Tweets or nighttime lights: Comparison for preeminence in estimating socioeconomic factors.

Abstract Nighttime lights (NTL) imagery is one of the most commonly used tools to quantitatively study socioeconomic systems over large areas. In this study we aim to use location-based social media big data to challenge the primacy of NTL imagery on estimating socioeconomic factors. Geo-tagged tweets posted in the contiguous United States in 2013 were retrieved to produce a tweet image with the same spatial resolution of the NTL imagery (i.e., 0.00833° × 0.00833°). Sum tweet (the total number of tweets) and sum light (summed DN value of the NTL image) of each state or county were obtained from the tweets and the NTL images, respectively, to estimate three important socioeconomic factors: personal income, electric power consumption, and fossil fuel carbon dioxide emissions. Results show that sum tweet is a better measure of personal income and electric power consumption while carbon dioxide emissions can be more accurately estimated by sum light. We further exploited that African-Americans adults are more likely than White seniors to post geotagged tweets in the US, yet did not find any significant correlations between proportions of the subpopulations and the estimation accuracy of the socioeconomic factors. Existence of saturated pixels and blooming effects and failure to remove gas flaring reduce quality of NTL imagery in estimating socioeconomic factors, however, such problems are nonexistent in the tweet images. This study reveals that the number of geo-tagged tweets has great potential to be deemed as a substitute of brightness of NTL to assess socioeconomic factors over large geographic areas. [ABSTRACT FROM AUTHOR]