# 36-613 Data Vis Rubric

**Total: 100 points**

This rubric evaluates the project as visual analysis. The central questions are whether
the graphics are well chosen, statistically honest, readable, and assembled into a
coherent evidence-based story.

**This is the 36-613 Data Vis Rubric.** It grades only the self-contained HTML client
report and its embedded reproducibility materials. The Professional Skills Rubric is a
separate working draft for the companion-course presentation and is not yet final; it
is not used to grade the Data Visualization report and is not part of the 36-613 grade.

## Required portfolio

The submission should include:

- 6–10 visualizations;
- at least three foundational forms from the first half of the course;
- at least three later-course or advanced forms where they genuinely help;
- no more than two one-variable plots;
- no more than three plots of the same type;
- at least one statistical analysis, model summary, sensitivity analysis, or uncertainty
  display that complements a graph; and
- one self-contained HTML report with an embedded reproducibility appendix.

The **best six eligible graphics** are scored individually. Additional graphics can help
the overall story, but do not earn extra graph points. If fewer than six eligible
graphics are submitted, the missing graph slots receive zero points.

## 1. Visual portfolio · 48 points

Each of the best six graphics is worth 8 points.

### A. Purpose and graph choice · 2 points per graphic

- **2:** The graphic directly addresses a decision-relevant question, and the chosen
  visual form fits the variables and comparison task.
- **1:** The graphic is relevant, but the question, comparison, or choice of form is only
  partly justified.
- **0:** The graphic is tangential, redundant, or poorly matched to the task.

### B. Encoding and statistical honesty · 2 points per graphic

- **2:** Encodings, scales, transformations, denominators, subsets, and comparisons are
  appropriate and do not overstate the evidence.
- **1:** The main pattern is usable, but an analytical or encoding choice is unclear or
  weakly justified.
- **0:** The graphic is materially misleading or does not support the stated claim.

### C. Readability and accessibility · 2 points per graphic

- **2:** The graphic is easy to read, properly labeled, legible at its final size, and
  uses color and formatting accessibly.
- **1:** The graphic is understandable but has noticeable labeling, clutter, contrast,
  sizing, or formatting problems.
- **0:** The graphic is difficult to interpret because essential labels, units, legends,
  or readable encodings are missing.

### D. Emphasis and interpretation · 2 points per graphic

- **2:** The title, annotation, caption, and surrounding text clearly identify the main
  comparison, finding, and important limitation.
- **1:** An interpretation is present, but the takeaway or limitation requires work from
  the reader.
- **0:** The graphic is presented without a clear or accurate interpretation.

## 2. Cohesive visual story · 25 points

- **5 pts:** Three clear client questions organize the analysis.
- **7 pts:** Graphics appear in a purposeful sequence rather than as a collection of EDA
  outputs.
- **7 pts:** The report uses consistent typography, color, labeling, spacing, and visual
  hierarchy.
- **6 pts:** The conclusion follows from the displayed evidence and distinguishes the
  recommendation, caveats, and next questions.

## 3. Data scope and evidence boundaries · 10 points

- **4 pts:** Clearly explains what a row represents, coverage, exact duplicates, and the
  consequences of cleaning or grouping decisions.
- **3 pts:** Displays relevant sample sizes, units, subsets, denominators, and sources
  where they affect interpretation.
- **3 pts:** Avoids unsupported causal, demand, market-representativeness, or geographic
  claims and states what the archive cannot establish.

## 4. Statistical support · 10 points

- **5 pts:** A statistical analysis, model, uncertainty display, or sensitivity analysis
  appropriately complements a visual question.
- **5 pts:** The result is interpreted correctly and integrated with the graphic without
  treating statistical significance as the whole conclusion.

## 5. Reproducibility and submission quality · 7 points

- **3 pts:** The single submitted HTML file opens correctly with figures, styling, and
  other required assets embedded.
- **2 pts:** Data links, access dates, cleaning decisions, role-grouping rules, and
  package/session information are documented.
- **2 pts:** Captions or alternative text, source notes, and the final output format make
  the work usable beyond the authors' own computers.

## Scoring notes

- A beautiful graphic cannot recover points for a misleading comparison.
- A technically correct graphic does not receive full credit if the reader cannot tell
  what matters.
- Repeated versions of the same graphic are not automatically distinct visual forms.
- Facets count as one graphic.
- Extra graphics that weaken the story can reduce the cohesive-story score.
- A submitted HTML report that cannot be opened may lose all reproducibility points and any graph
  points that cannot be evaluated reliably.