This paper examines the impacts of social factors on lenders' decision-making in online peer-to-peer (P2P) lending. Data collected from a major U.S. online loan marketplace, Prosper.com, have been analyzed. We propose a model based on preferential attachment and fragmentation to model the bidding behavior of lenders. Our data analysis presents strong empirical evidence that there were significant herding effects when lenders made their investment decisions on loan listings. The distribution of the number of bids put on loan listings exhibits a power law with an exponential cutoff, which matches what the model predicts. The paper concludes that lenders on Prosper did not make rational investment decisions based on risk and returns, but followed the herd.