Humanizing the AIGA Census

Spring 2020
6 weeks

Data Visualization
Python (NLTK)
Interaction Design
Print Design

Alice Fang
Rachel Lee
Jaclyn Saik
︎What are designers actually saying about issues in design? 

The AIGA Design Census is a comprehensive overview of the design industry, surveying salary, job satisfaction, future outlook, etc. in its census. Hidden in the raw dataset were the responses to the question “What critical issues face design?,” where respondents could either choose from multiple-choice answers, or more interestingly, write-in their own answers.

The 2700+ written responses were super interesting and expressive, and felt personal in a way that the discrete data did not. We wanted to use this corpus to tell a story, and to humanize the data set—to see reponses as people and not data points.

My roles on the project included parsing the text with python, analyzing responses, prototyping interactions, and laying out the booklet. We worked collaboratively in designing the final web and print interactions.

This project exists in two forms: an exploratory website, and an interactive book

01. The site

The book and site coexist within the same visual system, but have different affordances. A digital interface allows for quick transitions between content, with hyperlinks and visual feedback. This movement through layers of meta information, and the looping experience of continually clicking inspired our visual design and interactions.

I prototyped different iterations of content layout and web interactions, and organized the information architecture of the site, so that the team would remain on the same page as Rachel moved forward with web development. Jaclyn and I finalized the look and visual system of the site.
02. Site Architecture

03. Visual System

The site was designed to be visually busy, with multiple layers of interaction a user can have with the data. We wanted users to ‘physically’ engage with the content, by being able to receive visual feedback from the words themselves. 

04. Theme Highlights

Each quote has highlights on the insights that we pulled that related back to the broader category. Hovering over the outline of a highlight connects the quote to a different category, representing intersections in response topics. This prompts users to either follow a theme through one category, or jump to a different topic entirely. 

05. Quote Preview
The column on the left side provides previews of every quote within a category, and gives the user a way to mark their location and jump quotes without clicking every single one—they can browse based on interest.

06. Spot Words
Aside from the main ideas that we annotated, there are isolated words that, regardless of category, appear throughout the corpus. These spot words, upon hover, show stats about the frequency of use/mention.

07. External Articles
To relate the concerns and critiques that designers are writing about back to the industry and community, users can also hover over external links, which direct to an article on the AIGA Eye on Design site that we curated to match the quote’s topic. 

08. The book
With the book, we wanted to explore translating the digital experience into print. We were inspired by the different narrative paths a reader can take in choose-your-own adventure books. We turned to sticky notes as a method of layering an additional level of information that purposefully obstructs the content, and acts like a physical pop/up or hover-state.

I was in charge of laying out the book, and analyzed responses to pull out the individual keywords and associated quote, from which Jaclyn could find the content to place on each sticky note.

Featuring me as a hand model
09. The (guest) book

We wanted to collect our readers’ reactions and thoughts on the question as well!

I used python and NLTK to parse the corpus, collecting the most frequent 150 words (excluding function words like “I,” “the,” “only,” etc.), bi-grams and tri-grams. Looking at the lists, we picked out words and phrases that described themes or ideas, which we grouped into six broader categories: power, representation, education, quality, change, and community. I then used these dictionaries to parse through the quotes again in order to sort which responses fell in which category.

10. The code

This site is a prototype of a bigger collection. We selected ~20 quotes per category to analyze further, and picked phrases that described the main theme of each response. If quotes intersected with other categories, we made sure to highlight those phrases as well. (And we a few super sexy color-coded spreadsheets.) Jaclyn used NLTK to find statistics for the different spot words on the site.

11. Sorting categories
Code is great way to initially sort and parse data, but when dealing with language, there will still be a lot of manual labor (and love) going into the project. The three of us read lots of responses, annotated, highlighted, and cross-reference ~90 of them for the site, and curated a collection of key words, themes, and external sources.  

I’m really glad this is the direction we took for this project. Reading the responses that working designers in the field wrote gave us a different perspective on the realities of the industry that as students, we haven’t been as exposed to yet. It was disheartening at times to see the same consistent themes show up about diversity, or social media, but we’re hopeful that things are moving in a more inclusive, critical direction. Because the three of us are not developers, there were ideas that we had to make the experience more accessible that we could not produce. For future iterations, including the entire corpus of responses as well as making it a live dataset would make a more dynamic archive.