From manual to machine: assessing the efficacy of large language models in content analysis | Synapse