what would happen if work hours changed.
This methodological contrast was a core motivation for using the target trial emulation approach.
Methods
The study used a target trial emulation framework with machine-learning covariate adjustment and E-value sensitivity analyses to approximate a randomized experiment using observational longitudinal data.
Target trial emulation involves specifying the experiment one would like to run and then approximating it with observational data.
Three annual waves of NZAVS data (2020–2023) were used, covering N = 24,579 pre-retirement adults in New Zealand.
Machine-learning methods were used to adjust for baseline differences between those who did and did not change work hours.
Dropout was accounted for in the analysis.
E-values were used as sensitivity analyses to quantify robustness to unmeasured confounding.
What This Means
This research suggests that changing how many hours people work per week has a more limited effect on overall well-being than many observational studies imply. Using a sophisticated approach called 'target trial emulation,' the researchers analyzed data from nearly 25,000 New Zealand adults over three years to estimate what would actually happen—causally—if people worked 10 more or 10 fewer hours per week. They found that the clearest and most robust effects were on fatigue and sleep: working more hours increased fatigue and reduced sleep, while working fewer hours reduced fatigue. Effects on body weight (BMI) and perceived physical health pointed in the same direction but were less certain because they could more easily be explained by other unmeasured factors.
Importantly, most of the 28 well-being measures the researchers examined—spanning things like mental health, life satisfaction, and social connection—showed little to no change under either a 10-hour increase or decrease in work hours. This suggests that work-hour policies may primarily affect physical recovery rather than transforming broader quality of life. The researchers also found that simple correlations between work hours and well-being were often larger, wider, and sometimes pointed in the opposite direction compared to their more rigorous causal estimates, highlighting that observational associations can be misleading when trying to predict the effects of actual changes.
This research matters because it uses a more rigorous method than typical survey studies to get closer to answering the question: 'What would actually happen if this person's work hours changed?' The findings suggest that fatigue is a reliable and robust consequence of longer work hours, but that policies aimed at reducing work hours should not be expected to dramatically improve mental health or life satisfaction across the board. The results support the idea that adequate rest and physical recovery are the most direct benefits of reducing working time.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Chong B, Sibley C, Bulbulia J. (2026). How work hours affect well-being: A target trial emulation.. PloS one. https://doi.org/10.1371/journal.pone.0350816