i . 1!... “6
. Jmﬁaw kWh“

“2.1.3. , ..
Bzwwwmwhﬁ .'
wifmﬁ

5

tight“:
3. gaggiéz

t\ IanMMuxh

:5
H. “13):: a.

1....

.3...

. all... an.
a
.

it... u .0 .
Igmmwih

5!-
SA 2 i. 3.. ..
Hahn's-ll...)

OI

 

[I

It. . .
233
5:...
3..

 

 

7_
2009

 

LIBRARY
Michigan State
University

 

 

This is to certify that the
dissertation entitled

UNDERSTANDING MEDIA HABITS:
THE ROLE OF HABIT IN THE THEORY OF PLANNED
BEHAVIOR

presented by

RYAN LANGE

has been accepted towards fulﬁllment
of the requirements for the

PhD. degree in Mass Media

 

 

MaTor Professo‘rTS Signature
A 3c» a: par;

71 I J 1
Date

MSU is an Afﬁrmative Action/Equal Opportunity Employer

 

PLACE IN RETURN Box to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K'lProlecc&PresICIRCIDateDueindd

 

THE RI”)

 

UNDERSTANDING MEDIA HABITS:
THE ROLE OF HABIT IN THE THEORY OF PLANNED BEHAVIOR

Ryan Lange

A DISSERTATION
Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Mass Media

2009

 

 

H4133} i) ..

I
\-—-_)."‘ I‘ " '
5I_..\\?.L¢Julélll

I

‘I‘ L"" \V" - O
ﬂux).\$.-:A«}‘\.‘IA 15 u

r;-
'ﬁt

4“”; t \ ‘

.h11; «.

, u“)

' “Wur'r
‘3

r...

t"'.»‘5.i\ed l0 at? .

. .
J‘s‘
«L “e, - I

12‘3" Mr. A .

 

ABSTRACT

UNDERSTANDING MEDIA HABITS:
THE ROLE OF HABIT IN THE THEORY OF PLANNED BEHAVIOR

By
Ryan Lange

Habit is a variable of great importance to communication research. Due to poor
conceptualization and measurement, the true role of media habits in media consumption
behavior has often been obﬁiscated from media researchers. One of the goals of this
dissertation is to show an improved conceptualization of habit, as well as to propose more
conceptually consistent methods of measuring media habits.

This dissertation examines the media habit of YouTube use by using a more
strongly conceptualized version of habit. By conceptualizing habit in terms of its
automaticity rather than other aspects that may be confounded with other variables (such
as ﬁequency), the concept of habit becomes clearly discriminated from other variables
used to stand in for habit in the past, such as past behavior. Focusing on the important
aspects of habit, automaticity and uncontrollability, this dissertation demonstrates a
perspective on the habit construct based on an advanced understanding of habits that has
recently emerged in the ﬁeld of social psychology.

The Theory of Flamed Behavior was the theoretical framework of this
dissertation. In terms of expanding and testing the Theory of Flamed Behavior, the
interest of this dissertation is to examine the relationship between habit and control
beliefs, focusing on perceived behavioral control. Based on previous ﬁndings, habit was
proposed to act as a moderator to the relationship between perceived behavioral control

and behavior. A negative relationship for perceived behavioral control and habit was

I _
yep-53.1; .\.~ i.-.

"‘ I
$5.3.ou‘ '3 “(gt 1
tun-“u 'r'ishm

IV ‘ T.‘
..'*I' I?‘ ‘ "
h.-.. ...~.‘:. A»

I A h
k A; '
’,".‘I\ LL. 3 ..

. . _ . .av-u
‘t‘ - “MALDL5.
1:31
#4.“th \ "1"

"s ‘A- . Hit.

A:’\,\

 

‘0') V c
«’3. 5“ '
Td)\\. 3‘. 1" l
I ~‘ \

l
riﬁtif"\’1r
Mn C\‘\
I
‘I
In \I\_I\:"
:
'tv- :
ﬁelds;
- \_ 31‘ c\pl
b_-1‘.
‘tk‘ T‘ ‘
-.A)L§ 1n 1.“;
u,.h'\‘
45.3.. _
‘ [Ir-“I"
‘dtii
‘1.” ‘
“I. \ .14 .

proposed: As habit grows in strength, perceived behavioral control should become a less
effective predictor of behavior. A similar effect was proposed for the concept of self-
efﬁcacy, another concept driven by control beliefs, and its relationship with intention.
Habit itself was proposed to become stronger as the consistent context for behaviors
became stronger.

A web-based survey instrument was distributed to a sample of 1200 students at a
large Midwestern university, with 197 completed surveys submitted at the end of data
collection. Multiple regression and simple slope analyses were used to examine the data.

The Theory of Plamed Behavior was partially supported by the results of the
study. Habit was shown to have a negative relationship with perceived behavioral control,
with habit being a better predictor of both intention and behavior than PBC when habit
was present. Habit also had a negative relationship with self-efﬁcacy. Habit was
increased by greater levels of consistent context, with the possibility of a curvilinear
relationship existing between consistent context and habit not being supported.

In closing, this study’s conceptualization of habit is clear, parsimonious and
provides an explanation for automatic media behaviors that would assist reasoned action
theorists in understanding the great diversity of human media behavior. Habit appears to
have a moderating role on Theory of Plamed Behavior variables as it grows in strength

that should be acknowledged when studying potentially habitual media behaviors.

Copyright by
ITYAUQIAAbKSE

2009

To Harry G. Waterman, Jr.

X0 “I."-

‘ I

Ind -- p . ‘
1 prim,“-

Ldkd-i\|.£\.i‘\"

III
on.

I
"“ ‘1".: .1.
A__“L.. \~ :‘~\

g".

I 9'13) ‘U‘ﬂ .
.a~.\..\g_ 1‘11“

renter ofﬂ- p
H . i - —.
“3179‘“ 0? i .

."'t\0 “.‘ '
n l I , .
Pun“. OI 331‘ ._

\l,._ .
“.I|\\“ Lat '.. i '
II’ k‘x‘5 1k

,L'I' ‘L« 0'
. ., h.“
.a:.“ ”ILL? k\\“

f

v
.'

5i

'5-‘4‘40iinc d.

\

 

ACKNOWLEDGMENTS

No work of scholarship is done alone. Robert LaRose, my dissertation director,
had a tremendous impact on the development of this dissertation. On many occasions I
would have been completely stymied if not for his timely advice and nearly inﬁnite
patience. I am both a better writer and a better thinker because of Dr. LaRose’s rigorous
review of each draft of this dissertation. I would also like to thank my committee,
composed of Cliff Lampe, Hairong Li and Wei Peng, for their insight into the ﬁnal
product of this dissertation. I also owe a debt of gratitude to my peers Clay Dedeaux and
Missy Lewis for their ongoing moral support. I hope my own support helped them
through their own dissertations.

My time at Michigan State was also made considerably brighter by the helpful
efforts of the department’s office staff. Kim Croel-Kersten, Rachel Iseler and Denise
Mahoney have my undying gratitude for their assistance in acclimating me to the unique
culture of Michigan State. Nancy Ashley, MIS PhD Program Coordinator, also provided
me with considerable help and assistance in navigating the often conﬁising and alarming
bureaucracy of the university.

Last but certainly not least, I would like to thank my wife, Amanda Flowers
Lange, for her support during my lengthy dissertation process. Very few women would
have accepted their husband working up until they left to go to the altar on their data
collection project, nor lived with both me and the dissertation for nearly a year afterward.

Her love gives me the strength to prevail even against otherwise insurmountable odds.

vi

LIST OF HIS.
LIST OF FiGi

CHAPTER I

IXTRI’JD'L'CI.
Tire lat.
IOLIQ'
H351: .. I
The Iii.
Hii‘ll .1‘ I

..
53.7.2;

CHAPTER I
LITERXII RI
Hii‘li .
Curran;
ITS Clef:
The Th:
Hahn 3::
”0 rec.
Thaw},.
A few...l

The 5

 

A Ci
‘ n‘ﬂchA'
‘ “mi-7dr.

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................... ix
LIST OF FIGURES .............................................................................................. x
CHAPTER 1

INTRODUCTION ................................................................................................ l
The Internet .............................................................................................. 2
YouTube ................................................................................................... 3
Habit ......................................................................................................... 5
The Theory of Plamed Behavior ............................................................... 8
Habit and the Theory of Plamed Behavior ............................................... 10
Summary ................................................................................................. 12

CHAPTER 2 '

LITERATURE REVIEW .................................................................................... 13
Habit ........................................................................................................ 13
Current mass media research in habit ....................................................... 18
The deﬁcient self-regulation perspective on habit ..................................... 23
The Theory of Plamed Behavior .............................................................. 25
Habit and the Theory of Plamed Behavior ............................................... 28
Two recent examples of media habit studies using the
Theory of Plamed Behavior ..................................................................... 32
A re-conceptualization of habit measurement:

The Self Report Habit Index ..................................................................... 37
The unique role of control beliefs ............................................................. 4O
Approaching an understanding of the role of habit within the
Theory of Plamed Behavior ..................................................................... 46
Hypotheses .............................................................................................. 50
The Theory of Plamed Behavior and habit ................................... 50
What is the relationship between habit and PBC? ......................... 52
What is the role of consistent context in media habits? ................. 64
Will the relationship between habit strength and self-efﬁcacy
change media use intentions? ........................................................ 65
A model of habit in the Theory of Plamed Behavior ................................ 66
Summary ................................................................................................. 67

CHAPTER 3

METHODS. . . .. ................................................................................................... 70
Methods ................................................................................................... 70

Target behavior ............................................................................ 70
Survey overview ........................................................................... 71
Desired sample characteristics ...................................................... 72

vii

Operr

 

I

Habii Li:

 

Conwnl;

Habn_Pi
35N3n,
“‘15::

Self-c513.

A m :1]

[iv ,
“i re.

'5'

Q...
» ‘-L."A
{Hui ::‘\I;i

Ink‘iHQI'
uqkln- . '

Operational deﬁnitions ............................................................................. 73

Theory of Planned Behavior instruments ...................................... 73
Self-efﬁcacy and consistent context instruments ........................... 75
Habit instrument ........................................................................... 76
Recoding instructions ................................................................... 77
Addressing conceptual overlap between instruments ..................... 77
Perceived behavioral control ............................................. 78
PBC and the Self Report Habit Index ................................ 79
Consistent context and perceived behavioral control ......... 79
Analysis.............. ................................................................................. 80
CHAPTER 4
RESULTS.............. ......................................................................................... 82
Sample characteristics .................................................................. 82
Scale diagnostics .......................................................................... 82
The Theory of Plamed Behavior and habit ............................................... 89
Is the relationship between habit and perceived behavioral control
negative? ......................................................................................... 94
What is the role of consistent context in habit? ......................................... 98
Will the relationship between habit strength and self-efﬁcacy change
intentions? ............................................................................................... 102
Summary......... ..................................................................................... 107
CHAPTER 5
DISCUSSION........... ...................................................................................... 108
Habit and the Theory of Plamed Behavior .............................................. 108
The habit construct versus the “past behavior” construct ........................ 110
Habit and its relationship with the subjective norm .................................. 113
Control beliefs and habit ........................................................................ 114
Habit, PBC and behavior. . . .. .................................................................. 116
Habit and intention .................................................................................. 118
Consistent context and habit ................................................................... 120
Self-efﬁcacy and habit ............................................................................ 123
A rival proposal: Transaction costs and usability encourage
user retention .......................................................................................... 124
Limitations of current research ................................................................ 125
Theoretical issues ........................................................................ 125
Methodological issues ................................................................. 127
Suggestions for ﬁxture research ............................................................... 130
Conclusion................. ................................................................................... 132
APPENDICES .................................................................................................... 134
REFERENCES ................................................................................................... 143

viii

TABLE 4-1

~ '
n;' . I,
Tani“ 0‘ P“:

TABLE 4-:

7“" ~_ .
ml Mi? D,

\ i ‘ ‘
(€73.51:th II} B
\
I

"‘.)H' ‘vu V
Laqh~ll§jk!‘; and E‘:

7.13:5 4-5

tripsthsns l.-\

ABLE 4-6
u. ..'.
.._-I‘-l,'.;:€.\lS 18

TABLE 4-‘

Parh“ I
“fusion 5,"
~ . It»:
. r A

vi We”

44-511

. .I
.ik‘II 01‘. '

7-. 'I *
“ABLE 4TB
ZQSIOII Ar“
incited 8-“

B

‘-
| v
.. BEE L9
\. ‘- "I I
\J ‘ A ' n exny:
, "44:3.“ "
.312 {h

. C

:‘BuE 4-11.}

W

A. .‘
.. ..
\‘,

s ~

|."' .illf] A”
I “I ' L.
.. 3:“: 3‘“ ‘

Ad |\~1L k

 

LIST OF TABLES

TABLE 4-1
Theory of Plamed Behavior Scale Diagnostics .................................................... 83

TABLE 4-2
SRHI Scale Diagnostics ....................................................................................... 87

TABLE 4-3
Self-Efﬁcacy and Consistent Context Scale Diagnostics ...................................... 87

TABLE 4-4
Correlations Between TPB Scales, Habit, Self-Efﬁcacy, Consistent Context,
Intention and Behavior ........................................................................................ 88

TABLE 4-5
Hypothesis 1A —— Predicting Intention Using TPB Variables ................................ 90

TABLE 4-6
Hypothesis lB — Predicting Behavior Using TPB Variables ................................ 90

TABLE 4-7
Regression Analysis Predicting Behavior with Habit Moderating the Effect
of Intention on Behavior ..................................................................................... 92

TABLE 4-8
Regression Analysis Predicting Behavior with Habit Moderating the Effect
of Perceived Behavioral Control on Behavior ..................................................... 97

TABLE 4-9
Regression Analysis Predicting Behavior with Consistent Context
Moderating the Effect of Habit on Behavior ...................................................... 102

TABLE 4-10

Regression Analysis Predicting Intention with Habit Moderating the Effect
of Self-Efﬁcacy on Intention .............................................................................. 107

ix

 

RCIREI-i

nsnpvx~u
Ail\tt\\i i]

ran 2-:

\Ideioftbn!

L‘1\

T‘TFV- r—
5 '|
I.\J-,R;.1-:

I ‘ n ,.
fl Dxdlfq‘ﬂ: .

 

LIST OF FIGURES

FIGURE 2-1
The Theory of Planned Behavior ........................................................................ 28

FIGURE 2-2
Model of the Proposed Hypotheses ...................................................................... 67

FIGURE 4-1
A Simple Slope Analysis of Intention and Behavior Moderated by Habit ............ 91

FIGURE 4-2
A Scatterplot of PBC and Habit .......................................................................... 95

FIGURE 4-3
A Simple Slope Analysis of PBC and Behavior Moderated by Habit ................... 96

FIGURE 4-4
A Scatterplot of Habit and Consistent Context ................................................... 100

FIGURE 4-5.
A Simple Slope Analysis of Habit and Behavior Moderated by
Consistent Context ............................................................................................. 101

FIGURE 4-6.
A Scatterplot of Habit and Self-Efﬁcacy ............................................................. 105

FIGURE 4-7
A Simple Slope Analysis of Self-Efﬁcacy and Intention Moderated by Habit ..... 106

IS" .2114; III ’1

 

I
‘

[TV .J' T ’ ’
IAN—\LBU5- >5k‘L‘ 'x-

I I I
y... a .
ELLE-.13 Nil—I ‘5 i

   

3.35: of a hen.

I ' v c
‘Ul‘ r p“ .
.4 H .1‘
an. |3L “‘1‘ 1;)

 

I“. 1"
l
., k1 I:
xu§_\‘h_' ,. ‘1
t #15.
~
5.7“"?

i1:
. {~‘K‘ZF'I'.
1'

Chapter 1

Introduction

Introduction

Habit is a popular topic for social science research, but habit itself is often
misunderstood, poorly measured and overwhelmingly considered a negative aspect of
human behavior. This dissertation seeks to clarify the conceptualization of habit, provide
a test of a better instrument to measure habit, and explain the relationship between habit
and variables of interest to the Theory of Plamed Behavior.

For the purposes of this dissertation, habits are “a form of automaticity in
responding that develops as people repeat actions in stable circumstances” (Verplanken
& Aarts, 1999; Verplanken & Wood, 2006). The automatic nature of the behavior is an
essential aspect of habit, but the role of stable circumstances (an idea that will be referred
to as “environmental stability” or “stable context”) may be less certain in many cases,
particularly as it pertains to media habits.

The public has often perceived automatic media behaviors negatively. Television
has often been the victim of this negative image, with a great deal of research dedicated
to discussing an alleged phenomenon of “television addiction” (e. g. Anderson, Collins,
Schmitt & Jacobvitz, 1996; Csikszentmihalyi & Kubey, 1981; Kubey, 1984; Kubey,
1990; Kubey & Csikszentmihalyi, 1990; McIlwraith, 1998; McIlwraith, Jacobvitz, Kubey
& Alexander, 1991; Wim, 2002). However, normal media habits pose no threat to the

normal populations that are the subjects of most studies (cf. LaRose, Lin & Eastin, 2003).

zit.- DRUM-
:iu. its. 2

Of $65!." 13M Q

 

comers n; -'

9‘ - - .

3121:5303 :r. x

'7" _ V“

:Ct ”at ”WI
The [at

:1 .113 mat hax

#4.“: came-t 3;,

 

:«su splpc‘rs 3in
333'. at I? ii ”I“
prairie an mi:
.1,» . ‘ -
Maths 5; Ugd
\I. .

3:31.135 tr:ch
‘v-J."

Cijifrs ,
Mil emit

EDDIE "-

.{ 31H:

ewdi

Lu“
J 9
-.~ Q ‘
‘skkd It :I‘ _..
:\4t
‘
‘h- ‘
~.i§\;"y,.

“Internet addiction” has become a signiﬁcant area of discussion (e. g. Caplan,
2005; LaRose, Lin & Eastin, 2003; Shapira et a1. 2000; Song, LaRose, Eastin & Lin,
2004; Yee, 2004; Yee, 2006; Young, 1998) much as television became a signiﬁcant area
of scholarly consideration after it became prevalent in society. Much of the same
concerns about television are repeated about the Internet, but the Internet is dissimilar to
television in several important ways.

The Internet

The Internet is an electronic medium that has attributes of many of the mass
media that have come before it, but is better understood as its own medium. Like radio, it
can convey audio, and like television, it can convey video. It has similar attributes to
newspapers and other print media, allowing for pictures and text to be used together to
convey information. The primary difference that sets the Internet apart is its ability to
provide an interactive experience which other types of mass media camot easily replicate
(Morris & Ogan, 1996). This greater level of feedback makes the Internet different from
the mass media, which results in many Internet applications not truly being “mass” in the
traditional mass media perspective. Users have the capability to consume and create
content easily, which has created a media environment that is unlike any that has come
before it.

A great deal of Internet use may be habitual in nature. Fallows (2004) found that
up to 88% of Americans who go online say that the Internet plays a role in their daily
routines, and of those users, 64% believe their daily routines and activities would be
affected if they could no longer use the Internet. Online behaviors like getting news,

checking sports scores or communicating with others are common among these types of

 

 

_ .l‘. .
(.5313. E41103 .
21335 3 mill-

I ' '
other users. .x'

rats to br«_~.;

‘ t
n.- . , _ _
Lin .I‘U‘C ‘5‘

ISM to be C

IR‘

is \' ~
L55. Lm &

What 1.

'11", I.) ‘
.t..u\i'sii,lod 3\ .
Cain

 

users. Fallows also discovered that 30% of American Internet users claimed the Internet
plays a “major” role in their lives. These users do more everyday activities online than
other users, and are much more likely to do those activities exclusively online. Users with
access to broadband also had a far greater likelihood of doing common activities online
than those without broadband. Recent ﬁgures estimate that 42% of all American adults
have access to broadband at home, with the rate of growth continuing to rise over time
(Horrigan, 2006).

There are concerns about how this growth in online activity may affect
individuals and society. Use that could be argued to be excessive may have negative
effects on individuals (e. g. Caplan, 2002; Caplan, 2005; Grifﬁths & Wood, 2000;
Morahan-Martin & Schumacher, 2000, Yellowlees et al., 2007) but determining what is
“excessive” is a challenging issue. Using the Internet a great deal may appear to one
person to be excessive, but a person who simply uses the Internet with great regularity
may not be impacted negatively by their use (see LaRose, Maestro & Eastin, 2001 ,
LaRose, Lin & Eastin, 2003).

What is popularly believed to be “excessive” Internet use may be better
understood as the result of a state of deﬁcient self-regulation. Deﬁcient self-regulation as
deﬁned by LaRose et al. is a state in which conscious self-control is diminished. Habit
exists all along the spectrum of deﬁcient self-regulation, with most habits not presenting
any negative consequences to individuals.

YouTube
An Internet-related behavior that is growing in prominence is the use of the online

video site YouTube. YouTube is a free Internet site that allows users to upload their own

 

EliilUlS OI “ i
' "

u. . , l
,l “TEX. -.

I I

' V

‘ - ‘ t9?-

b.£>32c UL“
§

YUCI .

. .
ﬁrst} 071 ~‘~
,~: - , 9L)
ﬁxed} ‘3" “‘IL ‘\

‘1,"I‘I“ ﬂ

>5 .l.i\dIII UNL.'
TEL". I '
... .lI‘. :0 IRTL‘L‘

net's— .
lit-t AOJg-LII\ ‘3:

ﬁff‘ - I
~MMWHMddsn

 

27:60 He

TV?"
\ I \ \

‘al‘ 31— 5‘ II}

r..\\'

.:
[tn

IUUIuh

.F'
i "i" ‘
“'ch ‘11"
mt» \
IL
:1; " I.
i' " -
‘~41'\ . "“
‘~ Muir-WT} ‘I'
. t Li
it"‘h
‘AJjA .r'
\ \IC‘CC’ \

video content to share with others. It is claimed by YouTube that users watch hundreds of
millions of videos per day, and ten hours of video is uploaded to YouTube every minute
(YouTube, 2008). YouTube’s signiﬁcant user base uses the service to embed videos into
blogs and other web pages, making it a ubiquitous part of the modern Internet.

YouTube was founded in February of 2005. It is a video sharing site that relies
entirely on user—created content. Any one who makes a free YouTube account can upload
videos to the site, which allows videos to be seen by anyone whether they have a
YouTube account or not. YouTube claims its user base ranges between 18 and 55, with
an approximately even split between male and female viewers (YouTube, 2008). More
than half of YouTube’s user base is said to visit the site at least weekly.

YouTube’s claims appear to have a basis in fact. Rainie (2008) found that up to
48% of Internet users had visited a video sharing site like YouTube at least once, with
signiﬁcant user growth from previous studies in 2007 and 2006. Madden (2007) found
that up to three-quarters of young adults watch videos online on sites such as YouTube,
with roughly one in three Internet users between the ages of 18 and 29 claiming to watch
or download some type of video during a typical day. Overall, almost 30% of all online
video viewers claim to have watched or downloaded video from YouTube, which rises to
over 50% when only young adult online video viewers are considered.

YouTube use is a popular online social activity. Madden claims that 57% of
online video viewers shared links to videos they watch online with others, with young
adults sharing their links over two thirds of the time. Further, Madden found over half of

online video viewers say they have watched online videos with other people, with young

.. a-
H'A‘Jn? 1.'o_-.'.
s'.~.\\.IL .risi. \
.

--. _.. _ ’
ndi‘O 9,35. In.

3".3’33 :‘J i0 Dc

1‘
V

4.23;: SILLGCI“
i‘Tfreqsisitc I} '
MEN his:

To be?

Sis-3.510 be ‘IN

.IiC‘IEIUfe IIIJI i

more dztﬁcult :

 

“1.11;“?!
Habits ,
’1‘ --J .
I «Ned 111 UK
‘5,
:I‘ ",1,“
H‘EPQClg‘...
ii"? .1 A '
{ldi [he 1"
2:5,; _ ~
- d3 leafy.
“1"n' I
-. “IRA

 

adults tending to have said they watched online videos with others three-quarters of the
time.

This dissertation chooses to examine YouTube use as this study’s habitual
behavior of choice because of its ubiquity; many popular web sites use YouTube or
similar free video services to provide online video content. YouTube’s interactivity is
also an important difference from a similar electronic medium, television, which may
present different challenges for habit researchers. As one of the most popular online
video sites, use of YouTube among our chosen population of college students was
expected to be high per the Pew data previously reviewed. Use of YouTube among
college students was anticipated to be common, and while regular use is not always a
prerequisite for habit, repeated YouTube visits could create a situation in which a
YouTube habit would be more likely to exist.

To better understand the idea of Internet habits like YouTube use, some time
needs to be spent discussing what habits are and some of the controversies in the habit
literature that have helped to make discussion of this ﬁmdamental yet elusive concept
more difficult for academics to explain.

Habit

Habits are believed to be triggered by cues from the external environment, an idea
grounded in the behaviorist perspective (Dewey, 1924; Dunlap, 1949; Watson, 1924).
This perspective has softened over time to become more in line with Triandis’ (1979)
claim that the environment and the individual share responsibility for the creation of
habits as “learned patterns of acts”. These patterns of acts are encoded in higher-order

cognitive units called “schemas”, which organize reality in a way that is easier for us to

 

'
....J ..-,., ‘ I
I;«..ft'>t.£uu 1‘

is m. are:
.1; 7:503. I12."
arses-tics ask .-
Bmdtl'h.’ 3; '
E14553; A4715 (
intern Am.
E II 01:11 :5 1: 1r.
B-i‘iltj t."

'."._1~ . '
H‘.~\\.nz“ﬁ v p
‘5‘ _lL. . '

“‘ s.
‘

 

 

 

 

 

understand (Landman & Manis, 1983). It is believed that these cognitive structures can
be activated by environmental stimuli (Anderson, Berkowitz, Donnerstein, Huesmann,
Johnson, Linz, Malamuth & Wartella, 2003; Fiske & Taylor, 1984), even below one’s
conscious awareness (Bargh, Chen & Burrows, 1996; Bargh, Gollwitzer, Chai,
Bamdollar & Troschel, 2001; Bargh & Pietromonaco, 1982; Barker & Schoggen, 1978;
Hassin, Aarts & Ferguson, 2005; Kawada, Oettingen, Gollwitzer & Bargh, 2004;
Sheeran, Aarts, Custers, Rivis, Webb & Cooke, 2005; Quinn & Wood, 2004; Verplanken
& Wood, 2006; Wood, Tam & Witt, 2005).

Based on recent research, this dissertation deﬁnes habit as “a form of automaticity
in responding that develops as people repeat actions in stable circumstances”
(Verplanken & Aarts, 1999; Verplanken & Wood, 2006). This deﬁnition captures the two
most important aspects of habit the Iiterature considers; automaticity and the presence of
stable circumstances. Many studies do not deﬁne habit as clearly, relying only on how
often the behavior is repeated (6. g. Ajzen, 1991; Bagozzi, 1981; Landis et al., 1978;
Traﬁmow, 2000), only its automaticity (e. g. Bargh, 2002; Lindbladh & Lyttkens, 2002)
or do not deﬁne habit explicitly at all (e. g. Dahlstrand & Biel, I997; Guariglia & Rossi,
2002; Wittenbraker et al., 1983). Incomplete deﬁnitions of habit, or treating the construct
as a primitive term that does not need explicit deﬁnition, lead to inadequate consideration
of the multi-faceted nature of the construct and, as a result, measurement that does not
accurately account for the role of habit in behavior.

In many cases, habit is conceptualized as the frequency of previous behavior only.
Behaviors that are repeated more often are considered to be “habitual”, while behaviors

that are repeated less often are thought of as not being habits. The fallacy contained

 

l I-
stir; this IV

I
.

‘..-4o~,A_-‘ "
:1" am:

‘
. ss-buta

talk on .1
212': home '.
Limit:

RICE. FILIQQNN
Aft Opp

1'“

1

“Mn Tex-2' I”;T
' “‘ --.1.L

 

A \.
\q ‘
.i 35
_‘ ¢5&\ Irv“
~ u‘ .1.-“y ‘
C

1}.
~s
l‘ R (1'13.
‘ a\ Tr].
. ~. ~
\—
2';
~._ ‘4‘-,I .
wit-v u
‘liajit It»
f‘,-'Iv-_‘
“-J h—_
I itﬂlﬁeh
I .i\
I .
. 1&1!“ .
‘\‘K I..1F "P
V '
‘ T
‘hli
I T ,
H‘\-:.I‘.. ',
."~-.‘d ._
‘l-CE .J

within this type of conceptualization is that behaviors can be repeated often without being
habitual, and habitual behaviors can take place over an irregular period of time.

For example, a person may choose to watch YouTube videos published by a
single content provider and generally ignore other videos. This person may visit
YouTube on a regular basis (once or twice a week) to see if there have been updates by
their favorite video provider, but might not actively browse for new content for a long
period of time. The act of visiting the site may seem habitual, but in reality it is an active
search process being guided by a person who is looking for a speciﬁc type of content.

An opposite example presents itself when considering a YouTube habit that may
only recur infrequently. YouTube has previously highlighted videos produced by political
campaigns to encourage democratic discourse with candidates. A person who is
otherwise disinterested in YouTube may visit YouTube a great deal over a short period of
time to watch candidate videos, questions other people have posted to candidates, and so
on. Measuring only the frequency of that person’s visits to the site before or after the
political campaign season would detect no habit at all, but if the measurement was taken
during the height of the campaign, that person would seem to have a very strong
YouTube habit. Yet neither perception of that person’s use is accurate, since their strong
habit is only triggered by the presence of a type of desired content (political videos) that
are available only inﬁ'equently. The viewing of YouTube quickly becomes habitual over
several months, but then may taper off when the stimulus is no longer present.

A different type of habit conceptualization is found in the Self-Report Habit
Index, or SRHI (Verplanken & Orbell, 2003). The unique feature of the SRHI is it

conceptualizes what are believed to be the ﬁJndamental aspects of habit;

 

 

.. ....:'. kt
EEL‘IIIAHIJI I

.“ .

I
.‘.- or 4.
1'30 ...1\ iiJUJ

.'_. ,‘j‘ .
>.:..'uru 36 III...

1031; W” -°

IAIII.;LQ»‘
TL
..I- '-
i...‘ \.

‘9.- \0 \ . ‘. ‘ .
a.“ 1') in; 9"

Lula“

u. r
W ' .“J. -.
&A\'L)o 3.,\ :\:.\UP

 

 

I: "‘ J .- 7 '
an. .i 5. \ h - V-
. ._ K
I

.. x

‘9).- “H 3 N. ’
.._‘”» H; \ . .
t ‘ Juik'

carcass rr. 3cm.

_‘1 .
4a] 'E’ﬁ I
—:..' ‘Ale \l‘v‘
:- +-
1';' 'y‘.
I ‘ I”)... Hf If
H “19.,
4‘ lkg‘J
..§.A~'
Puma \l
I" ‘-
\ yi‘I‘J“ ‘3 ()1. ‘
su ‘0‘

I- ~
:l“V ‘
"1‘. 'C ‘
.
“411)
t-‘\
.n‘,
.u a
k hNF' ‘
‘ t NV.,
0 *‘ .
‘4
~
\
‘7‘.~_
“;\§p., \
1“ 1'
. ‘Kl‘ ‘1
\
It -.
\ a-
‘
- 4
‘ arka‘
P‘-‘y.r
C
u\\“
b H 3
"I. H:., a
4“ I. |v
‘ F
L11
~.

uncontrollability, lack of awareness, and cognitive efﬁciency (p. 1317). While the SRHI
also has traditional items that ask about behavioral frequency, its most important feature
is its approach to attempting to capture automatic behaviors. This conceptualization
should be much more robust in terms of ﬁnding habitual behaviors, as well as being able
to discriminate more readily between different degrees of habit strength.

This study seeks to underscore the importance of both external environmental
factors and internal goals in acting as the foundation for habitual behaviors. In some
cases, the external factors may be less useful than the internal factors, and vice versa. In
all cases, as habit grows in strength, both internal and external factors start to decline in
strength as automatic systems begin to take over. This loss of volitional control and
increase in automaticity may follow a predictable path, which can be understood initially
through the context of the Theory of Plamed Behavior.

The Theory of Planned Behavior

Habitual media use behavior can be understood through a reasoned action
perspective. Media use behavior, such as using YouTube, should begin as a process that
is controlled by volitional, active reasoning. How does this active reasoning process
become an unconscious, automatic behavior?

Most habits are not created by the intent to create a routinized behavior pattern by
an individual (Verplanken & Faes, 1999). Instead, an inherent need for cognitive
efficiency brings about a cognitive economization process, which eventually removes the
necessity for active thought processes about tasks humans engage in repeatedly. In other
words, a person may begin with a conscious goal and pursue that goal with conscious

effort, but if the action in question is repeated again and again over time, the necessary

 

t.‘ ‘0'.» ",‘Il‘
‘k‘lnhl‘e >ik:
b

m.‘ ,"1 '
ITALBdrnm

 

,.
~"-n-‘ " Np‘r‘v
itsl‘mé‘ t1. - .
L. M‘H‘ ' “-.-
u—ILIJ...‘ III-L a.

"'“ YIt I V‘

.. . H
“Eda \ 1.. i‘
“4‘ ”I' \‘ r) .
_.~-\1;)‘( Rkk

‘3'; 19 5.;
1‘ "l\.\u k (‘j

I.

fore\a

 

@3136 Sir-:3 .
“R "rfi; ‘

.‘Nx '
k‘1u\‘f‘
\
s-I"'.‘
\\ 1-3-‘ 1
.H‘.‘ *
J‘KA 'T"
-.‘\
|_ .2.
~)"--.
.ll‘g. £1- ‘
\‘I‘.
‘ l
‘5
‘ ‘-
‘C“)F ‘
\A\" ‘
‘7‘ .
"‘t

cognitive steps eventually become abbreviated through automaticity (Bargh, Gollwitzer,
Chai, Barndollar & Troschel, 2001). The process of actively thinking out a behavior
becomes abbreviated into what could be thought of as a cause and effect relationship that
humans internalize, sometimes called a heuristic, allowing individuals to reach goals
quickly without having to apply extra cognitive load on each occasion. Some behaviors
may also become automated without a specific goal in mind, being developed by chance
as a result of the environment.

For example, the placement of buttons on a web page a person regularly visits
may be used so oﬁen by a person that they eventually lead that person to attempt to use
that same area of the screen on other, different web pages. The placement of the buttons
was not under the control of the user, and using them in a particular order may be the
result of good (or poor) design rather than a conscious decision on the part of the user.

The Theory of Planned Behavior (Ajzen, 1991) can inform discussion as to how
an individual might start down the path of creating an automatic behavior. The Theory of
Planned Behavior consists of five major variables; attitude, subjective norm, perceived
behavioral control, intention and behavior. The first three variables predict intention, and
intention (as well as perceived behavioral control) then predicts behavior. It has been
referred to as a theory of the proximal determinants of behavior (Conner & Armitage,
1998) because variables that impact on the three major determinants of intention are not
considered inside the domain of the theory. However, the Theory of Planned Behavior
does not explicitly contain habit. The inclusion of habit within the theory has been a

controversial topic for many years.

 

l' l' ' .-
. . ‘4 ‘ y 4 'I‘
1’1“?“ “I‘L‘ .-nl
3 '.
TLiL ;\-

.. “~11” 1 r
.1 LOLA-40‘ e; _‘i“

. -
.1‘.‘L' “I"‘" . 2'
315‘. u.i.sb AU.

lb.) '\ I
., ‘L’t"i-- .
‘ “44.51“".

C ,

”i ’.M~ ..
.4 .gL‘JDZ‘VNl-y
.:.

K‘ :.:"\:H .V
. ‘ 3‘
101] i
"'hn‘ U ‘
\ " h,
:1 Nat F
.f... ~19:

 

 

 

Habit and the Theory of Planned Behavior

The idea of habit within the Theory of Planned Behavior is usually contained with
a controversial variable known as “past behavior”, which Ajzen (1991) claimed was a
“wastebasket” variable that contained error variance and temporal stability. Conner and
Armitage (1998) disputed Ajzen’s claim by noting observations that past behavior
accounted for more variance than error could reasonably explain. Part of this variance
was claimed to be the result of habit.

Ouellette and Wood (1998) proposed that stimulus consistency (or environmental
stability) was the missing link, as a stable environment would prompt the activation of
existing attitudes that would lead to a “reasoned” decision. Ajzen (2002a) disagreed with
the conclusions of Ouellette and Wood and focused the discussion on the active process
of reasoning, claiming that the environment was not the crucial piece of the problem, but
instead a person’s intentions drove their behavior. The stimulus was far less important
than an individual’s cognitions. However, an automated behavior would ultimately act to
perpetuate itself after a certain threshold, as the cognitions that initiated the behavior
would become inaccessible as a result of automation (J i & Wood, 2007).

The main complaint of Ajzen throughout the debate is the weak conceptualization
and poor measurement of past behavior, which is used as a proxy variable to reach habit
throughout the TPB literature. The stronger, separate conceptualization of habit used in
this dissertation that addresses its cognitive nature should be able to surpass the problems
inherent in past behavior and allow for a habit variable to be usefully integrated into

reasoned action theories like the Theory of Planned Behavior.

10

 

1:3 10711 C“
. I ~‘ -, ‘.
Jeers 330141

3
AN)“
0‘. ‘ ‘58.; k .-

.3 . I

“-.-‘It 1“,.” ‘

‘J35u531-o; L‘ '~
‘

355,} or. the

:ssemréur pr
:32: ones p:
33351: Moms
713:}: at it edit
staringﬁll} I.
37:37.12. habit SE

533?th it ill b

I
Ailt‘ Oi‘ '
d 4.lr}€d B.)L'.'
\....1\’

 

436 mil?»
3:?" 1'3. ,!
“‘J \ ‘-
‘ 0‘ n1

 

The main interaction of interest in this dissertation is between perceived
behavioral control and habit. Perceived behavioral control (PBC) is a variable created by
beliefs about one’s ability to control the external factors that will encourage or discourage
the execution of a behavior. The unique contribution of this dissertation to science is
discussing the shape of the interaction between habit and perceived behavioral control.
Based on the current scholarly literature on habit and perceived behavioral control, this
dissertation proposes that Internet habits may begin as behaviors that are very much
under one’s perceived behavioral control. As time passes, a habit may form, and that
habit becomes more important to determining ﬁiture behavior than external factors.
While at weak and moderate levels, perceived behavioral control still contributes
meaningﬁilly to the prediction of the future behavior, in situations where habit is very
strong, habit should be the primary predictor of interest. Evidence supporting this
assertion will be discussed throughout the course of this dissertation, but will be
discussed in particular throughout the second chapter.

The objective of this study is to apply the variable of habit to the Theory of
Planned Behavior to demonstrate how, over time, a behavior that begins as a reasoned
process can ultimately become dominated by habit. If theoretical understanding of this
process is correct, it should be clear that individuals who still have a low YouTube use
habit are more driven by conscious decision-making processes in their YouTube use than

individuals who have a strong YouTube use habit.

ll

 

5;;‘7ﬁ’72a'l'l'

Y T

\ J L

a, . ”ﬁli': "‘
fr mm

& ..45\U¢ "

2132 31:3 C(Z‘Ei. ,

 

Summary

YouTube habits are a type of human behavior that is becoming more important as
the medium matures. Exploring YouTube habits with a fresh perspective so to put them
into the context of the Theory of Planned Behavior will provide new information to help
address the heated debate over the risks of online activity. Additionally, exploration of
the topic should help to provide evidence for improved instrumentation in habit studies
that relies less on the environment and number of times a behavior is repeated, and more
on the attributes of habit that make it compelling to scholars; its uncontrollable,

subconscious nature.

 

 

v! '1
A ....,
2..) s:
V
".‘\
.Ln-
4;vJ'-'n«.
v5'\.lg._\ 1“},
l.

'H‘- v

.
4 Y‘” 1.1 .
u... “Pk ' ‘-

".‘ “Y‘v ‘ n
u l r‘ ~ .
:4... 1“ “> dhi,.

I
t.

 

 

 

 

Chapter 2

Literature Review

Habit

This dissertation deﬁnes habits as “a form of automaticity in responding that
develops as people repeat actions in stable circumstances” (Verplanken & Aarts, 1999;
Verplanken & Wood, 2006). Initially, a person may have accessible cognitions about
why they engage in an activity, but the cognitive economization of habit makes those
cognitions gradually less accessible as habit grows in strength. As time passes,
individuals may begin to respond to the average rewards over a long period of time rather
than immediate short-term rewards, further reinforcing habitual behavior (Wood & Neal,
2007). In other cases, a person may not have ever had a particular set of conscious
cognitions about why they engage in a behavior, with the behavior becoming routinized
within an “explanatory vacuum” (Oettingen, Grant, Smith, Skinner & Gollwitzer, 2006).
Regardless of how a habit is created, habits are difﬁcult to disrupt even when there are
conscious intentions to alter a given habit (Verplanken & Wood, 2006).

Habits can be triggered by many possible cues, such as the presence of interaction
partners, prior behaviors in a sequence, moods or environmental stimuli (Verplanken &
Wood, 2006). Environmental cuing was ﬁrst championed by the behaviorist perspective.
Behaviorists such as Dewey (1922), Watson (1924), Dollard and Miller (1941) and
Dunlap (1949) considered habit through their stimulus-response paradigm. To the

behaviorist, habits are conditioned responses to stimuli intended to resolve a need

13

 

1th3xiﬁriszr~

T‘
V' -n1 ,\v
“-5-: ILL 5.

‘
“' ‘J’JWynJ
51a.u\.~;1“‘_l~

.“-..-
‘~“\-‘ I“.— .. no
\ t‘d ‘ZIJ‘

‘9-
w.‘

rm‘, 3
““‘~\l'\ C \\

"‘12-..‘v. (j 9 W 1
It: ‘k k ‘
u

R‘?

l til '-

4y,
' ” 253?:1:

:‘IJ g ,
‘l' ‘mililum

 

 

 

presented by the stimulus (hunger, thirst, etc). Dunlap went as far as to claim that
individuals appear to each other to only be a collection of habits.

However, behaviorism does not account for any personal agency. Ajzen (1991)
and other reasoned action theorists believe that behavior is driven by intention rather than
by simple stimulus-response relationships. Intention should regulate how likely the
behavior is to occur so long as the behavior is under the actual, volitional control of the
person. The environment has an important role in this control over one’s actions, which
can determine how difﬁcult it is to complete a behavior. If the environment is favorable,
the behavior is more likely to be completed than if the environment is not favorable. The
importance of the environment in enacting behavior, especially habitual behavior, has
been a part of the academic literature for many years.

Triandis (1979) deﬁned habit as being an automatic situation-behavior sequence
that does not require conscious self-instruction, similar to James’ (1890)
conceptualization of habit over a hundred years before. A habitual behavior is activated
in a reﬂexive way, but is not itself a reﬂex because it must be learned. Triandis considers
learning to be a multifaceted experience, integrating both ability and previous experience.
Beyond learning patterns of acts, Triandis claims that habits integrate patterns of thought
and emotion.

Triandis and others have understood habitual behaviors as being part of schemas.
Scholars who support the idea of schemas claim that people have cognitive structures
which represent both general and speciﬁc knowledge in a single higher-order unit called
schemas, and these schemas help us organize reality in a way people can understand

more easily (Landman & Manis, 1983). These cognitive structures can be activated by

14

 

Hammett:
\lgfmtth 31'
or :rtrned. 1?.
P3303031}; .

:7 '1'~:.1_1.._
""~~ inu.lk

\\.1!1'~
‘5‘
y

a.‘

N1 ,4

-‘a';. J
_! ..X A‘ in“ K.

i,

:th1 V. .
ushﬂlJLq11§ r

"I‘Je . I

.‘1',s\.‘x)‘\ 1‘12,
\

T' I
1.6 IT...

- , l
u \§b. aLZK‘P‘ ‘
A...

X:- ‘ q"‘ I

L. k
59,.“ ’.‘
.'*-.“‘;-“ .'
'} JCIEK
,1
~.._,‘ -1
- 1-
X “ \l‘nc
.
ww- ..
. ¥\-\‘\e ‘.‘ ‘1
5:38;” .
.\h L() a\ ‘.
1 K
».

1
\: 1.
._ ‘3‘
“‘s,\ ”a.
J‘itcﬁ‘
01.51
.‘h.
”uh"
k. 1")”

 

environmental stimuli (Anderson, Berkowitz, Donnerstein, Huesmann, Johnson, Linz,
Malamuth & Wartella, 2003; Fiske & Taylor, 1984). When these schemas are activated,
or primed, their effects can occur even below one’s conscious awareness (Bargh &
Pietromonaco, 1982), which can be integrated into an understanding of habits as being
automatic behaviors.

Wegner and Bargh (1998) deﬁne an automatic behavior as a process that can run
by itself and does not need conscious guidance once it has begun. A more detailed
deﬁnition is provided by Bargh and Chartrand (1999), who describe automatic processes
as lacking intentionality, controllability, attention or awareness. These automatic
processes gain cognitive efﬁciency by not requiring active agency to execute.

The manner in which schemas can be activated below conscious awareness to
trigger automatic behaviors is the main argument used to support the conventional
scholarly understanding of how habits are activated. The argument scholars have made is
since humans encode their behaviors into these complex higher-order units, and these
schemas can be activated without their awareness, an automatic behavior like a habit is
primarily activated by the environment. Therefore, changing the environment should
allow someone to break free of a habit. For the purposes of this dissertation, this
perspective will be called the “environmental stability” perspective, sometimes also
referred to as “consistent context”, “stable context” or “stable circumstances” (Wood,
Tarn & Witt, 2005). It is important to note that the idea of “stable circumstances” also

includes internal states such as moods or other cognitions, whereas most deﬁnitions of

“environmental stability” only focus on the external, physical environment. Cues

15

 

‘.~ ‘
‘ k.‘ .
\ \n‘. Ir. .
Fje emﬁy‘ ,
.3"...
“\1)” 1" .__
\le
' H
13$?! -\
'45. .' H
a“, &
.m.
» A ‘
5"

«Ha

,. ‘
'smn"‘ ..
Frl_‘j‘.‘.'|‘u k v

1,... 1‘9i '
ET;...1}.Z~U I

:9" w tV‘f‘ﬁ-"ﬁ
\..‘A.\. l....s..

T— ' I
my; hr. .3 W
. 1515 u‘» \ \
.- L
.1.) . .1 ..
2>~.. ~;.~1 LA

r‘ l' ‘u—y]._‘,n0 y:
hu'Lk-... \ui .

3.31.315. 1994

:15.::e.'ed In

v""-.3m
H 5-;5.
136 m
‘

1 ail-telou

r
K

1:1

.2331: ad\ 8.".

 

‘-«1§:¢.;._
“‘de ‘
“-3.
.Vi‘y'I“ 1_._
M. c.
‘9’”); \

prompted by internal states can also be stable over time, but these have been less
emphasized in the habit literature than environmental factors.

The environmental stability based understanding of habitual behavior states that
habits that have been generated in the environment will be maintained until the
environment changes or they enter a different environment (Ouellette & Wood, 1998).
There have been studies (e.g. Quinn & Wood, 2004; Wood, Quinn & Kashy, 2002) that
assert that up to half of one’s daily activities may be habitual. Changing one’s
environment has been shown to be a useﬁil tool in breaking habits (Heatherton &
Nichols, 1994). Yet, environmental cues to engage in behavior are constantly
encountered in our daily lives, with only a handﬁil being acted on out of the great number
of them.

The most prevalent examples of these environmental prompts are advertisements.
The Yankelovich market research ﬁrm claims that the average American is exposed to up
to 5,000 advertisements a day (Story, January 15, 2007). Many of these advertisements
may attempt to invoke habitual behaviors to consume or purchase commonly used
products, but the average person would be hard-pressed to ﬁnd the time to react to them
all if the environmental stability perspective were as powerful as it would claim to be.

The environment may be an important cause for habitual behavior, but it is likely
not the only cause in most cases. The environment may be acting to prompt goals (e. g.
Hassin, Aarts & Ferguson, 2005; Kawada, Oettingen, Gollwitzer & Bargh, 2004) or
stereotypes (e. g. Bargh, Gollwitzer, Chai, Barndollar & Troschel, 2001). It is important to
emphasize that while the priming of a goal may lead to a behavior (Bargh, Chen &

Burrows, 1996; Sheeran, Aarts, Custers, Rivis, Web & Cooke, 2005) that does not

16

 

.' "
—- r ' ' f

. .

>v-'~~..:*v"r‘- '1 .-

Q.LASI‘J'- . *1
u; ~

-\ p1“
~¢3 ‘PD Y "“
3..» mkﬁ I It.
s-m .~.n '3'-
51.11.17“... p».;

131-171. use .'

'-—- ,.

1".ﬁm-C.
i 4 l
4*“ ud.ki.zd;.\.

~

.19.) 1" 1 .
““J‘ ‘ LWI d r\

"na 0.

.1}qu1, [
- S; “UKK j-d‘

“,3“, 1
e . ,3 .- .
kE‘ 5L drt‘:\\

‘1.

 

mi
1 I 3
.l C
, 4W.“-
'- \
.1 .
’5.”
\
~.._”.) A“?
\. \ v '
'u\ r‘
‘\
'~.
'-.~ ‘.
-‘~‘l"

 

necessarily force a behavior to be executed. However, goals can have an indirect role in
triggering habits by providing contextual triggers for action (Wood & Neal, 2007).

A possible way of addressing the question of whether the environment or goals
are more important to habitual cueing is to re-examine the idea of automaticity itself. The
common perception of the automaticity concept was questioned by Saling and Phillips
(2007), who proposed a broader interpretation of automatic behaviors. The authors stated
that automatic behaviors are not purely stimulus controlled, but can be modiﬁed by the
context of a behavior as well as an individual’s intentions. This idea of limited control
over automatic behaviors provides an explanation for why the environment is not the
absolute arbiter of automatic behaviors.

It is likely that the environment and individual goals (conscious and/or
unconscious) both have their say under different circumstances (Verplanken et al., 2007;
Verplanken, 2006). Environmental factors appear to be more important when a behavior
can only be executed in a speciﬁc circumstance, such as habitually wearing a seatbelt (J i
& Wood, 2007), but actions that can be executed across many different types of
circumstances, such as media habits, may be triggered by a number of different factors
that should be individually examined to see how they link behaviors and contexts (Wood
& Neal, 2007).

Habit can be understood as a form of automatic behavior that deve10ps as people
repeat actions in stable circumstances (Verplanken & Aarts, 1999; Verplanken & Wood,
2006). A behavior that begins as a reasoned process can become automatic as the
cognitive efﬁciency processes that create habitual behaviors take place. Habits are

contained within schemas, and these schemas can be activated by a variety of causes,

17

 

1" £31301
' . :1

1313113103. L _
1 '. ‘ '1"

3711'. .11....“

(‘51 n1

'1 .21. ‘ 1n
““M-nku H»
:w. “‘1 .1 -‘
"b “NLL Ll

I

 

including the environment and internally or externally prompted goals. Automatic
behaviors like habits are multifaceted, allowing for complex interactions between
reasoned and automatic behaviors.

The debate over the role of habits in human behavior remains a signiﬁcant topic
of discussion in the social sciences. Most mass media research into habit has focused on
television, but the observations made about one type of media habit should be able to
inform inquiry into other types of media habits.

Current mass media research in habit

A recent study of an automated response pattern in media habits is discussed in
Crawley, Anderson, Santomero, Wilder, Williams, Evans & Bryant (2002). Crawley et al.
examined viewers of the program Blue ’5 Clues. Blue ’s Clues has both recurrent content
and unique content in each episode, but its recurrent content becomes “highly
predictable” (p. 267) and so requires less direct attention from viewers. Experienced
viewers gradually do not need to watch the program as closely to see participation cues,
allowing them to perhaps focus on new content more than content which meets a form
they are already expecting. However, it is important to distinguish here between
behavioral habit and the process of developing a habit. Experienced viewers may not
necessarily have a behavioral habit to watch the program, but instead may have
developed a habit to respond to the form of the content itself, with behaviors being
triggered when some element of a habitual pattern is present (e. g. LaRose, 2008, Wood &
Neal, 2007).

Crawley et al. found that viewers appeared to learn how to watch the program.

Experienced viewers used less attention to view the program but seemed to learn just as

18

 

. '",|'

.135- -
3““1 \ {11$

~1
'7‘

:CillrL-EI k‘t “' ‘

I
I )«
5:131: \\ ..
I \u

3.3.1175 obs. “

”‘17".1'“ ‘1"-
h-~~;-A..A l
‘ ‘

.ln.-

' v
"Vi'~v¢ '

1%.}, ”e 'r
“u. ‘1‘ 416 p

5"»’ 71
““5431 11.11“}.

 

 

H"

r-i‘Jlrr. 0r 5;;

”3'4.- ,1 ‘
5“ \ D) 1
‘ ‘17» 3'31“
.1,
5.
1“ .
.. \V'! ,
“‘>~11'CI‘~Y 9
. n \K‘

N.“-
1,\ ..—,
"';‘ ﬂ 9
‘ ~N““ [K‘ r‘
l‘.‘ _ ‘
‘1”. ‘U
“Idle"; _ C
"4 l 9
l 1;
'-
1p1u. -
P». .11
‘~¢\ .
‘14: K?-
O J 1
‘1
x'-
H '~ V
542‘...“ .
'A
._._\ oh
1.“.3
"1
"x‘ In.
‘\ ‘pu"” ‘
.5. “‘1
~13

much as inexperienced viewers. Regular viewers also transferred their knowledge of the
format of the program to a new program, and were able to respond to the cues within the
program while paying less direct attention to it. The transfer of knowledge provides
clearer evidence of possible habituation rather than a behavioral habit. In this case, the
authors observe the habit of how to watch similar educational programming can occur in
a wide number of possible settings, and can be triggered by cues within the provided
programming.

A similar Internet-based example of habit formation can be found in how
individuals develop an email habit. Most popular email clients are very similar to each
other, as are popular web-mail interfaces. Learning how to check email using one
program allows one to transfer that knowledge into checking email using another
program or client later. The act of checking email can then become habitual more easily,
with many programs allowing a user to set a timer to determine when the program will
next attempt to ﬁnd new mail for an individual to review. This regular, repeating process
eases the human mind into automating the email checking process until it is entirely
transparent to the user.

Rosenstein and Grant (1997) discussed habitual television use using Nielsen
viewing data. They found that the role of lead-in programs on television was very
important to maintaining viewing into the following program, and were important for the
formation of television habits. It appeared that television viewing increased during
periods in which people tended to be either just waking up or coming home from work. It
is possible that television viewing was integrated into their behavioral schema during

these periods, as the television had become part of the daily experience (Langer, 1989).

19

 

|
swim f3. .
I
I
11:11:: 1‘} r.
I
73:26.1? ..
@5333} be.
3:113:32: f3.
::::‘ “L. . 01.1.
"r 52"“

V
.1.s .“““h>

i

\ll\
l~ ll":.| Q
.-~-~ ' v‘ 1
“fluent
__ .‘1

13‘ .
'1“: ‘(t 1
“ DC )r»
‘
Ii "‘
.77.. ”l ~
. 1‘31“: o',‘ _
x.‘-\. I
; -.'_ .
51‘4.‘ ’
“1- .fi
' 7'!
1.1.1“.

 

Rosenstein and Grant surmise that the act of watching television is thus the most
important factor rather than a conscious desire to fulﬁll needs.

At one point, the act of watching television could have been a motivated action,
driven by a reasoned action process. As television viewing became integrated into daily
routine, the original goals that drove a person to watch television in the ﬁrst place
gradually become encoded into a context -— in this case, time seems to be the most
important factor. A person may want to relax or consume time between other tasks, and
television acts as an efﬁcient way of achieving those goals. The goal is eventually
forgotten, but the reward provided is “cached” (Daw, Niv & Dayan, 2005), which links
the stimulus of “after work” or “waking up” with a potential reward provided by
watching television. As a result, “. .. performance environments in which goals are
reached or rewards are received may acquire the capacity to motivate historically
associated responses” (Wood & Neal, 2007), which shows how both the environment and
goals can cooperate to create a habit that is ultimately independent of a reasoned process.

Other aspects of television viewing may also act as important cues to a television
habit independent of simple environmental stability. Schmitt, Woolf and Anderson
(2003) undertook a study about how individuals watch television and found that almost
half of the subjects were engaged in some other activity while the television was on.
These other activities may provide an additional stable context for the viewing of
television, but it appears that the mere presence of a media device is adequate for that
device to be integrated into the scripts (Abelson, 1981) that people use in their daily lives.
With all of the other possible cues to choose from, a stable context does not appear to be

needed to maintain a media habit (Ajzen, 2002a).

20

 

ﬁrruzar113

‘-r- I]! 3‘
~ ? .3 . .
“June. sis

$1.4...
.s lu

st- 1.1y

N"
C44
(‘

9~ . §‘ '-
W‘Har 0‘ H

_-1.;x.5 AA\

'-‘_J"‘n'3_. _‘ ,‘
W'5-11¥L a~\'3\

Q
'2 $13.43;} I‘“

"‘21 u..- C _ ' “
Amntwecet

l v
'7‘; qh_“ ‘ —.
.31; _Q.> L]: I."

u..

I- 'HJ h
“VL‘“' die .
*1! ~., - ‘
~'-.33w-._11\.
~ 4“ F
~ -

3‘
“#334511 TL
—. ' .>}
A4\
I. 'k" .
uf‘x.‘ -
‘ \‘ 0")“.1
. \k a
.‘ ‘1 .
4. \ .h .
w. ‘ J -
A» \Ir\i‘Jr\
I
5.'.5‘u."
x.“ .3 ‘
.e TC\rP1
‘ 1“

 

The robustness of cuing of television habits appears to at least partially explain
why media habits can be executed in even “unstable” contexts; televisions can be found
almost anywhere, and devices that can act like televisions are ubiquitous in Western
culture. Television programs can be viewed outside of stable circumstances by means of
recording devices such as VCRs (video cassette recorders, which use physical magnetic
tape) and PVRs (personal video recorders, which use hard drives). With the growing
number of free wireless Internet connections available from businesses and even cities,
Internet access appears to be becoming even more prevalent than television access. All
one would need would be a media appliance capable of accessing wireless connections,
and those devices are also becoming prevalent. Many new cell phones are marketed on
the basis of their ability to access Internet services for a fee. A great number of cues exist
in Western popular culture to encourage habitual media consumption, both ofﬂine and
online. The media and its consumption are signiﬁcant aspects of daily discourse. If the
increasingly prevalent and culturally signiﬁcant medium of the Internet can be used
anywhere, Internet habits could arguably become a signiﬁcant part of our society for
good or ill.

In an attempt to answer this question, Newell (2003) as well as Newell and
LaRose (2004) directly examined whether or not lntemet habits rely on stable contexts to
function. The authors found in that study that media habits did not require a stable
context once a habit was established. Initial media consumption behaviors are associated
in stable circumstances per this dissertation’s deﬁnition of habit, but during the gradual
cognitive restructuring that occurs during automation, the trigger stimuli for the habit can

change to accommodate a less stable environment (LaRose, 2008; Saling & Phillips,

21

 

 

 

Iiifl'i. 11103

 

I .
“H.111...
Llama; u‘Jt

~
‘; 3 in !r_)} ’,'p
at .1. 15.31‘t.

1.1;... .
“assault-i \\ d‘

 

 

v
‘
NO~~- .
. . .3
l—uugJI J_\~ ‘

:3“ 3'1'1" l'

5.." . .

t b ‘AL1,. \
‘

\ .

. \ t \

5 4

‘
. ,‘
.....’.‘(‘ ‘,\‘

.,
3am...
1,_

{“3 1".
s_“j‘ .: x ‘
hu.‘\ \\

 

.’_
d. 1‘ .,
-,_ ﬂ.
1\
,-_.
1, _‘
-";'. ‘-

2007). In other words, a person may be able to trigger their media habit in a situation
similar but not identical to the one that was used to initially encode the habit because of
the increased cognitive efﬁciencies of habitual behavior, similar to C rawley et al.’s
(2002) learned viewing behavior.

Newell also found that media usage patterns showed that intention to use a
medium was linked to future use behavior. Newell also demonstrated a link between
habitual use and ﬁiture behavior, and proposed a model that showed gratiﬁcations
generating both intention and habit, with intention and habit driving future behavior.

Newell’s ﬁndings replicate those of Bentler and Speckhart (1979) and can be
explained within the context of Wood and Neal (2007) as well as Bem (1972). Because
individuals have only a small amount of insight into their own behavior, their habitual
behaviors can be used by them to infer their own goals and intentions. A desire to seem
internally consistent and competent leads individuals to claim that their habitual
behaviors are actually volitional behaviors.

Another study which shows the pitfalls of attempting to rely on intentions to
predict habitual behavior is J i and Wood (2007). The authors conducted two week-long
diary studies of university students to determine their fast food and television news
watching habits. Their intentions were recorded at the beginning of the week, along with
an estimate of their past performance of the behavior and the stability of their
circumstances. I i and Wood found that intentions and habit both guided behavior when
habit was weak, but as habit grew in strength, intentions became less helpful of an

indicator. Throughout the grth of habit, individuals claimed they were very certain

22

 

.1111 their intentions 1
3311,1131; most of the
Once again. .\1
122.131ng effect on 11'
33:1115Very strong. A
Eelua'ior to nuke ther
3339111335. but their
:3 become routine ft
Media habit is
561131. 101' can appear ,
:idia 1161);: Strong
mine to the initial
3.115;, The Waller):-

.3

13:;13 habits. Media

{hi \3 ()I1( ‘)

'3' "v-
‘1‘
5

"‘1'?“ of med.
The [1,.

flt’lt'Izz

ll Contra“ t“

e m<¥1rC

 

about their intentions related to television news viewing, but in reality their habit was
providing most of the explained variance.

Once again, Newell’s ﬁndings were supported. Habit appears to have an
enervating effect on intention to the point of making intention completely irrelevant when
habit is very strong. As a habit grows stronger, a person may begin to rationalize their
behavior to make themselves seem internally consistent. They may have strong intentions
to do things, but their habitual “autopilot” ultimately leads to a person doing whatever
has become routine for them to do in ordinary situations.

Media habit is often studied using television behavior. Triggers for habitual
behavior can appear in the content of media as well as the outer environment around a
media device. Strong habits can be maintained even in an unstable environment at least in
part due to the initial reward value of the behavior being "cached" (Daw, Niv & Dayan,
2005). The prevalence of cues in Western society to use media also helps to maintain
media habits. Media habits in particular do not appear to be bound to environmental cues
alone per the work of Newell and others. Scholars in the tradition of Bandura take a more
benign view of media habits than others have in the past.

The deﬁcient self regulation perspective on habit

In contrast to other media habit studies that treat media habits as problems,
LaRose, Lin and Eastin (2003) proposed a different method of discussing media habits.
The authors used the framework of social cognitive theory, and focused on lntemet use
behavior. The more precise term they prefer for “media addiction” is a situation in which
deﬁcient self-regulation has caused a problem (in terms of unwanted social

consequences) for an individual. Bandura (1991) discusses self-regulation as the method

23

 

“gt tthtch indi\ idud‘zs t
change their behax ior
ccztrol is diminished.

beﬁat‘ior (LaRose. Ll!
chemically Similar
3:2. Selt‘Report H .1
LaRose et a1.
rerageable \\ ithout
sailor-nothing stat
‘if??esston and lo“ 5
neitumt. As a resin.
“L ¢§3ed that ha':
renew the stimuli

3"- ‘

3-31" to the idea 0

ﬁery

alOl‘lS \\ CFC ll‘

..~.....dual. they \\ er

1\e to a \T 1
,bt'iil‘ fn’
A ~I.’ ’
-~‘t Man

I

by which individuals observe, judge and self-administer incentives or punishments to
change their behavior. Deﬁcient self—regulation is then a state in which conscious self-
control is diminished, and environmental factors become more of a factor in inﬂuencing
behavior (LaRose, Lin & Eastin, 2003). This concept is both conceptually and
operationally similar to a habit measure devised by Verplanken and Orbell (2003) known
as the Self Report Habit Index (SRHI), which will be elaborated on later.

LaRose et al. argued most media use problems are “benign problems” that are
manageable without professional interventions, and fall on a continuum rather than being
an all-or-nothing state of “addiction”. Media use problems appear to be worsened by
depression and low self-efﬁcacy (a belief in this case that one cannot stop using a given
medium). As a result of their study of possibly undesirable levels of lntemet use, LaRose
et al. argued that habits may develop through direct stimulus-response associations
between the stimuli (the media content) and the outcome (positive emotions). This is
similar to the idea of “caching” discussed by Daw et al. (2005). However, while these
associations were implemented via repetitive and conscious decisions made by an
individual, they were perhaps later forgotten, or made inaccessible, removing agency
from the hands of the user. In other words, the cognitive process of automation integrated
context and goals together, increasing cognitive efﬁciency at the cost of removing
accessibility of the relevant schema.

Media habits are generally not harmful to most people. Media habits may not be
as sensitive to a stable context as other types of habits (Ajzen, 2002a; Verplanken &
Orbell, 2003). Many cues exist in the environment and in the consumption of media itself

that can prompt a habitual response. These cues, such as time or the execution of other

activities, can be found in a variety of different environments. The mere presence of a
media device can act as a habitual cue even if no other cues are present. A stable
environment may only be important when a media habit is being established, which
implies that reasoned action processes may be able to explain behavior while a habitual
behavior is still weak.

Based on the recent mass media research into habit, being particularly informed
by both Newell and LaRose, this dissertation can begin to propose a broad theoretical
statement. This dissertation uses the Theory of Planned Behavior (TPB) as its foundation
for further discussion of habit. TPB variables (attitude, subjective norm, perceived
behavioral control) should still generate intentions, but based on how habit has been
conceptualized so far, habit would be separate ﬁom existing TPB variables, causally
preceding intentions — and likely making intention irrelevant as habit grows in strength.
Habit would generate behavior and may moderate other TPB variables, but what would
this relationship look like? Would habit have a moderating effect on TPB variables, or
would it operate as an entirely separate system?

Several studies have been done using the Theory of Planned Behavior as a
theoretical foundation to examine habitual behavior. Many of these studies have had
seemingly contradictory results that could be explained by the use of a stronger
conceptualization of habit. A short review of the Theory of Planned Behavior will be
provided to provide context for these studies.

The Theory of Planned Behavior
The Theory of Planned Behavior (Ajzen, 1991) can be understood as a theory of

the proximal determinants of behavior, as only the variables explicitly within the theory

25

,w

are considered within its domain (Conner & Armitage, 1998). The variables that are
present will be brieﬂy discussed here. An illustration of the theory can be found in Figure
2-1.

Attitude is created by beliefs about the behavior, speciﬁcally whether or not it
would be beneﬁcial or detrimental to execute it. Considering the contributions of
expectancy value theory on attitudes (Babrow & Swanson, 1988), it can be noted that
attitudes toward the behavior develop naturally from related ideas, and then become
crystallized as the behavior is repeated. Attitudes are generally measured by means of
bipolar scaling to measure both the intensity and direction of the relevant beliefs about
the behavior.

The subjective norm is created by beliefs about the possible outcomes of the
behavior. These beliefs speciﬁcally deal with how individuals perceive referent others
(friends, family) will feel about the behavior. The subjective norm is often considered the
weakest part of the Theory of Planned Behavior, as it often has the worst measurement
(Conner & Armitage, 1998). As a result, the subjective norm tends to explain the least
amount of variance and attracts extensive scholarly controversy. One aspect of this
controversy is that one’s motivation to comply with said referent others is often
overlooked in its measurement. The absence of this dimension may often be the reason
the subjective norm appears less robust. Ultimately, Conner and Armitage discarded the
subjective norm from their model when it failed to notably improve the predictive
validity of the model, focusing instead on attitude and perceived behavioral control.

The main variable of interest to this dissertation is perceived behavioral control.

Perceived behavioral control (PBC) is created by beliefs about one’s ability to control the

external factors that will encourage or discourage the execution ofa behavior. PBC is
often controversial because of efforts to integrate the self-efﬁcacy construct into the
variable. When successfully completing a behavior is not totally under someone’s
control, PBC becomes more signiﬁcant.

PBC is similar to Bandura’s (1982) idea of perceived self-efﬁcacy, but has been
problematic to conceptualize and measure (Ajzen, 2002b). Ajzen claimed that PBC can
be best measured by examining both perceived self-efﬁcacy and perceived controllability
to create an overall PBC construct. However, there has been a dispute about this in the
literature, with some scholars believing that controllability is the only item of interest to
PBC (Rhodes and Coumeya, 2003). Others believe that self-efﬁcacy is a better
instrument for measuring control than PBC, as PBC tends to draw on more global control
issues rather than internal, personal control issues. A discussion of the control beliefs that
create PBC in more detail will be provided later in this chapter.

These three factors determine one’s intention to complete the behavior.
Behavioral intention in turn tends to be related to actual behavior, though this is not
always the case. Both intention and behavior have a direct tie to perceived behavioral
control, which gives PBC a considerable amount of importance in the overall

understanding of what may lead to both the intention to act and what one actually does.

27

-'{.I

:5“—
H1
‘\

‘.

 

 

Behaﬁ Attitude

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Beliefs
Normative Subjective Intention . Behavior
Beliefs Norm ‘-
Control PBC
Beliefs "
Actual behavioral

 

 

control

 

Figure 2-1. The Theory of Planned Behavior.

Habit and the Theory of Planned Behavior

The Theory of Planned Behavior has had difﬁculty addressing the role of habit
within its model. As previously discussed, habitual behavior has been contained within
"past behavior", a problematic variable that may contain habit but also contains a variety
of other elements, including temporal stability and error variance (Ajzen, 1991). Habit
would be useful to the Theory of Planned Behavior, and various authors have attempted
to explain why habit would explain more variance than error (e. g. Conner & Armitage,
1998).

One solution to the problem may be to examine the role of environmental stability
in habit. Ouellette and Wood (1998) claimed that the environment cued existing attitudes
that could lead to a "reasoned" decision, which was opposed by Ajzen (2002a). Ajzen
stated that attitudes themselves were what led to behavior. The stimulus was far less

important than an individual’s cognitions. Yet if one operates under the assumption that

28

_.,‘
.41“ 1..

‘w'.n'
t - "
.C5\ .a

(6.131

I
“‘f1"H
1‘ .

sna3

 

\

«.5-13 h:

habits remove active reasoning from a behavior, Ajzen is also mistaken once habit
becomes established, as what was once a reasoned behavior has become automated.

The most signiﬁcant complaint of Ajzen throughout the debate is the weak
conceptualization and poor measurement of past behavior, which is used as a proxy
variable to reach habit throughout the TPB literature. Past behavior has been used as a
measure of habit throughout most of the contemporary literature, including Bentler and
Speckhart (1979) and continuing into the present. Past behavior does tend to explain a
signiﬁcant level of variance in the studies in which it is used, and has been shown to be
more useful in explaining future behavior in some cases than intention alone (e. g. Bentler
& Speckhart, 1979; Conner & Armitage, 1998; Kim & Malhotra, 2005; Ouellette &
Wood, 1998).

Despite the variance explained by past behavior, Ajzen’s complaint about past
behavior still rings true in that “past behavior” contains a great number of other possible
confounds and error points. Past behavior also does not address the important aspects of
habit per this dissertation’s deﬁnition; automaticity and a stable context. Past behavior
merely describes if and how often a behavior has been repeated in the past. It can capture
both habits and non-habits, and has no way of discriminating between the two. Past
behavior has also been criticized for containing common methods variance and other
unhelpful facets, such as temporal stability (Ajzen, 2002a). Past behavior is therefore not
suitable as a solitary measure of habit, though may be useful as an additional criterion in
some cases.

A recent example of the problem with using past behavior as a proxy variable for

habit is Kim and Malhotra (2005). The authors executed a two wave survey administered

29

.3";
u...

,.

.1

I '5.-

2. ”5

.‘W-

by email to 1000 university students to examine student use ofa personalized portal
website. Kim and Malhotra tested several theoretical frameworks, including the
Technology Acceptance Model (TAM) (Davis, 1989), the theory of belief updating
(Hogarth & Einhom, 1992), the self-perception theory (Bem, 1972), and an approximate
measure of habit using past behavior. The study found that past behavior was a
considerably better predictor of future behavior ([3 = .77, p < .01) than intention ([3 = .04,
n.s.).

What is interesting about this study is that the authors use a theoretically
consistent deﬁnition of “habit”; “performed frequently in a stable context” (p. 746), but
use past behavior as operationalized as “on average, how frequently have you visited (the
Web-based information system) over the past two months” and “on average, how much
time have you spent a day visiting (the Web-based information system) over the past two
months” as taken ﬁom Igbaria et al. ( 1997). At no time do the authors measure the other
half of their deﬁnition of habit, stable context. Context is assumed to be stable, but with
media habits like lntemet use may not necessarily need to occur in stable contexts after
the habit has formed. As such, while past behavior does appear to be a very strong
predictor of future behavior in this study, it cannot be said to be truly representative of
the authors’, or this dissertation’s, deﬁnition of habit.

Instead of using past behavior, a stronger, separate conceptualization of habit that
addresses its cognitive nature should be used. A clearly deﬁned construct of habit should

be able to surpass the problems inherent in past behavior and allow for habit to be

usefully integrated into reasoned action theories like the Theory of Planned Behavior.

30

 

hi;

 

 

' r:
...~~

 

 

ti:

 

Using a better conceptualization of habit and proving its validity is one of the goals of
this study.

An example of a better recent study of habit is Limayem, Hirt & Cheung (2007).
In their study, the authors examined general web usage by university students using
Bhattacherjee’s IS continuance model (2001). Limayem et al. disseminated self-
administered questionnaires to business students about their web usage, which was
followed later by two follow-up questionnaires. Like Kim & Malhotra, the authors took a
measure of past behavior (how often and how many times over the past 4 weeks), but
also used speciﬁc measures of habit (using the web is automatic/natural to me, when
faced with a particular task, using the web is an obvious choice for me).

The authors found that habit moderated the relationship between intention and
behavior, but their overall model did not explain as much variance as is usually expected
(26% versus 30% or more). One possible reason for the weaker effects than expected
may have been that the lntemet habit may not have been well established. While many
students may use the lntemet a great deal, some may have only recently gained high-
speed lntemet access at the university and may still be developing an lntemet habit.
Another reason for the weak variance explained may be that they did not measure a
speciﬁc web-related habit, such as YouTube use, and focused instead on broad habitual
trends about web use. The presence or absence of a stable context was also not measured,
which may have played a role for individuals with weaker habits. Limayem’s instrument
may not have captured all of the important elements of habitual automaticity relevant to
their target behavior, which could also explain why the variance explained was smaller

than expected.

31

 

 

HM .
.‘sis .

Pg“ v‘

i§‘-i

"P"

- h e¢-,

*0

9,.

\t ‘
3 _.u
- .i:
l... ‘
- '1
4t .4
' -»\

 

Despite these ﬂaws, Limayem et al.’s study can still be used to provide greater
insight to scholars who use other theoretical frameworks study habit. It shows that habit
can be measured in a conceptually valid way without using an experimental setting or
self-reports of past behavioral frequency, and demonstrates habit’s potential role as a
moderator in other theories of human behavior. With these assumptions in mind, a short
review of other Theory of Planned Behavior studies that have attempted to use habit may
be instructive in showing what mistakes may have been made previously, and how these
mistakes could be avoided in the future.

Two recent examples of media habit studies using the Theory of Planned Behavior

The ﬁrst study that will be discussed is Wood, Quinn and Kashy (2002). Wood et
al. examined cognitions and emotions and their relationship to habit. Wood et al..
collected data from undergraduates using diaries to compare their thoughts and emotions
when they were doing both habitual and non-habitual activities. The authors found that
thoughts were less likely to be relevant to behaviors when habitual behaviors were taking
place, and that emotional response to habitual behavior was lower than non-habitual
behavior.

Of interest to us is their claim that over half of their sample’s entertainment
behaviors were considered to be habitual (per their deﬁnition, performed repeatedly in a
stable environmental context), and that they speciﬁcally use the example of watching
television to illustrate some of the points they made in their study. Wood et al. explain,
that about half the sample felt that their television viewing was habitual, but half did not.
However, Wood et al. had to rely on direct self-report information about their behavior,

which may be prone to various types of unwanted biases.

For example, a subject may not be aware of their habitual behavior and believe it
does not exist, or may otherwise want to limit their admission of a habit for social
desirability reasons. This lack of awareness is to be expected with strong habits, as the
inaccessibility would be the result of the relevant schema being removed from a
conscious process of deliberation, and so asking a person a direct query about their habit
would not yield productive results. A subject’s estimate of how frequently they repeat a
habitual behavior is also impacted by the perceived importance of the behavior, as well as
how recently the behavior was executed (Verplanken, Myrbakk & Rudi, 2005).

Alternately, individuals who did not say habit may have been giving a socially
desirable answer, or incorrectly “remembering” reasons for use that may have been
operative once, but no longer actively guide their behavior. Recent research has also
found that intentions may follow actions rather than precede them, which may be
especially relevant to habitual behaviors (LaRose, 2004). Intentions also become less
accessible as a relevant behavior is automated, and individuals with strong habits may
believe very strongly in their intentions even if their behavior is much more inﬂuenced by
habit (J i & Wood, 2007).

Wood et al. may have also excluded media use behavior that occurred outside of a
stable context by their deﬁnition of “habit”, which may have missed other kinds of
television viewing (for instance, catching one’s favorite show at the bar because one’s
roommate has monopolized the television for a day). If a habit is strong enough, the
current environmental stability is not as signiﬁcant to the activation of a habitual
behavior as the strength of the habit itself. Stable circumstances appear to be most

important for new or weak habits, as previously discussed.

33

 

53' '°

.»._'_..

7n.- ‘!

 

'“Jw v]
..p ~ L

‘ :4“- o
.‘tggdﬁ
J

 

 

"u -

“H “.

 

For instance, a person may have a strong YouTube habit that manifests as
watching that day’s most popular videos after getting home from work or school. The
habit may have begun as a simple manifestation of a desire to be entertained, but the
person derived so much enjoyment from watching online videos that soon the act of
going to YouTube became automated. Every day, the person would come home, sit down
at their computer, and begin watching YouTube videos for a certain amount of time
before going on to other tasks. If that person’s lntemet is made unavailable at that usual
time, that person might be driven to attempt to ﬁnd other Internet access points in the
area (such as locating unsecured wireless routers, visiting a neighbor, etc.) because of
their need to maintain their habit.

A The preceding example shows an instance where a strong habit cannot be
executed in a person’s current environment. In that situation, a person may act to change
their environment to engage in the behavior (i.e. go to a different area than where they
customarily engage in the habitual behavior). The shift in setting may not be wholly
volitional, especially in the case of a strong habit. Such a phenomenon would
demonstrate the possibility for Ajzen’s perceived behavioral control to lose predictive
strength in a situation with a progressively stronger habit. Intentions should only be able
to counteract habit when circumstances change, the response from the habitual behavior
is no longer rewarded, or a person’s goals change enough to motivate them to resist their
habitual tendencies via “effortﬁrl self-control” (Limayem, Hirt & Cheung, 2007; Wood &
Neal, 2007).

Later, Wood, Tam and Witt (2005) examined the changes in habitual behavior

caused by a change in the environment; in this case, how college students who transferred

34

‘V‘.

.,‘
“he.
._ ‘
5“
.
'\\
.‘H
9.,
.3
.H‘
.i

 

\

.13..
\.r‘

to a new university continued or discontinued habitual behaviors. Data was collected by
self-report survey questionnaires. Questionnaire data was collected just before the
students entered the university, followed by a second wave ofquestionnaires collected
just after the beginning of the semester. Information about exercise, television viewing
and newspaper reading habits was collected in all questionnaires to allow for a
comparison across time.

Wood et al. found that the perceived change in context had more of an impact on
individuals who claimed to have a strong television habit than on those with weaker
television habits. This is a useful ﬁnding, but most changes in media context are much
less dramatic than a college student changing universities. Many factors may have
intervened to affect individuals with very strong television habits more signiﬁcantly than
those with a weaker television habit. Another ﬁnding of Wood et al. was that many
individuals in the study acted to re-create their previous viewing environment. These
individuals would assert active control over their environment, making their
circumstances more suitable to engaging in their desired behavior. In other words, the
students who changed universities would change their circumstances to continue
engaging in their old behaviors. This would allow them to perpetuate weaker habits, and
make their stronger habits more readily followed.

Also signiﬁcant for this discussion is the method Wood et al. used to measure if a
behavior was habitual. There is an acknowledged difficulty in habit studies in asking
someone directly “is this behavior of yours a habit” without encountering problematic
artifacts (Nisbett & Wilson, 1977). This discussion has already raised issues about social

desirability biases, the lack of awareness caused by strong automation of a behavior, and

potential problems with intentions not preceding behavior. Wood et al. were able to
address those concerns with their instrument to some extent.

Wood et al.’s instrument attempted to get around the problem of asking about
habit by using an instrument that measured habit strength by means of a simple 0 (never
perform) to 3 (perform every day) scale to measure frequency multiplied by another
simple scale designed to gauge how stable the circumstances were. However, behavioral
ﬁequency is a problematic measure of habit because of the way some habits may
function. Some very strong habits may only be done weekly, monthly or even less often.

For instance, a person may turn to the same Internet web site that discusses
politics during election season, use that site heavily during the election period, and then
not use the site again until the next election. Other behaviors may be done very regularly
but not become habitual, such as visiting a default home page whenever a person
activates their web browser but not necessarily interacting with the page at all. The
person does not choose to visit the page, but would likely claim they visited that page
very regularly because they have to do so whenever their web browser ﬁrst loads. Habits
can also be formed by active intentions to engage in a behavior repeatedly, such as
through the use of implementation intentions (e. g. Sheeran, Webb & Gollwitzer, 2005).
Returning to an earlier example, a person may have formed their habit to visit the
political web site out of a desire to stay informed, resolving to visit the web site during
the political season at a regular interval (daily, twice a day, etc.) until it became a
behavior that person did not need to actively consider any longer.

Based on the two recent examples used, this dissertation once again can reinforce

that many media behaviors are habitual. Learning about what media behaviors are

36

habitual can be challenging because of conceptualizations of habit that do not take into
account its automatic and subconscious nature. There are also social stigmas against
having some habits, as well as false recollections of rationales for behaviors that may be
habitual but were once volitional. Media habits may also occur outside of stable contexts,
which may lead to underreporting of some media habits.

The two studies discussed previously shared one attribute in common; both relied
on frequency-based measures of habit, and as a result may have over or under-
represented the presence of habitual behaviors in their sample. The use of only behavioral
frequency measures has been shown to be theoretically problematic for a variety of
reasons. Conceptually better measurement may be able to clarify the issue of habits in
general, and be of particular help to the study of media habits.

A re-conceptualization of habit measurement: The Self Report Habit Index

An instrument that has been proposed to address the measurement problems in
previous habit studies is the Self Report Habit Index (SRHI) (Verplanken & Orbell,
2003). The SRHI deals with the components of a habit in a more general way, focusing
on the uncontrollability, lack of awareness and efficiency that characterize habitual
behaviors. To accomplish this, the SRHI asks a series of questions about the behavior that
does not necessarily lead the subject to believe they are being asked directly if a behavior
is a habit (with all the cognitive baggage that entails). This permits strength to be gauged
without triggering other inappropriate responses from the subjects, and permits habit to
be tested independently of other environmental variables. Previous studies (e. g. Brug, de

Vet, de Nooijer, & Verplanken, 2006; Honkanen, Olsen, & Verplanken, 2005;

37

e- ‘ ‘v
.1. 3:.

[F

Dvw. ..

,.
u.-.

 

 

«I,»
—
«t.
‘>'
‘\ _.
-
.‘\
.
- I-t
L.x.
._ .
.
_
LA
- «1

1L.

Verplanken, 2006; Verplanken et al., 2005; Verplanken & Orbell, 2003) have supported
the SRHI’s content, discriminant and predictive validity.

The important contribution of the SRHI to the measurement of habit is its focus
on aspects of habit separate from frequency. While the SRHI does have troublesome
frequency-related items, the majority of the instrument examines elements of
automaticity that deﬁne habitual behaviors. Habits are by their nature uncontrollable,
automatic behaviors that are cognitively efficient and tend to be performed in stable
circumstances (at least initially). Habits are constructed within schemas, and these
schemas can be prompted by a variety of cues that we can receive consciously or
subconsciously. As a result, asking directly if a behavior is habitual is problematic for the
reasons discussed previously (misunderstanding of what “habit” means, social
desirability bias, a desire to be seen as competent and internally consistent, etc).

The SRHI addresses these issues by avoiding the word “habit” and focusing on
aspects of automaticity, asking subjects in different ways if their behavior is automatic, or
part of a routine that would be difﬁcult to disrupt. Addressing these aspects of habit gets
to the core of what the concept is without being reliant on only a “past behavior”
measure.

The SRHI also appears to discriminate well between behaviors that may be
habitual but are performed at differing intervals, such as daily and weekly habits
(Verplanken & Orbell, p. 1323). Returning to the election web site example, a person
may answer that they do not use the web site frequently, but they may answer that using
the web site is part of their routine of collecting political information. They may certainly

do the behavior automatically, or without having to consciously remember to visit the

38

web site, and those with very strong habits may answer that it may take effort not to visit
the page, and they would feel weird if they did not go there for their election information.
Because of the multiple overlapping items that discuss automaticity, it is easier for the
instrument to ﬁnd a habit that might not be repeated often, whereas an instrument that
relies only on frequency may miss the habit entirely.

Despite its apparent advantages over other types of instruments to measure habit,
the SRHI is not without conceptual problems of its own. While it does not ask for the
number of times a behavior is repeated explicitly, the instrument does ask if the behavior
is repeated as part of a routine, or if the behavior is done “frequently”. The former item
may be on shaky conceptual ground because of the lack of awareness that typiﬁes a very
strong habit (will a subject remember that their behavior is part of a routine?), while the
latter item encounters the same problem that all measures of frequency run into when
addressing habits that may not occur often but are nonetheless habitual. With these items
removed, the SRHI remains a functional instrument (Verplanken, 2006). Further, these
items do not entirely undermine the positive contributions the instrument makes in
measuring habits, because the instrument overall focuses on automaticity and
uncontrollability.

In summary, using behavioral frequency as a measure of habit is problematic
because some habits may only recur infrequently, and other behaviors that happen very
regularly may not be habitual. Fortunately, an instrument that may be able to address
these problems is the Self Report Habit Index (SRHI) devised by Verplanken and Orbell
(2003), but it also has a small number of problematic frequency-based items that should

be discarded.

39

 

 

be...

With a better measure of habit available to the study, the next task is to describe
what this dissertation wishes to examine in the Theory of Planned Behavior more
directly. While the entire Theory of Planned Behavior is useful to this dissertation, the
main variable of interest is perceived behavioral control. This variable is constructed
through an understanding of one’s control beliefs, which are fundamental to the Theory
of Planned Behavior. Perceived behavioral control has a direct role in the progression of
habitual behavior, and will be the focus of the remainder of the literature review.

The unique role of control beliefs

Control beliefs are discussed in the Theory of Planned Behavior (TPB), proposed
by Ajzen (1991). Control beliefs speciﬁcally refer to the presence or absence of resources
or opportunities. The more perceived resources or opportunities a person has, the more
control they feel they have over a given behavior or situation. Throughout the literature,
the nature of control beliefs and their interaction with other variables has prompted
vigorous debate. This debate has particular relevancy to our study because of its potential
contribution to habit.

Conner and Armitage (1998) discuss the interaction between past behavior and
control beliefs. Based on their meta-analysis, they determined that the effect size of the
past behavior and control belief interaction was very weak, with an r2 of only .02.
However, their analysis dealt with studies that examined self-reports of the frequency of
past behavior, and did not use our deﬁnition of habitual behavior, which is “a form of
automaticity in responding that develops as people repeat actions in stable

circumstances” (Verplanken & Aarts, 1999; Verplanken & Wood, 2006).

40

As previously discussed, the “past behavior” construct has problematic theoretical
baggage related to it, and is not as conceptually clear as later deﬁnitions of habit are.
Conner and Armitage note directly that . .we did not make the distinction here between
habit and past behavior that authors such as Triandis (1977) would argue for” (p. 1436).
This conceptual difﬁculty makes generalizing results found in early studies to be
difﬁcult, as a standard measure of habit is still not agreed upon within the literature.

A more direct discussion of habit took place in Aarts, Verplanken, and Van
Knippenberg (1998). In this article, the authors discussed a wide body of research about
the predictors of repeated behaviors. They questioned whether or not past behavior may
moderate the relationship between PBC and behavior. This question was partially
addressed by a later article by Verplanken, Aarts, van Knippenberg and Moonen (1998).
There, the authors attempted to use the TPB to predict travel mode choices, speciﬁcally
focusing on car use. Verplanken et al. also compared two habit measurements, one being
a self-report frequency of past behavior measure and the other a response time measure
(called “response-frequency” by the authors). Habit was found to have a statistically
signiﬁcant contribution to predicting behavior alongside other TPB variables, but the
interaction between habit and intention was only signiﬁcantly related to behavior when
habit was weak.

The ﬁndings of Verplanken et al. may begin to illustrate a trend within the
literature whenever habit is taken into account within a model directly. When habit is
weak or absent, behavior is still primarily under the volitional control of an individual.
Reasoned action processes are still functioning, and behaviors can be clearly explained

by looking at Theory of Planned Behavior variables. In the cases of the measures

41

Verplanken et al. used, measuring both schema accessibility and number of times a
behavior is repeated, a “weak habit” may actually only be a new behavior rather than an
actual automated process. In cases of a strong habit, the relevant schema should be less
accessible, and as a result may not be entirely visible within the instrumentation
Verplanken et al. used because of factors raised previously.

Elliott, Armitage and Baughan (2003) tested past behavior and its relationship
with TPB variables within the context of compliance with speed limits. Once more, the
TPB was supported, explaining a signiﬁcant amount of variance in the regression
equation. A measure of prior behavior was included that consisted of two 7-point Likert-
type items that asked subjects about their driving behavior in relationship to the speed
limit over the past three months. The measure speciﬁcally asked if and how often the
subject kept to the speed limit in “built-up” urban areas over the last three months, a
standard measure of past behavioral frequency. It was found that prior behavior
moderated the relationship between PBC and intention, but did not moderate the
relationship between the other TPB variables and. intention. Simple slope analysis
determined that the greater the role of past behavior, the weaker PBC became as a
predictor of intention.

Past behavior should not be considered perfectly analogous to habit. Measures of
habit have been shown to be independent of measures of past behavioral frequency
(Verplanken, 2006) .However, there is sufﬁcient similarity to how past behavior was
measured in Elliott et al. to other habit measures that have been previously discussed.
There is evidence within the study to demonstrate the potential moderating role of habit

within a framework of reasoned action. Control over the driving behavior of the subjects

diminished as past behavior increased. The people in the study developed a schema that
encouraged them to drive in a certain way, and when the habit was very strong, very little
could change their actual behavior. They did not feel they had control of their actions.
Conversely, it may be the case that people who may not have established driving habits
would be more receptive to interventions to encourage them to drive in a particular way.

The removal of perceived behavioral control from the driving behavior of
individuals in Elliot et al.’s study illustrates how habit can weaken personal agency over
one’s behavior over time. Initially, many behaviors are under conscious, volitional
control. In this case, learning how to drive takes a considerable amount of training and
practice before it becomes routine. As behavioral patterns are set in their ways, the
freedom to easily change those behaviors lessens. Automation processes act to encode
certain responses so to lower cognitive load on an individual, which result in some
behaviors going beyond the conscious control of a person. The customary way of doing
things becomes the only way of doing things, even if alternative behavior patterns would
work just as well. The barriers to changing the behavior are too high as a result of habit
making one way of responding far easier, leaving habit as the primary determinant of
behavior rather than PBC (by way of intention).

A contradictory account was provided by Armitage’s (2005) study of the
development of physical activity habits. Armitage examined the cumulative effects of
past gym attendance behavior on ﬁiture gym attendance behavior. TPB variables were
measured by standard means. Behavior was measured by gym attendance, and as the

Armitage study used a longitudinal design, past behavior could be observed over time.

43

Armitage found that gym attendance in ﬁiture weeks ofthe study was dependent
on past attendance in previous weeks. If a subject attended the gym for at least ﬁve
weeks, that early attendance would encourage gym attendance. It was found that the more
a subject attended the gym, the greater their PBC values became. Armitage found that
PBC was the most important statistically signiﬁcant independent predictor of exercise
behavior.

While the study focused on past behavior as an approximate indicator of habit,
and did not show a direct comparison of PBC levels at the baseline and follow-up data
points, Armitage found that successful past behavior tended to lead to an increase in the
mean value of perceived behavioral control in the follow-up ﬁ'om its baseline value,
while unsuccessful past behavior led to a decrease in the mean value of perceived
behavioral control. However, the effect of past behavior directly on the PBC construct
alone was small (a 13 of .49 at the follow-up versus a B of .46 at baseline) and may be
explained by error.

These seemingly contradictory results can be explained in the context of what this
dissertation has discussed so far in terms of response tendencies of individuals who have
strong habits. Individuals who dedicated themselves to going to the gym may have
remembered their initial control beliefs about being able to go to the gym and kept those
recollections as their behavior became habitual, an argument similar to Ajzen (2002a).
However, subjects who continued going to the gym a year after the baseline reading
likely had developed a gym-going habit of some strength, and maintained their behavior
because it would be more cognitively difﬁcult for them to stop going after a year of

steady attendance. The power of habit would compel them to maintain their behavior,

44

 

 
 

I.
“A.

\A'

-e

‘0

 
 

 

 

acting as a barrier to them developing perceived behavioral control to change or cease
their gym attendance. It is entirely logical that the subjects who continued going to the
gym gave strong positive answers to the perceived behavioral control item, as they
needed to be able to retroactively explain their irrational, automatic behavior rationally.

Unlike going to the gym, using YouTube is relatively easy for anyone to do.
Exercise offers numerous challenges to an individual, both external (traveling to the gym,
paying for a gym membership, and the act of exercising itself) and internal (concerns
about self-image, doubts about athletic ability, and apprehension about asking questions
about using equipment). YouTube use has few external barriers to anyone who has an
active high-speed lntemet connection. A person need only connect to YouTube to be able
to use the service. Internal barriers are much lower, as the use of YouTube is
intentionally uncomplicated. Any person with a functional web browser can use YouTube
without needing any additional expertise. Increasingly, personal electronic devices are
also capable of displaying YouTube videos, removing even the need for basic knowledge
of a web browser.

Control beliefs are the foundation of perceived behavioral control. Perceived
behavioral control is important to understanding habit, because how a person understands
external barriers or opportunities to engage in a behavior can inﬂuence their ability to
disrupt habitual behaviors. Many previous studies of control beliefs and habit have relied
on past behavior as a proxy measure of habit, which may be concealing the real nature of
the relationship. What has been apparent in most studies of control beliefs and past
behavior is that habit moderates the relationship between PBC and behavior. The reason

this moderated relationship is important is because if a person feels the external barriers

45

.

‘5

 

 

to changing their behavior is too high, they will be less capable of resisting or changing a
habitual behavior tendency.

One of the clearest studies of the PBC-habit relationship in recent memory dealt
with recreational drug use. This study can be used to more clearly illustrate the proposed
model of habitual behavior within the Theory of Planned Behavior.

Approaching an understanding of the role of habit within the Theory of Planned
Behavior

Orbell, Blair, Sherlock and Conner (2001) examined ecstasy use among young
adults in terms of habitual use and perceived control over taking as opposed to obtaining
the substance. It was found that ecstasy use was a Theory of Planned Behavior that could
be predicted using TPB variables. The introduction of a measure of habit which used two
automaticity-related measures (“Taking ecstasy is something I do automatically”;
“Taking ecstasy is something I do as a matter of habit”) added a small amount of
additional variance.

An immediate problem arises with Orbell et al.’s instrument with the use of the
phrase “as a matter of habit”. As previously discussed, misunderstandings of what “habit”
is by subjects may have made the item less helpful to the authors than the previous
automaticity question. Ignoring the “habit” question, there is only one automaticity-
related item to address the role of habit, which may not be enough to capture all of the
possible variance of the concept. Social desirability biases may have also thrown off the
results, as drug use is a much less acceptable behavior than watching television or using

the lntemet.

46

However, Orbell et al. observed that the effects of habit reduced PBC to non-
signiﬁcance in their regression equation. The moderating role of habit found by Orbell et
al. may provide a clue to how habit works over time. It appears to be the case that a weak
habit initially has little effect on reasoned action, but as habit grows stronger begins to
neutralize the effects of reasoned action variables until they are no longer viable
predictors of a behavior. Why does habit reduce the strength of reasoned action variables,
and how would this process occur?

One reason for habit’s ability to act against reasoned action variables may be the
nature of habit itself. It is more comfortable for people to follow a habit, and even when
the environment is disrupted, they will seek to rearrange the world to be more convenient
for their customary behaviors (e. g. Wood, Tam & Witt, 2005). The force of habit shapes
a situation, becoming capable of transcending the environment and its constraints as it
grows in strength. If the behavior is something easy to do, such as checking email or
visiting YouTube, repeating that behavior in a variety of contexts becomes elementary.
The automated behavior process enacts whenever the numerous cues that could prompt
the behavior appear, the actor making the environment more suitable to their habitual
needs if there are irregularities or problems. Reasoned action processes may have been
the starting point for the behavior, but as the behavior was repeated, standard automation
processes would remove the role of reasoned action from the behavior. Once a prompt for
the behavior appeared that could be followed within the context of the situation, the
behavior could be enacted without conscious thought — even if a person had behavioral
intentions counter to the habitual behavior (e. g. Baumeister & Heatherton, 1996; Wood &

Neal, 2007).

47

l

as teen st
neon 0t 1

‘ T ‘ I
1A

«:3 ."t n t

then-.1. 21.1:

‘rliH-JJ
S gr“ ‘&\
»

moi-l-‘
1‘1“;
MAL-10“

.o-‘ll'l‘a
, ,
t "MN 01

.4“

\i'gf'lll e.

The role of intention has been shown to diminish gradually as habit’s role
becomes stronger (e.g. Limayem, Hirt & Cheung, 2007; Kim & Malhotra, 2007), which
has been shown to occur once more in Orbell et al. ’5 ﬁndings. An overall enervation of
Theory of Planned Behavior variables as habit grows stronger makes sense in the context
of what habit is; an automatic behavior whose initial cognitions are made inaccessible by
cognitive efﬁciency processes.

The relevance of Orbell et al.’s ﬁndings to control beliefs and habit can be
understood by examining how recreational drug use appears to function. A person may
approach the behavior initially with a great deal of reasoned consideration. The initial
period of experimentation may lead the person to continue the behavior, reinforced by
positive outcomes. As time passes, it becomes harder to stop engaging in the behavior.
While there is a signiﬁcant biological component to this behavior, the cognitive aspect is
still important. It becomes more difﬁcult for a person to imagine not using the drug.
Barriers appear to ceasing use that did not exist before, and gradually a person’s
perceived control over their behavior disappears. They use the drug out of habit,
cognitive efficiencies appearing to remove the need for continual conscious thought
about the behavior. Certain contexts (locations, situations, time of day) act as prompts for
the habit to be initiated. A user may have ways of rationalizing their behavior to try to
hold onto the belief that their drug use is volitional, but in the end, they are dominated by
their habit.

This bleak view of how a drug habit forms can provide insight to the far more
benign and signiﬁcantly less dangerous domain of media habits. A person starts using a

medium like YouTube in a given way for some purpose; a goal (entertainment,

48

:ttmtio'
sore $211 a

writ:

it “”51 .

begtg‘me it

lifttitlr r

‘ 1
"P‘ (“.4

‘
I 'l
—'\..‘ .L‘J‘

3"""1'1".
Mmul

1
‘~ u-ul

- .7" .
J‘s. lit-t1
'S" 1T ll

“-51 u.C

information), or even without a particular goal in mind. The behavior is initiated, and
some reward for the behavior is generated that encourages it to be repeated. Information
or entertainment is provided, or perhaps YouTube only needs to provide distraction.
Either way, the use of YouTube is reinforced, and a person continues use of the service in
a given context (at work on lunch breaks, at home after getting back from work, etc). The
relevant context helps to encode the behavior into a schema that, as use continues, may
become automated. However, past a certain point of habitual behavior, the context of the
behavior no longer has the importance to activation of the habit as it once did, as
individuals ﬁnd ways of maintaining their habitual tendencies even in unstable
environments (e. g. Wood, Tam & Witt, 2005).

In the context of the Theory of Planned Behavior, a person’s attitudes, subjective
norm and perceived behavioral control over using YouTube drives their intention to use
YouTube. The intention to use YouTube, alongside perceived behavioral control, then
predicts their ongoing use of YouTube. This dissertation proposes adding a fourth
variable, habit, which moderates the interaction between PBC and behavior.

A YouTube habit may begin to form over time. The cognitions that are part of the
Theory of Planned Behavior begin to become less accessible as automation processes
begin to take place. Signiﬁcantly, perceived behavioral control, probably the most
relevant aspect of reasoned action to disrupting habits, is also weakened, helping to
ensure that the habit can continue to grow in strength over time. While the original
cognitions behind using YouTube may be falsely “remembered” if a person is asked

about why they engage in their YouTube use, they become less and less important to the

49

~KF"1 “31'1"

I
JUL“: qu\ “1'

22.15 ot‘et

' l . -1
331.11 307.11 1

1'5; '- 63 IT

3: the tutor

3] I 1".le

Laws», [0‘

#:17-

l
‘ ""- l’f(\(\

11‘ ‘E‘Jce‘u a

,.
. l
\",-.

actual initiation of the behavior until, if the habit becomes strong enough, they are totally
irrelevant.

When a YouTube habit is very strong, it is difficult to disrupt it. The normal
means of changing behavior by altering attitudes, subjective norm or especially perceived
behavioral control is less effective because of the entrenched nature of habit. A person
may very much want to change their behavior, but be unable to because of the efﬁciency
of the automated behaviors (Wood & Neal, 2007) and the rapid depletion of active
control over behaviors in “real life” (Muraven & Baumeister, 2000). Altemately a person
may believe their habit continues to reﬂect their active intentions, even if their habit is
counter to what they actually intend to do (Ji & Wood, 2007).

Hypotheses

The Theory of Planned Behavior and habit

The Theory of Planned Behavior has been discussed extensively in this
dissertation. Many scholars have proposed to add new indicators to the theory, such as
moral norms, self-identity, and past behavior (Conner & Armitage, 1998). Ajzen (1991)
has claimed that any new predictor that would capture a “signiﬁcant proportion of
variance” after the current variables are taken into account would be a welcome addition
to the theory.

Habit has been a problematic construct in the Theory of Planned Behavior. Many
scholars treat it as an aspect of past behavior, which is problematic for many theoretical
and methodological reasons previously discussed (Ajzen, 2002a). Attempts to include it
in the theory have typically not accounted for its unique interaction with variables within

the theory, resulting in contradictory or illogical results.

50

r-\‘-' 1‘
.. r\
L\\— a>\.

t _‘

..—- 47-0..
4 t

”Jul .rluu
..

113.1,}
‘1 .
.131. Lin

1?; ‘r .o ‘\r
..sut\.'..

~11va u ".
«Nubkt u

t”..‘r‘" r

hu\n.-‘\ 1..

.7... _
\_ll‘ '9,‘
~1...A¥. .

 

.h, ,
.1 t . ‘
."L‘Le\

If one goes outside of the Theory of Planned Behavior, one may be able to get
insight as to where habit may be best placed within the context of TPB. Newell (2003)
used uses and gratiﬁcations theory in his models of habit, and pr0posed in his ﬁnal model
that gratiﬁcations guided both habit and intention, and intention in turn also drove habit.
Habit and intention then led to future behavior, but intention was a much stronger
predictor of future behavior than habit. It may be the case that habit manifests as an
indirect inﬂuence on the main TPB variables rather than as an independent predictor of
intention. The reason for this is that habitual action, once a behavior has been automated,
is not reasoned. However, the initial selection of the habitual behavior may have begun
through a process of reasoned action. Habit has been noted to have an effect on intentions
(e.g. Bentler & Speckhart, 1979; Ji & Wood, 2007; Newell, 2003; Newell & LaRose,
2004; Wood & Neal, 2007), but its most interesting relationship in this dissertation is its
interaction with perceived behavioral control and behavior.

This dissertation supports the standard model of the Theory of Planned Behavior;
attitudes, subjective norm and perceived behavioral control should create intention, and
PBC should also guide behavior. These linkages have been proven time and time again to

be reliable, and should still apply in the study.

51

 

3., .
v ‘t 3" .
._ .
s- lti
-,

 

Hlal: Attitude about media use will be positively related to intention to
engage in media use.

Hla2: Subjective norm about media use will be positively related to
intention to engage in media use.

H1a3: Perceived behavioral control about media use will be positively
related to intention to engage in media use.

Hlblz Intention will be positively related to media use behavior.

H1b2: Perceived behavioral control will be positively related to media use
behavior.

What is the relationship between habit and PBC?

Many habit scholars have examined the role of control beliefs or the PBC
construct on habit, but have found seemingly contradictory results. Godin, Valois and
Lepage (1993) found intention and habit inﬂuenced behavior directly in the general
population, with PBC and habit only inﬂuencing behavior through their impact on
intention. Could habit be caused in some way by intention? Logically, habit has to have
some relationship with intention as an antecedent condition through the perspective of
reasoned action. Before a habit can be formed, a given behavior needs to be reinforced
enough to make it worth repeating. One type of reinforcement is goal achievement. The
process of achieving a goal (such as exercising) allows a person to determine if a
behavior pattern is going to meet with fruitﬁil results. If the behavior is successful, that
behavior becomes more likely to be repeated. In a weak habit, active reasoning still leads
to intentions driving behaviors. As habit becomes stronger, the intention loses predictive

power, and the habit becomes most important.

52

Godin et al. found in their study of pregnant women that habit influenced
behavior directly, while intention did not have a direct inﬂuence. Instead, attitude, PBC
and habit inﬂuenced behavior only through their impact on intention. However, it may be
the case that the disruption of old routines and the radical change in goals on the part of
the pregnant women led some habits to be retained (such as exercising or not exercising)
while other habits were necessarily discarded. Some of these discarded habits may have
been reliant on a context that no longer existed, while others may have been disrupted by
renewed perceived behavioral control as a result of environmental changes. In any event,
more reasoned action was taking place on the part of these subjects while they learned
how to adapt to their new life situations.

Using these two studies, a situation can be proposed in which a person in the
general population may have a moderate exercise habit which is still inﬂuenced both by
intention and habit. The full automaticity of habit has not yet been engaged, but neither is
the behavior entirely reasoned. After the ﬁrst few repetitions, it is clear to the person that
this behavior will yield a desired outcome, so thinking through the behavior completely is
no longer entirely necessary. Participating in exercise requires an expenditure of a
nontrivial amount of energy and time, which may be the reason this behavior has a mixed
result. The role of volitional action and automaticity appears to have equal importance in
this case. However, pregnant women may also be inﬂuenced to act based on their
previous habit. With their previous assumptions and actions altered by their new
situation, they look to past actions that have garnered success for them in the past to help
restructure their lives. If the habit was strong, they maintained the activity even if

environmental context changed, but if the habit was weak, the activity was discarded. A

53

marten."

‘.

I}

.\..

5k

\U

similar example found in the mass media might be an instance when a television schedule
is radically disrupted by a broadcaster. Programs one watched habitually may have been
moved to different days and times, and only strong loyalty to a program would drive
them to seek out the new placement of their chosen programs to view them.

This trend was seen in a different behavioral realm in a more recent study. Elliott,
Armitage and Baughan (2003) examined the compliance of drivers in the United
Kingdom with speed limits over two points in time (N = 598). As previously discussed,
Elliott et al. used the Theory of Planned Behavior, and measured past behavior in their
analysis. TPB variables added a signiﬁcant amount of explained variance in explaining
behavior. Elliott et al. also determined that PBC lost predictive power when past behavior
had a stronger role in guiding behavior. As most of the sample was made up of people
who were experienced drivers, with only a minority being inexperienced drivers, these
results may only speak to those with strong driving habits. For instance, examination of
inexperienced drivers may show just the reverse -- habits have not formed, so the role of
PBC should be much higher in inﬂuencing driving behavior.

A YouTube-related example of a similar situation would be to compare
experienced YouTube users to novices. People who have used YouTube for a long time
have developed response tendencies created by their habitual use of the service; either by
initially volitional search behaviors or chance behaviors created by interface design.
Experienced users do not need to actively think about how to get what they want out of
the service because they know how to get the interface to produce the reward they ﬁnd
valuable, whereas novices who have not yet developed YouTube habits have to

concentrate on what they are doing to get what they want out of the service.

54

‘ir

a}

I . 4
- .‘ ‘1
.4»- s A,

vC

‘-\~1
\

‘\.Qh

A similar ﬁnding was found in Mahon, Cowan and McCarthy (2006). Mahon et
al. examined the consumption of take-out food and ready meals using the TPB and a past
behavior based measurement of habit. They used a representative random sample of
individuals in Britain (N = 1004), and unlike other studies did not ﬁnd the expected
relationships among the TPB variables. Habit was the main item that signiﬁcantly
predicted both intention and behavior to eat ready meals (R = .509 for intention, R = .675
for behavior), with attitude being the only other signiﬁcant predictor (R = .079 for
intention, R = .656 for behavior). This trend continued in the prediction of intention to eat
take-out food (R = .522 for habit, R = .078 for attitude) and behavior (R = .403 for habit,
R = .381 for attitude).

It may be possible that the consumption of ready meals and take-out food can be
perceived as a strong habit for many people. It may be the case that the actual habitual
process is the act of going somewhere for lunch. Abelson (1981) may interpret the event
of eating fast food “habitually” as an activity that takes place within the “lunch script”.

The act of purchasing fast food from a drive-through window or a stand may have
become integrated into someone’s lunch script because of the availability of some types
of lunch food in the immediate environment over others. Initially, this decision-making
process may have been reasoned, but as time passes, the integration into the “lunch
script” removes the active reasoning component of the decision. Whatever lunch
selection is ultimately the habitual choice was rewarded at some earlier point in the
formation of the script as a result of active thought processes. Once the habit has been
fully integrated, the reward is assured, and there is no further need for active thought on

the matter. In other words, active thought processes may still exist, but as the “lunch

55

script” runs, habitual choices will override the necessity for volitional action as rewards
for behaviors are administered.

On the speciﬁc subject of perceived behavioral control, there may be an initial set
of external opportunities and/or barriers that inﬂuence a person’s volitional choice for
how to compose their “lunch script”. Some restaurants may be far away, too expensive or
not have the type of desired food a person wants; other outlets may meet these criteria
more easily. A person acts within these external constraints volitionally and chooses
where they get lunch, and then is either rewarded or not rewarded for their choice. As
time passes, and that restaurant is selected repeatedly, the need for consideration of those
external barriers or opportunities is diminished. The habitual process of going to the same
place within the “lunch script” again and again removes the need for consideration of
PBC until it is not useﬁrl as an indicator any longer.

Compare the process of ﬁnding a fast food restaurant to how individuals may
choose what YouTube content providers to subscribe to. Rather than continually seeking
out new content through lengthy YouTube searching, an initial period of exploration may
lead to a group of favorite content producers that will be visited repeatedly. The objective
to ﬁnd content providers that provide entertainment to a person’s tastes requires active,
volitional searching. Eventually a roster of providers that generate content preferred by
the user is found, and the active search process ends. The role of perceived behavioral
control diminishes as the challenges to ﬁnding desired content cease being relevant. The
user has found what that person wants, and the external barriers to locating the desired

content (not knowing what people to subscribe to) are no longer important.

56

5,?

511110

6v

' ‘33
.

\

~\

~'\

-A

313*

 

11“,“,

‘1‘

A more problematic case of unexpected ﬁndings occurs in Knussen, Yuke and
MacKenzie (2004). Knussen et al. examined intentions to recycle household waste
among the population of Glasgow (N = 252). They also used the TPB as their theoretical
construct, but added both a measurement of past behavior and a measurement of
perceived habit. Past behavior was measured as a series of four Likert-type items that
asked about the proportion of a speciﬁc type of household waste that a person had
recycled over the past three months, while perceived habit was calculated using four
Likert-type items that asked them to rate how much recycling a certain type of waste was
"a habit". While only 39% of individuals who recycled most or all of their waste claimed
to have a strong habit, this ﬁgure may not be entirely accurate. An alternative instrument
that asks about how automatic their recycling behavior is, such as the SRHI, would be
able to clarify how many people in the sample actually had a conceptually clear recycling
“habit”.

The reason more clarity in both conceptualization and instrumentation is
important in this case is because of the common perception of the term “habit”, which has
been discussed at length during the course of this dissertation. To brieﬂy reiterate
previous discussions, asking if someone has “a habit” without any measurements of
automaticity may prompt either false positives (in the sense of someone believing that
they have a recycling “habit” even if they have only started to engage in the behavior as
part of their daily/weekly/monthly/etc. activities and still consciously think through why
they are executing the given behavior) or false negatives (in the sense of someone having

such a strong recycling habit that they do not consciously process recycling, but know

57

they recycle — and therefore impute that they recycle their waste volitionally by post hoc
explanation).

In addition, Knussen et al. found that TPB variables were the greatest predictor of
the intention to recycle household waste (.R2 = .34). The next most important predictor
was past behavior (R2 = .54, R2 Change = .20), while perceived habit provided only a
trivial increase in prediction (R2 = .55, R2 Change =.01).

There is a signiﬁcant theoretical. insight available to us in this study. Knussen et

al. claim:
...in the current study, the consistency between past behavior and
intention was higher for those who had recycled none of their waste,
or who had recycled most of their waste, as it was for those in the middle
group, and overall, the strong correlation (r = .67) between past behavior
and intention is suggestive of a relatively stable situation, in line with
Ajzen's (1987, 2002) position on the role of past behavior in the TPB (p.
244)

This appears to provide us with another argument in favor of a negative
relationship between habit and PBC. In a situation where there was no habit, the role of
PBC is very strong. The no-habit group could speak to people who were only just

beginning to recycle their waste, or individuals who were not recycling their waste at all.

(It is also possible that individuals who did not recycle did not recycle out of a habit.)
New recyclers who had to consciously think through their recycling behavior if they were
going to choose to recycle, or individuals who actively decided not to recycle, are still

using volitional control over their actions to regulate them. This differs from individuals

58

who have a habit of not recycling they are not motivated to disrupt, as they may not want
to alter their behavior for a variety of reasons, foremost among them likely ease.
However, Knussen et al.’s focus is on recycling behavior, so we are uncertain as to how
these other groups may have behaved.

Compare recycling behavior to media behavior. Recycling behavior has a great
many external barriers (sorting waste into different containers, putting out containers on
the appropriate days for collection, etc.), while media use behavior has far fewer inherent
barriers to its own use. YouTube in particular has almost nothing in the way of itself and
the user except the need for a functional high-speed lntemet connection of some sort.
YouTube can be used on even slow computers, and can increasingly be accessed on
PDAs and cell phones. It should be generally very easy for the average person to use

YouTube, whereas recycling takes special effort to learn how to do correctly.

In Knussen et al.'s study, the majority of those who did not recycle did not claim
to have a recycling habit. As a habit is established, it would appear to be logical to
propose the role of volitional action declines until a threshold is reached in which the role
of habit and PBC is approximately equal. While this establishing step in the PBC-habit
relationship is not clearly measured in the study, it appears to be a valid assumption based
on the endpoint Knussen et al. found of a strong past behavior-intention linkage for
individuals with strong recycling habits. PBC is strongest when habit is weakest, and
PBC is weakest when habit is strongest.

As habit becomes very strong, people have less and less perceived control over

their actions. Alternately, a very strong habit does not permit barriers and limitations in

resources to stand in the way of enacting a habitual behavior. One could imagine the two

59

, spit.
(it. \V
.\ .

V

uh

. 1‘)

Jk“\\.
.L

J.
N‘DV-
A“

4*, \ .-

very different patterns of someone just learning how to use an internal email system at a
new job versus someone who had used that system for a long time. The novice has to
exert cognitive effort to check their email, but as their skill increases, the task becomes
simpler and requires less conscious action. A point may be gradually reached in which a
person chooses to check their email at certain times of the day, which becomes part of the
behavioral script (Ableson, 1981). As perceived control declines, habit strengthens until
the veteran user checks email at different points in the day without giving the task any
conscious thought. Even if something gets in the way of checking one’s email, additional
action can be taken by a veteran user to attempt to check it by other means (for instance,
accessing email through their cell phone) that may not necessarily be made in the case of
another type of lntemet outage.

While Knussen et al.'s study of recycling may have shown less strength of habit
than may have actually been present; there are other situations in which the role of
perceived behavioral control can be more clearly seen. As previously discussed, Orbell,
Blair, Sherlock and Conner (2001) examined youth use of ecstasy in Britain (N = 84).
Beyond the standard TPB variables, they added two automaticity-related measures of
habit: "Taking ecstasy is something I do automatically" and "Taking ecstasy is something
I do as a matter of habit", both 8-point Likert-type scales. Further, Orbell et al. examined
both the perceived behavioral control of obtaining ecstasy and the perceived behavioral
control of taking ecstasy.

Orbell et al. found that the model with the standard TPB variables of attitude and
subjective norm, as well as both PBC variables and their measure of habit explained a

signiﬁcant amount of variance (R2 = .86). However, the addition of habit, while

60

 

statistically signiﬁcant, did not add a large amount of new variance explained. Instead,
the importance of these ﬁndings is the ﬁirther evidence they provide to support a
potential negative model of the relationship between habit and PBC. Even though the
contribution of habit to the overall model was small, Orbell et al. found that habit
declined as control over taking ecstasy increased.

Strengthening PBC may be a reliable way of ﬁghting the inﬂuence of habit, while
weakening PBC would leave more room for habit to inﬂuence behavior. It is possible
based on this relationship that PBC could also be an inefﬁcient measure of self-
regulation. Bandura conceptualized self-regulation as having three steps; self-
observation, judgment and self-response. PBC addresses the ease or difficulty of
performing a behavior, which appears very similar to this self-regulatory function. For
example, a person may be uploading a video to YouTube for the ﬁrst time, and would
need to create a personal proﬁle to describe themselves to others. The user would need to
determine if ﬁlling out this proﬁle was consistent with their previous behavior, and then
compare what they would want to put into their proﬁle with the proﬁles of others. The
objective of ﬁlling out the personal proﬁle could then be completed or not completed
based on the cognitive process of self-reﬂection and comparing what one could produce
with existing proﬁles on YouTube.

Similar results to Orbell et al. were found by Verplanken, Aarts, van Knippenberg
and Moonen (1998). Verplanken et al. examined travel mode choices among residents of
a small Dutch village (N = 200). TPB variables were measured, as well as habit strength.
In this case, habit strength was measured both with a self-report measure of behavioral

frequency contained in two items (how frequently the car was used inside and outside the

61

, C
if.— .1631 ‘3
171/)”;
51‘
.9
20":

r .. p‘
\1 1"
.. L,

village), as well as a response-frequency measure of habit strength. The “response-
frequency” measure used by Verplanken is in actuality a measure of response latency,
which may be confusing to the reader. To assist the reader in distinguishing between
these measures, the addition of the word “latency” will be made at the end of the name so
to help to distinguish between the self-report measure of habit and the response latency
measure.

The response-frequency (latency) measure of strength presented the subjects with
15 imaginary trips that varied in distance and destination, and were asked to answer as
quickly as possible what type of travel mode they would use to reach each destination.
Consistent answers across trips were used to determine the strength of habit of using that
given mode. Adding the time constraint was intended to encourage responses from the
relevant travel schema rather than an invented answer created to please the experimenter.

Verplanken et al. found that the pure self-report measure of habit was not as
useﬁrl in the model as the response-frequency (latency) measure. Using the RF (latency)
measure with intention and perceived control generated a model with a multiple R of .31.
Overall, habit was a statistically signiﬁcant contributor to predicting behavior, but its
interaction with intention was only signiﬁcantly related to behavior when habit was not
strong. Actions, such as choosing to purchase a newspaper in the morning, are driven by
intention when an automatic behavior pattern is not in place to regulate action. The SRHI
later developed by Verplanken and Orbell was validated against this measure, helping to
support its overall validity (Verplanken, 2006).

These ﬁndings are another example of the overall trend in the literature. "When

habit is weak, intention predicts future behavior signiﬁcantly as habit strength

62

 

tu‘

«\-

a\..

E.

 

\

increases, the predictive power of intention decreases, and becomes non-signiﬁcant at
moderate and strong levels of habit strength" (p. 1 19). In other words, intention, partially
driven by PBC, will be less effective at prediction as habit increases in power; the
converse should be logically true. At the same time, PBC’s direct linkage to behavior
indicates that even if intention may be weakened by habit strength, actual behavior may
still be inﬂuenced by habit through PBC.

Based on the literature reviewed so far, there is a reasonable amount of theoretical
evidence to support PBC being moderated by habit. Empirically, there is also support for
this idea, though it has not been conclusively tested in a known study. An individual may
begin using YouTube by being very discriminating and attentive to what one selects,
considering the environment around them and choosing what videos to view or leave
comments on based on their own beliefs about their mastery of the world around them;
for instance, being able to determine quickly whether or not a video will be interesting to
them.

As YouTube use continues and becomes part of one’s behavioral scripts
(Abelson, 1981), the role of active reasoning becomes less important. Automation
processes used to make one’s behavior more cognitively efficient take over, removing the
need for active thought from enacting a given behavior. As a result, as habit becomes
stronger, reasoned action processes diminish in strength. Of particular interest here is
how perceived behavioral control weakens, as external barriers to action either act to

constrain a person to stay on a given behavioral path, or become irrelevant to execution

of habitual behavior.

63

We propose this hypothesis:
H2: PBC will be moderated by habit in predicting behavior. As habit
strength increases, PBC should become a weaker predictor of behavior.

What is the role of consistent context in media habits?

The studies discussed so far have raised the idea of consistent context, referred to
in the course of this dissertation as “environmental stability”, being important to habit.
Environmental stability, in the view of some scholars, should allow for the activation of
schema below conscious awareness, bringing about habitual behavior (Ouellette &
Wood, 1998). The potential power of the environment in this perspective has encouraged
scholars to study the potential negative impact of the environment in habitual behavior,
particularly in regards to media habits (e. g. Rosenstein & Grant, 1997; Schmitt, Woolf &
Anderson, 2003). There is some evidence that the environment does have an impact on
habitual behaviors, as changing the environment may make it easier to break ingrained
habits (e.g. Heatherton & Nichols, 1994; Quinn & Wood, 2004; Quinn & Wood, 2006;
Wood, Quinn & Kashy, 2002; Wood, Tam & Witt, 2005).

What is less clear from the scholarly work focused in this area is how much
impact environmental stability has on media habits. The literature so far appears to
indicate that altering the environment does have some impact on weakening or
eliminating habits, with Verplanken and Wood (2006) going as far as to claim that
control over the environment is “. .. key to the success of interventions designed to
change everyday habits and maintain new behavior” (pg. 95).

Even so, media habits do not appear to be as context-dependent as other types of

habitual behaviors because of the ease of using media in a wide variety of situations (e. g.

64

Ajzen, 2002a; Verplanken & Orbell, 2003). It may be that context is only important to
media habits early in their development, and not so much once they are established.
Testing this assumption would be useful to clarify discussion of media habits, and may
lead to a better explanation of what the real role of context in habitual behavior is.
Therefore:
H3: Habit will have a stronger effect on behavior when context is stable
than when context is not stable.

Will the relationship between habit strength and self-efficacy change YouTube use
intentions?

A related hypothesis to Hypothesis 2 is that YouTube use should be directly
related to YouTube habit strength and YouTube self-efﬁcacy. Individuals who habitually
use YouTube should not feel they have a great deal of personal control over using
YouTube. While they may have more control over the external factors that allow for
YouTube use, such as having access to a computer, a connection to the lntemet, and
programs on their computer that allow them to view the web or use other lntemet
services, their internal, self-regulatory factors to regulate their usage of YouTube may be
diminished.

Instead of a regimen of volitional control over their behavior, habitual YouTube
users are guided by behaviors that were encoded originally in pursuit of a goal (pleasure,
distraction, etc). The behavior has since become a goal in and of itself, resulting in a
repetitive behavior pattern. This is often thought of as being only referring to people who
use large volumes of media, but could also apply to someone who only visits a certain

group of web sites but does so automatically.

65

Following from this reasoning, people who do not have a strong YouTube use
habit should be guided more by active agency. They are in control of their actions, and
may be seeking out content to interact with for a variety of reasons. If they continue to
seek out that content, the potential for a YouTube habit may exist, but in situations where
there is low or no habit, volitional action should guide behavior.

The literature supports the idea of media consumption behavior in general
working in this way, but there has not been a test of this relationship in the media

literature itself. A direct test of this idea would be useful to expanding scientiﬁc

knowledge in the area of media consumption habits. Therefore:

H4: Self-efﬁcacy will be positively related to intentions when habit is low.

Self-efﬁcacy will become less related to intentions to use as habit becomes

moderate to high.

A model of habit in the Theory of Planned Behavior

Figure 2-2 is an illustration of the hypotheses listed in this chapter. The original

Theory of Planned Behavior hypotheses have been included along with the new

hypotheses provided by this dissertation. Habit has been included as a moderator, along

with consistent context and self-efﬁcacy.

The new hypotheses have had their arrows bolded to differentiate them from the

usual TPB model.

66

\uv
- Ml»

 

Attitude H 1 a1

 

 

 

 

 

 

 

 

Subjective H132 _ Intention Hlbl Behavior
Norm

 

 

 

 

 

 

 

 

    

Hla3

H4
PBC
H 1b2 l
H2 Habit

Consistent
Context

H3

 

     

 

 

Figure 2-2. Model of the proposed hypotheses.
Summary
Habits are contained within schemas. The activation of these schemas has been
claimed by some scholars to occur as a result of environmental prompts. Other
researchers have claimed that schema activation is primarily a result of an individual's
relevant goals. Concern over how schemas can activate is the primary dilemma faced by
researchers who study habits that cause unwanted consequences. A more constructive
social science framework for these undesirable habits is to examine them in the context of
being cases of deﬁcient self-regulation (LaRose, Lin & Eastin, 2003).
Confusion in the habit literature may be the result of ambiguous
conceptualization, which in turn leads to problematic instrumentation. Instruments that
only measure frequency of a behavior may miss habits that do not recur regularly in the

environment. Similarly, asking if a behavior is a habit may prompt subjects to respond

67

«.1.

inappropriately, creating false cognitions, prompting social desirability biases, or
otherwise generating error variance that clouds the real effect of habit. Strong habits
make cognitions related to the behavior less accessible because of schematic automation
processes, and instruments that do not attempt to measure automaticity may miss key
signs of a habitualized behavior. The Self-Report Habit Index (SRHI) devised by
Verplanken and Orbell (2003) appears to be a better instrument to measure habits, but
still contains problematic items that ask about frequency.

The Theory of Planned Behavior is a reasoned action theory that has been used in
the past to attempt to account for habitual behavior. A variable that is usually used as
habit, past behavior, is problematic because it contains many other aspects of the
environment outside of automated behavioral processes. The variable this dissertation
focuses on is perceived behavioral control (PBC) and its role in the development of
habits. PBC is created by control beliefs that deal with external factors surrounding a
person's ability to successfully complete a behavior.

Several recent TPB studies (Armitage, 2005; Orbell, Blair, Sherlock & Conner,
2001; Verplanken, Aarts, van Knippenberg & Moonen, 1998) have presented evidence
that appears to indicate that as habit becomes stronger, intentions lose their predictive
power. This weakening is particularly important in the case of PBC, which determines
obstacles to completing a behavior. If habit acts as an obstruction to a person's perceived
control over their behavior, changing or ceasing an unwanted behavior may seem to be
impossible until the effect of habit is confronted.

Overall, the literature supports an understanding of habit as acting as a moderator

to Theory of Planned Behavior variables. As habit grows stronger, Theory of Planned

68

w»

Behavior variables become less useful as predictors of behavior. A particularly important
linkage in the Theory of Planned Behavior to focus on is how habit moderates the role of
perceived behavioral control, which helps to ensure that the habit is perpetuated by
creating more external barriers to changing the behavior.

The goal of this study is to examine the role of habit in YouTube use, a common
lntemet behavior among college students and young people in general. This study seeks
to do what other studies have not done by using a strong conceptualization of habit to test
its effect on Theory of Planned Behavior (TPB) variables under conditions of weak,
moderate and strong habit. Directly examining the relationship between habit and
reasoned action variables, particularly perceived behavioral control, will provide insight
into the unusual ﬁndings found by other scholars. The ﬁndings of the study may act to
conﬁrm habit’s place in the TPB as a potential moderator variable.

The methodology for this study will be described in the next chapter.

69

Chapter 3

Methods

Methods

Target behavior

The general interest of this study is lntemet use. The behavior of interest to this
study is the use of the YouTube video sharing site, a popular web destination that has
become one of the most widely known free streaming video providers on the lntemet.

It was anticipated at the outset of the study that YouTube would be a behavior
college students would regularly engage in. YouTube is very popular with young adults,
and has a presence across almost every signiﬁcant demographic group who uses the
lntemet worldwide. Because of its ubiquitous presence on many popular web sites that
target college students such as MySpace and Facebook, it was anticipated that there
would be individuals in the sample who would have developed habitual YouTube usage
patterns of some strength.

Beyond the simple probability of habitual YouTube use in our target population,
YouTube has similarities to television, another media behavior that has encouraged
strong habitual tendencies in the past. Like television, YouTube allows users to, after
selecting a clip, sit back and watch without having to do any additional work. It is very
easy to go from clip to clip, requiring little mental overhead. The ease of the behavior
may encourage routinization, then habit formation. YouTube may become the “obvious”
choice for online video for its users because of its ease of use, and eventually even people

who initially were engagedin rational searches for video content may become acclimated

70

to how YouTube presents content. Strong design features may allow people to quickly
integrate YouTube into a video-seek and/or video-watch schema to replace volitional
processing with habitual goal-seeking processing.

YouTube is also different from television in important ways. The interactive
elements of YouTube add another layer to the use experience that television does not
have. Users can leave cements on video clips, rate the quality of a given clip, or
respond to a video with a clip of their own. Users may be more engaged by these
behaviors, which in turn might encourage more habitual use by giving users more
incentive to return to the site again and again, or might weaken habitual use by making
YouTube use a very active, volitional process.

Survey overview

An online survey was used to collect data about YouTube usage patterns by a
student sample recruited at a large Midwestern university. Variables of interest to the
Theory of Planned Behavior (attitude, subjective nomi, perceived behavioral control,
intention and behavior) as it pertains to YouTube use were measured. Additionally, two
measurements of habit regarding YouTube use were used; a past behavior measure of the
type used in previous habit studies and the Self Report Habit Index (SRHI, Verplanken &
Orbell, 2003). Standard demographic data was also collected.

Recruitment for the sample of students took place through an email list acquired
from the Bursar’s ofﬁce. A sample of 1200 undergraduate students out of 36,377 possible
undergraduates (as of Fall 2008) received emailed invitations to participate in the study.
Two reminder emails at the beginning and at the end of the week were sent after the

initial invitation to encourage participation. Additionally, participants who chose to take

71

the survey were entered into a rafﬂe to win a gift certiﬁcate to Amazon.com. Entering the
rafﬂe required consent to be re-contacted the following week for a follow-up
questionnaire, with two reminder emails sent to those participants at the beginning and at
the end of the week to take part in the follow-up.

Previous work (Ellison, Steinﬁeld & Lampe, 2007; LaRose, Kim, & Peng, 2008;
Lange & Lampe, 2007) had observed response rates to online surveys about Facebook,
another socially-oriented web site, of approximately 25% to 35%. A comparable response
rate in this case would have yielded an expected subject pool of at least 500 and possibly
up to 700. The collected information was used to re-contact the subjects for a follow-up
about their actual YouTube use behavior.

Desired sample characteristics

The study targeted YouTube use, and was targeted at YouTube users. A ﬁlter
question was placed at the beginning of the study that asked if a subject uses YouTube. If
not, demographic data was collected from the subject to determine a proﬁle of non-users
for comparison purposes.

Assuming a moderate effect size (d = .5) and an alpha level of .05, a power level
of .8 would be possible if there were 76 subjects per group (no/low habit, medium habit,
strong habit), or 228 total subjects (Cohen, 1992). If the comparison was only between
low habit and medium/high habit, there would only need to be 67 subjects per group for a

total of 134 total subjects to achieve .8 power under the above parameters.

72

Operational definitions
Theory of Planned Behavior instruments

Attitude toward YouTube use, subjective norm pertaining to YouTube use, and
perceived behavioral control over YouTube use were measured. The items were in
semantic differential form as is common in the prevailing Theory of Planned Behavior
literature. Beliefs relevant to the variables were elicited from a class of
telecommunication undergraduates at a large Midwestern university, and were used to
create each item.

All recoding and reﬂection instructions for the following scales appear in the next
section, “Recoding Instructions”. Readers are advised to review that section carefully
before attempting replication of the results described in the subsequent chapter, as several
items were reﬂected before analysis began.

The attitude scale for the study was based off previous work (Armitage, 2005;
Godin, Valois & Lepage, 1993; Verplanken, Aarts, van Knippenberg & Moonen, 1998),
as well as a beliefs elicitation of undergraduate telecommunications students. Attitude
was measured through a +3 to -3 semantic differential scale of seven items; good/bad,
unimportant/important, useful/useless, relaxing/stressful, conventional/diverse,
unrestricted/controlled and easy/hard. The attitude scale was then converted for the
purposes of the analyses to a 7-point Likert-type scale to be consistent with other
instruments. Higher means on the attitude instrument indicate a greater preference toward
a “positive” attitude.

The subjective norm scale was also based off of previous TPB work (Armitage,

2005; Bamberg, Ajzen & Schmidt, 2003; Elliott, Armitage, & Baughn, 2003) as well as a

73

beliefs elicitation of undergraduate telecommunications students. Subjective norm was
measured through a 7-point Likert-type scale ofthree items; “Most people who are
important to me approve of my use of YouTube”; “Most friends who are important to me
think I should use YouTube less often)”; and “Most family members who are important
to me think that I should use YouTube less often”. The subjective norm scale was
weighted toward more approval of YouTube behavior, with the latter two items reﬂected
to achieve this end.

The perceived behavioral control scale was created based on previous TPB work
(Armitage, 2005; Knussen et al., 2004), as well as a beliefs elicitation of undergraduate
students. Perceived behavioral control relied much more on the elicitation of beliefs by
undergraduate students than other items in an effort to probe speciﬁcally what students
believed were the most important barriers or opportunities to their YouTube use.
However, students primarily described possible barriers and did not enumerate enough
speciﬁc examples of opportunities to describe that dimension in great detail, causing the
item to be primarily biased toward barriers.

Perceived behavioral control was measured by a 7-point Likert-type scale of
twelve items, several of which were reﬂected before analysis began to be weighted
toward more perceived behavioral control; "I don’t have a lot of trouble ﬁnding time to
use YouTube" (reﬂected item); "I am able to use YouTube whenever I want to" (reﬂected
item); "If I lost my ﬁnancial support (job, scholarship, parental support, etc.), I might not
be able to keep using YouTube in the same way I do now"; "If I had to pay to use
YouTube, it would be a big problem for me"; "I would be prevented from using YouTube

if my computer had problems"; "I am able to use YouTube even when my Internet

74

connection is slow" (reﬂected item); "I am able to ﬁnd a way to use YouTube even when
I am not supposed to access it" (reﬂected item); "I can use YouTube even when my
school or job attempt to block access to it" (reﬂected item); "If I had to sign up or login to
be able to use YouTube, it would be a hassle"; "I can ﬁnd videos on YouTube even if
they aren’t supposed to be there (such as copyrighted content)" (reﬂected item); "I feel in
complete control over my YouTube use" (reﬂected item); and "My use of YouTube is
completely up to me" (reﬂected item).

Intention was measured by the use of two 7-point bipolar items ranging from -3 to
+3 (Elliot, Armitage & Baughn, 2003) asking subjects how likely they are to use
YouTube at home today and over the coming week.

Actual behavior was measured during the follow-up process by a single 7—point
Likert-type item asking subjects whether or not they had used YouTube often or rarely
over the past week.

Self-eﬁicacy and consistent context instruments

Self-efﬁcacy was measured by an instrument adapted from several authors
(Ajzen, 2002b; Chen, Gully & Eden, 2001; LaRose, Kim & Peng, 2008; Manstead & van
Eekelen, 1998), using a 7-point Likert type scale. Scale means for the self-efﬁcacy scale
should be interpreted as indicating greater self-efﬁcacy; "I believe I have the ability to
use YouTube"; "I am conﬁdent that I can use all the ﬁmctions of YouTube"; "I am
certain that I can use all the functions of YouTube"; "I ﬁnd YouTube to be easy to use";
and "I can get the things that are important to me out of my YouTube use". All of these

items were reﬂected.

75

Consistent context was measured by a ten item scale, three of which were
reﬂected before analysis began to reﬂect greater consistent context; "My family members
or roommates do not disrupt my YouTube use"; "I tend to use YouTube in the same
place"; "It would be unusual for me to use YouTube somewhere other than where I am
used to using it"; "The area where I tend to use YouTube the most very rarely changes";
"I always use YouTube with the same people"; "It would be uncomfortable for me to use
YouTube with different people" (reﬂected); "I tend to use YouTube as a way to change
my mood" (reﬂected); "I am usually in the same kind of mood whenever I decide to visit
YouTube"; "I always visit YouTube using the same computer"; and "It would be
uncomfortable for me to use YouTube from a different computer than the one I usually
use" (reﬂected).

Habit instrument

The Self Report Habit Index (SRHI) (Verplanken & Orbell, 2003) was used in
this study to measure habit. The SRHI includes 12 items under the heading of "Using
YouTube is something...", followed by the scale items. Three SRHI items were removed
from the analysis after data collection because of their references to behavioral
frequency; “I do frequently”; “that belongs to my daily routine”; and “I have been doing
, for a long time”. No recoding or reﬂection of SRHI items was necessary in the analysis.

Two secondary scales of past behavior were also included; one created by using
the three ﬁequency-related items removed from the SRHI, another created by asking
subjects how long in hours and minutes they used YouTube on an average weekday and

weekend day.

76

Recoding instructions

All instruments were recalibrated after data collection to be oriented toward
displaying higher values on their instrument. Attitude was recoded to represent 1 to 7
Likert type scaling, where higher values indicated higher levels of positive attitude
toward YouTube use (goodness, importance, usefulness, relaxingness, diverseness,
unrestrictedness, easiness). Subjective norm was recoded toward higher levels of
behavioral approval as well, with each item being recoded in favor of more approval of
YouTube use by family members, friends and most people who are important to that
person.

In the case of perceived behavioral control, the instrument was recoded on a 1 to 7
Likert type scaling in which low values represented less perceived behavioral control and
high values represented more. The items marked in the perceived behavioral control
section as reﬂected were reverse—coded so to continue to orient the instrument toward
being a measure of greater perceived behavioral control.

Both intention and behavior were also recoded in a standard I to 7 type scale in
which smaller values indicated less intention and behavior, and higher values represented
more intention and behavior.

The other scale items (consistent context, self-efﬁcacy and SRHI) were also
recoded as l to 7 "least to greatest" items for ease of interpretation.

Addressing conceptual overlap between instruments

It is important and useful to us to collect data on standard Theory of Planned

Behavior variables. This will allow us to test the theory across multiple media use

77

behaviors. However, there is potential overlap between several of the instruments that
needs to be brought to light.
Perceived behavioral control

Perceived behavioral control has been a difﬁcult construct to measure. There has
been lengthy debate within the Theory of Planned Behavior literature as to the best
methods to measure the construct (see Sparks, Guthrie & Shepherd, 1997; Armitage &
Conner, 2001 for a more in-depth discussion). The fundamental point of the behavioral
control construct, addressed in the earliest theorizing about the Theory of Planned
Behavior, is to address both the difﬁculty in completing the behavior and the amount of
resources available to complete the behavior (Ajzen, 1991). However, the abundant
resources available to use YouTube make ﬁnding items that would probe into possible
diminishment of resources to use the service difﬁcult. The standard elicitation of control
beliefs generated extensive discussion about things that could prevent a student ftom
using YouTube, but never raised a question about having the resources to use the service.
It was apparently assumed by the subjects that they would have the capacity to use
YouTube somehow, even if they were sometimes obstructed from using the service by a
barrier.

There have been efforts to integrate self-efﬁcacy into the PBC construct, but these
efforts may be theoretically detrimental. Bandura's conceptualization of self-efﬁcacy
addresses internal factors, while Ajzen's PBC should focus on external factors (see Terry
& O'Leary, 1995). A self-efﬁcacy instrument has been included to allow for a

comparison of PBC to common self-efﬁcacy items.

78

PBC and the SelfReport Habit Index

The reason for the clariﬁcation of perceived behavioral control is due to the use of
the Self Report Habit Index (SRHI). Some may say that items in the SRHI that talk about
loss of control may overlap with PBC, which is deﬁned as the feeling of control over a
given action. A recent factor analysis (LaRose, Kim & Peng, 2008) found that the SRHI
may have three dimensions; habit strength, deﬁcient self-regulation or compulsive use,
and negative outcomes. While some of the items on the SRHI may overlap with certain
PBC items, the instrument is as a whole concerned with other aspects of habit.

Consistent context and perceived behavioral control

Perceived behavioral control also has similarities to the variable of consistent
context. Both variables address elements of the external environment that pertain to a
given behavior, but their most important difference is speciﬁcity. Perceived behavioral
control refers to elements of the external environment that provide either obstacles or
opportunities to engage in a behavior, while consistent context addresses the external
environment in general as well as internal mood states. This is a subtle but important
difference.

For example, a person may use YouTube at a computer that is in a speciﬁc room
of their dwelling. The room itself in this case does not assist in the use of YouTube, but is
where the computer one uses YouTube at is kept. The computer could be moved to a
different room, or even a different dwelling, but the room the computer is in would not
have a direct effect on one’s ability to use YouTube so long as the computer one accessed
YouTube with could reach the service. However, if an lntemet connection was more or

less available in a given room, the room itself would become relevant to perceived

79

 

behavioral control; it would be providing an opportunity or a barrier to engaging in a
behavior, and would transcend context.

Internal mood is also a factor in consistent context that separates context from
perceived behavioral control. If someone regularly chooses to use YouTube to alleviate
boredom or remove a bad mood, that mood becomes part of their usage of YouTube.
However, mood states are internal forces, not external forces, and perceived behavioral
control exclusively addresses things in the external world that provide barriers or
opportunities to engage in a behavior.

Analysis

The ﬁrst hypothesis was tested by a multiple regression, with the use of simple
slope analysis used to test the rest of the hypotheses. Simple slope analysis was used to
test the remainder of the hypotheses. The division of individuals into different habitual
groups was made by ﬁnding the mean of the sample’s overall level of habit as determined
by their responses to the SRHI. Using that mean, an equal three-section partition (33%)
was made within the simple slope program to place users into low, medium and high
habit groups.

This method of division is based on mathematical parsimony, as there is no
theoretical argument for or against the division of individuals into different habitual
“moments”. Habit has been previously been treated as a binary condition in the literature
reviewed to this point, but different stages of habit should logically exist. The division of
individuals in this way may be crude, but it should allow for a better understanding of the

mentality of someone with a weak or a strong habit operates at the very least.

80

 

All statistical tests were completed in SPSS 10.1 for Windows. Simple slope

analysis was executed by the use of the SIMPLE program suite by O’Connor (I998).

81

 

Chapter 4

Results

Findings

Sample characteristics

Of the 1200 students sampled, 197 responded to the invitation to take the survey,
yielding a net response rate of 16.4%. Of those 197, 1 10 responded to the follow-up
survey, providing an internal response rate of 55.8% and an overall response rate of
9.1%. The main analyses will use the smaller internal sample of 110 subjects to maximize
continuity between the ﬁrst and second phases of the data collection.

The sample was 72.6% white, .5% African-American, 1% Hispanic and 4%
Asian. 9% identiﬁed themselves as “other”. 12.9% of the sample declined to provide any
racial or ethnic characteristics. 71.6% identiﬁed as male. The average age of the sample
was 21.5 years. While the ethnic characteristics of the sample were generally in line with
the ethnic makeup of the large Midwestern university that the sample came from, the
high sex bias toward males was not representative of the student body.

Scale diagnostics

Initial analyses indicated signs of ceiling effects in the instruments, so several
problematic items were removed. Items were also removed from scale instruments to
improve reliability. Any missing data was addressed by mean replacement. The surviving
indicators for key independent variables (attitude, subjective norm and PBC) are
summarized in Table 4-1. Scale values ranged from I to 7, with 1 being the lowest value

of the variable and 7 being the highest value.

82

Table 4-1

Theory of Planned Behavior Scale Diagnostics

 

 

Scale M SD or
Attitude 4.43 .58 .70
Good. 6.01 1.08
Important 4.79 1.45
Useful 5.49 1.58
Subjective Norm 6.22 1.22 .90
Most friends approve 6.25 1.32
Most family approve 6.19 1.23
Perceived Behavioral Control 4.82 1.01 .65
Able to ﬁnd a way to use when not allowed 4.77 1.66
Able to ﬁnd a way to get around blocks 4.94 3.25
Find videos that aren’t supposed to be there 4.75 1.80

 

There were problematic elements to the instruments. Because of the

overwhelming positive feelings toward YouTube, both the attitude and perceived

behavioral control instruments tended to have a high mean, leading to ceiling effects that

needed to be corrected by removing highly skewed items. Removing these items may

have caused the instruments to become less reliable. The removed items may have been

more reliable, but would have caused another type of problematic error variance in the

83

analyses. The reliability of perceived behavioral control was also lower than is normally
desirable (.7).

The habit instrument and a self-efﬁcacy instrument were included in the analyses.
In the collected sample, most people did not claim to have a strong YouTube habit,
though some appeared to have at least a moderate habit. Three ad hoc divisions of the
SRHI results were made after data was collected to complete analyses of the following
hypotheses. SRHI items that related to frequency of behavioral execution were removed
before the analyses were completed in an attempt to address theoretical concerns about
the validity of frequency measures in habit instruments.

The mean of the SRHI scale in this instance (M = 2.12) is notably lower than the
mean found in Verplanken and Orbell’s SRHI analysis of another media behavior,
regulme watching a popular Dutch soap opera (M = 3.47). This provides some evidence
that YouTube use may not be as habitual as other media use behaviors, as the standard
deviation of the SRHI in this study left little room for most people in the study to have a
strong habit (SD = 1.23). However, there were subjects in the study who did have a
moderate or stronger than moderate YouTube habit, which allowed for analysis of all of
the hypotheses.

The SRHI’s scale diagnostics are described in Table 4-2. Self-efﬁcacy and

consistent context scale diagnostics are listed in Table 4—3.

84

Table 4-2

SRHI Scale Diagnostics

 

 

Scale M SD or
SRHI 2.12 1.23 .90
I do automatically 3.18 1.88
I do without having to remember. 2.63 1.79
Makes me feel weird ifI don’t do it 1.58 1.14
I do without thinking 2.10 1.68
Require effort not to do it 1.80 1.39

Start doing before I realize I’m doing it 1.52 1.14

I would ﬁnd hard not to do 2.03 1.67
I have no need to think about doing 2.79 1.93
Typically “me” 2.10 1.48

 

The self-efﬁcacy instrument had strong ceiling effects, as one would expect in the
case of college students. The self-efficacy instrument was important to the study as it
deals primarily with internal factors, which provided contrast to the PBC instrument that
focused on external factors. While PBC was noted as being biased toward less PBC, self-
efﬁcacy was observed as being almost universally very high. In the case of YouTube

usage, most people in the sample appeared to feel conﬁdent in their own abilities to use

85

the system but found themselves at the mercy of external factors, which appears logically
consistent. The scale diagnostics for self-efﬁcacy are in Table 4-3.

The consistent context instrument performed very well, having almost a normal
distribution and an average alpha value of .8. While college students may have a great
deal of self-efﬁcacy, their circumstances did not appear to be entirely under their control.
The diagnostics for the consistent context instrument are in Table 4-3.

Overall, the adjusted scales were functional, but not optimal. The two most
important indicators in most TPB analyses, attitude and PBC, appeared to be generally in
line with the sample’s thoughts and concerns. The SRHI showed that many people in the
sample did not have a strong YouTube habit, but the behavior does appear to have a
habitual component. Standardization of the scales did not signiﬁcantly improve their
behavior in the analyses. The perceived behavioral control indicator had a low alpha
value, so should be interpreted cautiously.

Correlations of all of the relevant scales with intention and behavior included are

listed in Table 4-4.

86

Table 4—3

Self-Efficacy and Consistent Context Scale Diagnostics

 

 

Scale M SD a
Self Efﬁcacy 5.15 3.11 .94
Conﬁdent can use all the functions 5.54 1.51
Certain can use all the functions 5.31 1.68
Consistent Context 4.39 1.24 .80
Tend to use in the same place 2.84 1.64
Unusual to use in a different place 5.08 2.03
Area tends not to change 3.54 1.90
Always use with the same people 4.54 1.77
Always use in the same kind of mood 4.39 1.75
Always visit with the same computer 4.46 2.10
Uncomfortable to use a different computer 5.92 1.57

 

87

awe—a AL

chxmﬁzca magma: HEW Maﬁa. mags Mm§m§8§ megahmi Geimﬁ. $832.0: 98‘ @2839.

 

 

moan >383 m2 wwo wwwm mm on 58:30:. wormlig
338% .3

m2 .9: be

wwo “Ho: some .3

ma: .wmm: -.Nome .ummi .8

mm bee Low bum: .33. be

00 L8. L2... 3:0 .39.. -..om .8

5838: L31. ._ 3 on: .1. ad: .oS .SN -

waging .wom: L 3 b3: 93...... .15 .oB b3: -

 

22m. mz u 9.8.898 Zone. .50 n $4818 magic—d. 6256.. mg n ma: wove: Ive: Ema? mm M man—maomov» no u nonmmmnoa 02:2?

2 n :o. ._ h A .3. .3 h A .3. Cancun—..m at...» 5.58 m3 9a 838 98 38:8 o: :5 Emma—5.

88

The Theory of Planned Behavior and habit

A multiple linear regression was used to test whether the Theory of Planned
Behavior would continue to function normally for individuals who had a weak or
moderate habit. First, attitude, subjective norm and perceived behavioral control (PBC)
pertaining to YouTube usage were entered into a simple regression to predict intention to
use YouTube. The regression equation was statistically signiﬁcant (F (3,106) = 13.350, p
< .001) with an adjusted R2 of .254. Attitude was the best predictor of intention with a B
of .434 and a AR2 of .243, followed by perceived behavioral control with a B of. 162 and
a AR2 of .023. Even though perceived behavioral control was not statistically signiﬁcant
at the standard .05 level, it still had a small role to play in the prediction of intention.
Subjective norm was not a statistically signiﬁcant predictor of behavior in this analysis,
and contributed little explained variance. Based on these ﬁndings, HIA can only be
partially supported. The full output of the regression is shown in Table 4-5.

Second, intention to use YouTube and perceived behavioral control were entered
into a simple regression to predict actual YouTube behavior. Again, the regression
equation was statistically signiﬁcant (F (2,107) = 15.493, p < .001) with an adjusted R2 of
.210. Both intention and perceived behavioral control were important predictors of
behavior, with intention having a B of .378 and a AR2 of. 195, and PBC having a B of
.180 and a AR2 of .029. Based on these ﬁndings, HlB was supported. The full output of

the regression is shown in Table 4—6.

89

Table 4-5

Hypothesis 1A — Predicting Intention Using TPB Variables

 

 

 

 

 

Predictor B SE B 13 AR2 AF F Sig.
Attitude .748 .152 .434 .243 34.620 .001
Subjective Norm .168 .130 .108 .008 1.202 .275
PBC .236 .128 .162 .023 3.370 .069
Table 4-6

Hypothesis IB — Predicting Behavior Using TPB Variables

Predictor B SE B o ARL AF F Sig
Intention .405 .093 .378 .195 26.173 .001
PBC .273 .135 .180 .029 4.069 .046

 

A post hoc analysis using simple slope analysis further examined the relationship

between intention and behavior. Simple slope analysis is a form of multiple regression

that graphically displays slope values at given values of a moderator. This dissertation

used the SIMPLE SPSS syntax program by O’Connor (1998) to execute this and the

other simple slope analyses discussed in this chapter. Removal of outlier cases did not

90

signiﬁcantly affect the results ofthe analysis. Simple Slopes are shown in Figure 4-1,

with additional output relevant to the simple slope procedure contained in Table 4-7.

 

 

 

 

 

7

B i f . e

5 . ,

4 .

Habit

3 . Low
g Medium
co _
1%
[D 2 High

-.25 7.35

lnte ntion

Figure 4—1. A simple slope analysis of intention and behavior moderated by habit.

91

Table 4-7
Regression Analysis Predicting Behavior with Habit Moderating the E/fect of Intention

on Behavior

 

 

Variable B SE B 13 t F qu Chang
Step 1 14.009**
Intention .379 .110 .363 3.455**
Habit .221 .110 .136 1.296
Step 2 10.024** .014
Intention .496 .662 .475 3.129* *
Habit .005 .268 .036 .233

Intentioanabit .135 .101 .955 1.367

 

Note. Adjusted R2 = .199. * p < .05. ** p < .01.

The stepwise moderated regression tables that accompany each simple slope
ﬁgure follow the same format. Each moderated regression has two steps. The ﬁrst step
does not include a moderated relationship between the two variables used in predicting
the dependent variable, while the second step does include the moderated relationship.
The moderated relationship is demonstrated at three levels (low, medium and high) in the
table’s associated ﬁgure.

For example, Table 4-7 describes the stepwise moderated regression that

describes how habit moderates the relationship between intention and behavior. In the

92

ﬁrst step, intention and habit are used to predict behavior. Intention is the only
statistically signiﬁcant predictor, as is anticipated by theory. This is also true in the
second step, though the interaction between intention and habit should still be considered
important to the analysis because of its relative size. Figure 4-1 shows how this
moderated relationship looks at low, medium and high levels of media habit. The
difference between each level of habit is small, but there is a visible effect that may have
been more obvious with a different media behavior.

The expected relationship between intention and behavior when moderated by
habit would be for the link between intention and behavior to be strong when habit was
weak. In turn, when habit grew in strength, the link between intention and behavior
would diminish. The simple slope shown in Figure 4-1 supported this expected
relationship, with the relationship between intention and behavior weakening as habit
grows in strength.

To ﬁirther examine the results, the relationship between habit, intention and
behavior was examined by correlating intention and behavior at each of the three ad hoc
divisions of habit using the full dataset of 197 cases. It was found that intention and
behavior had the strongest correlation at low levels of habit (r(65) = .305, p < .05), with
the correlation declining at medium (r(63) = .241, n.s.) and low (r(63) = .240, n.s.) levels
of habit.

Further analysis using Steiger’s Z (Meng, Rosenthal & Rubin, 1992; Stieger,
1980) found that the impact of habit on intention and behavior was not statistically
signiﬁcant when habit was low (Z = -.532), was still not statistically signiﬁcant when

habit was moderate (Z = .999), but did become statistically signiﬁcant when habit was

93

strong (Z = 2.28). The overall trend of growth in statistical signiﬁcance appears to
provide further veriﬁcation of the results.

Is the relationship between habit and perceived behavioral control negative?

A review of the scatterplot followed by simple slope analysis was used to test the
relationship between behavior and perceived behavioral control (PBC) with respect to
habit strength. How did perceived behavioral control interact with habit?

Initial review of the scatterplot in Figure 4-2 showed that habit tended to be
weaker at high levels of perceived behavioral control, and became stronger as perceived
behavioral control weakens. Simple slope analysis was used to conﬁrm the ad hoc
analysis. Simple slopes are shown in Figure 4-3, with the output relevant to the simple
slope procedure contained in Table 4-8. Removal of outlier cases did not signiﬁcantly
affect the results of the analysis.

Table 4-8 shows mixed results. Habit has a strong impact on the results of the ﬁrst
step, showing itself to be both an absolutely stronger predictor and a more statistically
signiﬁcant predictor than PBC. However, none of the predictors in the second step rose to
standard levels of statistical signiﬁcance. The simple slopes shown in Figure 4-3 show
the broad pattern of relationship between PBC and habit, but these should be interpreted
conservatively. When habit is weak, perceived behavioral control has the expected
relationship; as PBC rises, behavior tends to become more common. However, as habit
grows in strength, the growth of PBC is reversed, leading to an almost ﬂat slope when

habit is very strong.

94

The expected relationship between PBC with respect to habit was conﬁrmed by
simple slope analysis. PBC’s impact on behavior weakened as habit grew in strength.

Based on these ﬁndings, H2 can be supported.

 

 

 

 

8
7‘ C]
6-
El
5- D [:1 DD D
D D u
[:1
4 c1 (3 3E 00
ODD c1 0 D
u a DC)
3 13 DD C1 DECIDE
u a B
B [31:18
0 DD EBB
D C] B
2 ”DEB 5003
EBEIEESEUD
1 Duo gag 0g
0 I
2 3 4 5 6 7 8

Perceived Behavioral Control

Figure 4-2. A scatterplot of PBC and habit.

95

 

5.0
5.5-
5.0-

4.5 «

 

 

 

4.0 ‘
Habit
3.5 x ' - - - -
Low
g 3‘0 ‘ Medium
cu _
g
m 2.5 High

 

1.59 5.94

Perceived Behavioral Control

Figure 4—3. A simple slope analysis of PBC and behavior moderated by habit.

96

Table 4-8
Regression Analysis Predicting Behavior with Habit Moderating the Effect of Perceived

Behavioral Control on Behavior

 

 

Variable B SE B [j t F qu Change
Step 1 9.547**

PBC .290 .146 .192 1.985*

Habit .437 .155 .273 2.829**
Step 2 6.663** .007

PBC .482 .822 .318 .586

Habit .066 .549 .041 . 120

PBC x Habit .127 .133 .699 .955

 

Note. Adjusted R2 = .135. * p <05. **p < .01.

In general, it appears that habit is a good predictor of behavior when habit exists
in any capacity at all. Granted, the mean of the habitual behavior used in this study was
lower than other studies that examined media use with the SRHI. For instance, compare
the YouTube SRHI mean of2. 12 (SD =1.23) to the SRHI mean of3.47 (SD = 2.41)
found by Verplanken and Orbell (2003) among soap opera viewers. Even so, habit still
contributed an important amount of variance to the regression.

It may be that any non-volitional aspect to a behavior that exists will tend to

weaken the role of reasoned action variables in predicting behaviors. In that case, the

97

greater the role of habit in a behavior, the more likely it appears that reasoned action
variables will lose predictive strength and ultimately become ineffective in guiding future
behavior. This trend is reflected in the remaining tests covered in this chapter.

What is the role of consistent context in habit?

A review of the scatterplot followed by simple slope analysis was used to test the
relationship between behavior and habit with respect to the strength of consistent context.
Would a consistent context moderate the relationship between habit and behavior?

Initial review of the scatterplot in Figure 4-4 appeared to show the expected
relationship; a higher level of consistent context tends to lead to stronger habits. Figure 4-
4 also shows what appears to be a curvilinear relationship between habit and consistent
context. Habit seems to increase with consistent context to about its midway point before
decreasing slightly. It could be the case that media habits are similar to other habits until
a moderate habit is formed, and then its reliance on consistent context diminishes because
of the wide availability of the media in the greater environment. This ﬁnding supported
the use of a post hoc test to test for curvilinearity.

Simple slope analysis was used to explore the preliminary ﬁndings of the
scatterplot. Simple slopes are shown in Figure 4-5, with the output relevant to the simple
slope procedure contained in Table 4—9. Removal of outlier cases did not signiﬁcantly
affect the results of the analysis.

Table 4-9 shows that in the ﬁrst step, consistent context is the most important
factor in predicting behavior, while habit's contribution is relatively unimportant.
However, in the second step, the interaction between habit and consistent context is

shown to be the most important factor. The contribution of the interaction is most clearly

98

shown in the simple slopes in Figure 4-5. As anticipated, consistent context appears to be
very important in the formation of habits, with the interaction term being much greater
than either habit or consistent context alone. Low consistent context weakens habit’s
impact on behavior, but as consistent context increases, habit’s relationship with behavior
becomes stronger.
These ﬁndings are consistent with our theoretical understanding of how habit and
consistent context should interact. Without a consistent context, it is difﬁcult to maintain
habitual behaviors, though not impossible. Once a habit becomes established, it may be
that the necessity for a consistent context is lost.
The data reﬂected the theoretical understanding that as the context for a behavior
becomes more consistent, habitual strength is likely to increase. A consistent context
allowed for a habitual behavior to be executed more easily. A behavior that is easy to
exe cute can be done more often, which would encourage habit formation. Aﬁerward, the
context would matter less, which is in line with our theoretical understanding of habit.
In all, as consistent context grew in strength, the relationship between habit and
behavior became stronger, whereas when consistent context was weaker, the relationship
between habit and behavior weakened. Based on these ﬁndings, H3 was therefore

SL1plborted.

99

 

 

 

 

Consistent Context

Figure 4-4. A scatterplot of habit and consistent context.

100

Behavior

 

 

 

Consistent Context

~ Low

Medium

High

 

 

1.91

Habit

5.89

Figure 4-5. A simple slope analysis of habit and behavior moderated by consistent

Cont ext.

101

Table 4-9

Regression Analysis Predicting Behavior with Consistent Context Moderating the Effect

 

 

of Habit on Behavior
Variable B SE B B t F qu Change
Step 1 7.350**
Habit .006 .147 .042 .456

Consistent Context .571 .149 .353 3.827**
Step 2 6.604** .037
Habit .623 .575 .385 1.085
Consistent Context .51 1 .306 .320 1.673
Habit x CC .291 .135 .785 2.148*

\

NOte- Adjusted R2 = .134. *p< .05. **p < .01.

A post hoc regression analysis was attempted to locate a curvilinear relationship
be tw een consistent context and habit. However, the regression analysis did not
s11 e <3 essfully demonstrate a curvilinear relationship, with the additional term not being
stat i Stically signiﬁcant (p > .05). While the relationship may be nonlinear, it does not
app ear to follow a traditional curvilinear path.

Will the relationship between habit strength and self-efﬁcacy change intentions?

102

A review of the scatterplot followed by simple slope analysis was used to test the
relationship between intention and self-efﬁcacy with respect to habit strength. Would
habit diminish the role of self-efﬁcacy as habit grew in strength?
Initial review of the scatterplot in Figure 4-6 appeared to show that as habit
increased, self-efﬁcacy tended to decrease. While most of the highest self-efﬁcacy scores
were among individuals with relatively low levels of habit, the overall trend appeared to
support the idea that habit and self-efﬁcacy have an adversarial relationship. Simple slope
analysis was used to conﬁrm the ﬁndings in the scatterplot.
Simple slopes are shown in Figure 4-7, with the output relevant to the simple
Slope procedure contained in Table 4-10. Removal of outlier cases did not signiﬁcantly
affect the results of the analysis.
Habit was the most important predictor in the ﬁrst step of the stepwise regression,
0 V6 rpowering self-efﬁcacy itself. Per the scatterplot in Figure 4—6, self-efﬁcacy tended to
be Weaker when habit was strong and stronger when habit was weak. The strong effect of
hab it in the ﬁrst step of the regression did not carry over to the second step, in which
One e again none of the predictors rose to statistical signiﬁcance. Further, the simple
S 1013 es in Figure 4-7 show only very mild effects between self-efﬁcacy and habit. There
are differences between the different habit conditions, but these differences are not nearly
as dramatic as in previous hypotheses.
Habitual behaviors tend to be repeated often, and behaviors that are repeatedly
axe Q11th tend to be behaviors a person becomes competent executing. This could be why

th
e automatic nature of habit does not appear to have lowered perceived self-efﬁcacy

103

 

signiﬁcantly. Non-volitional aspects to a behavior appear to weaken the role of reasoned

action variables in predicting behavior. Based on these ﬁndings, H4 can be supported.

 

 

 

 

8
7' :1
6.
5 :0 C) D B
D 1:1
8 u
44 1:10 a DD 0088 0
1:1 B 1:1
:1 :1 C10 0
3 130 CID Dog I] :gEE
U u 8 E D D D a a B
2 Sad SUEDE 09
1:1
01: 8 9 08058005
1 I31:1 a g 88:13:10 1:]
O i
3 4 5 6 7 8

Se” Efﬁcacy

Figure 4-6. A scatterplot of habit and self-efﬁcacy.

104

5.5

 

 

 

 

5.0 .
4.5 « --------------
4.0 4
3.5 -_ T ___________________
Habit
3.0 ~ ' ' ' '
Low
-5 2'5 / Medium
E —'
3 2.0 High
-.54 5.59
Self Efﬁcacy

pigZ-l re 4- 7. A simple slope analysis of self-efﬁcacy and intention moderated by habit.

105

Table 4-10
Regression Analysis Predicting Intention with Habit Moderating the Effect of Self—

Efficacy on Intention

 

 

Variable B SE B 13 t F qu C hange
Step 1 26.549”
Self-Efﬁcacy .002 .098 .222 1.005
Habit .888 .123 .655 3656*
Step 2 l7.707** .011
Self-Efﬁcacy .272 .517 .222 .955
lHabit .564 .233 .655 .938

Self-Efﬁcacnyabit .004 .082 .270 1.167

\

Note. R2: .115. *p< .05. **p<.01.

82‘ ’77 maty

After collecting data, a multiple regression and simple slope analyses were used
to t 93 st the hypotheses. The expected results were found, albeit not strongly. Habit was
ab 1 e to override reasoned action as habit became stronger, particularly variables that dealt
W ith external control (PBC) and internal control (self-efficacy). Greater levels of
Q Q l:ISistent context increased the strength of habit, and habit appeared to at least partially

W e'élken the role of self-efﬁcacy in predicting both intention and behavior as habit grew in

106

strength. Overall, as habit became stronger, reasoned, volitional action became less useful

in predicting behavior. Theoretical discussion of the results follows in the next chapter.

107

Chapter 5

Discussion

H51!) it and the Theory of Planned Behavior
This dissertation found that the Theory of Planned Behavior can be used
co nstructively in an examination of media habits. Even though the Theory of Planned
Behavior’s variables did not have the level of predictive power they normally have in
stud ies of human behavior, the variables were still able to be used to predict both
intention and behavior. The predictive power of the theory may have been weakened in
th is case because of the inaccessibility of moderate to strong YouTube habits. YouTube
use may also be a semi-automatic process that has periods of active agency (locating new
Cont ent, adding comments to videos, etc), allowing the Theory of Planned Behavior to
partially explain YouTube use, but at the cost of losing some predictive power from the
au tOnrlatic portions of the behavior.
The habit variable has value to the Theory of Planned Behavior. A clearly
e0 IIceptualized and well-expressed habit variable can add to the predictive power of the
theory. Habit has been shown in this data to have important interactions with planned
behavior variables, particularly with perceived behavioral control. Including habit in the
Theory of Planned Behavior and accounting for its moderating relationship with
lbertteived behavioral control should only increase its predictive power, and allow the
T110 Gel to address situations where automaticity may be at work.
At present, some behaviors that the theory attempts to account for may be

utQmatic for their users, even if subjects can generate false cognitions or retroactively

108

provide imaginary intentions to attempt to put their behavior in a logical context. By
add ing habit to the Themy of Planned Behavior, scholars will be better able to address
t he many behaviors that human beings transform into habits over the course of their
1i fet imes, especially media behaviors.

Previous studies discussed during the course of this study have used various
mea sures of habit in Theory of Planned Behavior studies to argue in favor of its inclusion
in the model. This study went one step ﬁirther and demonstrated how the habit variable

has an important interaction with existing Theory of Planned Behavior variables, as well
as showing the utility of a well-conceptualized habit variable in predicting behaviors.
Th is study has shown that the habit variable and TPB variables can — and should — work
together in modeling human behavior.

This study, along with the others that have been done previously, shows that
inc luding a measure of habit in Theory of Planned Behavior studies can directly assist
researchers determine whether or not a behavior is still being actively reasoned or if its
Sehe ma has become part of an automated habit system. If a behavior has not been
automated, or has only been somewhat automated, Theory of Planned Behavior variables
SI‘Ollld still be a good predictor of intentions and behavior. If a behavior has been
automated, researchers can set aside Theory of Planned Behavior variables which will not
he 11) them predict the continued repetition of a behavior and focus instead on measuring
VPariables of interest to habit.

Even partial automation of a behavior suggests that habit should be integrated into

the 'Theory of Planned Behavior. As behavior became more automated, intention became

l
e SS effective at guiding behavior (e. g. Limayem, Hirt & Cheung, 2007; Kim & Malhotra,

109

200 7), resulting in situations in which a person may say they intend to engage in a
behavior counter to their established habit but ultimately fail in executing the anti-
hab itual behavior (Baumeister & Heatherton, 1996). This dissertation observed that as
hab it increased in strength, the correlation between intention and behavior gradually
deC lined and lost statistical signiﬁcance.

The standard test of TPB without habit used ﬁrst in this study showed that the
theory was able to predict intentions to use YouTube as well as actual YouTube behavior,
but may not have been an accurate picture of the true nature of the behavior. There are
various reasons why this may have been the case. For example, individuals with a very
strong YouTube habit who are asked whether they intend to engage in the behavior
would say they had a very strong intention to use YouTube. Their behavior may be
habitual, but they want to be perceived as a rational, internally consistent actor rather than
SOmeone who is driven by non-volitional cognitive processes. Without the presence of
the habit variable in the theory of planned behavior, it would be possible to draw the
c()l'lcmsion that YouTube use is an entirely rational behavior driven by thoughtful
col'lSideration of attitudes, subjective norms and perceived behavioral control. With the
a(1(1ition of habit to the theory, the non-volitional aspects of the behavior become more

Vis it3le, and are shown to have signiﬁcant effects on the explanatory power of the
corllponents of the theory.
Additional discussion of habit and intention can be found later in this chapter.
The habit construct versus the “past behavior” construct
The question of whether the variable of habit is superior to the past behavior

0 . . .
Qlilstruct has been addressed throughout the course of th1s dissertation. The data shown

110

in this dissertation demonstrates the use of the construct in predicting behavior, but is this
d issertation’s strongly conceptualized depiction of habit better than previous past
beh avior measurements?

The habit variable as conceptualized in this dissertation should address the
pro blem of Ajzen’s objectionable “wastebasket” variable of past behavior. As discussed
earl ier in this study, Ajzen’s primary objection to past behavior was that it contained a
number of concepts beyond mere behavioral repetition that could not be controlled for,
add ing a great deal of error variance to the measurement and thus being theoretically
11 se 1 es s.

This dissertation supported that habit can be clearly deﬁned and measured, and
has an instrument, the Self Report Habit Index (SRHI), that is reasonably effective at
measuring the construct via self-report with some minor alterations. Mere repetition is
not a key criterion for determining whether or not a behavior is a habit using the modiﬁed
S l{I—II used in this dissertation. Instead, the automatic nature of the behavior is of primary
C()Ilcern. This shift in focus to the cognitive nature of the behavior makes the habit
measurement distinct from behavioral frequency, and superior to the idea of “past
behavior”.

It may be the case that YouTube use is a behavior that may be repeated a great
deal, but with focused and volitional cognitions driving the behavior. The low means of
the SRHI instrument for this behavior support that many people in the sample did not use

ube automatically, or at least did not use it Within the facets of automatiCity
11Flei‘tsured by the SRHI. In this case, a past behavior measurement would also not be

1‘ . . . . .
epresentative of the habitual nature of the behaVior, because even if the behav1or was

111

repeated frequently, it would not be repeated automatically, which is one of the most
important facets of this dissertation’s deﬁnition of habitual behavior. Mere repetition may
be an important part of establishing a habit, but some behaviors that are repeated a great
dea l are not habits.
The distinction between past behavioral frequency and habitual behavior presents
a c lear reason for past behavior to be avoided as a proxy for measuring habit. In the case
0 f this study, a person could have used YouTube regularly, but only when they were
provided links from a friend on a social networking site or a blog. This person would be
considered a regular user of YouTube with a strong YouTube habit, but in reality their
use of the service would be driven by volitional browsing behavior rather than habitual
behavior prompted by a certain context or stimulus. Conversely, a person might use
YouTube only once a week, but their use of YouTube is to review new submissions by
CO ntent producers they have speciﬁcally subscribed to that might not update on a daily or
Weekly basis. A pure past behavior measure would indicate this person would have only a
Weak YouTube habit, even though a SRHI-like instrument would show a strong
alJtOmatic tendency to the behavior.
Instead, this dissertation used the SRHI and its items that inquired about
an tOmaticity. As a result of using a more conceptually clear instrument, a picture of
Y o1.1Tube use as a potentially automatic behavior was provided that would have been
inV isible to a less sophisticated instrument. The SRHI mean for YouTube use was lower
than in the case of another media behavior (soap opera viewing), but this could be
e)CIDIained by its newness as a behavior. It may be that YouTube use is also not

thsidered an independent behavior from “web use” or “browsing F acebook” in the

112

minds of the users studied, which may have depressed the degree to which YouTube has
been automated in their lives.
Further, we still do not have an explanation for how much repetition is needed to
F0 rm a habit, and it may be that media behaviors may require more or less repetition than
other non-media behaviors to become truly automated. Other behaviors may be more
eas i ly identiﬁed as being automated by individuals than media use behaviors, which
would explain the lower mean of soap opera viewing compared to another behavior
Verplanken and Orbell reviewed, eating candy (M = 5.31). Alternatively, YouTube may
actually be primarily a volitional behavior driven by active searching more often than it is
an automatic behavior driven by a stimulus-response relationship of some kind.
Habit and its relationship with the subjective norm
Previous studies (e. g. Ouellette, 1996; Ouellette & Wood, 1998) have found that
measures of past behavior and subjective norm tend to have positive, if small,
COI‘I‘elations, ranging between .15 and .25. This study also found a small correlation
betWeen our habit instrument and subjective norm, but in this case the correlation was
Small and negative (r(l 10) = -.205, p < .001). A separate correlation between past
bel“havior and subjective norm found a similar, but smaller result (r(110) = -.131, p <
-00 1 ).

In the case of this dissertation, the subjective norm toward YouTube use was
overwhelmingly positive. Subjects did not perceive any resistance from their friends or
fat7‘1in about their amount of YouTube use, and as a result the items were
overwhelmingly biased in favor of using YouTube. While this bias made the instrument

Very reliable, it may not be a very helpful predictor of YouTube use. The predictive

113

povver of subjective norm was shown to be trivial in the regression in Table 4-5, with all
of the important variance being consumed by attitude. One’s personal feelings about
us in g YouTube appear to be more important in determining whether or not someone uses
it rather than the subjective norm or even perceived behavioral control.
Subjective norm is a problematic indicator in the Theory of Planned Behavior,
Kno an for its volatility and weak predictive power. Various proposals to redress the
problem of the subjective norm’s deﬁciencies have been suggested (Armitage & Connor,
200 l), but even the best multi-item measures of subjective norm tend to only have
moderate correlations with intention (r(30) = .38, p <.001). Attitude tends to be a better
overall predictor than the subjective norm, along with perceived behavioral control.
Departing from subjective norm, the most interesting aspect of this dissertation is
how habit interacted with control beliefs through the variable of perceived behavioral
control. The ﬁndings of the dissertation in this area have important ramiﬁcations for the
forlilation and maintenance of habits.
Control beliefs and habit
The focus of this dissertation was primarily on how control beliefs may be
iIlfIllenced by habit. It was found that PBC decreased as habit grew in strength, and that
habit was a better predictor of behavior than PBC alone. While intention may still. be an
OveI‘all better predictor of behavior than PBC and habit together, the moderated
re lationship between PBC and habit had a signiﬁcant impact on the prediction of
be=1‘lavior.
In the case of YouTube use, a person who has just started using YouTube may

qu iekly develop the ability to address both external obstacles and internal control over

114

their actions, developing the sense of conﬁdence and control needed to become a
seasoned user of the application. These newer users consider their actions and select

act i‘vities based on their goals, whether it is entertainment, information or self-promotion.
[)8 C becomes less relevant as habit increases because of the standard automation
pro cesses which take place as habit grows in strength. Yet, it is important to note that
there are relatively few barriers to using YouTube and numerous opportunities to engage
in the behavior. The Internet can be used in a wide variety of contexts, and YouTube is
accessible through almost all of them.

PBC is a much less important predictor of behavior than habit when habit is
present. While a person may perceive that their opportunities to engage in a behavior are
abundant, and barriers to the behavior are low, individuals with a YouTube habit are
u 11: imately being compelled to engage in the behavior by automated processes. The
gr Ovvth of perceived behavioral control is not as signiﬁcant to predicting behavior as the
increased power of habit over time and repetition.

If it is not important what someone expects to get out of a behavior when they are
acting through a habitual process, the growth of perceived behavioral control is not very
t1'i‘eOretically helpful in explaining behavior. If habit is of appreciable strength, the

Strength of perceived behavioral control is not ultimately affecting whether or not
Sol"Incone engages in a behavior. Subjects are likely invoking their high perceived
be havioral control falsely, as their control over their external environment is far less
i111Ibortant than the role of automated behavior. Automated behavior, prompted by

QEtched" recall of rewards for using YouTube (Daw, Niv & Dayan, 2005), is the ultimate

115

determinant of whether or not someone with a YouTube habit continues engaging in the
be h avior.

Perceived behavioral control may be very high as a habit is being established. By
the time habit has become strong enough to eliminate the potential impact of perceived
behavioral control, most barriers to the behavior and most opportunities to engage in the
behavior have been identiﬁed. Almost nothing will stand in the way of someone with
suc h a strong habit from continuing to engage in the habitual behavior. It may be possible
to alter the behavior by changing the context the behavior is executed in, bringing about
either goal change or at the very least making it more difﬁcult to engage in an old
habitual pattern, but well-established habits will be resistant to disruption attempts
because of their subconscious, non-volitional nature.

Habit, PBC and behavior

The data supported the claim that when habit exists, it has a greater impact on
predicting both media use intention and behavior than perceived behavioral control in
1110 St cases. The proposed model in which habit moderates PBC appears to be a
pars imonious explanation for why habit is so much stronger than PBC in these situations.

As a behavior becomes automated, cognitive efﬁciency processes take place
which make the cognitions that may have led to the behavior inaccessible to the actor.

H abitual processes overrule active, volitional thought, and individuals engage in the
automated behavior based on internal or external prompts. Very strong habits will
e1lQourage individuals to ﬁnd ways around any exterior barriers that may exist, and
e)‘llbloit any external opportunities to engage in the behavior in the correct context. In the

ease of YouTube use, the sheer number of possible methods of using the service and the

116

trivial barriers to engaging in the behavior make PBC a relatively unhelpful indicator of

YouTube intentions and YouTube behavior, especially when habitual processes become

involved.

When prompted to explain their actions, an individual may be able to recall those
initial cognitions that led to the creation of a habitual behavior, but those cognitions are
not actually responsible for the behavior being repeated habitually. Rather, they are being
used as a retroactive explanation for behavior to make themselves appear internally
consistent and rational rather than being driven solely by habitual processes.
Throughout this dissertation, it has become clearer that habitual processes become
stronger at the expense of volitional, reasoned action, leading to the continuance of a
habitual behavior as a non-reasoned action. What reasoned action processes may have
existed cannot ﬁmction any longer because they have been integrated into a habitual
sc hema, which is transparent to the person with the habit. A person may still believe their
behavior is reasoned by inventing post hoc reasons for why they engaged in a specific

behavior, or falsely recalling initial reasons for beginning the behavior, but in reality the

habit has become the primary force behind the given behavior.

Previous models that have proposed that past behavior can explain a great deal of
Variance may have been attempting to account for the role of habit in human behavior.
F l‘-10t11ations in the ability of past behavior to account for certain types of actions may be
dUe to more or less habit being functional for a given behavior. However, without
C()l'lceptual speciﬁcity, Ajzen’s comments about past behavior being a “wastebasket” still
app 1y. Only by describing habit explicitly and measuring its most important aspects can

SQ ieIltists use the concept to explain human behavior in a way that is theoretically sound

117

and conceptually coherent. Otherwise, past behavior may account for a variety of non-

habitual factors such as temporal stability or behavioral frequency, which are not as

helpful for understanding human behavior as other variables.

Habit and intention
Habits are difﬁcult to disrupt even when a person has intentions to disrupt the

habit (Verplanken & Wood, 2006). While intention may be able to modify the execution
o f an automatic behavior (Saling & Phillips, 2007), in general habits override intention
once a habit is established. Intentions have been shown in the past to be less relevant to ‘
guiding behavior as habit grows in strength (e. g. Limayem, Hirt & Cheung, 2007; Kim &
Malhotra, 2007), and habit can override intention entirely when habit is strong (e.g.

B aumeister & Heatherton, 1996; Wood & Neal, 2007). Intentions may even appear after

a behavior has been done rather than preceding it (LaRose, 2004).

In this study, it was found that habit and intention had a strong positive
correlation (r(l 10) = .575, p < .001). On its face, this correlation is unexpected, as habit
Should have little to do with intention per our conceptualization of habit. Habit is an
automatic behavior pattern in a consistent context, while intention deals with a volitional

thought process based on attitude, subjective norm and perceived behavioral control.

Why would these two variables have such a strong correlation?
A growing literature (e. g. Bem, 1972; Bentler & Speckhart, 1979; Ji & Wood,

2()07; LaRose, 2004; Wood & Neal, 2007) claims that people infer the reasons for their
be havior based on what they have already done. As individuals seek to appear rational
and internally consistent, they will claim that even non-volitional behavior patterns like

habits are the result of rational thought. Even if a person claims their intentions are very

11.8

certain and strong, such as in the case of J i and Wood, when habit is established, their

habit is more likely to guide their behavior.

In other cases, the behavior could still be being learned by an individual. As the
behavior is learned, the components of the Theory of Planned Behavior are being
constructed as a rewarded behavior is successfully repeated. The immediate outcomes
generated by engaging in the behavior are reﬂected in attitudes and intentions, which may
lead to a habit being created by repeated successes. Intention to engage in the behavior
could then be logically linked to the automation of the behavior, even if habit ultimately
o vex-rides intention’s power to predict behavior. Alternately, a developing habit could still
be being learned, and people still could be responding to their immediate expected
out comes rather than the long term average outcomes of a behavior indicative of habit.

Another reason that intention and habit may be so strongly correlated is the desire

o f people to be internally consistent. While many individuals in the sample did not have a
strong YouTube habit, those who had any habit at all would want to explain to others
Why they engaged in a behavior. Not wanting to appear weak-minded or lazy, people who
had a YouTube habit would explain their ongoing use based on a retroactive assessment
0 f their behavior. Since they had used YouTube so much over the past week, they would
Cone lude that they enjoyed using YouTube, and would likely continue using the service.
Those people who had a very strong YouTube habit may also falsely recall cognitions

that initially guided them to a YouTube habit, and impute those false reasons into their

“intention” to use YouTube. In reality, the main variable of interest in predicting

YouTube behavior in that circumstance would be habit. Habit has a signiﬁcant impact on

their "intention" to engage in the behavior because of this retroactive explanation process

119

Their actual behavior may have become more inaccessible to the person because of the
transparency involved in habit creation and continuance. Intentions are much more
ambiguous, and so can be more directly attributed to past remembrance of engaging in a
given behavior, such as YouTube use.

This strong correlation may also help to explain why when intention was included
in regression equations to predict behavior alongside habit and PBC, consistent context
and habit, and habit and self-efﬁcacy, the impact of habit appeared to be blunted. Without
adding the sophistication of a moderated relationship between habit and other variables of
interest, habit’s variance is consumed by intention because of the close relationship
between the two variables. When the role of habit in another variable is accounted for,
habit’s true inﬂuence becomes more obvious, and it becomes more easily discriminated
from the overall variance of intention.

Consistent context and habit

This dissertation found evidence that consistent context has an important role to
play in habit. Individuals who had a more stable context for their behavior tended to have
stronger habits, with even moderate consistent context having a sharp impact on habit
strength. This dissertation supports the assertion that stable, consistent context allows for
a behavior pattern to be more quickly automated, as it provides convenient prompts for
semi-automatic behavior pattems. Further, this dissertation found evidence indicating that
once a media habit develops, the role of context may be less important to maintaining a
habit. A stable context appears to make developing stronger media habits easier, but does

not appear to be necessary once a media habit is established.

Simple slope analysis allowed for a powerful display of how consistent context
influences habit strength. Initial correlations showed a weak but statistically signiﬁcant
correlation between habit and consistent context (r(l 10) = .190, p < .05). The prevailing
wisdom would indicate that this correlation would be higher and more statistically
signiﬁcant. It may be possible that media habits are generally less context-dependent than
other types of habits, which could have depressed the strength of this correlation. The
context independence of media habits has been discussed throughout this dissertation,
though a stable context does appear to be useful in the creation of all types of habits.

Most people in the study had a moderately consistent context for their YouTube
behavior. This appears to be consistent with previous studies (e.g. Ji & Wood, 2007;
Wood, Tam & Witt, 2005) in which the media use of individuals appears to have, at least
initially, a strong contextual component. Once a media habit develops, the role of context
may be less important to maintaining a habit, but a stable context appears to make
developing stronger habits easier. The point at which stable context stops being necessary
to maintain a habit is still an open question, but it is clear from the data collected that it
does have an important role in habit creation.

One reason a stable context might be important to maintaining and developing a
habit beyond merely making it easier to engage in a behavior would be a “cached”
memory of a reward associated with a behavior linked to a speciﬁc circumstance (Daw,
Niv & Dayan, 2005). This “cached” reward would encourage a person to continue
engaging in a behavior in that particular circumstance, which may eventually be
transferable to similar circumstances someone may encounter later (Crawley et al., 2002;

LaRose, 2008; Saling & Phillips, 2007).

The data reported in this dissertation supports Verplanken and Wood (2006) in.
their assertion about the importance of stable contexts to habitual behaviors, at least in
situations where habits are still developing. The authors claimed that the best way to
approach an unwanted habit is to disrupt the context the habit takes place in. This may be
the only way to alter a very strong habit that is no longer driven by a person making
volitional decisions in an effort to reach expected outcomes, whereas a weak habit may
be more easily altered by having a person change their goals to make the problematic
behavior unattractive.

For example, a beginning YouTube user will be much easier to disrupt from using
YouTube excessively by providing them with competing outlets for their goals than
someone who has developed a strong YouTube habit. Conversely, a person with a strong
YouTube habit will likely ignore competing outlets to achieve the same ends as their
YouTube use, likely fabricating false cognitions to explain their automatic behavior to
others if prompted.

As such, consistent context may be most important in situations where habits are
being established, or in cases in which habit is weak. When habit is established,
consistent context appears to become less necessary to maintenance of a habit, though
having a consistent context should have a positive impact on habit strength. Some
behaviors are very tied to a speciﬁc context (seat belt use) and other behaviors, like
media habits, may be more independent of many situational variables.

An important weakness of consistent context to consider is that there is no
theoretical explanation for what aspects of the environment are most important to keep

stable. While this study attempted to derive scenarios that appeared to have strong face

validity, it is possible that the most relevant aspects of the environment to YouTube
habits were not discussed. There are no theoretical explanations for what aspects of the
environment are most important, only that changing the environment in general will
cause changes to habit. More speciﬁcity is needed to be able to determine what is most
constructive for interventions either for or against habit.

Self-eﬂicacy and habit

This dissertation found evidence to support that self-efﬁcacy has an important
role to play in habitual behavior. Throughout the data set, self-efﬁcacy was generally
highest when habit was lowest. Habit appeared to only have slight effects on the
relationship between self-efﬁcacy and intention. When the moderating power of habit
was not accounted for, habit was a better predictor of intention than self-efficacy.

It is likely that, initially, YouTube self-efﬁcacy was a factor in encouraging the
formation of a YouTube habit. Having a sense of personal control, enhanced by mastery
experiences, would encourage a person to use YouTube more. Greater YouTube use in
turn allows for more opportunities for behaviors to become automated in the name of
more efﬁcient pursuit of desired outcomes. As cognitive automation processes become
greater, these outcomes become "cached" and no longer actively encourage behavior.
Instead, the habit itself, prompted by internal or external prompts, becomes the main
predictor of ongoing behavior rather than intentions.

Self-efﬁcacy over media use has been shown to decline as a media habit became
stronger. Understanding how to engage in a behavior, and feeling a sense of agency and
ability over that behavior, would become less relevant by that behavior becoming

habitual. This negative relationship would likely be clearer in a different behavioral

domain than YouTube use, which is not as habitual in nature as other types of media use
behavior, such as television viewing.
A rival proposal: Transaction costs and usability encourage user retention

It may be the case that YouTube use is influenced more by a combination of
rational transaction costs and usability features. A user could ﬁnd their way to YouTube
by some means (external link, web search, etc.) and ﬁnd the service to be to their liking.
YouTube has a high level of ease of use and is organized in such a way to encourage a
user to learn the basics of its operation in a short time. As YouTube provides a service
that is desirable, a user would be inclined to stay with YouTube rather than going to a
rival site. Further, the transaction costs of learning a new site are higher than the costs of
using YouTube to ﬁnd whatever video content they are seeking out. This encourages user
retention on YouTube, and has nothing to do with any automated processes.

While it may be possible that a user’s regular use of YouTube may be entirely
rational, this is at odds with interpreting the data as describing habitual processes. Some
people may be driven by volitional and rational processes in using YouTube, but other
people would not be inclined to remain thoughtful and rational about their YouTube use.
After a period of rational searching, habitual processes would begin to take place.
YouTube would become their automatic choice for ﬁnding video content rather than
other rivals. YouTube’s strong design features would predispose people to quickly
integrate its features into a habitualized video-seek and/or video-watch schema, removing
rational, volitional processing and replacing it with habitual, uncontrolled behavior

executed to achieve a goal.

124

Limitations of current research

Theoretical issues

The lower than anticipated adjusted R2 values found in the Theory of Planned
Behavior analyses may be due to a variety of factors. Attitude appeared to be the primary
predictor of YouTube behavior, with other concerns being less relevant to the majority of
the sample. Subjective norm is typically a poor predictor in the Theory of Planned
Behavior, and may have especially been so in this case because of prevalence of
YouTube use in this population. A recent Pew lntemet and American Life memo showed
that 70% of lntemet users between the ages of 18-29 had visited video sharing websites
like YouTube, with 30% using video sharing sites like YouTube on a typical day (Rainie,
2008). The ceiling effect on the subjective norm items was likely a factor in depressing
the R2 value of the regression, as there was no way to address the inﬂated values without
removing the scale from consideration entirely. Relatively high PBC values may have
made the impact of possible external barriers lower in this case than in other types of
behavior which have more challenges to their execution. Alternately, there may be so few
barriers to successfully engaging in this media use behavior that PBC may not be a
helpful predictor at all.

It may also be the case that YouTube use is not a behavior that is heavily
reasoned, making it not as well suited to being predicted by a reasoned action theory.
Engaging in YouTube use as a distraction, or to merely to view a speciﬁc video one was
linked to and then not using it ﬁrrther until the next novel incident, would limit the role of

reasoned action in the behavior.

125

The Self-Report Habit Index encapsulates some but not all of the relevant aspects
of automaticity. As automaticity is a multi-faceted concept (Saling & Phillips, 2007), it
will be important for future habit scholars to attempt to tap other elements of automaticity
that the SRHI does not examine as closely, such as lack of intentionality. By making a
habit instrument that addresses a variety of aspects of automaticity, the many different
varieties of habit can be more clearly examined. For instance, habits that rely more on
context, such as seatbelt use, can be more clearly separated from habits that may be more
independent from context, such as media habits.

YouTube use itself may present unique challenges for individuals studying habit.
While YouTube habits appear to favor environmental stability as other habits do,
YouTube use may not necessarily rely on the environment as much as non-media habits.
The increasing ubiquity of lntemet access has allowed people to take YouTube with them
almost anywhere they could imagine, and individuals with very strong YouTube habits
may not be deterred from its use even in the most unstable environment. Conversely,
college students — the primary population of this study — are known for having unstable
environmental circumstances. It may be that college students can only sustain the oldest,
strongest habits they have in the wildly changing environment they face over four years
of undergraduate education. Class schedules change ﬁom semester to semester,
dormitory space opens and closes, and off-campus housing becomes more or less
available. High-speed Internet may always be available, or it may become difﬁcult to
acquire depending on a given person’s circumstances.

The ever-changing environment of college students may lead to an artiﬁcially

deﬂated sense of perceived behavioral control. Individuals who have a more stable life

126

situation may believe that there are fewer obstacles in the way of any given behavior than
college students, who may not be able to predict how able they will be to engage in a
given behavior next month, next week or tomorrow. This depression of perceived
behavioral control does not appear to impact the self-efﬁcacy of college students, as it is
easier to control one’s internal environment than one’s external environment. Lowered
perceived behavioral control may lead to a habit being maintained, or it may lead to a
habit not being formed at all. A more “normal” population may have provided a much
different picture of YouTube use.

YouTube use may not lend itself well to the creation of strong habits, which may
have impaired the ability of the study to extract strong habitual effects. While many
people use YouTube on a daily basis, their use patterns may not truly be habitual. A
person may seek out a YouTube video presented to them via a link, view it, and then not
use YouTube again until presented with another link. Alternately, a person may use
YouTube occasionally to pass the time along with other lntemet sites as part of a broader
Internet habit, but without speciﬁcally developing a directed YouTube habit. Other new
media behaviors may lend themselves to developing stronger, clearer habits.

Methodological issues

The data has several notable problems. First, the sample was predominantly male;
while this is historically representative of the average Internet user in the late 20th
century, Internet use today is more evenly divided between genders. 53% of men and
43% of women have visited video sharing web sites like YouTube, with 20% of men and
11% of women using sites like YouTube on a typical day (Rainie, 2008). As a random

sample of students at a large Midwestern university, it would be hoped that more gender

127

parity would have been achieved by this dissertation’s sampling method. Future studies
should endeavor to have more gender balance in their analyses.

Some of the indicators were problematic. Perceived behavioral control had an
alpha value slightly below the conventional threshold of .7, which raises the possibility of
greater Type I error. There were also ceiling effects in many of the indicators that were
addressed by removing much of the instrument ﬁ'om analysis, leaving only potentially
less helpful indicators behind. By studying college students rather than the general
population, measures of self-efﬁcacy and of YouTube usage in general tended to become
biased toward the high end rather than being evenly distributed. These violations of the
expectations of the statistical methods used are likely the cause of the slightly lower
variance explained in the Theory of Planned Behavior. The initial elicitation that
generated the items may have also caused ﬂaws in the instrument by generating
responses from a subset of the student body (telecommunication students) that may have
different beliefs about media use than the general population of students.

The study was conducted near the end of an academic term, which may have
artiﬁcially created a situation of above-average context instability. Measurements that
relied on a stable context to work successfully may be attenuated and should be
considered with caution when attempting to generalize the ﬁndings of this study across
the entire population of college students. Another study should be run during the middle
of an academic term to determine if the end-of-term pressures depressed context stability
measures.

A related problem was a small number of participants in the study, likely also

caused by end-of-term pressures. While TPB research tends to be accepting of small

sample sizes, the low number of respondents limited the study’s ability to use more
powerful statistical methods. With a larger respondent pool, it may have been possible to
use structural equation modeling to analyze the data. Some of the more ambiguous
ﬁndings of the study may have become more visible methodologically if a more powerﬁil
method could have been employed. As the number of usable responses was below the
acceptable lower bounds of SEM, that method of data analysis could not be used.
However, even if SEM had been used, it may have been problematic to use with the
Theory of Planned Behavior because of as-yet unexplained multidimensionality and
complexity within the model’s variables (Rhodes & Blanchard, 2006) that is beyond the
scope of this study.

Simple slope analysis was used to test the hypotheses. However, the main
moderator of interest, habit, differed signiﬁcantly from a normal distribution. A majority
of the cases demonstrated only very low habits with a far smaller number of moderate
and high habits. While the skew of the habit distribution was within acceptable limits
(1.4), the results should be interpreted conservatively. It is suggested that replication of
the study be done with a different media behavior with established attitudes, norms and
so on. A larger number of cases would also be preferable to ease in the analysis of results.

The subjective norm was once again problematic in this study, but the subjective
norm is customarily a poor predictor in TPB studies. The type of measurement used for
the subjective norm was taken from other similar TPB studies for the sake of being
consistent with past research. A more thorough examination of subjective norm,
addressing the concerns raised by Conner and Armitage (1998), should be done in future

habit studies.

Despite these difficulties, important theoretical ﬁndings about the relationship of
habit to the theory of planned behavior were discovered. Improved instrumentation and
better data collection procedure should provide further evidence in support of the
hypotheses. The presented ﬁndings are promising, but a stronger test is required to
validate the results.

Suggestions for future research
This dissertation found its expected theory-driven results for the most part. The

results of the data collection should act as a foundation for future work in the habit area,

 

particularly in the area of the Theory of Planned Behavior. The results of this study have
shown that habit can be usefully integrated into the Theory of Planned Behavior in a
conceptually coherent way. A robust measure of habit, such as the modiﬁed SRHI,
should be used in situations where a reasoned behavior may have become habitual. Use
of the habit variable as a diagnostic tool should provide scholars with help in
understanding behaviors that should be reasoned but apparently are being executed
without actual volitional control. Additional instruments should be added to the SRHI to
measure other aspects of automaticity that the SRHI overlooks.

This study should also provide an important piece of evidence to show that many
other, earlier measures of habit are conceptually ﬂawed and should no longer be used.
The conceptually and methodologically strong measure of habit used in this study is
parsimonious and easy to implement for almost any behavior imaginable. Using measures
that rely on frequency, or measures that misunderstand how to measure an automatic
behavior, only slows down the process of scientiﬁc understanding of habits. The SRHI as

written is not perfect, but it is superior to other measures observed in the social sciences.

130

Media habits appear to be similar to other habits in favoring stable environments
over less stable environments. The degree to which media habits may be independent of
stable environments is not addressed in this study. Future work should compare media
habits and non-media habits directly to determine if there are differences to how much
instability in the environment habits of equivalent strength can tolerate before being
disrupted. Additionally, more work should be done on exploring what aspects of the
environment are most and least important to supporting habitual behaviors. A theoretical
explanation for why some aspects of the environment can (or cannot) be disrupted would
greatly improve the quality of literature in this area.

Understanding of YouTube use has improved from the study. Most people in the
sample did not have a very strong YouTube habit. This appears to be at odds with how
common YouTube use appears to be in the population studied. Additional studies on
larger student populations should be done to get a more accurate picture of who uses
YouTube and for what purpose. Further elicitations of the underlying belief structure of
YouTube use may be in order, as the initial elicitation appeared to yield only mixed
results. Are the beliefs that underlie YouTube use similar to beliefs that underlie other
new media behaviors? Understanding how and why people use services like YouTube
may provide insight into how to shape these services to encourage constructive, pro-
social habits.

Exploration of the relationship between consistent context and habit should
continue. There is a strong possibility of a nonlinear relationship between consistent
context and habit based on how habitual variables develop. Current theory would provide

support for a curvilinear relationship, wherein consistent context is very important while

131

habits are being established and becomes less important once habits become moderate or
strong. Additional examination of how consistent context interacts with the habit variable
may provide more insight into what aspects of consistent context are most important to
habit formation, helping to resolve that ongoing theoretical problem in the habit
literature.
Conclusion

If many human behaviors are habitual, understanding habitual behavior will give
society a better understanding of the human condition. Better understanding habitual
behavior also provides scholars with a means to put the numerous automatic behaviors
people engage in every day into a coherent framework. Behaviors that seem irrational
may have had their grounding in rational, volitional goal pursuit. These once conscious
behaviors were eventually made cognitively efﬁcient, a process of automation that is
normally a boon to human begins. When their schemas became inaccessible to the person
who had originally developed them, the behaviors became uncontrolled and automatic.

Understanding how media habits function provides society with a rational answer
to its irrational desire to blame complex societal problems on simple causes. The Internet
is no more to blame for antisocial behavior than the types of media that came before it.
Media consumption allows people who have goals that they cannot otherwise ﬁrlﬁll to
pursue those goals easily and inexpensively.

This study provided an illustration of how habit can diminish reasoned action
processes. The decline in predictive power of reasoned action variables that appears to
accompany a strengthening habit is a result of the cognitions that lead to volitional

control of behavior becoming more and more inaccessible as automation processes take

132

over goal pursuit. When a behavior is entirely automated, the real reasoning behind why a
behavior is done is lost in favor of false cognitions and retroactive explanations. Instead
of being able to control one’s environment, one’s environment becomes a prison that
constrains volitional action in favor of the habitual course. Resisting the habit seems to be
too much work for too little reward.

There is an appropriate place for concern about strong media habits, but there are
means to address media habits that might be unwanted. Context change, though it
remains poorly understood, can lead to even strong habits being disrupted, as can the
restoration of self-regulation.

In closing, habitual behavior can be both friend and foe. Most habits are benign
and act to make complex behaviors easier for us to execute. While there is always the
possibility of a habit becoming dysfunctional, habits should not be seen as a threat to
human happiness, but a powerful tool for creating cognitive efﬁciencies that can improve

the quality of our lives.

133

Appendix A

Survey Instrument

134

Do you use You Tube ?

Yes No
Do you use YouTube? I

IF NO: SURVEY SKIPS TO END PAGE
IF YES: SURVEY CONTINUES AS BELOW

How do you use YouTube?

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

I ﬁnd YouTube. ..

Good -3 -2 - l 0 l 2 3 Bad
Unimportant -3 -2 - ’l 0 l 2 3 Important
Useful -3 -2 - l 0 l 2 3 Useless
Relaxing -3 —2 -1 0 l 2 3 Stressful
Conventional -3 -2 - l 0 l 2 3 Diverse
Unrestricted -3 -2 - l 0 l. 2 3 Controlled
Easy -3 -2 - l 0 l 2 3 Hard
Family and Peers

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

Most people who are Agree 1 2 3 4 5 6 7 Disagree
important to me approve
of my use of YouTube.

 

Most friends who are Agree 1 2 3 4 5 6 7 Disagree
important to me think that
I should use YouTube less
often.

 

 

Most family members Agree 1 2 3 4 5 6 7 Disagree
who are important to me
want me to use YouTube
less often.

 

 

 

135

The You Tube Lifestyle

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

 

 

My family members or Agree 1 2 3 4 5 6 7 Disagree
roommates do not disrupt

my YouTube use.

[tend to use YouTube in Agree 1 2 3 4 5 6 7 Disagree
the same place.

It would be unusual for me Agree 1 2 3 4 5 6 7 Disagree

to use YouTube somewhere
other than where I am used
to using it.

 

The area where I tend to Agree 1 2 3 4 5 6 7 Disagree
use YouTube the most very
rarely changes.

 

 

I always use YouTube with Agree 1 2 3 4 5 6 7 Disagree
the same people.
It would be uncomfortable Agree 1 2 3 4 5 6 7 Disagree

for me to use YouTube
with different people.

 

 

 

 

 

I tend to use YouTube as a Agree 1 2 3 4 5 6 7 Disagree
way to change my mood.

I am usually in the same Agree 1 2 3 4 5 6 7 Disagree
kind of mood whenever I

decide to visit YouTube.

[always visit YouTube Agree 1 2 3 4 5 6 7 Disagree
using the same computer.

It would be uncomfortable Agree 1 2 3 4 5 6 7 Disagree

for me to use YouTube
from a different computer
than the one I usually use.

 

 

136

 

YouTube and Your Life

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

I don’t have a lot of

trouble ﬁnding time to use
YouTube.

Agree

1

2

3

4

5

6

7 Disagree

 

I am able to use YouTube
whenever I want to.

Agree

Disagree

 

If I lost my ﬁnancial
support (job, scholarship,
parental support, etc.), I
might not be able to keep
using YouTube in the
same way I do now.

Agree

Disagree

 

If I had to pay to use
YouTube, it would be a
big problem for me.

Agree

Disagree

 

I would be prevented from
using YouTube if my
computer had problems.

Agree

Disagree

 

I am able to use YouTube
even when my lntemet
connection is slow.

Agree

Disagree

 

I am able to ﬁnd a way to

use YouTube even when I
am not supposed to access
it.

Agree

Disagree

 

I can use YouTube even
when my school or job
attempt to block access to
it.

Agree

Disagree

 

If I had to sign up or login
to be able to use YouTube,
it would be a hassle.

Agree

Disagree

 

I can ﬁnd videos on
YouTube even if they
aren’t supposed to be there
(such as copyrighted
content).

Agree

Disagree

 

I feel in complete control
over my YouTube use.

Agree

Disagree

 

 

My use of YouTube is
completely up to me.

 

Agree

Disagree

 

137

 

You Tube Skills

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

I believe I have the ability Agree 1 2 3 4 5 6 7 Disagree
to use YouTube.

 

I am conﬁdent that I can Agree 1 2 3 4 5 6 7 Disagree
use all the functions of
YouTube.

 

I am certain that I can use Agree 1 2 3 4 5 6 7 Disagree
all the functions of
YouTube.

 

I ﬁnd YouTube to be easy Agree 1 2 3 4 5 6 7 Disagree
to use.

 

 

I can get the things that are Agree 1 2 3 4 5 6 7 Disagree
important to me out of my
YouTube use.

 

 

I38

 

 

How Do You Actually Use YouTube?

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

Using YouTube is
something I do frequently.

Agree

1

2

3

4

5

Disagree

 

Using YouTube is
something I do
automatically.

Agree

1

2

3

4

5

Disagree

 

I sometimes try to conceal
how much time I spend on
YouTube from my family
and friends.

Agree

Disagree

 

Using YouTube is
something I do without
having to consciously
remember.

Agree

Disagree

 

Using YouTube is
something that makes me
feel weird if I do not do it.

Agree

Disagree

 

I feel guilty about the

amount of time I spend on
YouTube.

Agree

Disagree

 

Using YouTube is
something I have to keep
doing more and more to get
my thrill.

Agree

Disagree

 

Using YouTube is
something I have a hard
time keeping under control.

Agree

Disagree

 

Using YouTube is
something I do without
thinking.

Agree

Disagree

 

Using YouTube is
something that would
require effort not to do it.

Agree

Disagree

 

I would go out of my way
to satisfy my urge to use
YouTube.

Agree

Disagree

 

Using YouTube is
something that belongs to
my daily routine.

Agree

Disagree

 

Using YouTube is
something I start doing
before I realize I’m doing
it.

Agree

Disagree

 

Using YouTube is
something I would ﬁnd
hard not to do.

Agree

Disagree

 

 

Using YouTube is
something I feel guilty
about spending so much
time on.

 

Agree

Disagree

 

I39

 

 

 

Using YouTube is Agree 1 2 3 4 5 6 7 Disagree
something I have no need
to think about doing.

 

Using YouTube is Agree 1 2 3 4 5 6 7 Disagree
something I would go out
of my way to satisfy my
upge to use.

 

Using YouTube is Agree 1 2 3 4 5 6 7 Disagree
something that’s typically

5 9

me.

 

 

Using YouTube is Agree 1 2 3 4 5 6 7 Disagree
something I have been
doing for a long time.

 

 

Your Usage Of The Electronic Media

Instructions: Please select one number in each row. The closer the number is to a descriptive word,
the more strongly you feel that word represents your answer to the statement.

 

 

 

Do you intend to use Deﬁnitely 1 2 3 4 5 6 7 Not At All
YouTube at home today?
Do you intend to continue Deﬁnitely 1 2 3 4 5 6 7 Not At All

to use YouTube at home
over this coming week?

 

 

Instructions: Please answer the following questions in hours and minutes in the boxes provided.
Hrs. Mins.

How long do you use YouTube on the average week day? |_| |_|

How long do you use YouTube on the average weekend day? |_| |_|

140

 

 

You’re almost done! These last questions are about you. Again, all of your responses will be kept
strictly conﬁdential, so please answer accurately and honestly.

Are you male OR female?
What is your year of birth?

Are you Spanish, Hispanic or Latino?

 

|_| Yes

Ll Mexican, Mexican-American, Chicano

|_| Puerto Rican

|_| Cuban

l_| Other Spanish/Hispanic/ Latino — please ﬁll in the blank
H No

What is your race?

|_| White
|_| Black, African American or Negro
|_| American Indian or Alaska Native
|_| Asian Indian
|_| Chinese
l_| Filipino
|_| Japanese

|_| Korean

|_| Vietnamese

|_| Other — please ﬁll in the blank

 

Please enter your email to be entered into our rafﬂe for an iPod Shufﬂe:

 

Thank you for completing our survey!

141

A ﬂer completing the questiommire. this briefinstrument was emailed to participants who answered ”yes "
to the You Tube ﬁlter question to complete the next week to compare their actual behavior to their
previously stated intention.

YouTube Usage Followup

 

 

How often did you use Often I 2 3 4 5 6 7 Rarely
YouTube over the last
week?

 

 

Please enter your email address to verify your participation:

 

Thank you for your participation!

142

 

References

Aarts, H., Verplanken, .B., & van Knippenberg, A. (1998). Predicting behavior from
actions in the past: Repeated decision-making or a matter of habit? Journal of
Applied Social Psychology, 28, 1355—1374.

Abelson, R. P. (1981). Psychological status of the script concept. American Psychologist,
36, 715-729.

Ajzen, I. (1991). The Theory of Planned Behavior. Organizational Behavior and Human
Decision Processes, 50 (2), 179-211.

Ajzen, I. (2002a). Residual effects of past on later behavior: Habituation and reasoned
action perspectives. Personality and Social Psychology Review, 6 (2), 107-122.

Ajzen, I. (2002b). Perceived behavioral control, self-efﬁcacy, locus of control. and the

Theory of Planned Behavior. Journal of Applied Social Psychology, 32 (4), 665-
683.

Armitage, C. J. (2005). Can the Theory of Planned Behavior predict the maintenance of
physical activity? Health Psychology, 24 (3), 235-245.

Armitage, C. J ., & Conner, M. (2001). Efficacy of the Theory of Planned Behavior: A
meta-analytic review. British Journal of Social Psychology, 40, 471-499.

Anderson, C. A., Berkowitz, L., Donnerstein, E., Huesmann, L. R., Johnson, J. D., Linz,
D., Malamuth, N. M., & Wartella, E. (2003). The inﬂuence of media violence on
youth. Psychological Science in the Public Interest, 4 (3), 81—110.

Anderson, D. R., Collins, P. A., Schmitt, K. L., & Jacobvitz, R. S. (1996). Stressful life
events and television viewing. Communication Research, 23 (3), 243-260.

Armitage, C. J. (2005). Can the Theory of Planned Behavior predict the maintenance of
physical activity? Health Psychology, 24 (3), 235-245.

Babrow, A. S., & Swanson, D. L. (1988). Disentangling the antecedents of audience
exposure levels: Extending expectancy-value analyses of gratiﬁcations sought

from television news. Communication Monographs, 55, 1-21.

Bagozzi, R. P. (1981). Attitudes, intentions and behavior: A test of some key hypotheses.
Journal of Personality and Social Psychology, 41 (4), 607-627.

Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist,
37, 122-147.

143

 

 

Bandura, A. (1991). Social cognitive theory of self-regulation. Organizational Behavior
and Human Decision Processes, 50, 248-287.

Bargh, J. A. (2002). Losing consciousness: Automatic inﬂuences on consumer judgment,
behavior and motivation. Journal of Consumer Research, 29 (2), 280-285.

Bargh, J. A., & Chartrand, T. L. (1999). The unbearable automaticity of being. American
Psychologist, 54, 462-479.

Bargh, J. A., Chen, M., & Burrows, L. (1996). Automaticity of social behavior: Direct
effects of trait construct and stereotype activation on action. Journal of
Personality and Social Psychology, 71 (2), 230-244.

Bargh, J. A., Gollwitzer, P. M., Chai, A., Bamdollar, K., & Troschel, R. (2001). The
automated will: Nonconscious activation and pursuit of behavioral goals. Journal
of Personality and Social Psychology, 81 (6), 1014-1027.

Bargh, J. A., & Pietromonaco, P. (1982). Automatic information processing and social
perception: The inﬂuence of trait information presented outside of conscious
awareness on impression formation. Journal of Personality and Social
Psychology, 43, 437—449.

Barker, R. G., & Schoggen, P. (1978). Measures of habitat and behavior output. In R. G.
Barker (Ed.), Habitats, environments, and human behavior: Studies in eco-
behavioral science from the Midwest Psychological Field Station (pp. 229-244).
Stanford, CA: Stanford University Press.

Baumeister, R. F ., & Heatherton, T. D. (1996). Self-regulation failure: An
overview. Psychological Inquiry, 7, 1—15.

Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in
experimental social psychology (Vol. 6, pp. 1-62). New York: Academic Press.

Bentler, P. M., & Speckart, G. (1979). Models of attitude-behavior relations.
Psychological Review, 86, 452-464.

Bhattacherjee, A. (2001). Understanding information systems continuance: An
expectation-conﬁrmation model. MS Quarterly, 25 (3), 351-370.

Brug, J., de Vet, E., de Nooijer, J., & Verplanken, B. (2005). Predicting fruit
consumption: Cognition, motivation, and habits. Journal of Nutrition and
Behavior, 38, 73—81.

Caplan, S. E. (2002). Problematic intemet use and psychosocial well-being: Development
of a theory-based cognitive-behavioral measurement instrument. Computers in
Human Behavior, 18, 553-575.

144

Caplan, S. E. (2005). A social skill account of problematic internet use. Journal of
Communication, 55 (4), 721-736.

Chen, G., Gully, S. M., & Eden, D. (2001). Validation of a new general self-efﬁcacy
scale. Organizational Research Methods, 4 (1), 62-83.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112 (1), 155-159.

Conner, M., & Armitage, C. J. (1998). Extending the Theory of Planned Behavior: A
review and avenues for ﬁrrther research. Journal of Applied Social Psychology, 28
(15), 1429-1464.

Crawley, A. M., Anderson, D. R., Santomero, A., Wilder, A., Williams, M., Evans, M.
K., & Bryant, J. (2002). Do children learn how to watch television? The impact of

extensive experience with Blue ’s Clues on preschool children’s television viewing
behavior. Journal of Communication, 52 (2), 264-280.

Csikszentmihalyi, M., & Kubey, R. (1981). Television and the rest of life: A systematic
comparison of subjective experience. Public Opinion Quarterly, 45 (3), 317-328.

Dahlstrand, U., & Biel, A. (1997). Pro-environmental habits: Propensity levels in
behavioral change. Journal of Applied Social Psychology, 27 (7), 588-601.

Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of
information technology. MIS Quarterly, 13 (3), 319—341.

Daw, N. D., Niv, Y., & Dayan, P. (2005). Uncertainty based competition between
prefrontal and dorsolateral striatal systems for behavioral control. Nature
Neuroscience, 8, 1704—1711.

Dewey, J. (1922). Human Nature and Conduct. New York: Henry Holt.

Dollard, J., & Miller, N. E. (1941). Social learning and imitation. London, England:
Oxford University Press.

Dunlap, K. (1949). Habits: Their Making and Unmaking. New York: Liveright.

Elliott, M. A., Armitage, C. J., & Baughan, C. J. (2003). Drivers’ compliance with speed
limits: An application of the Theory of Planned Behavior. Journal of Applied
Psychology, 88 (5), 964-972.

Ellison, N. B., Steinﬁeld, C., & Lampe, C. (2007). The beneﬁts of Facebook "ﬁiendsz"
Social capital and college students' use of online social network sites. Journal of
Computer-Mediated Communication, 12 (4). Available online:
http:/Licmcindianaedu/vol12/issue4/ellisonhtml

 

145

Fallows, D. (2004, August). The lntemet in daily life. Available online:
http://www.pewinternet.orglpdfs/PlP_lnternet_and_Daily__Life.pdf

F iske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, MA: Addison-Wesley.

Godin, G., Valois, P., & Lepage, L. (1993). The pattern of inﬂuence of perceived

behavioral control upon exercising behavior: An application of Ajzen's Theory of
Planned Behavior. Journal of Behavioral Medicine, I 6 (1), 81-102.

Griffiths, M., & Wood, R. T. (2000). Risk factors in adolescence: The case of gambling,
video game playing, and the lntemet. Journal of Gambling Studies, 16 (2-3), 199-
225.

Guariglia, A., & Rossi, M. (2002). Consumption, habit formation, and precautionary
saving: Evidence from the British Household Panel Survey. Oxford Economic
Papers, 54(1), 1-19.

Hassin, R. R., Aarts, H., & Ferguson, M. J. (2005). Automatic goal inferences. Journal of
Experimental Social Psychology, 41 (2), 129-140.

Heatherton, T. F., & Nichols, P. A. (1994). Personal accounts of successful versus failed
attempts at life change. Personality and Social Psychology Bulletin, 20, 664-675.

Hogarth, R. M., & Einhom, H. J. (1992). Order effects in belief updating: The belief
adjustment model. Cognitive Psychology, 24 (1), 1—55.

Honkanen, P., Olsen, S.O., & Verplanken, B. (2005). Intention to consume sea food:
The importance of habit strength. Appetite, 45, 161—168.

Horrigan, J. B. (2006, May). Home broadband adoption 2006. Available online:
http ://www.pewintemet.org/pdfs/PI P_Broadband_trend52006.pdf

I gbaria, M., Guimaraes, T., & Davis, G. B. (1995). Testing the determinants of
microcomputer usage via a structural equation model. Journal of Management
Information Systems, 11 (4), 87-114.

James, W. (1890). Principles of psychology. Available online:
http://psychclassicsyorku.ca/James/Pringples/index.htm

J i, A. W., & Wood, W. (2007). Purchase and consumption habits: Not necessarily what
you intend. Journal of Consumer Psychology, 1 7, 261-276.

Kawada, C., Oettingen, G., Gollwitzer, P. M., & Bargh, J. A. (2004). The projection of

implicit and explicit goals. Journal of Personality and Social Psychology, 86 (4),
545-559.

146

Kim, S. S., & Malhotra, N. K. (2005). A longitudinal model of continued is use: An

integrative view of four mechanisms underlying postadoption phenomena.
Management Science, 51, 741-755.

Knussen, C., Yule, F ., MacKenzie, J., & Wells, M. (2004). An analysis of intentions to
recycle household waste: The roles of past behaviour, perceived habit, and
perceived lack of facilities. Journal of Environmental Psychology, 24, 237-246.

Kubey, R. (1984). Leisure, television and subjective experience. Unpublished doctoral
dissertation, University of Chicago.

Kubey, R. (1990). Psychological dependence on television: Application of DSM-III—R
criteria and experience sampling method ﬁndings. Paper presented at the annual
meeting of the American Psychological Association, Boston.

Kubey, R., & Csikszentmihalyi, M. (1990). Television and the quality of life: How
viewing structures everyday experience. Hillsdale, NJ: Lawrence Erlbaurn.

Landis, D., Triandis, H. C., & Adamopoulos, J. (1978). Habit and behavioral intentions
as predictors of social behavior. Journal of Social Psychology, 106 (2), 227-237.

Landman, J. & Manis, M. (1983). Social cognition: Some historical and theoretical

perspectives. In L. Berkowitz (Ed), Advances in experimental social psychology
(Vol. 16, pp. 49-123). New York: Academic Press.

Lange, R. & Lampe, C. (2008). Feeding the privacy debate: An examination of
F acebook. Paper presented at the 58th annual meeting of the International
Communication Association, Montreal, Quebec, Canada.

Langer, E. (1989). Mindfulness. Reading, Mass: Addison-Wesley.

LaRose, R. (2004). Cybercompulsions: Media habits, media addictions and the intemet.
In P. Lee, L. Leung & C. So (Eds), Impact and issues in new media: Toward
intelligent societies. Cresskill, NJ: Hampton Press.

LaRose, R. (2008). Media habits. Unpublished manuscript.

LaRose, R., Kim, J ., & Peng, W. (2008). The problem of problematic Internet use: Anti-
social disease, dangerous habit, or just one of our favorites? Paper presented at the
94th annual meeting of the National Communication Association, San Diego,
California.

LaRose, R., Lin, C. A., & Eastin, M. S. (2003). Unregulated lntemet usage: Addiction,
habit or deﬁcient self-regulation? Media Psychology, 5, 225-253.

147

LaRose, R., Mastro, D., & Eastin, M. (2001). Understanding lntemet usage. Social
Science Computer Review, 19 (4), 395-413.

Limayem, M., Hirt, S. G., & Cheung, C. M. K. (2007). How habit limits the predictive
power of intention: The case of information systems continuance. MIS Quarterly,
31, 705-738.

Lindbladh, E., & Lyttkens, C. H. (2002). Habit versus choice: The process of decision-
making in health-related behavior. Social Science and Medicine, 55 (3), 451-463.

Madden, M. (2007, July). Online video. Available online:
http://www.pewintemet.org/pdfs/PlPmOnline_Video__2007.pdf

Mahon, D., Cowan, C., & McCarthy, M. (2006). The role of attitudes, subjective norm,

perceived control and habit in the consumption of ready meals and takeaways in
Great Britain. Food Quality and Preference, 1 7, 474-481.

Manstead, A., & van Eekelen, S. (1998). Distinguishing between perceived behavioral
control and self-efﬁcacy in the domain of academic achievement intentions and
behaviors. Journal of Applied Social Psychology, 28 (15), 1375-1392.

McIlwraith, R. (1998). ‘I’m addicted to television’: The personality, imagination, and TV
watching patterns of self-identiﬁed TV addicts. Journal of Broadcasting and
Electronic Media, 42 (3), 371-387.

McIlwraith, R., Jacobvitz, R., Kubey, R., & Alexander, A. (1991). Television addiction:
Theories and data behind the ubiquitous metaphor. American Behavioral
Scientist, 35 (2), 104-121.

Meng, X., Rosenthal, R., & Rubin, D. B. (1992). Comparing correlated correlation
coefficients. Psychological Bulletin, 111 (1), 172-175.

Morahan—Martin, J ., & Schumacher, P. (2000). Incidence and correlates of pathological
lntemet use among college students. Computers in Human Behavior, 16, 13-29.

Morris, M., & Ogan, C. (1996). The Internet as mass medium. Journal of Computer-
Mediated Communication, 1 (4). Available online:
mpzl/jcmcjndianaedu/vol1/issue4/morris.html

Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of limited
resources: Does self-control resemble a muscle? Psychological Bulletin, 126,
247—259.

Newell, J. (2003). The role of habit in the selection of electronic media. Unpublished
doctoral dissertation, Michigan State University.

148

Newell, J., & LaRose, R. (2004). Habitual and intentional consumption of electronic
media. Paper presented at AEJMC, Toronto, Canada.

Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports
on mental processes. Psychological Review, 84, 231-259.

O'Connor, B. P. (1998). SIMPLE: All-in-one programs for exploring interactions in
moderated multiple regression. Educational and Psychological Measurement, 58,
833-837.

Oettigen, G., Grant, H., Smith, P. K., Skinner, M., & Gollwitzer, P. M. (2006).
Nonconscious goal pursuit: Acting in an explanatory vacuum. Journal of
Experimental Social Psychology, 42, 668-675.

Orbell, 8., Blair, C., Sherlock, K., & Conner, M. (2001). The Theory of Planned Behavior
and ecstasy use: Roles for habit and perceived control over taking versus
obtaining substances. Journal of Applied Social Psychology, 31 (1), 31-47.

Ouellette, J. A. (1996). How to measure habit? Subjective experience and past behavior.
Unpublished doctoral dissertation, Texas A&M University.

Ouellette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple
processes by which past behavior predicts ﬁrture behavior. Psychological
Bulletin, 124 (1), 54-74.

Quinn, J. M., & Wood, W. (2004). Habits across the lifespan. Manuscript accepted for
publication.

Quinn, J. M. & Wood, W. (2006). Inhibiting habits and temptations: Depends on
motivational orientation. Manuscript under editorial review.

Rainie, L. (2008, January). Video sharing websites. Available online:
http://www.pewintemet.org/gdfs/Pew_Videosharing_memo_Jan08.pdf

Rainie, L., & Horrigan, J. (2005, January). lntemet: The mainstreaming of online life.
Available online: http://www.pewinternetorg’pdfs/Intemet_Status_2005.pdf

Rhodes, R. E., & Blanchard, C. M. (2006). Conceptual categories or operational
constructs? Evaluating higher order Theory of Planned Behavior structures in the
exercise domain. Behavioral Medicine, 31 (4), 141-150.

Rhodes, R. E., & Coumeya, K. S. (2003). Self-efﬁcacy, controllability and intention in

the Theory of Planned Behavior: Measurement redundancy or causal
independence? Psychology and Health, 18 (1), 79-91.

149

Rosenstein, A. W., & Grant, A. E. (1997). Reconceptualizing the role of habit: A new
model of television audience activity. Journal of Broadcasting and Electronic
Media, 41 (3), 324-344.

Saling, L. L., & Phillips, J. G. (2007). Automatic behaviour: Efﬁcient not mindless. Brain
Research Bulletin, 73, 1-20.

Schmitt, K. L., Woolf, K. D., & Anderson, D. R. (2003). Viewing the viewers: Viewing
behaviors by children and adults during television programs and commercials.
Journal of Communication, 53 (2), 265-281.

Shapira, N. A., Goldsmith, T. D., Keck, P. E., Khosla, U. M., & McElroy, S. L. (2000).
Psychiatric features of individuals with problematic internet use. Journal of
Affective Disorders, 57, 267-272.

Sheeran, P., Aarts, H., Custers, R., Rivis, A., Webb, T. L., & Cooke, R. (2005). The goal-
dependent automaticity of drinking habits. British Journal of Social Psychology,
44, 47-63.

Sheeran, P., Webb, T. L., & Gollwitzer, P. (2005). The interplay between goal intentions

and implementation intentions. Personality and Social Psychology Bulletin, 31,
87—98.

Song, 1., LaRose, R., Eastin, M. S., & Lin, C. A. (2004). Internet gratiﬁcations and
Internet addiction: On the uses and abuses of new media. CyberPsychology and
Behavior, 7 (4), 384-394.

Sparks, P., Guthrie, C. A., & Shepherd, R. (1997). The dimensional structure of the

perceived behavioral control construct. Journal of Applied Social Psychology, 27
(5), 418-438.

Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix. Psychological
Bulletin, 87, 245-251.

Story, L. (2007, January 15). Anywhere the eye can see, it's now likely to see an ad.
Available online:
htfp://www.nytimes.com/2007/0 l /l 5/business/media/l 5evervﬁwhere.html?_r=l &s
g=arlywhere%20the%20eye%20can%205ee&st=cse&oref=slogin&scp=l &pagew

anted=print

Terry, D. J ., & O’Leary, J. E. (1995). The Theory of Planned Behavior: The effects of
perceived behavioral control and self-efficacy. British Journal of Social
Psychology, 34 (2), 199-220.

150

Traﬁmow, D. (2000). Habit as both a direct cause of intention to use a condom and as a
moderator of the attitude-intention and subjective norm-intention relations.
Psychology and Health, 15 (3), 383-393.

Triandis, H. C. (1979). Values, attitudes and interpersonal behavior. Nebraska
Symposium on Motivation, 27, 195-259.

Verplanken, B. (2006). Beyond ﬁequency: Habit as mental construct. British Journal of
Social Psychology, 45, 639-656.

Verplanken, B., & Aarts, H. (1999). Habit, attitude, and planned behaviour: Is habit an
empty construct or an interesting case of goal-directed automaticity? European
Review of Social Psychology, 29, 591-604.

Verplanken, B., Aarts, H., van Knippenberg, A., & Moonen, A. (1998). Habit versus
Planned behaviour: A ﬁeld experiment. Unpublished manuscript, University of
Nijmegen.

Verplanken, B., Aarts, H., van Knippenberg, A., & van Knippenberg, C. (1994). Attitude
versus general habit. Journal of Applied Social Psychology, 24, 285-300.

Verplanken, B., & Faes, S. (1999). Good intentions, bad habits and effects of forming

implementation intentions on healthy eating. European Journal of Social
Psychology, 29, 591-604.

Verplanken, B., Friborg, 0., Wang, C. E., Traﬁmow, D, & Woolf, K. (2007). Mental
habits: Metacognitive reﬂection on negative self-thinking. Journal of Personality
and Social Psychology, 92, 526-541.

Verplanken, B., Herabadi, A. G., Perry, J. A., & Silvera, D. H. (2005). Consumer style
and health: The role of impulsive buying in unhealthy eating. Psychology and
Health, 20, 429—441.

Verplanken, B., Myrbakk, V., & Rudi, E. (2005). The measurement of habit. In T. Betsch
& S. l-Iaberstroh (Eds), The routines of decision making (pp. 231-247). Mahwah,
NJ: Erlbaum.

Verplanken, B., & Orbell, S. (2003). Reﬂections on past behavior: A self-report index of
habit strength. Journal of Applied Social Psychology, 33 (6), 1313—1330.

Verplanken, B., & Wood, W. (2006). Interventions to break and create consumer habits.
Journal of Public Policy & Marketing, 25 (1), 90-103.

Watson, J. B. (1924). Behaviorism. New York: Norton.

151

Wegner, D. M., & Bargh, J. A. (1998). Control and automaticity in social life. In D.
Gilbert, S. T. F iske & G. Lindzey (Eds), Handbook ofsocial psychology (4th ed.,
pp. 446-496). New York: McGraw-Hill.

Winn, M. (2002). The plug-in drug: Television, computers and family life. Penguin
Books: New York.

Wittenbraker, J ., Gibbs, B. L., & Kahle, L. R. (1983). Seat belt attitudes, habits and
behaviors: An adaptive amendment to the F ishbein model. Journal of Applied
Social Psychology, 13 (5), 406—421.

Wood, W., & Neal, D. T. (2007). A new look at habits and the habit-goal interface.
Psychological Review, 114, 843-863.

Wood, W., Tam, L., & Witt, M. G. (2005). Changing circumstances, disrupting habits.
Journal of Personality and Social Psychology, 88 (6), 918-933.

Wood, W., Quinn, J. M., & Kashy, D. (2002). Habits in everyday life: The thought and
feel of action. Journal of Personality and Social Psychology, 83 (6), 1281-1297.

Yee, N. (2004). Addiction. The Daedalus Project: The psychology of MMORPGS.
Retrieved on July 10, 2006 from
http://www.nickyeecom/daedalus/archives/000818.php

 

Yee, N. (2006). A new disorder is born. Downloaded on May 24, 2006 from
http://wwwnickyee.com’daedalus/archives/OO1494th

Yellowlees, P. M., & Marks, 8., (2007). Problematic lntemet use or lntemet addiction?
Computers in Human Behavior, 23 (3), 1447-1453.

Young, K. S. (1998). Internet addiction: The emergence of a new clinical disorder.
CyberPsychology and Behavior, 1 (3), 237-244.

YouTube. (2008). YouTube Fact Sheet. Available online:
http://wwwyoutube.com/t/fact_ sheet

 

152

   

M'llllllllljjlllﬁllljlllllllllllllles