To reduce survey length, researchers have respondents answer only a subset of questions. The missing data is then imputed using advanced data imputation techniques. We evaluate the effectiveness of LLMs� imputation capabilities against traditional methods, using real human responses as the benchmark. Our findings offer valuable insights into how and under what conditions LLMs can enhance data quality and efficiency in survey analysis.