Evaluating gender representativeness of location-based social media: a case study of Weibo

Abstract

Researchers have utilized location-based social media (LBSM) as potential resources to characterize daily mobility patterns and social perceptions of place. Similar to other types of big data, LBSM data also have differential data-quality issues such as accuracy, precision, temporal resolution, and sampling biases across various population groups. However, these issues have not been investigated sufficiently for LBSM users. This research aims to quantitatively examine the sampling biases of a Chinese microblogging site, Weibo, which is functionally similar to Twitter. The analysis focuses on investigating the bias in gender groups, and how this bias varies/autocorrelates in different provinces of China. The results indicate that in general, women are more likely to use Weibo in China. We also detected a strong regional pattern for Weibo gender ratios. The results provide valuable input in quantifying demographic biases in Weibo, and the methodology can be applied to other LBSM to analyse sample biases. This study also offers a data preprocessing strategy to identify potential research questions in sociology, regional science, and gender studies.