PYSC Final Review 思维导图模板_ProcessOn思维导图、流程图

1. Human Abilities

Function

Sensation and Perception

seeing

hearing

Motor

Communication

Mental

Factors that impact functions

Impairment

Physical Environment Demands

Social Enviroment Demands

Task Demands

Impairment and Disabilities

Impairment = disability?

No impairment = no disability?

What is disability?

Consider both Personal + Environmental = Contextual Factors

Disability is a social construct created by the interaction of one's abilities with context

Design Strategies to Support Abilities and Function

Specialized Design

compensates for special abilities and functional limitations

Assistive Technologies

increase functional capabilities of individuals with disabilities

Accessible Designs

compensate for functional limitations by minimizing environmental demands on individuals with disabilities

子主题

Universal Design

promotes use by people with various abilities and large range of function levels though better design overall

Principles of Universal Design

Equitable use

Flexibility in use

Simple and Intuitive use

Perceptible information

Tolerance for error

Low physical effort

Size and space for approach and use

Accessible Design VS. Universal Design

What determines usability?

people's ability

design characteristics

task

2. Ethics

IRB issues

consent forms

statement that it's a research

purpose and procedures

health and financial risk

compensation/costs

confidentiality

in case of injury

subject rights

appropriate reading level

What is research?

Typically defined as a systematic investigation on a subject that generally leads to the production of explicit knowledge adding to the existing body of knowledge about the subject

what is human subject

3. HCI Research

Found on empirical principles: rely on or derived from observation or experiments

HCI Research 特点

Inherent conflicts in HCI

complex

there is not one optimal solution

tradeoffs and multiple stakeholders with conflicting goals

Interdisciplinary nature

Each discipline has its own methods

Modified methods and created new ones

What does it look like

Emphasizes systematic observations of a sample of

individual behavior

interactions - among people, between people and objects

Uses various degree of "control"

descriptive or comparative research

matin reliability and validity

Uses common methods

Questions in general

what's useful

what's usable

Forms of research

Formative

predictive

through the lifecycle, early and iterative, evaluate the desigh

Summative

make judgments after it's finished, evaluate the implementation

Types of user research

descrptive

Describe a situation

relational

Identify relations between A and B

experimental

Identify causes of a situation

Research Process

Identify a problem

Type of variables

Independent Variables: independent of a user's behavior

Dependent Variables: dependent on a user's behavior or the changes of IVs

Third(confounding) Variable: anything else that you are not manipulating

Research Questions

Hypothesis and predictions

needs to be testable, operationally defines your variables, describes provisonal relationships between IVs and DVs

Research Plan

Samples

Other concerns

Experimental Bias - see what you only want to see

Participant Bias: participants want to please the researcher

Placebo Effect: Tends to think something better when told so

Gather info/data/evidence

what to measure, will them help evaluate your hypothesis

validity / reliability

external validity: will it work outside like it did in a controlled lab

Analyze

Answer the research questions

4. Existing Knowledge

literature review

patent searching

competitive analysis

documentation mining

How the current tasks should be done, standard policies, manuals, histories, best practices

data logging/ analytics

As part of usability testing

5. Oberservation

Various Forms

Standalone / Part of a method

Controlled environments / Field visits

Casual / Structured

Aware / Unaware

More/ Less intrusive

In-person/ Remote

Avoid seeing only what you want to see

Techniques

Try not to interfere

Notes and sketches

Position yourself to really see what happens

Create a plan

Task Analysis (to optimize procedures)

Begin with observation, to see what tasks you are going to support

Where and when - prior to design

Data you need to describe current tasks, methods to gather them, how to present the data, use the data to improve the design

Think about

Requirements to perform a task

info, equipment need to have when they start

Input and output interfacing

sensory, motor, cognitive, communication

Conditions under which task is done

Results and Outcomes - how will they know when the task is complete

Representing Task Analyses

Task Outlines

Hierarchical Task Analysis / Entity-Relationship Diagrams

Flow Charts

Narratives

6. Interviews

Various forms

Stand alone/ combined with other methods

Controlled environments / field vists

In-person / Remote

Individual / Group

Users/ Other stakeholders

Traditional Interview (Decreasing control. potential richness)

Structure Interview

exact the same questions in the same order

more like a survey

Semi-structured Interview

followed a guide

go with the flow

Unstructured Interview

inefficient

good for exploration

Contextual Inquiry

Talk to customers about their work while they work
Users and researcher collaborate to understand the user's work

Stay concrete, don't abstract

we usually get reports by mail - do you have one, can I see it?

Collect ongoing work, not summary experience

Partnership

the user is the expert

Help the user articulate and see their work practice

Intepretation
Assign meanings to observations

Create a shared understanding

Offer interpretations, not just open-ended questions

Inquire into the meaning of customer action and words

Be honest

Focus

Know your focus

Challenge your assumptions

Dos and don'ts

Where and when in research

Good for exploratory work

Gain feedback about design concepts
Broadly understand information architecture and content strategy needs
As part of usability testing

To find out - attitudes, behaviors, how, opinions, comparison to others
Not to find out - predict the future, how they design, hypothetical situations

Interview Conditions

Planning

Research Questions

Avoid

Jargon, Slang

Close-ended questions - if so, followed with "why"

Biased or leading questions

suggest an answer to the question
provide reasons for something doesn't work

Predictions about future

Emotion-focused question

ask about behaviors, not feelings

Imagining themselves in hypothetical scenarios

Guidance on specific design needs

Example

1. Would you use X feature?
• Typically answered with a yes or no
• People are bad at predicting their future behavior
2. What would you need in X app?
• People don’t typically ask for the “right” thing, especially when their need is abstract
• Focus on “why” questions instead. “What feature would solve this problem,” should be phrased as
“tell me more about this problem.”
• It is the designer’s role to extract insights from the interviewee
3. What are your goals, motivations, and pain points?
What makes this question bad?
• The purpose of a user interview is to derive a user’s goals, motivations, and pain points, but you
should never ask this directly – these things lay beneath the surface of a person’s consciousness

Ask: Open Ended Questions : can't be answered with yes or no

Three awesome questions

What are you trying to get done?

How do you currently do this

What could be better about how you do this?

Recruiting - screening

Turn off your Assumptions

Rapport

Listening

Notetaking tips

annotate your notes

Respond to their responses

Support interviews with objects and artifacts

Photo Elicitation

use photos to stimulate vivid, concrete, meaningful information

Whats the first thing comes to your mind

what you would like to use to describe how you could feel if you were part of this scene

7. Focus Group

purpose is to collect data

Just use group interaction to elicit information from the members

Homogeneity= participants share similar relevant characteristics ⤴️ backgrounds and experiences, success of the group ⤴️

Pros and Cons

Pro

Understand perceptions, beliefs, opinions of wide variety of participants

Depth of information

Influence of the group context

Flexible and dynamic with a relatively low cost

Useful for exploratory initiatives

Con

Requires skills

Make sense of the data

Time and effort of researchers

challenging group dynamics

focus group can influence individual responses

format can be challenging

Not so good for

Closed-ended questions

assessments with statistical data

participants are not comfortable with each other

group may sway individual opinion

unsafe or confidential enviroments

situations that are emotionally changed

Location

Structure

Ground rules

Moderator role

Wants/Needs Analysis

"moderated" focus group with brainstorming and prioritization

i wish / how to " represents needs and ideas

8. Qualitative Data Analysis

What is qualitative data?

transcript of interviews, focus groups, observations

related to concepts, attitudes, opinions, values and behavirous

not numbers

What is qualitative data analysis?

to make researchers familiar with the data

Analysis Approaches

deductive 演绎

use your research questions, theory to group the data or develop codes

inductive 归纳

use an emergent framework to group data and then look for relationships

Most common approaches

grounded theory

aims to derive theory from systematic analysis of data

Coding: categorization approach

Three levels of coding

Open

Identify categories

Axial

flesh out and link to subcategories

Selective

from theoretical scheme

contextual inquiry

Contextual design process

Contextual interviews- interpretation sessions- affinity mapping- visioning

Affinity Process

Affinity notes are about what participants do, how they feel and what they want

Affinity mapping is hierarchically group affinity notes from interpretation sessions

The process

1. Raw data : data cleaning

2. Data reduction: Chunking and coding

3. Interpretation: coding and clustering

4. Representation: making sense and storytelling

code and count

etc

9. Quantitative data analysis

have your analysis plan beforehand

data types

discrete data

nominal (categorical data - number of occurrences of each qualitative data code)

ordinal (rank order data)

continuous data

interval

without a meaningful zero point - rating scale responses

ratio

zero has a meaning - heart rate

population and samples

parameters and statistics - 母体特征和样本统计量

descriptive and inferential statistics - 描述性统计-characterize a sample, but don't draw conclusions about the wider population 推论统计-predict the values of features of an entire population

histograms - 直方图

a diagram consisting of rectangles whose area is proportional to the frequency of a variable and whose width is equal to the class interval.

some descriptive statistics

central tendency

mean

median

center value

mode

most commonly occuring balue

spread/dispersion

variance

standard deviation

shape

bimodality/unimodality

two or one peak

skewness - how asymmetric is the distribution 偏斜

positive

long tail in the positive direction

negative

kurtosis 峰度

the flatness of the distribution

visually represent multiple distribution - Box and Whisker Plots

outliers - extreme data points

error

algorithmic classification

how to treat them

Bar graph with standard deviation

standard deviation (样本的), plotting what we expect a population to look like

correlational statistics

direction

person correlation p

-1 perfect negative correlation
1 perfect positive correlation
0 no relationship

strength

R^2 balue

1 - A fully explains B. 0 - no relationship

influential outliers

remove

form of relationship

！

assumptions when using a survey

respondent will give truthful answers

they can give truthful answers

they understand your questions

pros and cons

Advantage

cost

large sample size

ease of completion

possible less bias than face to face interview

timely

uniform question presentation

anonymity and confidentiality possible

anonymity

can't associate any response with a specific participant

confidentiality

can - used to track data, but only group results are reported, individual data not reported

if anonymity not guaranteed

result in misrepresentation of income, job, age, education level. and special issues like drug use, hours worked

disadvantage

low response rate

reliability and validity issues

question limitations

prejudice against questinnaires

impersonal 不客观

possible sample limitations

knowing who completes questions items interdependence

questions

wording

avoid leading and loaded questions

avoid agreement bias

full sentences are more likely to elicit agreement because they are posted as statement of fact
Georgia Tech is a satisfactory program - agree or disagree

avoid use of negation - 双重否定

avoid asking double questions

avoid "giveaway" words, undefined terms and ambiguous questions

be more specific about frequency and duration

common scales

open ended

always ask them to explain why

Pros and cons

Pro

raise issues not addressed by survey designer

useful to determine range of responses

help to determine respondents' frame to reference

Con

Time consuming to answer

Difficult to scope

Response can be misinterpreted

may generate repetitive material

fill in the blanks

Binary/multiple
(by adding "additional options")

pro

easy to complete

everyone has a same reference point

easily scored

con

can't say how much better one choice is than another

response bias set

response alternative may overlap

designers need to know full range of responses

Forced choice items

pro

more resistant to biases and response sets

can force thoughtful response

con

frustrating

hard to develop properly

time consuming to complete

Rating scales

pro

reflects both direction and magnitude of opinion

some relative measure of distance between ratings

relatively easily answered

applicable to many issues

con

vulnerable to bias built into stem and other biases

recommendations

response options must be clearly different

use standard tems

use a natural rating order

whichever order you choose, be consistent throughout the questionnaire

ranking

pro

familiar to respondent

easy to score

con

produce high error rate

doesn't allow respondent to tell the difference between items

less precise than ratings

must read the entire list

generally better to use rating scales unless rank data are adequate
use the minimun number of response categories per question

checklist

semantic differential

put an X in the rows

Better accessibility

Content of the survey questions themselves, exclude

only have a visual referent if not necessary

might be difficult for AT to interpret

Avoid JavaScript based questions

grid questions

avoid hidden questions

inclusive language

Instructions govern the survey's behavior

visual design

hight contrast

larger font size

obvious buttons for actions

are respondents who complete the survey representative?

better approach

distribution methods

internet-based

do they have internet access?

bias towards internet users ?

perceptions of internet security and privacy concerns

designed for larger displays

Pre-testing / pilot

Test your survey

check for accessibility

check for readability and interpretation

depends on its content and typography

check for timing

survey to inform survey

use an open-ended survey to discover most common responses

add "other", if more than 10% of respondents use "other"

plot the question strategy and design

11. Behavioral Persona

12. Statistics II

four steps in hypothesis testing

Type I Error and Type II Error

Type I: false positive 算出显著性差异拒绝了H0 但是实际上H0 IS TRUE

Type II: error/miss 算出没有显著性差异接受了 H0 但是实际上有 Ho is false

Null and alternative hypotheses

Null (H0): There is in fact no difference between the different levels of an independent variable in terms of their effect on a dependent variable

Alternative (H1): there is in fact one

sampling distribution

central limit theorem

平均值就是Mean，满足正态分布

足够多的样本就会变成population distribution

P value

the probability that we would make a Type I Error if we claim the alternative hypothesis is sture
the probability that the new sample is in fact from the original population, but we mistakenly conclude that it is from the alternative distribution
0.05 - decidion threshold

We mean that it is at least 95% likely that two groups, perhaps Design A and Design
B, are different in the population in terms of the dependent variables. We accept that
there is a < 1 in 20 chance that we are mistaken, and they are not in fact different in
the population.

Increasing statistical power
lower the risk of Type II Error

increase the sample size

increase the size of an effect (make design A more different from design B)

decrease the random error component of the sample standard deviation (design a better scale rating survry)

difference between a significant effect and large effect size

an effect size is a quantitative measure of the strength of a phenomenon.

if the result is meaningful

significance testing is all-or-nothing. a different is either statistically significant or not. but the fact that two groups are statistically significantly different doesn't mean there is necessarily that large of a difference

13. Persona

Rather than traditional personas, we built behavioral ones ^[14].
The reason was that this project focused on
improving and getting a better understanding of an existing system’s
performance for a given audience ^[15]. Our users are
relatively well-defined. Behavioral personas tell us what our users do and how
well they work with the current information gathering process when selecting a
course.

typical user concluded from user profile

14. Design Iteration - From evidence to design

Five panel of UX

Strategy

user needs, product objectives

strategy document

market research, competitive analysis, focus group, interviews, call logs, surveys,

Scope

features and functions

functional spefications.

content requirements: text images audio video, management system, etc

requirement document - be positive, be specific, avoid subjective language

contextual inquiry, contene analysis, interviews, focus group, surveys, comptietive analsyis, analytics

Structure

flow

interaction design

perform and complete task

describe user behavior and how system responds to that behavior

conceptual model: goal: consistency

information architecture

convey info to user

paper prototype, storyboard

paper prototype interview, card sorting, treejac

efficient and inexpensive

Skeleton

placement of buttons and controls
information design

interface design

navigation design

wireframe prototype. specification

user testing

Surface

visual, images, aesthetics

sensory design

functional prototype, design documents

user testing

Functionality or Information

15. Research Design Research

Research

Design

Research through Design - use design to ask questions.

Iteration design research

Design practice

does so with a research in mind

Design studies

what if - design becomes a statement of what's possible

Design exploration

discussions about design theory, methodology and history, philosophy

Design to create knowledge

lab: rich interaction design

Showroom: critical design

Field: participatory design

bring users into the creative process. to see the world through their perspective

PD as research

to understand how people truly think about a given problem or technology

when you think what users say and actually do are not the same

when you feel that there is a disconnect between you and the end user

PD Tools. methods. slides

16. Design a Study

Bias

Primary Effect

Recency Effect

Memory

the order in which you present or ask things can affect the results

Order Effect

people get tired, knowledge of the earlier prototype may affect later performance,

randomize the order of tasks and prototypes

Effect of the Researcher

Placebo & Bias

Unwanted Bias

Hawthorne Effect (Observer Bias)

participants try to improve their performance because they know they are being observed

Training and Practice Effect

Between subjects or within subjects

Between

Each participants receives a different condition, a comparison is made from the data between subjects

fewer recency effect, simple design and analysisi, easier to recruit participants, less efficient

Within

Each participant receives both conditions, so that they act as their own control and comparison can be amde between different levels of IV without worry for individual diffferences

more efficient, stastistical power, require fewer participant, more complicated, design to avoid recency effects

A/B testing

Include

research question

participants

procedure

measures

17. Accessibility

WCAG 2.0

P perceivable

small screen size

zoom/magnification

contrast

O operable

keyboard control for touchscreen devices

touch target size and spacing

touch screen gestures

placing buttons where they are easy to access

U understandable

changing screen orientation

consistent layout

position important page

group operable elements perform same action

provide clear indication that elements are actionable

provide instructions for custom touchscreen and device manipulation gestures

R robust

set virtual keyboard to type of data entry required

provide easy methods for data entry

support the characteristic properties of platform

basic principles of accessible web content

1. accessibility statement

2. Alt Tags

3. Color and contrast

high contrast color scheme

background does not overpower text

color schemes used consistently

avoid color coding

4. hyperlinks

should make sense out of context, describe the destination, unique for each unique destination

5. accessible multimedia

text transcript

video description

closed captions

accessible media player

6. readability

7. tables

ARIA

javascript cannot communicate with accessibility tree - add ARIA to the code to identify properties, relationships and states

mobile more important than ever

screen reader

magnifier

color settings

text settings

captioning and video

description

18. Evaluation

what is evaluation

gather data about the usability of a design for a particular activity by a specific group of users within a specified enviroment

goals of evluation

assess extent of system's functionality

assess effect of interface on user

identify specific problems with system

styles of evaluation

formative (predictive)

all through the lifecycle

summative evluation

make judgments about the final item

experimental/empirical approach

lab studies, quantitative results

manipulate IV to see effect on DV

replicable but expensive

naturalistic approach

field studies, qualitative results

observation occurs in real life setting

ecologically valid

cheap and quick

not reproducible, yield user-specific results

not quantitative

predictive- without users

interpretive evaluation (naturalistic)

predictive evaluation

user used

heuristic evaluation

several experts assess system based on simple and general heuristics

visibility of system status

aesthetic and minimalist design

user control and freedom

consistency and standards

error prevention

recognition rather than recall

flexibility and efficiency of use

recognition, diagnosis and recovery from errors

help and documentation

match between system and real world

perform 2+ passes through the system, find problems. inspect each screen and between screens

then decide what are or not problems, group and structure

severity ranking 0-4 based on frequency, impact, persistence and market impact

pro and con

pro

cheap and good for small companies who can't afford more

get someone practiced in method is valuable

con

a little subjective

why are these controversial

some identified problems really aren't

discounting usability testing

hybrid empirical usability testing and heuristic evaluation

have two or three think-aloud user sessions with paper or prototype mock-ups

cognitive walkthrough

assess learnability and usability through simulation of the way users explore and become familiar with interactive system

task-specified

review actions needed for task and predict how users would behave and what problems they will encounter

need to have

users description and their profile

tasks

complete list of the actions needed to complete tasks

prototype or description of the system

ask four questions to construct a believability story

will users be trying to produce whatever effect the given action has

will users be able to notice that the correct action is abailale

once found, will they know it's the right action for the desired effect

will users understand feedback after the action

literature-based evaluation

many system exist have been evaluated. experiments have shown performance abilities and limits. apply what is already known

model-based evaluation

computer simulate what users would do and how they would respond

not so effective at modeling perception

Fitts Law

formal evaluation

questionnaires

think-aloud / cooperative evaluation

experimental/usability test

empirical evaluation

determine the task

goal

task scenarios

benchmark tasks

gather quantitative data

specific, clearly stated tasks for users to carry out

representative tasks

add breadth, can help understand process

make the task realistic

make the task actionable

to do the action, instead of asking a question

avoid clues and steps

performance measures

experiment

Variables

hypotheses

IRB

participants

must fit in user population - validity

screening

set inclusion/exclusion criteria

include distractor answers and questions

data

analyze

conclusion

redesign and implement

19. Metrics and Physiology

Task performance metrics

task success

frequency of task completion

task time

time taken for completion

have a ceiling effect

errors

frequency of errors by task

efficiency

mental and physical effort required

often # of steps

learnability

performance improvement over time

some performance metrics are more diagnostic indicators of the variable pf interest than others

Ceiling or floor effect can occur when users never perform higher or lower than a certain threshold

Issued-based metrics

informal

Subjective reaction metrics

reflect on their interaction with a system.

think-aloud protocol

difficult to get people to actually do this- especially when the task gets hard

can be combined with other metrics to help observed data tell a story

self-report metrics

users reflect and rate aspects of their interaction with the system

e.g Semantic differential scale

surveys

usability surveys

workload surveys

time, mental effort, psychological stress load

engagement surveys

behavioral & physiological metrics

eye tracking

eye movement

goals

visually salient elements

fixation: moments the eye pause to take in and process info

saccade: quick, partially involuntary eye movement used to shift the eye toward a nearby area that may hold useful info. followed by another fixation

a visit: humans often saccade into a general area, then saccade and fixate on lots of nearby areas within that area. once sufficient info has been acquired, the eye saccade away from that area
AOIs: areas that users visit tend to correspond with elements of visual display.

procedure

define AOIs

calibrate eye tracker

record data

generate visualizations as well as numerical data for each AOI

automated cluster AOIs

pros and cons

metrics generated for each AOI

fixation count

total number of times the eye fixated within an AOI

are users spending time to consider content in an AOI? are there a lot of separate info elements in that AOI

average fixation length

are users spending a lot of time processing each info element in an AOI

time until first fixation

is an AOI being seen as soon or as late as the designer wants?

average visit length

how important? how condusing? how much valuable info?

total visit time

visit sequence

hit rate

percentage of users who looked at least once in an AOI

visit count

visualiztion

focus on sequence

focus on time amount users view different content

combining subjective reactions and eye tracking

retrospective think aloud

users are shown the video of their eye movements during a task and aksed to comment on what they were thinking at that time

physiological measurements

continuous, unobtrusive, quantitative and unbiased reflection of the user's internal state during a task

measure physiological arousal, which comes about through the activity of the autonomic nervous system

sympathetic nervous system

easy to measure

increase heart rate, increase rate of breathing,
blood go to muscle and brain: constricts blood flow to nonessential organs like digestive organs and the skin, sweat response

perceived threat

electodermal activity EDA - increased sweat causes increased conduction of electricity

most important!

difficult when hand and foot moves near the sensor

Skin temperature SCN

cooler - blood flow to skin is decreased

difficult : change in enviromental temperature

Respiration rate

increase the rate of breathing

difficult: physical activity

Heart rate variability HRV

increase

Heart Rate ECG.EKG

increase

difficult - muscle movements

pupil size ⤴️

light changes

arousal vs workload

increase in cognitive workload co-occur with sympathetic nervous system response, raising physiological arousal

measure cognitive activity directly

EEG - pick up brain activity as a whole in real time

beta/ alpha, more beta waves, higher mental activity

difficult- hair. lack of solid skin contact

FMRI - temporal resolution, spatial resolution, but you are in a giant magnet

fNIRS- portable without giant magnet - difficulty: sunlight

sense affect: valence

facial expression recogntion

biometric storyboards

live website data

Google Analytics

information architecture metrics

Tree testing, card sorting

accessibility metrics

子主题

composite scores

20. Usability Testing

The systematic observation of end users attempting to complete a task or set of tasks with your product based on representative scenarios.

task should be a representative sample of typical things a user might do, reasonable

tasks should be described in terms of a person's goals and motivations not the system's

tasks must be possible, with a definable success/fail conditions

F: They want to give up, they reaches wrong but think they are right, or the opposite

define errors

have a specific end goal

in a realistic sequence

require an appropriate amount of expertise in the domain

verbal protocol

think-aloud

problems: awkward, modify wau users perform a task, won't describe all the thoghts

post-event protocol (retrospective think-aloud)

difficult to recall

notes

paper

miss things, slow, cheap and easy, ultimately less work

record

good for talk aloud,, hard to tie to the interface, multiple cameras probably needed, good, rich record of session, intrusive

software logging

too low-level, massive amount of data, need analysis tools

sensors and physiological logging

good at reflect users' internal state, hard to interpret, requires participants wear strange devices

Hawthorne Effect (Observer Bias)

remote testing

can't probe into unexpected results and insight

can't change the script on the fly

can't corral the participant if she gets off track

ay not able to give them the design because of security concetn

can't observe body language

distracted

cheaper, reduce moderator bias, more natural environment, get more data quicker, easier to get a large sample size, wider geographical reach

21. Business Model Canvas