Design Exercise 1: Sketching Solutions to Real Data Problems

in Design-Exercises

For Design Exercise 1, you will sketch ideas for designs for visualization problems using two sample data sets. Later in the semester, you will use tools (or programs) to make real visualizations. For this assignment we want you to sketch. The goal is for you to think of how visualization designs can answer questions, and to practice sketching to explore designs.

Sketching

Important: for this design exercise you should sketch. This means drawing things by hand. You should not write programs to make visualizations. If you want to draw using a tablet or drawing program, that’s OK - but the spirit of the assignment is for you to sketch on paper (and submit a picture of what you’ve drawn).

For scanning your sketches: do the best you can to make a digital version. A picture taken using your phone is OK, but if possible, use a scanning app that rectifies (straightens the page, adjusts levels, …). Apple Notes does this, as does the phone version of Adobe Acrobat. We will not be grading your ability to scan (or your artistic talent) - but the better the scan quality, the easier it will be for us to appreciate what you have done.

Do not turn in .heic files! If your iPhone makes these by default, please convert them.

The goal in this assignment is to sketch: you need to convey your design ideas well enough that we can understand what you are trying to show. We don’t expect precise depictions of the data. If you feel like you need to annotate/explain what your design is, that’s OK. We want to get a sense of your design idea - we are not testing your artistic skill.

There are two aspects to sketching: (1) you have freedom of design since you can imagine what you want to have, without worrying about how to make it; and (2) you have freedom of data - you can think about whether the design would show things if they were there.

Deadlines

Like most things in class, you will turn this in on Canvas at Design Exercise 1: Sketches and Questions. For design exercises, the due time (12:15pm Friday) actually matters: we will try to look at the assignments promptly and give students opportunities to give and receive feedback (if they turn it in on time). Students can then resubmit, based on this feedback, before the cutoff. We will give students who choose this option opportunities for both learning (through the feedback), but also points (through better assignments and additional questions and providing feedback to others).

The only penalty for a late submission (before the cutoff) is that you miss the opportunity to react to feedback and do the “extra stuff” that might get you extra points (providing peer critiques and/or answering questions on how you improved your design). The idea is there is a strong incentive for you to do things on time and iterate.

This early hand-in is an experiment. It isn’t clear how we can have feedback loops in a class where things keep moving.

Catch 1: Academic Honesty

We hope that students won’t try to game the system. For an assignment to be considered “on time” the student must have turned in a complete enough and good enough assignment that it deserves feedback.

Rather than trying to create policy to avoid this, we’ll try using common sense. This is an experiment. We will look at what gets handed in at the deadline and decide. It is the staff’s discretion. Hopefully, it will be easy to tell legitimate assignments from intermediate-works-in-progress.

We ask that you do not upload/submit assignments to Canvas unless you really believe that it is complete enough for grading.

Catch 2: Feedback on DE1
Because of some time constraints, we may not have the mechanisms for peer feedback in place for DE1. That’s why we aren’t giving lots of details for it now.

The Data

The two data sets are described at A Tale of Two Data Sets.... The first questions deal with the census data, and the latter questions are from the life expectancy data.

We recommend that you start by familiarizing yourself with the data sets. For these exercises, we aren’t asking you about the data itself - but its form. (in fact, we will ask you to imagine different possible data in some questions). You will see these data sets again.

The questions are grouped - Canvas numbers questions, we have no way to say “Part A Question 1”.

Question 1,2,3 - A simple question, one visualization, two answers

We’ll start with a “simple” question to make a visualization for. There is a “standard answer” (which is OK), but consider what designs might be effective.

Design a visualization that contrasts how the unemployment rates have changed across the three different levels of “urbanness” (urban, sub-urban, rural) over the years (2000-2019). Sketch your design two ways: (1) imagining that there are interesting differences in the patterns between the levels and (2) imagining that the trends are similar across the three levels. Question 3 will ask you to explain your designs.

Remember: we are asking you to sketch, which means you need to imagine the data. You may not be an expert in unemployment rates in the US. That’s OK - you can imagine something. But, for this example, we want you to imagine that there is a difference in the patterns (for question 1) and a counter-pattern (similar trends) (for question 2). The goal is to use this “data imagination” to see how the design is effective at showing that there are (or are not) differences in the trends between the levels.

A bit about the actual data

The data set labels each county with a “Rural-Urban Continuum” code for each county. There are 9 levels (this is ordinal, not interval), but you can assume they are grouped into 3 levels (again, ordinal, not interval): 1=urban, 2=suburban, 3=rural.

The data set provides unemployment rates for every year from 2000-2022. We’re going to stop at 2019 (since 2020 unemployment is weird because of the pandemic).

In Canvas, there are three questions:

  1. Upload your sketch of your design with “data” where there is a difference in the trends.
  2. Upload your sketch of your design with “data” where the trends are similar.
  3. (a) Explain your design in terms of how the variables are encoded. (a few sentences), and (b) explain how your design makes it easy to see the differences between the data in 1 and 2.
Aside: Why this would be tricky in Tableau (or other tools)

Remember: we are asking you to sketch, not to make an actual picture from the data. To actually make this picture correctly is a bit tricky.

Problem 1: To compute the unemployment rate for a group of counties (e.g., the counties with each urban/rural level), you cannot simply average the rates in the different counties. Some counties have lots of people, some have very few. You could do this with a weighted average (weighting each county by its size). Or you could compute the rates by summing the number of unemployed people and the total population for each level, and then find the ratio for each.

Problem 2: Because each year is a separate variable (or measure in Tableau), making many kinds of graphs requires putting a list of measures together. This is trickier than it should be in Tableau.

These problems are beyond basic Tableau skills (or at least my skills). This is a case where I probably would have written a script to reorganize the data in a format that would make things easy for Tableau.

Question 4,5,6 - A Harder Question

The data includes how many people migrate to each county (this is split into domestic and international migration). It also has the education levels and employment levels (numbers and rates).

Your task is to explain the kinds of places people migrate to. Do they go to places where there are jobs (low unemployment, or a large existing workforce)? Do they go to places where the existing population is more or less educated? Do they go to more rural or urban areas?

Note: you probably want to consider this in terms of rates. And you may not be able to consider all possible relationships at once.

Again, you are sketching, so you get to “make up” the data. Imagine that there are interesting patterns to find, so that they come out in your designs.

For this task, create two different visualizations. Try to imagine similar data, but create two different visual designs that emphasize different aspects of the data.

Upload your designs as Questions 4 and 5. For question 6, describe your two designs and what you can see in each. How do the designs differ to make different things easy to see?

Question 7,8,9 - Biggest Changes in Life Expectancies

With the Worldwide Life Expectancy Data…

Let’s consider the countries with the largest improvements over the time period where we have complete data (1950-2021). In the tutorial, I made something showing this (that was simple, but hard to implement in Tableau). Of course, this is only part of the story.

Create a design that shows that the biggest improvements don’t necessarily lead to the longest life expectancies (i.e., the biggest improvements were in places that were bad to begin with). You want to show where things ended up, but in a way that emphasizes how they got there.

Even describing the possibilities might bias how you think about this… but there are lots of things to possibly show. You might highlight the largest changes, or the diversity of changes, or the correlation, or …

Imagine the data. And then sketch 2 different designs that “tell stories” about the relationship between amount of change and where things end up.

Turn in the designs as questions 7 and 8. For question 9, describe your designs, and explain what each one makes easy to see (the story it is designed to tell).

Question 10,11,12 - Progress in Life Expectancies

In the previous questions, I asked about the overall change. But, progress probably doesn’t happen uniformly. There are probably periods of faster and slower advances. Each country might have a distinct trend, or there might be periods where all parts of the world experience similar things.

Question 10: List 5 questions about this (trends in progress) that could (1) be answered by the data and (2) could be answered with a visualization. Each question should be a sentence. Please provide it as a numbered list.

Question 11: Pick one of those questions and sketch a visualization that “answers” it.

Question 12: Explain the encodings of your visualization, and how it makes the answer to the question “easy to see”. (it should be clear which of the questions you are answering).

Question 13,14,15 - Differences in Lifespan

Consider working with the version of the data that has the rates by sex for 1960-2023.

Historically, in much of the world, females have lived longer than males. Does this vary by country?

Sketch two different visualizations that provide different insights into this (assuming the same data). That is - your visualization should highlight different aspects of the “story”, or show it in different ways.

Question 13: Provide your first visualization

Question 14: Provide your second visualization

Question 15: Describe what each of those visualizations makes easy to see.