Genetic Algorithms are a popular set of optimization algorithms often used to aid software testing. However, no work has been done to apply systematic software testing techniques to genetic algorithms because of the stochasticity and the lack of known outputs for genetic algorithms. Statistical metamorphic testing is a useful technique for testing programs when the output is unknown or when the program has random elements. In this paper, we identify 17 metamorphic relations for testing a genetic algorithm and show, through mutation testing, that these relations are more effective at finding defects than traditional unit tests based on known outputs. We examined the failure rates of the system-level relations when initialized with various fitness functions. We found three relations failed excessively and we then modified these relations so that they failed less often. We also identified some metamorphic relations for genetic algorithms that are generalizable across different types of evolutionary algorithms and showed that our relations had similar mutation scores between two implementations. This is the first time statistical metamorphic testing has been applied for testing genetic algorithms.