Design & Re-Design

CSE 512 - Data Visualization

Jeffrey Heer University of Washington

Design Criteria [Mackinlay 86]

Expressiveness

A set of facts is expressible in a visual language if

the sentences (i.e. the visualizations) in the language express all the facts in the set of data, and only the facts in the data.

​

Effectiveness

A visualization is more effective than another visualization if the information conveyed by one visualization is more readily perceived than the information in the other visualization.

Design Criteria Translated

Tell the truth and nothing but the truth

(don’t lie, and don’t lie by omission)

​

Use encodings that people decode better

(where better = faster and/or more accurate)

Design Considerations

Title, labels, legend, captions, source! Expressiveness and Effectiveness

Avoid unexpressive marks (lines? gradients?) Use perceptually effective encodings

Don’t distract: faint gridlines, pastel highlights/fills The “elimination diet” approach – start minimal

Support comparison and pattern perception

Between elements, to a reference line, or to totals

Elimination Diet approach

• The elimination diet in visualization design is a concept that involves starting with a minimal approach and gradually adding elements to support relevant comparisons.

​

• This approach is akin to the process of eliminating certain foods in a diet to identify triggers for adverse reactions.

​

• In visualization design, the elimination diet strategy focuses on creating visualizations that are clear, concise, and effective by starting with minimal elements and then adding only what is necessary to convey the intended information

​

• This method ensures that the visualizations are not cluttered with unnecessary elements, allowing users to focus on the key data points and comparisons within the visualization.

Design Considerations

Transform data (e.g., invert, log, normalize)

Are model choices (regression lines) appropriate?

Group / sort data by meaningful dimensions

Reduce cognitive overhead

Minimize visual search, minimize ambiguity Avoid legend lookups if direct labeling works Avoid color mappings with indiscernible colors

Be consistent! Visual inferences should consistently support data inferences.

Bar Charts

Counts

CoIIege Admissions: Where is the Gender Gap?

Number of Applicanrs

90O

800

700

600

​

500

​

400

​

300

​

200

​

100

Male Applicants

Rejected Admitted

Female Applicants

Rejected Admitted

Astronomy

Biology

Psychology Sociology

Law Physics

Department

500

400

​

300

​

200

​

100

100

200

400

500

600

How does the pro po rtion of applicants va ry by depa rtment?

Astronomy

Physics

Psychology

Sociology

Biology

Men Women

Law

Rates

Admission Rate

0.9

0.8

0.4

​

0.3

​

0.2

​

​

0.1

^0.0 Astronomy

Biology

Admission Rates Per Department

Law

Men

VVomen

Rychology

Sociology

Physics

Department

·

How Does Gender Play Roles In Admission?

65 1 %

65 9%

76 1 %

• 94 1%

66 9%

63 1%

Admission Rate

Applicant Num.

Astronomy

Biology

Law

Physics

Z5

325

Sociology

191

Admission Rate

Status

Admit

Reiect

​

Gender

Difference

Gender Gaps in Graduate Acceptance

accep. rate

-

Astronomy

Biology Law Physics

Psychology Sociology

0

accep. rate

-4% 0 4%

favoring men

1

favoring women

Q: How do the rates of admission per gender at this university dkfer, how equitable are they, and how do they compare to the proportion of degrees granted nationally?

Astronomy

​

Biology

Law Physics Psychology

Sociology Overall

20

30

People Admitted into Major by Gender

40 50 60

70

80

00

100

Hybrids

Fern a I e

Fern a I e

Are college admissions by deparment equally competitive for men and women?

La.v

Ph\si ce

Percent of applicants admitted

(width of bars is within department percentage of applications)

Rej eQed

Do departments attempt to balance gender during admissions?

Applications & Ad missions

Acce ptance Diłfe ren ce

20%

20%

​

1009

​

80°A

40%

​

20%

Astronomy

Biology

Law

Female Applicants

Females Accepted

Physias

Psychology

Male Applicants

Mates Accepted

sociology

Total

Do Department Admissions Differ by Gender?

Gender Composition of Admits

Admissions Rate by Gender

100%

​

80%

​

60%

​

40%

​

20%

Males Females

​

48% 52% 51% 49% 95% 5% 37% 63%

36% 64%

Astronomy (601 admits)

Biology (46 admits)

Law (269 admits)

Physics (370 admits)

Psychology (322 admits)

Sociology (147 admits)

University Department

Dot Plots

Do Departments Correct}or App£îcatîon Çender Ratio?

Percentages By Gender

Gender Acceptance Deviation From Normal

1 ŒÏ

G Male Applicants

o Female Applicants

Departments

Aqpllcenls

Acœptance Rate Aqpllcants

Acceptent Rate 4ppllcants

Acceplance Rate Agpllcants

Acceptance Rate Appllcants

Acceptance Rate ApgllcanD

Acceptance Rate ApglicanD

Acœptance Rate

8.80°.

3.23°/.

4.ù1°/‹

1.35°/.

0.32%

1.86°.

3.3ù°.

Scatter Plots

Does the proportion of women applicants affect admission rates within each gender?

15

ADMISSION RATE RATIO

% of Females Admitted : 1

% of Males Admitted

0.5

0

1

APPLICANT RATIO

# Female Applicants : # Male Applicants

2

RatI0 0fA

eptance RatIOS

Admissions are biased toward the underrepresented gender

in some departments and less selective departments are heavily dominated by males

"i

Sd,ciology

l.l

p^- j

-_s 0.9-

Psychology

0.7—

Asbonomy

Admission Status

TotalAdmibed

Toal Rejected

Total Applications

584

7o0 800

933

Physics

0 2 4 6 8 10 12 14 16 18 20 22

Raâo of Enering Class (Male : Female)

Department

100

75

50

25

0

Average Salary vs. Gender Ratio

0

Astronomy

Biology

Law Physics Psychology

Sociology

20

O

40

% Female Majors

​

​

Gender Breakdown by Department

0io

20%

40%

60%

Percentage

Astronomy - 903

Biology - 734

Law - 792

Physics - 585

Psychology - 897

Sociology 584

60

80%

100%

80

Male Female

Misc

Simpson’s Paradox

Is there any Gender bias in admissions at DG College ?

a.

Admissions Dashboard

Applications Received (Total - 4,526)

c,

Sociology -

Physics -

E

”c

Biology -

Psychology -

Astronomy -

0 5

10 15

Percent

Gender

Female

Male

College Overalll Admit Rate

Female

Male

Gender

​

Admit Rate

d,

sociology -

Physics -

E

”c

Biology -

Psychology -

Astronomy -

0 20 40 60 80

Percent

Abhishek Pratap • CSE512 Assignment• 1 yySpring’16

Gender

Female

Male

Figure 1. a.) Admissions application statistics. Comparing figure 1.b and 1.d one can see the confounding pattern in admissions data. While overall admission rates show significant difference for females and males (chi-squared p-val <.001) (1.b). department wise number of females and males admitted are seen to be more balanced. Physics and Astronomy departments receives the least amount of applications by females(J.c) yet admit more percent of temales than males (1.d). More males are applying to easier to get-in departments.

·

ALL Μ……

^ή HI

^{E m} ς

Ψ… ^δΌ"?Η

COLL ΕςΕ

ψ[>/μ(ς;|οΝς

Η DE^PARTAAENT

ΑΝ» 6|ΞΝΌΞρ

& ΜιφΜ

& Ά)'ιΜ

Μ ^Ρ

ΛςΤΚοώΜΥ

.

.

Μι… ' 'Μο……

ς"Ο 010… ^ΜΘχ

αν

^Ντακ_ …΄΄

Ρ..

…Μ.

Γ…

Μ

Ρ

Μ

Re-Design Exercise

Re-Design Exercise

Task: Analyze and Re-design visualization Identify data variables (N/O/Q) and encodings Critique the design: what works, what doesn’t Sketch a re-design to improve communication Be ready to share your thoughts with the class

​

Break into groups with those sitting near you (~4 people per group)

Mackinlay’s Ranking

Conjectured effectiveness of encodings by data type

Source: Good Magazine

Source: The Atlantic 300 no. 2 (September 2007) Number of Classified U.S. Documents

Washington Dulles Airport Map Source: United Airlines Hemispheres

Source: National Geographic, September, 2008, p. 22. Silver, Mark. "High School Give-and-Take."

Source: Business Week, June 18, 2007

Preparing for a Pandemic

Source: Scientific American, 293(5). November, 2005, p. 50

Source: Wired Magazine, September 2008 Edition Music: Super Cuts (page 92)