A recent study by the Centers for Disease Control and Prevention found that employees with paid sick leave are 28 less likely to be involved in a work related accident than employees who do not receive payment for sick leave Researchers hypothesize that e

The author provides two pieces of evidence in support of the researchers’ contention that workers without paid sick leave are more likely to work when ill, and that such workers are more likely to experience a job-related accident as a result. The results of the study cannot be disputed—we must accept as true, for example, that those with paid sick leave are significantly less likely to have work-related accidents than those without paid sick leave. We might, however, question whether the study was large enough, or representative enough, to draw broad conclusions. Further, the study does not demonstrate causation: it does not tie the incidence of work-related accidents to illness. While the researchers’ hypothesis is certainly one possibility, more research is needed to eliminate other possibilities and to bolster the strength of this argument.

The largest leap in the argument is the assumption that those without paid sick leave feel pressured to work when ill. No evidence is presented to establish this supposition. In order to strengthen this part of the argument, the researchers might ask study participants whether they have actually come to work ill during the same time frame covered by the original study and, if so, why they chose to come to work when ill. If the study participants who did experience a work-related accident were also more likely to come to work ill for fear of lack of pay, then the hypothesis would be much more strongly supported, particularly if this occurred with a correspondingly large proportion of workers (to match the 28% greater incidence of accidents in the original study).

The researchers would also strengthen their case by addressing alternative explanations for the data in the original study. For instance, are there differences between the two groups based upon industry or job performed that might explain the data? For example, are hourly workers more likely to lack paid sick leave, while salaried workers are more likely to receive it? Are hourly workers more likely to work in blue-collar or more manual occupations, where on-thejob accidents are more frequent? If so, then we would expect a correlation between unpaid sick leave and a higher incidence of workplace accidents because the work itself is inherently more dangerous, not because people are choosing to work when ill. Further, if it is the case that higher-risk occupations in general are more likely to lack paid leave, then the second piece of evidence also loses its significance. In such a case, the researchers’ hypothesis would be significantly weakened.

The data presented also lacks a depth of detail that would help us to evaluate the significance of the study results. How many people were surveyed? What is the margin of error and how was the study conducted? Is the 28% figure statistically significant? If the study represented a large enough survey group to extrapolate to the general population, across regions, industries, and job responsibilities, then the study results may be conveying something significant. If, alternatively, few people were surveyed or the incidence of job-related accidents were very low, then perhaps the 28% difference represented a small number of people, well within the statistical variance expected.

While the argument presents an interesting hypothesis, the data presented is not strong enough to establish the validity of the conclusion to even a small degree. First, at a basic level, we need to know whether the existing data is statistically significant and sufficiently representative. There are also several gaps in the logic chain, assertions made without supporting evidence. Finally, the researchers could strengthen their case by examining, and dismissing, alternative explanations for the data presented thus far. These steps might not be enough to establish the validity of the hypothesis beyond a doubt, but they would allow the researchers to determine whether the pursuit of the hypothesis is a good use of time, funds, and attention.