Challenges when identifying migration from geo-located Twitter data
Bibliographic record
Abstract
Abstract Given the challenges in collecting up-to-date, comparable data on migrant populations the potential of digital trace data to study migration and migrants has sparked considerable interest among researchers and policy makers. In this paper we assess the reliability of one such data source that is heavily used within the research community: geolocated tweets. We assess strategies used in previous work to identify migrants based on their geolocation histories. We apply these approaches to infer the travel history of a set of Twitter users who regularly posted geolocated tweets between July 2012 and June 2015. In a second step we hand-code the entire tweet histories of a subset of the accounts identified as migrants by these methods. Upon close inspection very few of the accounts that are classified as migrants appear to be migrants in any conventional sense or international students. Rather we find these approaches identify other highly mobile populations such as frequent business or leisure travellers, or people who might best be described as “transnationals”. For demographic research that draws on this kind of data to generate estimates of migration flows this high mis-classification rate implies that findings are likely sensitive to the adjustment model used. For most research trying to use these data to study migrant populations, the data will be of limited utility. We suspect that increasing the correct classification rate substantially will not be easy and may introduce other biases.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.034 | 0.227 |
| Meta-epidemiology (narrow) | 0.000 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.006 | 0.010 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.002 | 0.004 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".