The Face of Internet Recruitment: Evaluating the Labor Markets of Online Crowdsourcing Platforms in China

Zhubajie/Witmart and other online crowdsourcing platforms have proliferated in China, and researchers have increasingly used them for subject recruitment. One critical question remains, however: what is the generalizability of the findings based on these online samples? In this study, we benchmark the demography of an online sample from Zhubajie to nationally representative samples and replicate commonly asked attitudinal questions in national surveys. We find that online respondents differ from the general population in many respects. Yet, the differences become smaller when comparison is made with the internet users in benchmark surveys. Importantly, when predicting attitudes, our online sample with post-stratification weights is able to produce similar coefficients in most cases as these internet-active subsamples. Our study suggests that online crowdsourcing platforms can be a useful tool for subject recruitment, especially when researchers are interested in making inferences about Chinese netizens. We further analyze the political and social desirability issues of online subjects. Finally we discuss caveats of using crowdsourcing samples in China.

猪八戒网（Zhubajie/Witmart）及其他在线众包平台在中国迅速兴起，研究者亦愈发多地借助其招募研究被试。但仍有一个核心问题有待解答：基于此类在线样本的研究结论，其可推广性究竟如何？
本研究以猪八戒网在线样本的人口统计学特征与全国代表性样本进行对标，并复刻了全国调查中常见的态度类问题。研究发现，在线受访者与普通大众在诸多维度上存在差异。但当与基准调查中的互联网用户群体进行对比时，此类差异会明显收窄。关键在于，在开展态度预测时，经过后分层加权的本研究在线样本，在多数场景下可得到与这些活跃互联网用户子样本相近的回归系数。
本研究表明，在线众包平台可作为招募研究被试的有效工具，尤其当研究者希望针对中国网民群体进行统计推断时。我们进一步分析了在线被试的政治与社会期许偏差问题。最后，本文探讨了在中国使用众包样本的各项注意事项。