Recount: Revisiting the 42nd Canadian Federal Election to Evaluate the Efficacy of Retroactive Tweet Collection

Abstract

In this paper, we report the development and testing of a methodology for collecting tweets from periods beyond the Twitter API’s seven-to-nine day limitation. To accomplish this, we used Twitter’s advanced search feature to search for tweets from past the seven to nine day limit, and then used JavaScript to automatically scan the resulting webpage for tweet IDs. These IDs were then rehydrated (tweet metadata retrieved) using twarc. To examine the efficacy of this method for retrospective collection, we revisited the case study of the 42nd Canadian Federal Election. Using comparisons between the two datasets, we found that our methodology does not produce as robust results as real-time streaming, but that it might be useful as a starting point for researchers or collectors. We close by discussing the implications of these findings.