Issues of Generalization From Unreliable or Unrepresentative Stimuli: Broad Lessons From Lexical Ambiguity
Bibliographic record
Abstract
Abstract The reliability and representativeness of the stimuli used in psychological experiments plays a critical role in the generalizability of their findings. To evaluate the potential impact of reliability and representativeness in psycholinguistics and the cognitive sciences more broadly, we conducted a case study using the domain of lexical ambiguity as a foil. We examined how often studies agreed on the ambiguity types assigned to a word (i.e., homonymy, polysemy, and monosemy), and how well the words represented the populations underlying each ambiguity type. These analyses involved 3597 unique words (14792 tokens) from 240 studies. We observed that (1) there is substantial, albeit imperfect agreement in words being assigned to ambiguity types; (2) that coverage of the underlying populations is relatively poor and biased, with substantial re-use of some stimuli across studies; (3) some clusters of studies engage in substantial stimulus re-use, which although beneficial in some respects, may impact generalizability; and (4) in a series of pseudo-experiments, the aforementioned issues of reliability and representativeness could conceivably alter the reported patterns of effects observed in lexical decision, a popular experimental task. Taken together, our findings raise questions about issues of reliability and generalizability that could impact prior theoretical claims. We discuss our findings with respect to specific considerations related to lexical ambiguity, such as the challenge of ambiguity type labeling, as well as broader considerations relevant to the cognitive sciences, such as the theoretical basis for generalizing, and how we optimize the trade-off between replication and generalization. We close by offering targeted directions to improve research practices.
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 | 0.135 | 0.422 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.002 | 0.016 |
| Scholarly communication | 0.005 | 0.014 |
| Open science | 0.004 | 0.008 |
| Research integrity | 0.003 | 0.004 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
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".