chi research and submission process · le g, ee n al ch ch g e ey es n in . er in er d ts, n-n xt...

CHI Research and Submission Process

Leah Findlater, Jon Froehlich, Jen Golbeck July 9, 2012

Goals for Today

•  Get people excited about submitting to CHI! •  Learn about CHI and what makes it unique

•  Figure out concrete plans to advance current projects toward CHI submissions

•  Identify collaborators within HCIL to increase the strength of individual projects

CHI Attendees TODO

•  ~2500-3000 attendees •  Mainly US, Canada, and Western Europe, then

Japan and Korea, and more recently China •  Academics: undergrads, grads, professors in CS,

iSchools, Psychology, Art/Design, and emerging IxD programs; practitioners: interactive design, usability testing;

•  Human Factors Engineering—tend to go to HF conference

Past 12 Years of CHI

0

400

800

1200

1600

2000

1980 1985 1990 1995 2000 2005 2010 2015

Num of Papers Submitted

[ACM DL, h2p://dl.acm.org/citaAon.cfm?id=2207676&CFID=124494017&CFTOKEN=46356610]

Acceptance Rate

0%

20%

40%

60%

80%

100%

0

400

800

1200

1600

2000

1980 1985 1990 1995 2000 2005 2010 2015

# of Papers Submi2ed

Acceptance Rate

[ACM DL, h2p://dl.acm.org/citaAon.cfm?id=2207676&CFID=124494017&CFTOKEN=46356610]

Disclaimer

These notes are based on our experiences and there are other perspectives.

Don’t do research for a venue (e.g., CHI); do research for research.

Who has submitted to CHI?

#1 Way to Get Into CHI

High quality research +

High quality write-up

Ways to Participate at CHI

•  Papers & Notes (10 pages; 4 pages) •  Works-in-Progress •  Doctoral Consortium •  Workshops •  Student Competition •  Alt.Chi •  Case Studies •  Courses •  Panels

Submission and Reviewing Process��(Papers & Notes)

•  Initial submissions due in September (in 2012: Sept 17th)

•  Rebuttal period: ~1 week, early November

•  Program committee meeting: early December, ~250 people

•  Decisions: early to mid-December

•  Camera ready: due mid-January

More on Reviewing…

•  Reviewing: rigorous!! •  Submissions reviewed “as-is”

•  Rebuttals are important

•  (Rare) shepherding process

Subcommittees 1.  Usability, Accessibility and User Experience

2.  Specific Application Areas

3.  Interaction Beyond the Individual

4.  Design

5.  Interaction Using Specific Capabilities or Modalities

6.  Understanding People: Theory, Concepts, Methods

7.  Interaction Techniques and Devices

8.  Expanding Interaction Through Technology, Systems and Tools

Choosing a Subcommittee

•  Read the description •  Look at the Chairs and Associate Chairs

•  As for advice from people with experience

The HCIL Process

•  Submit to CHI – even if it’s not your main venue, it’s a great inspiring one

•  Find a collaborator – if you’re not sure how to fit your work at CHI, find someone who frequently publishes there. We love helping and collaborating!

•  HCIL CHI Workshop in September

Contributions (and Evaluations)

Empirical Artifact

Methodological

Theoretical

Dataset

Survey

Opinion

[from Wobbrock, Research Contribution Types in Human-Computer Interaction]

Contributions and Evaluations

Empirical Description •  New findings based on

systematically observed data •  Common methods include

experiments, case studies, interviews, and many more

Evaluation

•  Methods must be rigorous and precise


A

Thseptoufwcinkcainapetos

A

T

A

Hin

G

H

I

TthtaMtafthp

PpnbosCC

Typing

{ l

ABSTRACT

Touch screen have become screens pales examine typingphysical keyboo inform future

ultimate goal ofeedback. To swe asked 20 conditions: (1) nput errors,

keyboard, but contact points aas natural fingndividual keys

also show that press distributievidence that ouch surfaces

such a solution

Author Keywo

Touch-typing, m

ACM Classific

H.5.2. [Informnterfaces—inp

General Terms

Human factors,

NTRODUCTIO

Touch surfaceshe form of mablets and in

Microsoft Surfaablets and tabl

finger text inpuhese devices, h

pale in effecti

Permission to makpersonal or classronot made or distribbear this notice anor republish, to pospecific permissionCHI 2011, May 7–Copyright 2011 AC

on Fla

L1 The InformUniversity oleahkf, wobb

surfaces largeincreasingly pin comparison

g patterns that oards touch-type designs of tof supporting to

study the issueexpert typist

with no visualthen (2) withwith some f

and hand contoger positionings, and the patteexpert typists eions within aneyes-free toucand points to

. We conclude

ords

multi-touch inp

ation Keyword

mation interfacput devices and

s

, design, exper

ON

s have becomemobile phonesnteractive tablace). Unlike thletops are largeut. Despite thehowever, commiveness compa

ke digital or hard oom use is grantedbuted for profit or nd the full citation ost on servers or n and/or a fee. –12, 2011, VancouCM 978-1-4503-0

at Glass

Patte

Leah Findlamation Schoof Washingtobrock }@uw

e enough for popular, yet tn to physical

emerge whenpe on a flat suruch screen key

ouch-typing wis inherent to fts to enter tel keyboard andh and (3) wfeedback. We ours, looking ag, the spread ern of non-fingexhibit spatialln individual,

ch-typing may the role of pewith implicati

put, tabletop co

ds

ces and presd strategies.

imentation.

pervasive, mos, and increasetops (e.g., A

heir mobile phoe enough to ace importance mon methods ared to typing

copies of all or pd without fee provcommercial advanon the first page.to redistribute to

uver, BC, Canada. 0267-8/11/05....$10

s: Exam

erns on

ater,1 Jacobol on

w.edu

ten-finger inptyping on tou

keyboards. Wn expert users rface. Our aim yboards, with thith limited tactiflat-glass typinext under thrd no feedback o

without a visuanalyzed tou

at attributes suof hits amon

ger touches. Wly consistent kwhich providbe possible o

ersonalization ions for design

omputing.

sentation]: Us

ost commonly singly as largApple iPad anone counterparccommodate teof text entry ofor entering te

g on a physic

part of this work fvided that copies antage and that copi To copy otherwislists, requires pri

0.00.

mining T

Touch

O. Wobbro

put ch

We of is

he ile ng, ree on ual ch ch ng

We ey

des on in

n.

ser

in ger nd ts,

en-on ext cal

keyboamajor [3,25])and evPhysictabletothe fail

A majotactile depresAs a retimes. [25], wspuriouHowevfor cusbased. of tentraditioSurfacinclude[6], bu

In this expert surfacetouch supporemploy(e.g., [

Figure unrestrseparatforearm

forareiesse,ior

Ten-Fin

Surfac

ock,1 Daniel2 M

Rdwigd

ard [2,12]. Inddeficiency of

), potentially dven leading uscal keyboards op interaction [lings of curren

or challenge ofeedback, not sions but also esult, users muInadvertent to

which makes us input, espever, touch screstomization and

Despite this pn-finger keybonal QWERTYe, Canesta ke adaptive key

ut have not yet

paper, we exausers of phy

e (Figure 1). Oscreen keyb

rting touch-typyed a techniqu33]): we asked

1. Finger (bluericted typing cte left and rims and heels of

nger Ex

ces

l Wigdor1,2

Microsoft ResRedmond, Wdor@micros

deed, text entryf tabletops indetracting fromsers to alter thhave been pro

[10], which, alnt software-bas

of typing on flat just in the phin the ability t

ust visually atteouches occur touch screen ecially from teen keyboards d adaptation bepotential, comboards have keyboard (e.g

keyboard [24])yboards [5,11] offered high p

amine typing pysical keyboarOur aim is to boards, with ping with limitue evocative ofd 20 touch-typi

e) and non-fingcondition, showight space barf the hands resti

xpert Ty

search WA soft.com

y has been calln multiple stum sustained uheir linguistic oposed as a solthough useful,ed methods.

at surfaces is thysical perceptto feel the homend to the keyb

frequently onkeyboards vul

the user’s armalso offer ric

ecause they aremmercial imple

largely mimg., Apple iPad,). Research aand wearable

erformance tex

atterns that emrds touch-typeinform future the ultimate

ted tactile feedf a user-defineists to enter phr

ger (yellow) touwing space betwr areas, and eing on the scree

yping

ed out as a udies (e.g. use [22,32] style [32]. olution for , highlights

the lack of tion of key me row [2]. board at all n tabletops lnerable to

m or palm. h potential e software-mentations

micked the , Microsoft approaches keyboards

xt input.

merge when on a flat designs of

e goal of dback. We d interface rases using

uches in the ween hands, evidence of en. (N = 20)


Empirical Artifact

Description •  Inventions: new systems, architectures,

tools, techniques or designs

Evaluation

•  Often (not necessarily) accompanied by empirical evaluations

•  Systems: more holistic evaluation on what they enable

•  Techniques: more formal evaluation



Description •  Add or refine methods by which

researchers or practitioners carry out HCI work

Evaluation

•  Based on novelty and utility of the new or improved method

•  Validation usually required to show method is useful and reliable


Empirical Artifact Methodological

Computing Political Preference among Twitter Followers Jennifer Golbeck

Human-Computer Interaction Lab College of Information Studies

University of Maryland College Park, MD 20742

[email protected]

Derek L. Hansen CASCI, Human-Computer Interaction Lab

College of Information Studies University of Maryland

College Park, MD 20742 [email protected]

ABSTRACT There is great interest in understanding media bias and political information seeking preferences. As many media outlets create online personas, we seek to automatically estimate the political preferences of their audience, rather than of the outlet itself. In this paper, we present a novel method for computing preference among an organization’s Twitter followers. We present an application of this technique to estimate political preference of the audiences of U.S. media outlets. We also discuss how these results may be used and extended.

Author Keywords Twitter, politics, liberal, conservative, news, journalism

ACM Classification Keywords H5.m. Information interfaces and presentation (e.g., HCI): Miscellaneous.

General Terms Human Factors

INTRODUCTION As major media outlets establish online presences in social media, understanding the characteristics of those audiences is an important task. It has implications for how information is presented in this environment where personalization is expected. Furthermore, it can provide valuable information to marketers and social media analysts.

As a first step toward understanding audiences, we present a technique for estimating audience preferences in a given domain on the microblogging service Twitter. We use U.S. politics as our motivating example by estimating the political preferences of media outlets’ audiences.

BACKGROUND While we are not studying media bias, but rather the political preferences of audiences, it is worth briefly discussing the extensive research on analyzing media bias.

A subset of this work uses automated methods to infer liberal/conservative bias of news stories and outlets. These automated methods do not depend on subjective measurements of bias, although the specific techniques used to infer bias can be problematic and are highly contested. One approach is to compute a media bias score based on citations in the news story – news outlets that cite “think tanks” that are also cited by Congressperson’s with a known liberal bias are assumed to be more liberal [8]. Another approach is to compare keywords and phrases used by Congresspeople of known political persuasions with those used in news articles – news outlets that use terms like “death tax” and “illegal immigration” are more likely to be conservative [7]. A final approach assigns a liberal/conservative score to web documents based on the number of times they are co-cited with other web documents that have a known political bias [3].

In contrast to these approaches, we estimate the political preferences of news outlet audiences, not the news outlet content itself. Our strategy is similar to [8] in that we use Congresspeople’s American for Democratic Action’s (ADA) scores as a starting point for our scoring; however, we use Twitter Follow relationships rather than article citations. Using Follow relationships avoids the concern with [8] that results rely too much on the citation practices of journalists and Congresspeople. Our approach does not require coding of data (as in [8]) or access to large corpuses of news stories and congressional speeches; it relies instead on freely available and open access data from Twitter.

METHOD AND SAMPLING Unlike [8], we are not interested in predicting media bias. Instead, we are interested in predicting the political preference of the audience of different media outlets and organizations by using sites like Twitter that embed social ties. Our examples and applications are in the political domain, but the technique is generalizable when the right background information is available. Our approach includes the following steps:

Step 1: Apply known scores to a seed group, in this case Congresspeople using Twitter. The base data of liberal/conservative scores are obtained from Americans for Democratic Action (ADA), who puts out an annual report that considers the voting record of members of Congress [1]. ADA defines a key set of votes that indicate liberal and

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CHI 2011, May 7–12, 2011, Vancouver, BC, Canada. Copyright 2011 ACM 978-1-4503-0267-8/11/05....$10.00.

CHI 2011 • Session: Microblogging Behavior May 7–12, 2011 • Vancouver, BC, Canada

1105

Curator: A Game with a Purpose for CollectionRecommendation

Greg Walsh, Jennifer Golbeck

Human-Computer Interaction LabUniversity of Maryland, College Park, MD

{gwalsh,jgolbeck}@umd.edu

ABSTRACTCollection recommender systems suggest groups ofitems that work well as a whole. The interaction ef-fects between items is an important consideration, butthe vast space of possible collections makes it di�cultto analyze. In this paper, we present a class of gameswith a purpose for building collections where users cre-ate collections and, using an output agreement model,they are awarded points based on the collections thatmatch. The data from these games will help researchersdevelop guidelines for collection recommender systemsamong other applications. We conducted a pilot studyof the game prototype which indicated that it was funand challenging for users, and that the data obtainedhad the characteristics necessary to gain insights intothe interaction e↵ects among items. We present thegame and these results followed by a discussion of thenext steps necessary to bring games to bear on the prob-lem of creating harmonious groups.

Author Keywordshuman computation, games with a purpose, seriousgames, recommender systems

ACM Classification KeywordsH5.m Information interfaces and presentation: Miscel-laneous

General TermsDesign, Human Factors

INTRODUCTIONCollection recommender systems [2] are similar to exist-ing recommender systems but instead of recommendingindividual items to the user, they recommend groups ofitems that work well together as a whole unit. Thereare many factors to consider when creating a collection.The size of the collection, diversity, potential order, and,quality of items all have an impact. One of the most

Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise, orrepublish, to post on servers or to redistribute to lists, requires prior specificpermission and/or a fee.CHI 2010, April 10 – 15, 2010, Atlanta, Georgia, USACopyright 2010 ACM 978-1-60558-929-9/10/04...$10.00.

challenging features to consider is the interaction e↵ectsbetween items.

Regardless of the domain, some items work well to-gether and others do not. These co-occurrence e↵ectsare one of the most important factors in the success orfailure of many collections.

It can be a complex task to evaluate co-occurrence ef-fects. Even two items that both have high individ-ual item ratings may not work well together. Some-one might have a deep love for chocolate and also forpickles, but not for the two together. This is a ratherintuitive e↵ect when considering pairs, but gets morecomplicated when considering the quality of larger setsof items such as a triple.

For example, chocolate bars and graham crackers area fine combination; marshmallows and chocolate barsare also; and marshmallows and graham crackers are aswell. None of these pairs are poor but neither are theyexceptional. However, the combination of all three intoa “s’more” makes a much beloved snack for many peo-ple. The combination of all three items is better thanwould be indicated by looking at the three pairs. On theother hand, three items that are very good pairwise canmake a bad triple. Consider building a research teamof two professors and one graduate student. The pro-fessors may work well together, and each may work wellwith the student. However, all three may have troubleworking together. The presence of a student may bringout some tension between the faculty members aboutwho is in control, and the student may have troublebalancing work or contradictory instructions from thefaculty.

Similar scenarios can be made moving up from groups ofthree to four, and so on. While it is useful to look at thecompatibility of groups of two or even three items, thisapproach quickly becomes computationally di�cult, re-quiring O(nk) comparisons for groups of size k.

Even with extensive data on users’ preferences for itemsand groups of items, this space is vast enough that gen-eral rules will almost certainly be necessary for collec-tion recommender systems to be successful. To derivethese rules, too, will require a large set of data. Oneway to obtain that data is through collection-oriented

CHI 2010: Sharing in Specific Communities April 10–15, 2010, Atlanta, GA, USA

2079



Theoretical

Description •  New models, principles, concepts, or

frameworks, or variations on those that already exist

•  Quantitative or qualitative

Evaluation

•  Validated for novelty, importance, descriptive and/or predictive power

•  Almost always accompanied by empirical observation


An Error Model for Pointing Based on Fitts’ Law Jacob O. Wobbrock,1 Edward Cutrell,2 Susumu Harada3 and I. Scott MacKenzie4

1The Information School 3Computer Science & Engineering

DUB Group University of Washington Seattle, WA 98195 USA

[email protected] [email protected]

2Microsoft Research One Microsoft Way

Redmond, WA 98050 USA [email protected]

4Computer Science & Engineering York University

Toronto, ON, Canada M3J 1P3 [email protected]

ABSTRACT For decades, Fitts’ law (1954) has been used to model pointing time in user interfaces. As with any rapid motor act, faster pointing movements result in increased errors. But although prior work has examined accuracy as the “spread of hits,” no work has formulated a predictive model for error rates (0-100%) based on Fitts’ law parameters. We show that Fitts’ law mathematically implies a predictive error rate model, which we derive. We then describe an experiment in which target size, target distance, and movement time are manipulated. Our results show a strong model fit: a regression analysis of observed vs. predicted error rates yields a correlation of R2 = .959 for N = 90 points. Furthermore, we show that the effect on error rate of target size (W) is greater than that of target distance (A), indicating a departure from Fitts’ law, which maintains that W and A contribute proportionally to index of difficulty (ID). Our error model can be used with Fitts’ law to estimate and predict error rates along with speeds, providing a framework for unifying this dichotomy.

ACM Categories & Subject Descriptors: H.5.2 [Information interfaces and presentation]: User interfaces – theory and methods; H.1.2 [Models and principles]: User/machine systems – human factors.

General Terms: Experimentation, Human Factors, Theory.

Author Keywords: Movement time, pointing time, pointing errors, mousing errors, clicking errors, error rates, speed-accuracy tradeoff, Fitts’ law, Schmidt’s law, error model.

INTRODUCTION Even before Newell and Card advocated for a “hardening of the science” of human-computer interaction (HCI) [23], researchers sought quantitative models of human action to explain behavior and inform design. Although there are relatively few such models in HCI, those we do have are highly influential.

Perhaps the most influential of these is Fitts’ law [8,17]. Since its first application in HCI in 1978 to predict pointing times in a text editor [3,19], Fitts’ law has facilitated design innovations [2,10,36], informed aggregate models of computer use [4,13], and been a tool for modeling and evaluation [1,16,18,24,28]. This is no surprise given the law’s robustness, ease of use, and the prevalence of pointing in graphical user interfaces.

However, although Fitts’ law supports the prediction of speeds, it does not readily support the prediction of errors. In fact, to date, there is no equivalent “error law” that predicts the probability of a user hitting or missing a target using Fitts’ law parameters. Although speed-accuracy tradeoffs have been studied (see [12,22,25] for reviews), this work almost universally regards accuracy as the “spread of hits,” which is of limited use in predicting error rates in user interfaces. Post hoc corrections can be used to normalize differences in speed-accuracy performance among a pool of human subjects [5,17,29,31], but these adjustments lack the predictive power of an error model.

Why predict errors? Error prediction should be as useful as time prediction given the diametric relationship of these two entities: where one increases, the other decreases. Thus, “rounding out” the theory requires a predictive model for errors. Also, if a Fitts-based error model is shown to hold, it contributes to the soundness of the law itself. If it is shown not to hold, it motivates a deeper investigation into the assumptions underlying Fitts’ law, since, as we show, a Fitts-based error model is mathematically implied.

An error model also has practical applications. For example, it allows us to estimate text entry error rates given different tapping speeds on a stylus keyboard, or to ensure that buttons are big enough in a safety-critical system where speed is crucial. In computer games, as another example, designers may want to predict how many targets a player can hit in a given amount of time.

As we demonstrate, Fitts’ law mathematically implies an equation for pointing errors. To our knowledge, this equation has not been derived in the literature. Instead, prior work focuses on motor-control theories accounting for endpoint variability in human movement [6,22,25,26].

Permission to make digital or hard copies of all or part of this work forpersonal or classroom use is granted without fee provided that copies arenot made or distributed for profit or commercial advantage and that copiesbear this notice and the full citation on the first page. To copy otherwise,or republish, to post on servers or to redistribute to lists, requires priorspecific permission and/or a fee. CHI 2008, April 5–10, 2008, Florence, Italy. Copyright 2008 ACM 978-1-60558-011-1/08/04…$5.00.



Theoretical Dataset

Description •  New and useful corpus for the

benefit of the research community

Evaluation

•  Utility for future evaluations, with explanation on how to use it


Phrase Sets for Evaluating Text Entry Techniques I. Scott MacKenzie1,2 and R. William Soukoreff 1

1 Dept. of Computer Science York University

Toronto, Ontario, Canada M3J 1P3 +1 416-736-2100

{smackenzie,will}@acm.org

2 Unit for Computer-Human Interaction (TAUCHI) Dept. of Computer & Information Sciences

FIN-33014 University of Tampere Tampere, Finland +358 3 215 8566

ABSTRACT In evaluations of text entry methods, participants enter phrases of text using a technique of interest while performance data are collected. This paper describes and publishes (via the internet) a collection of 500 phrases for such evaluations. Utility programs are also provided to compute statistical properties of the phrase set, or any other phrase set. The merits of using a pre-defined phrase set are described as are methodological considerations, such as attaining results that are generalizable and the possible addition of punctuation and other characters.

TEXT ENTRY EVALUATIONS Among the desirable properties of experimental research are internal validity and external validity. Internal validity is attained if the effects observed are attributable to controlled variables. External validity means the results are generalizable to other subjects and situations. Simple as this seems, these attributes are typically at odds with one another. That is, too strictly attending to one tends to compromise the other. This paper pertains to one such point of tension between internal and external validity: the text entered by the participants in evaluations of text entry techniques. Text entry research typically pits one entry method against another. Thus, entry method is the controlled variable, and it is manipulated over two or more levels, for example, Multitap vs. Letterwise in an experiment comparing text entry techniques for mobile phones [2], or Qwerty vs. Opti in an experiment comparing soft keyboard layouts [3]. Allowing participants to freely enter “whatever comes to mind” seems desirable, since this mimics typical usage. Such a procedure improves external validity since the results are generalizable. Although of unquestionable merit in gauging the overall usability of a system or implementation, such methodology also has problems. For one, accuracy is difficult to gauge since there is no source

text with which to compare the entered text. Also, the lack of control means performance measurements are coincident with spurious behaviours, such as pondering or secondary tasks. Thus, sources of variation are present in the dependent variables (e.g., speed or accuracy) that are not attributable to the controlled variable. This compromises internal validly because variations in measurements are, in part, due to other effects. On balance, the preferred procedure – that used in the majority of research studies – is to present participants with pre-selected phrases of text. Phrases are retrieved randomly from a set and are presented to participants one by one to enter.

Creating a Phrase Set In creating a phrase set, the goal is to use phrases that are moderate in length, easy to remember, and representative of the target language. In a recent paper comparing two soft keyboards, MacKenzie and Zhang [3] used a set of 70 phrases. We recently expanded this set to 500 phrases. A few examples from the set follow:

video camera with a zoom lens have a good weekend what a monkey sees a monkey will do that is very unfortunate the back yard of our house I can see the rings on Saturn this is a very good idea

We have used the new phrase set with good results in recent studies [1, 5], and wish to share them with the community of text entry researchers via this paper. The phrases contain no punctuation symbols, and just a few instances of uppercase characters. (Participants may be instructed to ignore case and to enter all characters in lowercase.) The complete set is available from the authors or directly in http://www.yorku.ca/mack/PhraseSets.zip. Some minor modifications may be necessary to convert spellings to a local dialect (e.g., colour vs. color). A phrase set should be representative of the target language. The analysis of phrase sets is automated

Copyright is held by the author/owner(s). CHI 2003, April 5–10, 2003, Ft. Lauderdale, Florida, USA. ACM 1-58113-630-7/03/0004.

Short Talks: Specialized Section CHI 2003: NEW HORIZONS

754

Short Talk: Fitt's Law & Text Input CHI 2003: NEW HORIZONS

754

Copyright is held by the author/owner(s). CHI 2003, April 5-10, 2003, Ft. Lauderdale, Florida, USA. ACM 1-58113-637-4/03/0004



Theoretical Dataset Survey

Description •  Review and synthesize work done in

a research field, to expose trends, themes, and gaps in the literature

Evaluation

•  Synthesis of previous work and extraction of emergent themes / trends / gaps




Theoretical Dataset Survey

Opinion

Description •  Goal is to persuade rather than

simply to inform

Evaluation

•  Evaluated on credibility and use of supporting evidence, as well as interest to the community


Often: Intersections of Contributions

The Design of Eco-Feedback Technology Jon Froehlich1, Leah Findlater2, James Landay1

1

Computer Science and Engineering, 2

The Information School

DUB Institute, University of Washington

{ jfroehli, leahkf, landay}@uw.edu

ABSTRACT Eco-feedback technology provides feedback on individual

or group behaviors with a goal of reducing environmental

impact. The history of eco-feedback extends back more

than 40 years to the origins of environmental psychology.

Despite its stated purpose, few HCI eco-feedback studies

have attempted to measure behavior change. This leads to

two overarching questions: (1) what can HCI learn from

environmental psychology and (2) what role should HCI

have in designing and evaluating eco-feedback technology?

To help answer these questions, this paper conducts a

comparative survey of eco-feedback technology, including

89 papers from environmental psychology and 44 papers

from the HCI and UbiComp literature. We also provide an

overview of predominant models of proenvironmental

behaviors and a summary of key motivation techniques to

promote this behavior.

Author Keywords Eco-feedback, Environmental HCI, Reflective HCI, Survey

ACM Classification Keywords H5.m. Information interfaces and presentation (e.g., HCI)

General Terms Design, Human Factors

INTRODUCTION As environmental issues such as climate change, air

pollution, and water scarcity become more salient in the

global consciousness, so too have they become more active

targets of research within HCI and Ubiquitous Computing

[6, 19, 57]. One particularly popular form of environmental

HCI research is the design and study of eco-feedback technology, which we define as technology that provides

feedback on individual or group behaviors with a goal of

reducing environmental impact (adapted from [39] and

[28], see Figure 1 for examples). Despite this goal, few HCI

eco-feedback studies have even attempted to measure

behavior change. Although eco-feedback may be seen as an

extension of research in persuasive technology [17], it

actually extends back much further to over 40 years of

research in environmental psychology. This leads to two

interrelated questions: (1) What can HCI learn from

environmental psychology and (2) what should be the role

of the HCI community in contributing to eco-feedback

research? To explore these questions in detail, we present a

review of the related environmental psychology literature as

well as a comparative survey of eco-feedback studies in

both HCI and environmental psychology.

Eco-feedback technology is based on the working

hypothesis that most people lack awareness and

understanding about how their everyday behaviors such as

driving to work or showering affect the environment;

technology may bridge this “environmental literacy gap” by

automatically sensing these activities and feeding related

information back through computerized means (e.g., mobile

phones, ambient displays, or online visualizations). HCI

and UbiComp researchers have built eco-feedback

technologies for a variety of domains including energy

consumption [28], water usage [3], transportation [19], and

waste disposal practices [29].

Contributing to this growing interest in eco-feedback

technology is the parallel advancement and availability of

sensing systems for environmentally related activities (e.g.,

human activity inference [35]) and interactive displays to

feedback this data (e.g., iPods and mobile phones). Such

advances provide a rich space of opportunities for new

types of eco-feedback that could not be considered in the

past. Moreover, the next generation of resource

measurement systems (often referred to as “smart meters”)

will soon provide real-time (or near real-time) data on

electricity, gas, and water usage in homes and businesses.

This will produce tremendous amounts of data that can be

Figure 1. Examples of eco-feedback technology. (left-to-right) The Infotropism display uses sensors and living plants to provide feedback about recycling and waste disposal [29]. WaterBot provides ambient feedback information about water usage [3]. The UbiGreen Transportation Display semi-automatically senses and feeds back information about transportation to encourage green transit [19].

Permission to make digital or hard copies of all or part of this work for

personal or classroom use is granted without fee provided that copies are

not made or distributed for profit or commercial advantage and that copies

bear this notice and the full citation on the first page. To copy otherwise,

or republish, to post on servers or to redistribute to lists, requires prior

specific permission and/or a fee.

CHI 2010, April 10–15, 2010, Atlanta, Georgia, USA.

Copyright 2010 ACM 978-1-60558-929-9/10/04....$10.00.

CHI 2010: Home Eco Behavior April 10–15, 2010, Atlanta, GA, USA

1999

Appropriate evaluation depends on type of contribution

CHI Contribution Types

•  Development or refinement of interface artifacts or techniques •  Understanding users

•  Systems, tools, architectures and infrastructure •  Methodology

•  Theory

•  Innovation, creativity and vision •  Argument / opinion

Specifically

h2p://chi2012.acm.org/cfp-‐contribuAon-‐types.shtml

Writing the Paper

From CHI 2012 paper writing guide: •  Offer benefit to the reader

•  Ensure results are valid

•  Gain credit for originality

•  Describe the work clearly and concisely

Writing the Paper

•  Format of the paper depends on the type of paper and contribution

•  Find a few examples of good papers with similar contribution types and follow their pattern

•  Iteration and feedback is critical

0

10

20

30

40

50

Camera Ready Versions Submission Versions

Num Versions of Jon’s Papers��A selection

Interaction Technique with Experimental Evaluation

Ephemeral Adaptation: The Use of Gradual Onset to Improve Menu Selection Performance

Leah Findlater*, Karyn Moffatt*, Joanna McGrenere, Jessica Dawson Department of Computer Science

University of British Columbia, Vancouver, Canada {lkf, kmoffatt, joanna}@cs.ubc.ca

ABSTRACT We introduce ephemeral adaptation, a new adaptive GUI technique that improves performance by reducing visual search time while maintaining spatial consistency. Ephemeral adaptive interfaces employ gradual onset to draw the user’s attention to predicted items: adaptively predicted items appear abruptly when the menu is opened, but non-predicted items fade in gradually. To demonstrate the benefit of ephemeral adaptation we conducted two experiments with a total of 48 users to show: (1) that ephemeral adaptive menus are faster than static menus when accuracy is high, and are not significantly slower when it is low and (2) that ephemeral adaptive menus are also faster than adaptive highlighting. While we focused on user-adaptive GUIs, ephemeral adaptation should be applicable to a broad range of visually complex tasks.

Author Keywords Adaptive interfaces, personalization, abrupt visual onset, menu design, user study, interaction techniques.

ACM Classification Keywords H.5.2 [User Interfaces]: Evaluation/methodology, interaction styles.

INTRODUCTION Adaptive graphical user interfaces (GUIs) automatically tailor features to better suit the individual user’s needs. To date, these interfaces have tended to rely on one of two forms of adaptation: spatial or graphical. Spatial techniques reorganize items to reduce navigation time and, to a lesser degree, to aid visual search [2,12,14]. An adaptive split menu, for example, moves or copies the most frequently and/or recently used items to the top of the menu for easier access [14]. Graphical techniques, on the other hand, reduce visual search time, for example, through changing the background colour of predicted items [7,8,18]. Some techniques use a combination of both spatial and graphical elements [18,19].

As an alternative to spatial and graphical adaptation, we propose the use of a temporal dimension and introduce

ephemeral adaptation as a new adaptive interaction technique that uses this dimension to reduce visual search time. Ephemeral adaptive interfaces use a combination of abrupt and gradual onset to provide initial adaptive support, which then gradually fades away. The goal is to draw the user’s attention to a subset of adaptively predicted items, in turn reducing visual search time. Figure 1 applies ephemeral adaptation to a menu: adaptively predicted items appear abruptly when the menu is opened, after which the remaining items gradually fade in.

Ephemeral adaptation maintains spatial consistency, thus addressing one of the main drawbacks of spatial adaptation techniques [2]. An adaptive menu that reorganizes features, for example, by promoting the most frequently used ones, offers theoretical performance benefits over a traditional static menu. In practice, however, spatially adaptive interfaces are not often faster than their static counterparts because the user needs to constantly adapt to the altered layout, wiping out any potential gains [2,4,5,12]. Successes have tended to occur only when the adaptive approach greatly reduces the number of steps to reach desired functionality, for example, through a hierarchical menu structure [8,10,19], or when limited screen real estate necessitates scrolling [5]. 1

Similarly to ephemeral adaptation, graphical techniques also maintain spatial consistency and focus on reducing visual search. Several researchers have proposed techniques to highlight predicted items with a different background

* The first two authors are equal contributors to this work.

Figure 1. Ephemeral adaptation applied to menus: predicted items appear immediately, while remaining items gradually fade in.

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CHI 2009, April 4–9, 2009, Boston, Massachusetts, USA. Copyright 2009 ACM 978-1-60558-246-7/09/04...$5.00.

CHI 2009 ~ Desktop Techniques April 8th, 2009 ~ Boston, MA, USA

1655

Intro Intro

Results


Related Work

The Technique

Results

Discussion Applications

Future Work

Conclu- sion

Study Method

Study Method

System Contribution with Qualitative Evaluation

UbiGreen: Investigating a Mobile Tool for Tracking and Supporting Green Transportation Habits

Jon Froehlich1, Tawanna Dillahunt

2, Predrag Klasnja

3,4, Jennifer Mankoff

2, Sunny Consolvo

4,

Beverly Harrison4, James A. Landay

1,4

1CSE; 3The Information School

DUB Institute, U. of Washington

Seattle, WA 98195 USA

{jfroehli, landay}@cs.washington.edu;

[email protected]

2HCI Institute Carnegie Mellon University

Pittsburgh, PA 15213 USA

{tdillahu, jmankoff}@cs.cmu.edu

4Intel Research Seattle Seattle, WA 98105 USA

{sunny.consolvo, beverly.harrison}

@intel.com

.

ABSTRACT

The greatest contributor of CO2 emissions in the average

American household is personal transportation. Because

transportation is inherently a mobile activity, mobile

devices are well suited to sense and provide feedback about

these activities. In this paper, we explore the use of personal

ambient displays on mobile phones to give users feedback

about sensed and self-reported transportation behaviors. We

first present results from a set of formative studies

exploring our respondents’ existing transportation routines,

willingness to engage in and maintain green transportation

behavior, and reactions to early mobile phone “green”

application design concepts. We then describe the results of a 3-week field study (N=13) of the UbiGreen

Transportation Display prototype, a mobile phone

application that semi-automatically senses and reveals

information about transportation behavior. Our

contributions include a working system for semi-

automatically tracking transit activity, a visual design

capable of engaging users in the goal of increasing green

transportation, and the results of our studies, which have

implications for the design of future green applications.

Author Keywords

Sustainability, transportation, ubicomp, ambient displays

ACM Classification Keywords

H5.m. Information interfaces and presentation (e.g., HCI):

Miscellaneous.

INTRODUCTION

In 2005, Americans consumed 100 quadrillion British

thermal units (BTUs) of energy [32], almost six times the

worldwide average per person [20]. This in turn caused the

release of 2.2 billion metric tons of carbon dioxide (CO2), a

greenhouse gas assumed to be a major cause of adverse

climate change. To reverse this trend, action will be

required on many levels, including policy, infrastructure,

and individual change. Given the growing prevalence of

mobile phones with sensing capabilities, one compelling

opportunity to potentially impact human behavior is to offer

immediate feedback about how currently sensed behaviors

affect the environment. In this paper, we explore the use of personal ambient displays on mobile phones to give users

feedback about their sensed and self-reported transportation

behaviors (Figure 1).

Researchers have identified three areas responsible for a

majority of energy consumption in American households:

home heating and cooling; shopping and eating (and the associated transportation of goods); and commuting, flying

and other daily transportation activities [3,35]. In this paper,

we focus on the latter (personal transportation), the greatest

individual contributor of CO2 emissions (26%) in the

average American household [35].

There is extensive literature in the areas of environmental

sociology, public policy, and more recently, conservation

psychology that discuss the promotion of environmentally

responsible behavior [1,2,26,33]. Past work has shown that

motivators such as public commitment, frequent feedback,

and personalization can positively impact environmentally

responsible behavior [1]. Since the 1990s, information

campaigns and other programs have attempted to engage

individuals in voluntary greening of transportation behavior

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or

republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CHI 2009, April 4–9, 2009, Boston, Massachusetts, USA. Copyright 2009 ACM 978-1-60558-246-7/09/04...$5.00.

Figure 1 (left) The UbiGreen Transportation Display shows transit behavior as “wallpaper” on a phone’s screen. Here the tree is nearly full of leaves, indicating that the user has completed several green trips for the

week. (top) The MSP sensor worn near the waist and the phone’s GSM cell tower data are used to semi-automatically infer transportation mode.


Intro

Intro

Two Formative Studies + Results



System Description System Description

System Description

Study Description

Study Description

Study Description

Study Results & Implications

Study Results & Implications Study Results &

Implications

Conclusion

Formative Interview Study

In the Shadow of Misperception: Assistive Technology Use and Social Interactions

Kristen Shinohara and Jacob O. Wobbrock The Information School

DUB Group University of Washington

Seattle, WA 98195 {kshino, wobbrock}@uw.edu

ABSTRACT Few research studies focus on how the use of assistive technologies is affected by social interaction among people. We present an interview study of 20 individuals to determine how assistive technology use is affected by social and professional contexts and interactions. We found that specific assistive devices sometimes marked their users as having disabilities; that functional access took priority over feeling self-conscious when using assistive technologies; and that two misperceptions pervaded assistive technology use: (1) that assistive devices could functionally eliminate a disability, and (2) that people with disabilities would be helpless without their devices. Our findings provide further evidence that accessibility should be built into mainstream technologies. When this is not feasible, assistive devices should incorporate cutting edge technologies and strive to be designed for social acceptability, a new design approach we propose here.

Author Keywords: Accessibility, product design, interface design, stigma, social interactions, assistive devices.

ACM Classification Keywords: K.4.2 [Computers and society]: Social issues—assistive technologies for persons with disabilities. General Terms: Design, Human Factors. INTRODUCTION People with disabilities use assistive technologies for various tasks in their everyday lives. Assistive technologies are defined by the Technical Assistance to the States Act as “any item, piece of equipment, or product system, whether acquired commercially off the shelf, modified, or customized, that is used to increase, maintain or improve functional capabilities of individuals with disabilities” [6]. This definition describes any technology appropriated for the specific purpose of aiding people with disabilities, created with the express purpose of enabling access to environment, technology, information, and services. But assistive technologies are abandoned at high rates [20,22], and surprisingly, this is often due to personal meaning

associated with such devices [18]. If assistive technologies are built to be functional and usable, but people are abandoning them, how effective are they in helping people with disabilities accomplish daily tasks? Research involving assistive technologies generally focuses on functionality and usability [10], yet technology use does not happen in a social vacuum. Rather, personal preferences in social contexts may dictate whether and how a device is used [24]. This implies an effect on technology use arising from social contexts. We present this study investigating the effects of assistive technology use in social and professional contexts from the perspective of people with disabilities, a perspective often lacking in assistive technology research.

We conducted an interview study to find how 20 people with disabilities feel about using assistive technologies in social and professional contexts. We found that while assistive technology empowers and enables people to work, socialize, and orchestrate their lives, it still lives in the shadow of social misperceptions. These misperceptions may perpetuate social barriers to accessibility. Assistive technologies are used in social situations and not in isolated laboratories; therefore, design of such technologies must be assessed for impacts on social and professional interactions.

BACKGROUND In previous work [24], it was found that unwanted attention brought on by an assistive device made the individual feel self-conscious in some social contexts. Indeed, assistive technologies often bridge functionally and socially-situated experiences. To understand how social contexts may affect assistive technology use, we turn to literature on the disabilities rights movement, the meaning of objects, and the psychology of stigma.

Disability Rights: An Enabling Environment As researchers who do not have disabilities, we believe it is important to endeavor to understand social and cultural issues of disability rights, and how these affect assistive technology adoption and use. In Nothing About Us Without Us, Charlton describes the social issues which drove the disabilities rights movement [5]. Fueled by the Civil Rights Movements of the 1960s in the United States, people with disabilities rallied for equal rights and equal access. The movement rejected the idea of disability as a medical condition, and instead adopted a socially constructed view, emphasizing that disability lies not in the person, but in the

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. CHI 2011, May 7–12, 2011, Vancouver, BC, Canada. Copyright 2011 ACM 978-1-4503-0267-8/11/05....$10.00.

CHI 2011 • Session: HCI for all May 7–12, 2011 • Vancouver, BC, Canada

705

Intro Intro Related

Work

Study Method

Future Work

Conclu- sion


Back- ground

Results

Discussion

Don’t do this….

•  Being overly dismissive of related work •  Not justifying your evaluation decisions •  Not providing enough detail to reasonably replicate

system / approach / study •  Unclear contributions—if you’re not clear about it,

don’t expect the reviewer to be clear •  Invalid data analysis •  Overstating results •  Including useless figures and/or figures with small

writing!

Small Group Breakouts Full paper or note. Why?

Articulate / refine your ideas (1-3 sentences each): •  Motivation: What's the research context and why should we care? Why is it

important?

•  Problem: What's the specific problem you are trying to address within this context?

•  Approach/Solution: What approach will you take / have you taken to address the problem? [For an evaluation, why is this an appropriate evaluation? Justify]

•  Outcomes: What were the main results/findings or what do you predict the main results/findings will be?

•  Contribution: What is/are the main contribution(s)? [within our framework]

Specific CHI submission details: •  CHI contribution type

•  Subcommittee choice

Google: CHI 2012 paper submission

chi research and submission process · le g, ee n al ch ch g e ey es n in . er in er d ts, n-n xt...

Documents