Probability-Based Samples on Twitter: Methodology and Application

Social media platforms such as Facebook and Twitter can be excellent resources for collecting data, especially when targeting a hard-to-reach or rare population. However, data collection through social media has generally relied on non-probability methods, which limits inference from the sample to the sample set. This paper demonstrates a methodology to select a random, probability-based sample from Twitter. We apply our methods to a survey of youth (persons aged 14 to 21) with the goal of oversampling sexual and gender minorities. We offer recommendations for how our methodology can be reproduced with other populations and provide suggestions on how to improve the methodology to ensure response targets are achieved.