UNDERSTANDING USERS: MODELING TASKS AND LOW- LEVEL INTERACTION Human-Computer Interaction 10.13.2012.

UNDERSTANDING USERS: MODELING TASKS AND LOW-LEVEL INTERACTIONHuman-Computer Interaction 10.13.2012

Agenda

Project Part 1 notes Task analysis review An overview of HCI design Fitts’s Law Project group time

Project Part 1

Problem or task description, revised and refined Description and justification of how the information

above was collected Description of the users and their tasks related to the

problem, including a task analysis of a single task Description of the larger system and environment Analysis of the existing systems and tools used in the

problem Initial list of three usability criteria you will focus on,

with justification Support your claims with specific data and examples! Prepare a poster about your work

Task Analysis review

Task analysis

Hierarchical Task Analysis (HTA) Knowledge-Based Task Analysis

Hierarchical Task Analysis (HTA) Two parts

Task breakdown – a listing of all tasks broken down into subtasks

Plans – specifications of the order of subtasks within the supertask

Often represented as a graphical diagram for clarity

Example: changing a light bulb

Hierarchical Task Analysis

Knowledge-based analysis: TAKD

Focused on categorizing objects by action, function, or other properties

List all objects associated with the task Build taxonomy using AND, OR, and

XOR branching

Knowledge Based Analysis

Kitchen item ANDshape XOR

dishedmixing bowl, saucepan, soup bowl, glass

flatplate, chopping board, frying pan

function ORpreparation

mixing bowl, saucepaneating XOR

for foodplate, soup bowl

for drinkglass

Knowledge-based analysis

TAKD – Task Analysis for Knowledge Description

Taxonomic - unique categorizations of items, characteristics, and functions

Good for understanding a problem language or an environment

Task analysis

When to use which?

Design process basics

Why is HCI Design Difficult?

Difficult to deeply analyze human behavior

May be too close to the domainMultiple clients with different needsCo-evolution of technology and users

Software life cycles – Waterfall Model

RequirementsSpecification

ArchitecturalDesign

Detailed Design

Coding andUnit Testing

Integrationand Testing

Operation and Maintenance

Limitations of the waterfall model

Limitations of the waterfall model You can’t determine all requirements

from the start Some tasks will only be known after the

user has interacted with the system Users will perform tasks that weren’t

intended by the designer Doesn’t support the user’s perspective

of the system

Software Life cycles – Iterative Waterfall Model

RequirementsSpecification

ArchitecturalDesign

Detailed Design

Coding andUnit Testing

Integrationand Testing

Operation and Maintenance

Iterative design

Applying HCI in the cycle

Formative Strategies to build a better interface prior

to creating the technology

Summative Assessing an existing interface after

creating the technology

Formative techniques

Apply principles “Don’t assume the user is right-handed”

Build prototypes Apply design rules / standards

Java look and feel Create usability specifications

The XYZ dialog takes < 5 sec. Study potential users to understand their

needs

Summative techniques

Empirical / laboratory evaluation Expert review Field study or deployment

Iterative design

Understanding users: Modeling

Modeling takes many forms

Interaction Low-level/physical actions Complex activities/tasks Cognitive Contextual

What is a model?

What is a model?

A constructed representation intended to help understand and reason about the world Abstracted and simplified Generalized Not necessarily reflective of how the world

actually works

Fitts’s Law – Modeling physical actions

Physical modeling: Using Fitts’s Law

Models movement time for selection tasks

Quantitative modeling technique A summative technique

Fitts’s Law Live!

Fitts’s Law demo

Tap back and forth between the two rectangles as quickly as you can! Don’t worry about where in the rectangle

you tap- just tap as many times as you can somewhere within the shape

Basics

Movement time (MT) is proportional to Index of Difficulty (ID) of a selection task

The big picture

The movement time for a well-rehearsed selection task:increases as the distance A to the target

increases; and decreases as the size of the target W

increases

Index of difficulty (ID)

Measure difficulty of selection taskID = log2(2A/W)

“bits”

A = distance between targetsW = target width

Movement time (MT)

MT = a + b ID(a=0 if line

passes through the origin)

MT

Difficulty

How MT is determined

Empirical measurement establishes constants a and b

Different for different devices and different ways the same device is used.

Original application of Fitts’s Law

1-dimensional selection task

Original application of Fitts’s Law

W

A

Extending to 2D

What is W when we consider 2 dimensions of movement?

θ

Extending to 2D

What is W when we consider 2 dimensions of movement? Same as usual?

θ

W’

Extending to 2D

What is W when we consider 2 dimensions of movement? Smallest dimension?

θ

W’

Extending to 2D

What is W when we consider 2 dimensions of movement? Distance from edge to

centroid?

θ

W’

Application of Fitts’s law?

When does it apply? When does it not?

Application of Fitt’s law?

When does it apply? When does it not? Used for predicting performance low-

level physical actions Automated tasks and actions Minimal cognition – you don’t have to

“think” about it

A possible example

How is Fitts’s Law used in UI design? Predicting

performance with an interface May substitute

for empirical testing, particularly in early stages

Comparing alternative UI layouts

KLM- another way of modeling physical action

Keystroke-Level Model (KLM) Another way of

doing physical modeling

Decompose tasks into low-level elements with time values

Calculate prediction for total execution time

Best for automated behavior

Keystroke-Level Model (KLM) K – striking keys B – pressing a mouse button P – pointing (dragging a pointer to a

target) H – homing – switching the hand

between the mouse and keyboard D – drawing lines using the mouse M – mentally preparing for a physical

action R – system response time

Keystroke-Level Model (KLM) Calculate time required for individual

generic actions Decompose tasks into individual actions Calculate the total time for a task as a

sum of the time for each action Can be used for comparing alternate

ways of executing a task Does not take time for cognition into

account

For next week

Read “The Model Human Processor” by Card, Moran, and Newell