Prompt Refinement or Fine-tuning? Best Practices for using LLMs in Computational Social Science Tasks | Read Paper on Bytez