MétaCan
Menu
Back to cohort
Record W4362634965 · doi:10.1093/jrsssb/qkad046

Tao Wang's contribution to the Discussion of ‘Vintage Factor Analysis with Varimax Performs Statistical Inference’ by Rohe & Zeng

2023· article· en· W4362634965 on OpenAlexaff
Tao Wang

Bibliographic record

VenueJournal of the Royal Statistical Society Series B (Statistical Methodology) · 2023
Typearticle
Languageen
FieldMathematics
TopicAdvanced Statistical Methods and Models
Canadian institutionsUniversity of Victoria
Fundersnot available
KeywordsVarimax rotationVintageInferenceFactor (programming language)Statistical inferenceComputer scienceMathematicsStatisticsEconometricsArtificial intelligenceGeographyArchaeologyProgramming language

Abstract

fetched live from OpenAlex

I congratulate the authors on their excellent paper studying the fundamental mechanism of PCA with the Varimax rotation. The paper thoughtfully establishes theorems to demonstrate that the Varimax rotation can supply an unified estimating technique for a wide range of semiparametric factor models. My discussion will be primarily centred on clarification and potential extensions. The paper illustrates that if PCA is performed on a matrix with independent elements, PCA with the Varimax rotation can be utilized to fit the semiparametric model under certain assumptions. Is the proposed technique applicable in the case of Quartimax or Equamax rotation? In comparison to oblique rotation, the interpretability of the resulting components from orthogonal rotation is not always satisfied (may oversimplify the data). Although the paper reveals that the developed technique can handle correlated factors, to broaden the scope of the paper’s application, I wonder if the built theorems apply to oblique rotations such as Oblimin and Promax. Also, the paper employs the scree plot (better if using parallel analysis scree plot), which is somewhat subjective, to argue the number of PCs to extract in order to obtain the most parsimonious factor structure. According to the paper, ‘there is not a single correct answer for the choice of k’. Provided that the primary purpose of the paper is to lay the theoretical groundwork for a widely used model, I am intrigued whether the authors can deliver a more theoretically justified method for selecting which factors to retain. In addition, the paper demonstrates that the procedure as a whole is effective on condition that the principal component matrix entries are reasonably nonnormal. Practically, we may assume that the original dataset is normally distributed to ensure that the PCs are independent and the results are more robust. In this case, I am concerned with the implications of the nonnormal condition on factor analysis using a normally distributed dataset. Given the leptokurtic condition on the elements of Z that must be satisfied for Varimax rotation to function, the validity of the multivariate normality assumption regarding the distribution of observable variables or latent constructs appears to be compromised. Aside from that, the variation around the mean for symmetric distributions is a useful measure of dispersion. Nonetheless, it can fail when dealing with skewed or asymmetric distributions; see (Tran et al., 2019). Is the developed method capable of accommodating asymmetric distributions or distributions lacking moments (or with undefined or infinite variance)? There has been substantial research towards the robustification of PCA in the field of robust statistics; see (Candès et al., 2011). Will the Varimax rotation work if we consider a more robust PCA method or are interested in capturing the tail behaviour of the data such as quantile estimation? Data sharing not applicable–no new data generated.

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 imitation

Not 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.

metaresearch head score (Codex)0.012
metaresearch head score (Gemma)0.068
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Not applicable · Consensus signal: none
GenreCandidate signal: Commentary · Consensus signal: none
Teacher disagreement score0.012
Threshold uncertainty score0.063

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0120.068
Meta-epidemiology (narrow)0.0020.001
Meta-epidemiology (broad)0.0010.002
Bibliometrics0.0020.005
Science and technology studies0.0020.005
Scholarly communication0.0030.005
Open science0.0030.003
Research integrity0.0050.008
Insufficient payload (model declined to judge)0.0060.002

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.

Opus teacher head0.114
GPT teacher head0.427
Teacher spread0.313 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designNot applicable
Domainnot available
GenreCommentary

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".

Quick stats

Citations0
Published2023
Admission routes1
Has abstractno

Explore more

Same venueJournal of the Royal Statistical Society Series B (Statistical Methodology)Same topicAdvanced Statistical Methods and ModelsFrench-language works237,207