Design Exercise 2: Life Expectancy (stories): Grading Information V0

in Posts

This page explains the feedback for Design Exercise 2: Life Expectancy (stories). Your feedback is a comment on your Canvas submission. For each item it gives the assessment, any codes, and comments, then an overall comment, the score, and the grade. This is an initial version, prepared to grade the AI prompts.

Items

Some items cover several questions in the Canvas quiz; they are numbered Item 1, Item 2, and so on.

Item 1: Part A: Women and Men

  • Question 1: (Part A) provide your visualization
  • Question 2: (Part A) explain what you see in your visualization, and give the rationale for the desig…

Also rated for: Design Type.

Item 2: Part B: Exploration Loop

  • Question 3: (Part B) Your initial question. (It shouldn’t be a question from another part of the assi…
  • Question 4: (Part B) Your initial visualization, one that attempts to “answer” the question from Ques…
  • Question 5: (Part B) What you see in that visualization, and how your design choices expose it. Then …
  • Question 6: (Part B) A later visualization that addresses the question from Question 5. It doesn’t ha…
  • Question 7: (Part B) What you see in this visualization, the question or story it is meant to tell, a…

Also rated for: Design Type.

Item 3: Part C: Two Stories 1

  • Question 8: (Part C) Provide your first visualization
  • Question 9: (Part C) Explain what you see in this visualization - and how and why your design exposes…

Also rated for: Design Type.

Item 4: Part C: Two Stories 2

  • Question 10: (Part C) Provide your second visualization
  • Question 11: (Part C) Explain what you see in this visualization - and how and why your design exposes…

Also rated for: Design Type.

Item 5: Summary: Process and AI Use

  • Question 12: (Summary Questions) Explain how you made these visualizations. Give details about your to…
  • Question 13: (Summary Questions) How did you use AI tools in completing the assignment? Please be spec…
  • Question 14: (Summary Questions) Provide the prompts that you used. If it’s too hard to put it into th…
  • Question 15: (Summary Questions) What did you do to prevent AI from getting in the way of your learnin…

Assessment scale

Each item is given one of these levels:

  • 0: Missing
  • 1: Poor
  • 2: Adequate
  • 3: Good
  • 4: Excellent
  • 5: Superlative

Criteria

Some items are also classified by these criteria:

Design Type

  • DD: Default Design

    The obvious choice (when there is one). Generally implies a lack of attention to the details to specialize it for the problem. Although, a superlative version of a default design can be achieved through good details. Not a clear distinction with SD (it is usually a subset of SD).

  • SD: Standard Design

    A standard chart type, in its basic form and typical usage. Sometimes DD (if its the obvious solution) and sometimes AD (if alterations or novel usages).

  • AD: Adapted Design

    A standard design that is sufficiently modified. This might be by alterations (e.g., annotations, extra encodings, …) or by a non-standard (but still valid) use.

  • NSD: Non-Standard Design

    Not obviously a named design. Sometimes an uncommon (but known named) design might qualify here. Non-standard doesn’t necessarily mean novel.

  • CD: Compound Design

    A design made by piecing together more standard pieces. Usually, this is a multi-chart layout (e.g., dashboard), but it could also be a layered design.

Codes

Codes are short labels for things we noticed, so the same observation gets the same name for everyone. A code is written like G:ANNOTE: the letter before the colon is its valence (B = Bad, U = Ugly, N = Neutral, G = Good, S = Superlative, A = AI Only). The letter shows how the code applies to your answer, which can differ from the code’s usual valence listed here. A code written without a letter, like :ANNOTE, has its usual valence.

Bad

  • INVALID Invalid Representation: Not a valid representation of the data: a fundamental flaw that violates basic things.
  • NORAT No Rationale: The rationale is missing entirely.
  • PARTWH Incorrect Part Whole: Part whole design used for non-part-whole scenario.
  • SCALES Different Scales: Charts meant to be compared use different scales, making the alignment misleading.

Ugly

  • BARS Too Many Bars: So many bars that they have no labels or meaning. A specific kind of Noisy.
  • CALISBIG California is Bigger: Showing a statistic as an absolute, which ends up really showing results correlated to the overall size, rather than the point at hand.
  • CAPTION Caption - worth mentioning: For commenting on a caption. A missing caption might be Ugly, or Bad (if it really needs a caption). An exceptionally effective caption might earn a Good. Generally, we give a Caption code if the visualization really needs it (visualizations that don’t need captions generally aren’t be penalized for not having them).
  • CHTS Comparison is Hard To See
  • CLAIM Beware of Claims: Declaring that something is good or effective in the rationale doesn’t make it true. A claim of effectiveness that is suspect.
  • COLORS Too Many Colors: Too many colors to keep track of.
  • DATADUMP Don’t Dump Data!: The visualization provides a large amount of data without any guidance to the viewer of how to extract meaning from it. A specific kind of No Clear Story.
  • FUSE Hard to Fuse: Hard to fuse multiple views into a single story.
  • IA Inappropriate Aggregation: Poor choice of aggregation. Generally Ugly, but might be Bad for an actual misuse.
  • IE Ineffective Encoding: Poor choice of encoding. Generally Ugly, but might be Bad for an actual misuse.
  • LABELS No Labels: No labels, and it needs them. Might be Bad if the chart can’t be read without them.
  • MISMATCH Mis-Matched Charts: There are multiple charts without consistency which makes them difficult to interpret together.
  • NCS No Clear Story: The visualization does not help make a story emerge.
  • NEEDRAT Needs Rationale: The choices require the rationale to explain (generally, a poor choice that you had to convince us why it is reasonable). Positive valence for especially compelling arguments.
  • NMV Not Multi-Variate: Doesn’t have a sufficient degree of data richness in the story.
  • NOISY Noisy: Excess data and small features make the overall picture hard to see; usually results in too much overdraw. Consider a scalability strategy, or some way to help the viewer see through the noise.
  • PARTIAL Not the Whole Story: Doesn’t provide a complete answer to its own question.
  • POORX Poor Execution: Poor execution or image quality (e.g., left the Tableau interface in the image).
  • RAMP Ineffective Color Ramp: The color ramp doesn’t fit the data or makes differences hard to see.
  • TEXTSTRY Story Told in Rationale: The story only emerges from the text: you get a lot more after reading. Different from Needs Rationale: the visualization should show the story.
  • TITLE Title - worth mentioning: For commenting on a title. A missing title might be Ugly, or Bad (if it really needs a title). An exceptionally effective title might earn a Good. Absence of this code doesn’t mean the title is there or imply any value judgment: it just means the grader didn’t comment on it.
  • WEAKRAT Weak Rationale: Substandard or incomplete rationale (the opposite of Notable rationale).
  • WHYSEL Why This Selection?: The selection of data is unexplained or seems arbitrary (the opposite of Selection).

Neutral

  • CBE Consider Better Encodings: Not wrong, but a better choice of encodings would help (milder than Ineffective Encoding).
  • MINIMAL Minimal: Meets the requirements, but not much else.
  • NONE Nothing to note: No specific issues or highlights for this part.
  • UNUSUAL Unusual Chart Type: A known (named) chart type that isn’t common. This might be good or bad (many uncommon chart types are uncommon for a reason).
  • WEIRD Weird Design - beyond non-standard: This is for something that is not immediately recognizable, and leads to an “I don’t recognize that” response. It might turn out to be an adaptation of a known design, or something novel. It isn’t necessarily positive or negative.

Good

  • ANNOTE Good Annotation: Good use of annotation for storytelling.
  • CLEAR Clear Effectiveness: The design is clearly effective at communicating its story.
  • DETAILS Detail: Good use of details to make the chart work effectively. Might also be ugly for insufficient attention to detail.
  • LIT Design from the Literature: Good use of a design from the readings.
  • PROGRESS Shows Progression - Question Follows: The second question follows well from what the first answer showed.
  • RATIONAL Notable rationale: Beyond expectations.
  • RICH Exceptional Richness: Exceptional data richness in the story (the positive counterpart of Not Multi-Variate).
  • SCALECTL Scale Control: Focusing (via selection or other means) to manage scalability (the positive counterpart of Noisy).
  • SELECT Selection: Good use of selection as a key/creative part of storytelling. Superlative use is key to storytelling, Good use makes for effective guidance.
  • SELFCRIT Good Self-Critique: Good critique of one’s own design: recognizing its limitations and what could be better.
  • STORY Notable Story: Really shows something interesting in the data, and brings it out through design choices.
  • VISUAL Notably Nice Visuals: Artistic merit: notably well done visually.
  • XFORM Interesting Transform: Interesting data transform or aggregation, e.g., a reduction that makes the problem more tractable.

Superlative

  • NOVEL Novel: Something that we don’t see much of, but makes sense to tell the story it tells.

AI Only

  • NOTHUMAN Not a design a human would likely choose: This is admittedly a judgment call - but a flag to think about what the reasoning is.