Accessibility is a Must in the Data Science Curriculum
Talk at Data Science and Society Initiative, Vassar College
Mine Dogucu University of California Irvine
2023-09-20
My Background
Have been teaching in the Department of Statistics at University of California Irvine since July 2019 (with one year detour in Department of Statistics at University College London).
Department of Statistics is in Donald Bren School of Information and Computer Sciences (ICS).
Teach (with) R mostly.
Some statistics from tech world
Only 4.8 % of technical roles are held by Hispanic employees, 2.4 % by Black employees and 25.8 % by women at Meta (2022)
Of Google’s technical roles in the US, 1.2 % are held by Black women, 1.3 % by Latina women, and 0.2 % by Native American women (2023)
Only 6.5% of Google’s global employees(technical and nontechnical) report having a disability, whereas World Health Organization estimates that 16% of the world’s population experience significant disability (2023).
Some Problems with Tech Tools
Statistical translation tools (e.g., Google Translate) yield male pronoun defaults more frequently (Prates et al., 2019).
Facial recognition algorithms are much less accurate in identifying women than men, and darker-skinned people than lighter-skinned people(Raji and Buolamwini, 2019).
Autonomous vehicles fail to recognize behaviors of people with disabilities (Disability Rights UK, 2021).
Cost of books and supplies for a student at a four-year university in 2021 was estimated to be $1240 per year (Hanson, 2021).
About 11% of the students indicated that they skipped meals in order to afford books and course materials (Hanson, 2021).
The bayesrulesbook.com page has been accessed from more than 150 countries.
Assistive Technology
Assistive Technology is any form of technology (software, device) that helps people with disabilities perform certain activities.
Examples:
walking sticks
wheel chairs
screen readers
Screen reader
A screen reader is an assistive technology that supports blind or visually impaired people in using their computer.
The video shows use of a screen reader briefly.
Alternate Text
“Alt text” describes contents of an image.
It is used in HTML pages.
Screen-readers cannot read images but can read alt text.
Alt text has to be provided.
Making math accessible
Calculus is not a prerequisite.
Checking intuition with quizzes.
Supporting mathematical concepts with computing.
Making learning fun
How can we live if we don’t change? —Beyoncé. Lyric from “Satellites.”
vs.
What is probability?
Embracing mistakes as part of learning
As you read the book and put Bayesian methodology into practice, you will make mistakes. Many mistakes. Making and learning from mistakes is simply part of learning. We hope that you persist through the struggle of learning so that you can contribute your unique insights, perspectives, and experiences to the Bayesian community.
v.s.
The proof is obvious.
Making learning relevant with diverse set of applications
84% of industry respondents say it is important or very important to their company to hire developers and designers with accessible technology skills.
60% said it was difficult or very difficult for their organization to find job candidates with accessibility skills. 0% said it was easy or very easy.
41% of companies consider accessibility certification credentials as a differentiator in hiring.
Curriculum Goal
As educators, we have to ensure that the current and the next generation of data scientists, provide public-facing outputs (websites, analysis reports, etc.) that are accessible.
Learning Objectives
Students should get familiar with Americans with Disabilities Act and/or United Nations Convention on the Rights of Persons with Disabilities.
Students should use at least one assistive technology (i.e. screen reader).
Students should consider different representations of data.
Students can use screen reader on Windows and Mac
If you are on a Windows machine go to Settings > Ease of Access > Narrator, and then turn on the toggle under Use Narrator.
If you are on a Mac go to System Preferences > Accessibility > Voice Over > Enable Voice Over
Data Visualization - Colors
Color Blindness Simulator
colorblindr::cvd_grid(fig)
Okabe-Ito Color Palette
palette.colors(palette ="Okabe-Ito")
black orange skyblue bluishgreen yellow
"#000000" "#E69F00" "#56B4E9" "#009E73" "#F0E442"
blue vermillion reddishpurple gray
"#0072B2" "#D55E00" "#CC79A7" "#999999"
fig + colorblindr::scale_color_OkabeIto()
Resources
Color Blindness Simulator You can upload any image and in return get an image with colors that would be visible to color-blind people with specific color vision deficiency.
Okabe-Ito 2008 Color Universal Design. Color palette that is color-blind friendly.
The speed of cars (mph) and the distances (ft) taken to stop
```{r}#| fig-cap: "The speed of cars (mph) and the distances (ft) taken to stop"#| fig-alt: "The scatterplot shows speed ranging from about 0 to 25 mph on the x-axis and dist ranging from 0 to 120 ft on the y-axis. There is a moderate positive linear relationship."plot(cars)```
The speed of cars (mph) and the distances (ft) taken to stop
Manual Alternate Text
Chart type
Type of data
Reason for including the chart
Link to data or source (not in alt text but in main text)
The automated alternate text for this plot is on the next slide.
Data Verbalization - Automated
BrailleR::VI(example_barplot)
This is an untitled chart with no subtitle or caption.
It has x-axis 'transmission' with labels automatic and manual.
It has y-axis 'count' with labels 0, 5, 10 and 15.
The chart is a bar chart with 2 vertical bars.
Bar 1 is centered at 1, and length is from 0 to 19.
Bar 2 is centered at 2, and length is from 0 to 13.
Drawback of using VI() is that not every type of plot can be verbalized yet.
Alternate Texts are Under Utilized in the Data Science Community
“Over the 3 years of TidyTuesday, there were 7,136 data viz tweets and only 215 (3%) of them had alt-text”
Data sonification is the presentation of data as sound.
x <-1:10y <-1:10plot(x, y)sonify::sonify(x,y)
Data Tactualization
Data tactulization refers to making data visualization in a form so that it can be touchable. The video shows printing of a tactile boxplot.
Data Tactualization
library(tactileR)brl_begin(file ='tactile.pdf', pt =11, paper ='special', font='BRL')hist(airquality$Ozone)brl_end()
Different representations of data is one way of introducing accessibility to students but it is not meant to be a complete introduction to accessibility.
Prates, M. O., Avelar, P. H., and Lamb, L. C. (2019), “Assessing Gender Biasin Machine Translation: A Case Study with Google Translate,” NeuralComputing and Applications, 32, 6363–6381.
Raji, I. D., and Buolamwini, J. (2019), “Actionable Auditing: Investigatingthe Impact of Publicly Naming Biased Performance Results of Commer-cial AI Products,” in Proceedings of the 2019 AAAI/ACM Conference onAI, Ethics, and Society, pp. 429–435.
Disability Rights UK. (2021), “Self-Driving Cars Pose Threat for DisabledPeople,” Available at https://www.disabilityrightsuk.org/news/2021/april/self-driving-cars-pose-threat-disabled-people.
Meta Diversity Report (2022) “Embracing Change Through Inclusion: Meta’s 2022 Diversity Report” Available at https://about.fb.com/wp-content/uploads/2022/07/Meta_Embracing-Change-Through-Inclusion_2022-Diversity-Report.pdf.
Google Diversity Annual Report (2023) Available at https://static.googleusercontent.com/media/about.google/en//belonging/diversity-annual-report/2023/static/pdfs/google_2023_diversity_annual_report.pdf?cachebust=2943cac.
World Health Organization (2023). “Disability” Available at https://www.who.int/news-room/fact-sheets/detail/disability-and-health.
