Unveiling Selection Biases: Exploring Order and Token Sensitivity in Large Language Models