Bibliographic record
Abstract
Microtask platforms are becoming commonplace tools for performing human research, producing gold-standard data, and annotating large datasets. These platforms connect requesters (researchers or companies) with large populations (crowds) of workers, who perform small tasks, typically taking less than five minutes each. A topic of ongoing research concerns the design of tasks that elicit high quality annotations. Here we identify a seemingly banal feature of nearly all crowdsourcing workflows that profoundly impacts workers' responses. Microtask assignments typically consist of a sequence of tasks sharing a common format (e.g., circle galaxies in an image). Using image-labeling, a canonical microtask format, we show that earlier tasks can have a strong influence on responses to later tasks, shifting the distribution of future responses by 30-50% (total variational distance). Specifically, prior tasks influence the content that workers focus on, as well as the richness and specialization of responses. We call this phenomenon intertask effects. We compare intertask effects to framing, effected by stating the requester's research interest, and find that intertask effects are on par or stronger. If uncontrolled, intertask effects could be a source of systematic bias, but our results suggest that, with appropriate task design, they might be leveraged to hone worker focus and acuity, helping to elicit reproducible, expert-level judgments. Intertask effects are a crucial aspect of human computation that should be considered in the design of any crowdsourced study.
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 | 0.018 | 0.113 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.005 | 0.003 |
| Open science | 0.001 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.008 | 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".