GDES-360 spring 2023 / David Ramos, American University Design
Examine techniques for comparing the magnitude of individual observations against the whole. The 2017 Design Census, created by AIGA and Google, gives us an unusually fine-grained look at the people who practice design—surveys since then do not have as much data available. For this project, provide insight into the survey’s salary data, producing a visualization or suite of visualizations.
Look at these issues in depth. Go farther than just, say, the median salary for designers, or medians for men/women/everyone. You might make deeper comparisons—between age groups, perhaps—or look at the shape of the data. (For ways to approach this task, see chapter 7 from The Truthful Art.) See Archie Bagnall’s introductory look at the Design Census data for some early exploration.
Your goal is to explore these data and to practice methods of making visualizations. Your project need not offer grand conclusions, and it does not yet need a polished presentation.
On to design
As with the previous assignment, produce two versions, one that is intended to be as legible and understandable as possible, and one that expresses the meaning in the most compelling or intriguing way.
Your final project can be rough in the sense that type, color, and rendering need not be refined. (You can produce this by hand.) Details still matter—they just don’t need to be finessed in software.
Each version might consist of a single visualization (that is, one figure), but that is unlikely to work well and you should strongly consider arranging several visualizations on the same page.
- Trim size: up to 11×17 in., any orientation.
- Color, grayscale, or black-and-white.
- The reader is generally curious person with at least a high school education, holding your piece at a normal reading distance for a book or a magazine.
- Provide labels and numerical values, but not necessarily for everything.
- A title is necessary, and an introduction will probably help. Please put your name on this.
- Provide notes about your sources, enough that we know where you got the data from. “Data from 2017 Design Census” would suffice.
- Print this for critique.
Tools and methods
Anything, really. A spreadsheet will probably prove useful for exploring and for creating a rough visualization, which you can then edit or trace over. You can make the final versions with Illustrator, by hand, or by another method.
Update: there is now an Observable notebook with initial Vega-Lite visualizations. Start there.
You can also download preprocessed data (ZIP) to work with. This folder contains several subfolders:
summary data/ — Tables of salary data for all respondents, and for respondents identifying as male, female, and nonbinary, broken out by age brackets.
- lo: lower bound of age range, inclusive
- hi: upper bound of age range, inclusive
- min: minimum value (of doubtful significance here)
- qt1: first quartile
- qt3: third quartile
- max: maximum value (also of doubtful significance)
visualizations/ — Prebuilt visualizations, to give you a sense of the shape of the data and, perhaps, to form the basis of some of your work. You can trace over these or edit them. (You probably shouldn’t invest much time in trying to improve the scatterplots.)
filtered data/ — Original survey responses, broken out by gender, and also by gender and age brackets. There were fewer than 100 respondents who described themselves as nonbinary or third gender, so they are not separated by age ranges.
You can also download a CSV file of responses (ZIP), cleaned but unfiltered by demographic group.
Data cleaning and analysis
You’re looking at data that I’ve already cleaned. The original dataset contained 13,158 responses. I cleaned the data by:
- Removing entries that did not contain a salary response.
- Removing responses in the top and bottom 1% of salaries ($210,800 and $0).
- Removing responses with an age of 95 or older.
This left 12,470 responses.
There is still a peculiarity: people are reporting salaries of as low as $1. I’m not sure what to do with these responses, since a $15,000/year salary is plausible, if low; where should you draw the line? (The responses do raise questions about the quality of the data in the Design Census, and about its broader applicability—but we have few surveys about the design field.)
I also renamed several of the most interesting columns so that the names are more reasonable lengths, and removed several columns whose exceptionally long responses were making the data hard to read. (Also, asking about music is fun and all, but what are we going to do with that? The project is asking about salaries.)
Caveat: median pay looks strangely high
The median pay, $57,000, is dramatically out of line with the BLS Occupational Outlook Handbook salary figures for graphic designers, of $48,700 in May 2017.
It is possible that the Design Census captured an unusually large number of high-paid designers, or perhaps this is a consequence of the BLS splitting off web developers (who “design and create websites”) as a separate category, with a May 2017 median salary of $67,990. Perhaps the survey captures people whose jobs fall outside regular design practice.
Earlier, I was discarding responses with salaries over $200,000. The Design Census median salary fell closer to the BLS median salary, and nothing looked surprising.
Despite these explanations, the disparity still concerns me. It is possible that my methodology or R script contain an error. Nevertheless, I’d like to give you these figures to look at for this project, with the caveat that there’s something strange going on.
Other designer surveys, for background
The BLS Occupational Outlook Handbook offers authoritative salary figures for jobs, but it suffers from some omissions, like job classifications that do not reflect contemporary practice. See Arts and Design Occupations and Web Developers.
See also prior surveys of designers:
- AIGA/Aquent Survey of Design Salaries, 2014 (cf. past surveys)
- A List Apart “People Who Make Websites” survey, 2011
- AIGA Design Census 2017 (AIGA has been pulling down the Design Census sites)