Peer review from Group61

[Yuechen-Zhang603]

Summary
This paper provides a forecast for the 2024 U.S. presidential election based on polling data, aggregating data from national polls and focusing primarily on polls conducted by the Washington Post and ABC News. The paper also discusses the ideal survey and methodological design for election forecasting within a $100,000 budget.

Strong positive point

• Well-structured appendices: The appendices contain a well-organized breakdown of both the pollster’s methodology and the idealized survey design, including a comprehensive budget.

• Detailed methodology analysis: The paper provides a thorough analysis of The Washington Post’s polling methodology, which offers insight into the strengths and weaknesses of the pollster’s approach.

Critical Improvements Needed

1. Lack Model set up and explanation.

The paper mentions using methodology but lacks a detailed explanation of the specific model used for forecasting. For example, it is unclear whether a linear model, generalised linear model , or Bayesian approach was employed, and the assumptions behind the model are not discussed.

2. R and data not cited

The reference page needs to be updated with the dataset which download from the original website, and it also needs to be cited correctly in the text.

3. Insufficient justification for the selection of pollsters:
   Although the Washington Post was selected for the in-depth methodological analysis, the newspaper did not explain why this particular pollster was chosen over others.

4. There is no visual representation of polling data:
   Polling data is usually presented through time series charts, bar graphs, etc. to show changes in candidate support over time or across regions. In this article, polling data is discussed only in text form, which makes it difficult for readers to track trends or compare results across states or demographics.

5. Introduction, data, and discussion sections not completed

Suggestions for Improvement
Provide a more detailed description of the forecasting model, including the type of regression or aggregation model used. Explain why this model was chosen, and discuss any assumptions made in the modeling process (e.g., linearity, independence of errors).

Include the reasons for choosing the Washington Post as the focus of the analysis of the survey methodology. This can be based on grade,the reliability of the methodologywhich can provide a stronger basis for analysis.

Includes charts that summarize key polling data, such as average support for each candidate, sample sizes and important variables. As well as graphs to analyse how different factors have influenced general election trends.

Evaluation
R appropriately cited: 0- Not cited in the main content
Data appropriately cited: 0-Not cited in the main content
Class paper: 1 - Yes. There is no indication that this is a class project.
LLM usage is documented: 0 - No. There is no mention of language models or auto-complete tools used.
Title: 1 - Yes. The title is acceptable but could be more informative.
Author, date, and repo: 2 - Yes. The author and GitHub repository are clearly linked.
Abstract: 0
Introduction:0
Estimand: 0
Data: 6 - Acceptable. The data section is sufficient but could benefit from more details on the weighting of polls and voter turnout factors.
Measurement: 2- Measurement is discussed, but more could be done to explain the rationale for the chosen pollster and model.
Model: 3 - Some issues. The model lacks explanation, and figure analysis
Results: 0
Discussion: 0
Prose: 4 - Acceptable. The writing is clear but could be more concise in some sections.
Cross-references: 0
Captions: 0
Graphs/tables/etc.: 0
Idealized methodology: 8 - Exceeds expectations. The methodology is thoughtful and well-structured but could include more on trade-offs.
Idealized survey: 4 - Exceptional. The survey design is comprehensive and includes all necessary details, including budget breakdown.
Pollster methodology overview and evaluation: 6 - Acceptable. More justification for the pollster selection is needed.
Referencing: 2-not complete
Commits - 2, commits
Sketches - 0 , not updated
Simulation - 4, completed
Tests - 8 well done
Parquet - 1
Reproducible workflow - 2 , README not completed changes
Miscellaneous -2, the paper logic is not bad

Estimated overall mark
52/126

Any other comments:
The appendix section is well done, but further completion is needed in the introduction, data and modelling sections.