If people are not that great at judging their own fatigue levels without a validated questionnaire, should we even consider getting them to judge how well they will perform at their current level of fatigue?
The problem with most of the research that looks at subjective estimates of performance and fatigue is that the participants are usually very fatigued. Researchers will often keep them awake for over 24 hours. In these studies, you will see moderate correlations between people’s judgements of their future performance and actual performance. In other words, people are only somewhat accurate at predicting how well they will perform.
It gets a little more complicated when you look at the tasks for which you ask the person to predict their performance. For example, one study found that after being awake for 28 hours, people were quite good at predicting how quickly they would respond, but not nearly as good at predicting how accurately they would respond, and not good at all at predicting how well they would be able to use logical reasoning.
Whether a person’s accuracy in predicting performance is as accurate with lower levels of fatigue, like when a person has slept for only one hour less than needed is hard to say. I suspect that it would follow the same pattern as accuracy does with subjective estimates of fatigue. That is, predictions of performance would be even less accurate at lower levels of fatigue because if you don’t recognize that you are fatigued, you likely won’t recognize that your performance could be impaired.
Rather than simply asking people to predict how well they will perform, a better approach is to use a validated test of performance. But not just any test of performance, use a test with results that are highly correlated to fatigue levels. That way, if the test results show reduced performance, and a validated fatigue questionnaire indicates that the person is fatigued, you can safely assume that the reduced performance is likely due to fatigue. This will give you a good idea of how well the person will perform on-duty tasks that tap into the same abilities as the performance test (e.g., reacting quickly and accurately). Not only that, since we know that fatigue generally does not lessen over time without sleep, you can also safely assume that the person’s performance will likely become worse than measured on the performance test as the duty progresses.
Take Home Point
Given the difficulty that people experience in predicting their performance, especially on specific tasks, it is probably not a good idea to rely solely on subjective judgements of performance in your fitness for duty assessments. A better approach is to use a validated test that measures performance and is highly correlated to fatigue.
 Dorrian, J., Lamond, N., Dawson, D. (1999). The ability to self-monitor performance when fatigued. Sleep Research, 9, 137-144.
 Two tests of performance that are highly correlated to fatigue are the Psychomotor Vigilance Test and the Digit Symbol Substitution Test (DSST). See: Honn, K., & Van Dongen, H. (2017). Sleep deprivation effects on the digit symbol substitution test: General cognitive slowing or wake state instability? Sleep, 40, Supplement, A95.