POSTER: The Impact of Group Imbalance on Logistic Regression Analyses with Assessment Data
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Introduction Logistic regression (LogReg) is widely used in analyzing educational assessment data, a special case of which is LogReg for differential item functioning (DIF). The model of interest in this paper is a LogReg akin to an analysis of covariance analysis with a dichotomous outcome variable and three predictors: a continuous covariate, a (skewed) dichotomous grouping variable and the interaction. The skewed grouping variable reflects an imbalance in sample sizes of the two groups. Little to no research has been done to examine the effects of skewed predictors on parameter estimates in logistic regression. Objectives The present simulation study investigates the impact of unbalanced group membership on the Type I error rate and statistical power of the Wald tests in this model. Methods To examine Type I error of the Wald tests: a 4A—4A—10 completely crossed factorial design, varying three factors: sample size (from 200 to 5000), skewness of the dichotomous predictor (from 50:50 to 1:99), skewness of the dependent variable (from 50:50 to 1:99). To examine power, a 4A—4A—10A—2 completely crossed factorial design, varying four factors: the same three as studied for Type I error, and effect size of dichotomous predictor and interaction (odds ratios of 2 and 4). Results and Conclusions The Type I error and power findings are a complicated interaction of the skewness of the dependent variable, the imbalance of the group sizes (i.e., skewness of the grouping predictor variable), and sample size. As a general statement, the Type I error rate and power are negatively affected by severe imbalance in group sizes. In cases wherein the Type I error rate of the Wald test of the grouping variable is effected, it is consistently deflated and close to zero. The complicated findings will be interpreted focusing on providing advice for data analysts and practitioners.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 it