How many employees are there? Seems like you'd need to know that to disprove the null hypothesis, which would be due to a disproportionately small sample size.