Stereotypes, Discrimination, and Inclusion Case Study: Eating - - PowerPoint PPT Presentation

CMSC 20370/30370 Winter 2020 Stereotypes, Discrimination, and Inclusion Case Study: Eating Disorders

Feb 19, 2020

Quiz Time (5-7 minutes).

Quiz on “Notjustgirls”

Principles of Good Design

Administrivia

• GP2 posters – please bring in for Friday if

your group did not get to present to all three of us

• If your group did present to all of us,

please come in for a regular check- in/progress report

Today’s Agenda

• Midterm summary
• Stereotypes versus inclusion
• Case Study: Eating disorders and

#notjustgirls

Midterm Summary #1

• Too easy ☺
• Average was 37/40
• Major missteps:

– Children on autism spectrum

• Hard to use methods requiring interaction
• Need to help children feel comfortable
• Can observe and cannot observe -> depend on how

you described it

– Discount usability methods

• WoZ is not an inspection method or discount usability

method

Midterm Summary #2

• Survey questions

– Should be closed-ended but did not penalize for this oversight – Filter vs survey questions

• Attention check

– E.g. Please select “A” so we know you are paying attention and can keep your survey response. – Or not an attention check question at all

Case Study: #Notjustgirls

• Looked at posts on Eating Disorders (ED)

associated with males on Tumblr, Twitter, Instagram

• Researchers based classification of male on appearance
• Web scraping of posts over time
• Qualitative analysis of subset of posts
• Implications

– Design automated detection of ED on social media for mental health data collection or intervention – What terms to use?

Trigger warning

Eating Disorders

Types of Eating Disorders Covered In Case Study

• Anorexia Nervosa (5%)
• Bulimia Nervosa (15%)
• Binge Eating
• Avoidant/Restrictive Food Intake Disorder (up to 67%)
• All have disrupted eating habits and weight control

behaviors

• All have physical or psychosocial impairments
• Bigorexia -> body dismorphic disorder – affects

bodybuilders

• Males represent significant portion of these disorders

HCI prior work

• Characterizes ED in online space
• Methods to classify and predict ED (for

interventions)

Research Methods

• Web scraping of posts over 6 months in 2018

from Tumblr, Twitter, Instagram

• Examined 2016 data for terms needed

MenWithED

• Collected new data from Jan 1 to Aug 1 2018

– Manorexia,, menwithanorexia, manorexic, malethinspo, bigorexia – Public posts, no IRB *What do you think about this?*

Privacy and Ethics

• De-identified posts
• Paraphrased
• Did not use images without alteration or

looked for example images instead

Hashtag analysis

Implications

• ED effects are broader than females only
• Tech to detect or predict won’t work unless

it includes these cases

Other gender related studies in HCI

• Low income moms
• Moms and breastfeeding
• Studies of menstruation
• Dads and parenting
• Sex workers
• Intimate partner violence
• Feminist HCI

So be inclusive but also, design is sometimes easier with constraints…

Technique: Personas

• Archetypal character representing a group of

users with common goals, attitudes, and behaviors when interacting with a particular product or service

• Goal directed persona
• Role based persona
• Combination of two
• Can be fictional or created based on user

research

Examples

Source: https://venngage-wordpress.s3.amazonaws.com/uploads/2018/03/user-persona-examples-16.png

Are personas useful then for inclusive technology?

• Yes but likely need other research methods

to avoid stereotypes

Also, is inclusive technology limited

• nly to gender considerations?

What if you’re female and black?

And on that note stereotypes can get us into trouble…

Russian Disinformation campaign on Twitter

So what theories and methods can shape our designs?

• Intersectionality -> person has race,

gender, class-> all affect their day to day lives and lead to intersecting disadvantage

• r discrimination
• Interesting theory -> controversy around

use depending on who you are

• I won’t try to explain but let’s take a look

at how different aspects of your life affects you…

Methods to mitigate these issues?

• Participatory design (see Muller et all, CHI 91)
• Started in European in Scandinavian workplace

– Empower workers – Allow workers to influence tech introduced into workplace – Workers seen as expert and source for innovation

• Involve all stakeholders in the design process
• Pros:

– Shared ownership – End-users are experts – Equalizes power dynamics

• Cons:

– Not always easy to involve everyone – Stakeholders have to be engaged – Resource intensive

Action Research

• Kurt Lewin (1948)

– “research that produces nothing but books will not suffice”

• Produces interventions and then assess how they work
• Researcher highly collaborative with stakeholders
• Critical action research

– Empower groups and communities

• Pros: Shared ownership of ideas
• Cons:

– Not repeatable – Researchers not impartial

Value Sensitive Design

• Batja Friedman et al. at UW
• Considers values of users of technology

and all that are affected by a technology

• Consider values of direct and indirect

stakeholders

• Considers values in all places the tech is

situated in (e.g. work/home/school etc)

• And many other methods such as

reflective design (Sengers et al)….

What does that mean for inclusive technology?

• Technology can’t solve every problem
• Technology is inherently biased
• Technology influences people and people influence

technology

• Working with under-served and marginalized

communities you will encounter these issues

• You can’t and likely should not aim to be a savior
• But you *can* be mindful of biases and design your

systems to be inclusive

– Study your target groups – Try to understand how current designs are helping/hindering people – Evaluate interventions

Summary

• Gender plays an important role in people’s lives
• But can’t focus on just one aspect of a person
• Intersectionality recognizes that race, gender,

sexual orientation and so forth all affect a person

• Inclusive systems cannot always support everyone

and sometimes designs have to be specific

• There are research processes for making systems

more inclusive – these go beyond user-centered design

• Overall, we should be mindful that technologies

are not going to exist in a vacuum

Coming up…

• GP2 poster session continued on Friday

Get in touch:

Office hours: Fridays 2-4pm (Sign up in advance) or by appointment JCL 355 Email: marshini@uchicago.edu