International Journal of Human-Computer Interaction

ISSN: 1044-7318 (Print) 1532-7590 (Online) Journal homepage: http://www.tandfonline.com/loi/hihc20

Measuring Attitudes Towards the Internet: The

General Internet Attitude Scale

Mary Joyce & Jurek Kirakowski

To cite this article: Mary Joyce & Jurek Kirakowski (2015) Measuring Attitudes Towards the
Internet: The General Internet Attitude Scale, International Journal of Human-Computer
Interaction, 31:8, 506-517, DOI: 10.1080/10447318.2015.1064657

To link to this article: http://dx.doi.org/10.1080/10447318.2015.1064657

Accepted author version posted online: 24

Jun 2015.

Submit your article to this journal

Article views: 131

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=hihc20

Download by: [Monash University Library] Date: 25 November 2015, At: 09:01
 Intl. Journal of Human–Computer Interaction, 31: 506–517, 2015
Copyright © Taylor & Francis Group, LLC
ISSN: 1044-7318 print / 1532-7590 online
DOI: 10.1080/10447318.2015.1064657

Measuring Attitudes Towards the Internet: The General Internet

Attitude Scale
Mary Joyce and Jurek Kirakowski
School of Applied Psychology, University College Cork, Cork, Ireland

psychometric scales. However, these studies failed to employ

The General Internet Attitude Scale (GIAS) is a questionnaire conventionally accepted methodologies for the measurement of
Downloaded by [Monash University Library] at 09:01 25 November 2015

designed to explore the underlying components of the attitudes of self-efficacy (a theoretical critique of this research may be found
individuals to the Internet, and to measure individuals on these in Joyce & Kirakowski, 2014). As a result, the validity of these
attitude components. Previous Internet attitude research is cri-
tiqued for its lack of a clear definition of constructs. GIAS was
scales and emerging conclusions are difficult to ascertain.
developed starting from the well-established three-component psy- Similar issues arose in studies of concepts related to self-
chological model of attitude (affect, behavior, cognition) into which efficacy and individual differences in online behavior (e.g.,
applicable statements found in previous Internet attitude measures Durndell & Haag, 2002; Tsai, Lin, & Tsai, 2001). Some
were fitted. GIAS was developed using an iterative psychometric researchers (e.g., Jackson et al., 2003; Schumacher & Morahan-
process with four independent samples (N = 2,200). During iter-
ations, the wordings of the items were refined, and exploratory
Martin, 2001) who were interested in patterns of Internet usage
and confirmatory factor analyses identified four underlying fac- hypothesised that Internet attitudes may influence Internet use,
tors in the scale: Internet Affect, Internet Exhilaration, Social which would be consistent with a general social psychological
Benefit of the Internet, and Internet Detriment, all of which explanation (e.g., Aronson, Wilson, & Akert, 2002). As a result,
had acceptable internal reliabilities. The final instrument con- a number of researchers went about developing scales to mea-
tains 21 items and demonstrates strong reliability achieving an
overall Cronbach’s alpha value of 0.85. The behavioral com-
sure Internet attitudes. However, in attempts at developing an
ponent of the three-factor attitude model could not be repli- Internet attitude measure, the major issue of concern was the
cated, although there was a medium, positive correlation between lack of a recognized theoretical framework. On one hand, con-
GIAS and a measure of Internet self-efficacy. Attitude and self- structs that are not normally considered to be attitudinal were
efficacy are important personal constructs and may well con- attributed to attitude (e.g., Tsai & Lin, 2004; Zhang, 2007),
tribute to the large variance that usability metrics are known to
exhibit.
and on the other, items that did not relate to any recognized
attitudinal construct were included in initial item pools that pri-
marily describe Internet uses rather than attitudes (e.g., Morse,
Gullekson, Morris, & Popovich, 2011; Tsai et al., 2001). Thus,
1. INTRODUCTION whatever such scales were measuring, it was not attitude as
Our observations of individuals’ reactions when completing conventionally understood within psychology (a more detailed
online tasks for a website evaluation led us to this research topic. critique of previous research on Internet attitude may be found
It was observed that some individuals attributed their difficulties in Joyce & Kirakowski, 2013).
navigating the website to themselves, stating that it was their Taking the aforementioned issues into account, it became
skills (or lack thereof) that contributed to their failure in locat- clear that there was a need for a standardised psychometric
ing the required information to complete the task. Others were scale, recognizable as such from well-established social psy-
inclined to attribute their difficulties to the website, sometimes chological theory, which would measure attitude to, and self-
commenting that the website was difficult to use or poorly pre- efficacy in an Internet environment. The major objective of this
sented. Examination of the literature yielded surprisingly few article is thus to outline the development of a scale that mea-
explanations in any quantitative sense for such discrepancies sures the attitudes of individuals to the Internet. We differentiate
in individual responses. Some studies (e.g., Eastin & LaRose, ourselves from research that attempts to predict specific sys-
2000; Torkzadeh & Van Dyke, 2001) attempted to measure indi- tem usage (e.g., SUMI, Kirakowski & Corbett, 1993; TAM,
viduals’ confidence in using the Internet through the use of Davis, Bagozzi, & Warshaw, 1989), to measure the possibly
related concept of computer anxiety (e.g., Heinssen, Glass, &
Address correspondence to Mary Joyce, School of Applied Knight, 1987), or to assess attitudes to computer technology
Psychology, University College Cork, Cork Enterprise Centre, North
in general (e.g., Nickell & Pinto, 1986). We began by drawing
Mall, T23 K208 Cork, Ireland. E-mail: marychjoyce@gmail.com

506
 MEASURING ATTITUDES TOWARDS THE INTERNET 507

on the social psychological literature on attitudes to provide a social psychological definition of attitude and then furthermore
theoretical foundation for this research. attempted to measure self-efficacy in terms of past behavior.
We sought therefore to find a classic attitudinal component
relating to behavioral intention in our questionnaire, which is
2. THE CONSTRUCT OF ATTITUDE
in line with recent social psychological theories. As outlined in
Since 1918, when Thomas and Znaniecki (as cited in Allport, section 1 however, it was also noted that attempts at measur-
1935) initiated the discussion about attitudes, much debate ing Internet self-efficacy independent of Internet attitude was
has taken place among psychologists about the definition and vitiated by methodological issues. Thus, because self-efficacy
structure of this construct. However, most psychologists today itself is also of concern (although is not a recognizable aspect
now agree that the emphasis lies on the feeling and evalua- of attitude), we also considered that a tool for the measurement
tive elements of attitude. With this in mind, an attitude can of Internet self-efficacy, uncontaminated by attitude, should be
be defined as “a psychological tendency that is expressed by developed in parallel. While the theory behind the construct of
evaluating a particular entity with some degree of favour or self-efficacy and its measurement has been described in detail
disfavour” (Eagly & Chaiken, 2007, p. 598). One-, two-, and elsewhere (Joyce & Kirakowski, 2014), in brief, self-efficacy
three-component models of attitude were proposed over the focuses on judgments of perceived capabilities to perform a
earlier half of the 20th century (e.g., Allport, 1935; Katz & task or activity and is measured by asking individuals to rate
Downloaded by [Monash University Library] at 09:01 25 November 2015

Stotland, 1959; Thurstone, 1931). However, researchers were the strength of their belief in completing a specific activity on a
swift to adopt Katz and Stotland’s (1959) proposition of a self-rating scale. By closely following recommendations for the
three-component attitude model (e.g., Krech, Crutchfield, & creation of a self-efficacy scale, an Internet Self-Efficacy Scale
Ballachey, 1962; Rosenberg & Hovland, 1960), and this has was also developed for the purposes of this research. The essen-
since remained the modal model of attitude in social psycho- tial features of this method are that respondents are invited to
logical theory. The three components are briefly described as rate a list of activities carried out on the Internet on (a) how
affect—an emotion that charges the idea, a feeling that may be frequently they engage in the activities and (b) how confident
good or bad when thinking about the attitude object; behav- they felt they were on these same activities. The Internet Self-
ioral intention—the individual’s predisposition to action with Efficacy score is then an average of the self-reported confidence
regard to the attitude object; and cognition—the beliefs and ratings about completing the activities, weighted by frequency
ideas a person has about the attitude object. However, there of use.
has been a lack of empirical evidence investigating such theo- The Internet Self-Efficacy Scale was developed in parallel
ries. In fact, very few studies have explored the existence of the with the Internet attitude measure in the later stages of the
three-component model of attitudes in any domain. The existing research reported here (Studies 3 and 4). Our hypothesis was
research that did explore the three components and their rela- that Internet Self-Efficacy, if measured correctly, would form a
tionship with one another (e.g., Breckler, 1984; Woodmansee & component moderately correlated with the measure of attitude
Cook, 1967) found inconsistent evidence for the presence of the when conventional methodologies for attitude measurement
three components. Nonetheless, although there may not be suf- were followed. The Internet Self-Efficacy Scale is presented in
ficient evidence to prove this model beyond dispute, it is the best Appendix B.
working hypothesis that social psychology at present possesses
for relating dispositions to behaviors.
4. DEVELOPMENT OF THE GENERAL INTERNET
ATTITUDE SCALE
3. ATTITUDE AND SELF-EFFICACY
The research reported in this article therefore took as its 4.1. Scale Development
starting point the three-factor model, which social psychologists The three-component model was therefore the starting point
agree is the ‘modal model’ of attitude. Earlier two-factor mod- for the general internet attitude scale (GIAS). It initially con-
els, which the three-component model effectively supplanted in sisted of items relating to affect (feelings, likes/dislikes about
social psychology, did not explicitly incorporate a ‘behavioral the Internet), behavioral intention (the intention to act a certain
intention’ component but concentrated on ‘affect’ and to some way on the Internet), and cognition (beliefs and cognitions of
extent ‘cognition’ or beliefs. individuals about the Internet).
The motivation for following the three-component model The scale takes the form of a Likert summated ratings
was reinforced by the view that many of the studies that investi- scale—one of the most frequently used tools in the social sci-
gated Internet attitude (see section 1) obviously considered that ences when measuring attitudes (Spector, 1992). There are four
behavior in some form was indeed a component of attitude. characteristics of a summated rating scale: (a) the scale must
Past research (e.g., Tsai & Lin, 2004; Zhang, 2007) incor- contain multiple items; (b) each item must measure something
rectly included self-efficacy as a subscale of Internet attitude, that has an underlying, quantitative measurement continuum;
and then incorrectly attempted to measure the self-efficacy con- (c) there are no “right” or “wrong” answers to items; and (d)
struct in terms of past behavior. By conceptualizing attitude each item is a statement and respondents are invited to give a
as self-efficacy, however, these researchers stepped out of the rating for each statement (Spector, 1992).
 508 M. JOYCE AND J. KIRAKOWSKI

4.2. Item Generation 4.3. Overview of Studies

Existing questionnaires for the measurement of Internet atti- The creation of a psychometric scale is an iterative pro-
tudes were consulted when generating items for the GIAS. cess; thus the development of the GIAS took place across four
An initial item pool of 97 statements was used in the first studies. Study 1 involved the testing of Version 1 of the scale.
stage of scale development. Statements from four previous The objectives of Study 1 were (a) to discover whether a factor
questionnaires that exemplified the best attempts at creating structure underlay the individual items; (b) if a factor structure
items depicting Internet attitudes in the past decade were col- emerged, to delete items that clearly did not belong to a factor;
lated. There were 19 items from Weiser’s (2000) Internet and (c) to identify items that could be retained after rewording
attitude survey, 18 items from Tsai et al.’s (2001) Internet to fit a factor that they loaded on, albeit weakly. If Study 1 was
attitude scale, 20 items from Durndell and Haag’s (2002) successful, it was planned to continue iterating a test–analyze–
Internet attitude scale, and 40 items from Zhang’s (2007) revise cycle until a moderately stable factor structure had been
Internet attitude scale. These were the total number of items achieved. This comprised Studies 2 and 3. The final iteration
in each of the scales, and there was some overlap among (Study 4) with an independent sample was then planned, for
items. which a confirmatory factor analysis would be applied. (The
The creation of Version 1 of the questionnaire proceeded Internet self-efficacy measure was introduced as part of the test
Downloaded by [Monash University Library] at 09:01 25 November 2015

in two redactive stages. The first stage of redaction was to battery for Studies 3 and 4 for the reasons outlined in section 3
address the issues outlined earlier (see section 1) in Internet of this article.)
attitude scale development. Statements that addressed specific
uses of the Internet were removed, as they simply describe pat-
terns of use that may reflect individual circumstances and do 5. STUDY 1
not represent an underlying attitude. Examples of these state- Version 1 of the GIAS consisted of 27 items. These items
ments include “I like to use the Internet to communicate with were identified following the redactive process outlined in
my friends” or “I use the Internet regularly throughout school.” section 4.2 of this article. Version 1 of the GIAS was completed
Other problematic items were statements that referred to feel- by a sample of 415 Irish participants recruited using ad hoc
ings of confidence with using the Internet, for example, “I feel sampling methods. There were 220 female and 193 male partici-
confident using an Internet browser” or “I feel confident dis- pants; two participants did not indicate their gender. Participants
cussing questions with others through the Internet.” Such items were between 17 and 65 years of age. The majority of par-
are rudimentary attempts at capturing feelings of self-worth (or ticipants (66.27%) were 18 to 24 years of age. The remainder
perhaps self-efficacy) rather than attitudes and do not relate to of participants were spread relatively consistently across age
the modal attitude model. groups. With regard to occupational composition, 278 (66.98%)
The second stage of redaction was to apply the modal the- participants in this sample were students. The remaining par-
oretical framework of attitudes to each item in the final item ticipants (33.02%) came from a wide range of occupations
pool. Each remaining item was examined to identify whether including teaching, nursing, managerial positions, librarians,
the statement represented (or could represent) one of the three engineering, and accountancy. Frequency of use of the Internet
components of an attitude: affect, behavioral intention, or cog- was also asked. Participants reported that they used the Internet
nition. If a statement could be reworded to represent one of the regularly with the majority of participants (90.60%), indicating
components of an attitude, then this was done. If however, a that they used the Internet on a daily basis and a further 7.22%
statement did not in any way represent one of the components of participants using the Internet “a few times a week.”
of an attitude, it was deleted. Initial screening to assess the suitability of the data for fac-
After the redactive process, the final item pool consisted tor analysis was first carried out. Inspection of the correlation
of 27 items. There were eight statements representing the matrix revealed the presence of many coefficients of .3 and
affect component, eight statements representing the behavioral above. The Kaiser-Meyer-Olkin statistic was .90 and exceeded
intention component, and 11 statements representing the cog- the minimum recommended value of .60 (Tabachnick & Fidell,
nition component of an Internet attitude. All 27 items were 2007), indicating that the sample size was appropriate. Bartlett’s
attitudinal statements, some positively and some negatively Test of Sphericity reached statistical significance (p < .001),
worded. With regard to the response surface, Spector (1992) supporting the factorability of the matrix.
denoted that “agreement response choices are usually bipolar Bandalos and Finney (2010) suggested that methods for
and symmetrical around a neutral point” (p. 19). Five response determining the number of factors should be statistically, math-
anchors were thus employed: strongly disagree, slightly dis- ematically, and heuristically based. Statistical procedures for
agree, no opinion, slightly agree, and strongly agree. The scale determining the number of factors to retain include Bartlett’s
examines the extent to which participants agree with each (1950, 1951) chi-square test, Velicer’s (1976) Minimum
statement. Average Partial procedure, and Horn’s (1965) Parallel Analysis;
 MEASURING ATTITUDES TOWARDS THE INTERNET 509

mathematically based tests refer to the use of eigenvalues When interpreting the factor solution, Bandalos and Finney
(Kaiser, 1960), whereas heuristic procedures refer to items (2010) recommended attending to the structure coefficients ini-
such as Cattell’s (1966) Scree test (Bandalos & Finney, 2010). tially, followed by an evaluation of the pattern coefficients to
Cliff (1987) argued strongly against a strict application of the comprehend the unique factor-variable relationships. Keeping
eigenvalues greater than 1 strategy. Henson and Roberts (2006) this in mind, all items that loaded on each of the factors on both
reported that of all of these methods, parallel analysis has been the structure matrix and the pattern matrix resulting from the
found to be the most accurate procedure. oblimin rotation were examined. All of the items that loaded on
The following criteria for factor retention were used: (a) each of the four extracted factors in the analysis were exam-
eigenvalues greater than 1, (b) Horn’s Parallel Analysis, and (c) ined. Typically, items were included on a particular factor if
inspection of Cattell’s Scree Plot (for the point of inflection). they had high factor loadings (> .30) and fit with the defini-
Of the three criteria, Horn’s Parallel Analysis was given the tion of that particular factor (Bandalos & Finney, 2010). Items
greatest consideration for the reasons outlined in the previous were deleted if an item achieved a factor loading across several
paragraph. factors, if the factor loading was less than .30, or if the item did
not semantically fit on the factor.
In total, six items were deleted from the analysis because of
1. Eigenvalues: The extraction analysis identified six factors inconsistent loadings and definitional fit, reducing the item pool
Downloaded by [Monash University Library] at 09:01 25 November 2015

with eigenvalues of greater than 1. These six factors repre- to 21 items.

sented 59.15% of the cumulative variance.
2. Horn’s Parallel Analysis: The parallel analysis was con-
ducted using a software program developed by Watkins 6. STUDY 2
(2000). Watkins’s software program generates 1,000 sets of Following the deletion of items from Version 1 of the scale,
random data and calculates the average eigenvalues for these it was felt that some of the subscales lacked an adequate num-
1,000 generated samples based on the real data file in this ber of items. In particular, Subscale 2 (Internet Affect) was of
study. The eigenvalues obtained in the real data file in this concern, as affect is thought to be the central component of an
study are then compared with the corresponding values from attitude, and the subscale in its current form contained only five
the random results generated by the parallel analysis. If the items representing this component. As a result, statements of
obtained value is larger than the random value from the par- an affective nature were added in hopes of strengthening the
allel analysis, then the factor is retained; if it is less, then it is overall scale. The affective words that were added described the
rejected. For this analysis, four obtained values were larger emotional responses of an individual but did not describe their
than the random values, therefore suggesting that four factors actions. In total, seven new statements of an affect nature were
be retained for further analysis. added to the scale. Examples of these statements include “I feel
3. Inspection of Cattell’s Scree plot indicated a four-factor overwhelmed by the Internet” and “The Internet excites me.”
solution, consistent with the results of the parallel analysis. As well as the addition of the new affect items, one other item
was amended to emphasise its affective nature.
Two previously deleted “behavioral intention” items were
As a result, it was decided to retain four factors (which
reworded and included to determine if they would achieve
explained 50.78% of the variance). The factors were named as
stronger factor loadings in the current analysis.1 This resulted
follows:
in a 31-item scale for Version 2 of the GIAS.
Social Benefit of the Internet (the positive influence that the
There were 705 participants in this study (448 female,
Internet has on society and the benefits it creates in societal liv-
257 male), recruited in University College Cork, Ireland.
ing), Internet Affect (feelings, both positive and negative, that
Therefore, the vast majority of the sample (94.18%) was uni-
individuals may have about the Internet), Internet Exhilaration
versity students with 73.62% of the sample in the 19–24 years
(the excitement and personal thrills that people experience when
age group. Most participants used the Internet on a regular basis,
using, or are about to use, the Internet), and Internet Detriment
with 95.75% of participants indicating that they use the Internet
(the negative effects of the Internet at both an individual and
daily and the remaining 4.25% of participants stating that they
societal level).
use the Internet a few times a week.
Common factor analysis was the chosen method for this
The same preliminary tests were carried out for factor reten-
research as it is the most appropriate analysis when the
tion as for Study 1. Results from one of the tests suggested
researcher’s intention is to interpret the components as latent
the initial retention of five meaningful factors. When five fac-
dimensions or factors (Bandalos & Finney, 2010). An oblique
tors were extracted from the data, a meaningful interpretation
rotation was applied to the data (Bandalos & Finney, 2010).
of the data was not produced however and the fifth factor had
For the current and subsequent analyses, direct oblimin was the
just the two behavioral intention items loading on that factor.
oblique rotation utilized as it simplifies factors by minimizing
cross-products of loadings and can handle a wide range of factor 1
These two items proved to be troublesome to the very final
intercorrelations (Tabachnick & Fidell, 2007). analysis, and in Study 4 they were finally deleted.
 510 M. JOYCE AND J. KIRAKOWSKI

As a result, it was decided to retain four factors that explained

48.15% of the variance and to see if the fifth factor items could
be reassigned meaningfully to one of the other four factors.
Common factor analysis with an oblique rotation was car-
ried out on the 31-item questionnaire. Inspection of the structure
coefficients identified a number of items that had factor load-
ings across several factors. Six items achieved factor loadings
across three factors in the structure matrix; however, these
items achieved high factor loadings on one factor only in the
pattern matrix, so these items were retained pending rewriting—
keeping the modal theoretical model of attitude in mind. In total,
four items were deleted for reasons outlined in Study 1, reduc-
ing the item pool to 27 items. Enough rewriting and deleting
had been done to warrant a third iteration.
Downloaded by [Monash University Library] at 09:01 25 November 2015

7. STUDY 3 FIG. 1. Cattell’s scree plot for the General Internet Attitude Scale Version 3.
In addition to rephrasing statements for this stage of scale
development, the response anchors for Internet frequency were
also revised, as they arose as an issue of concern. Up until now,
participants had been provided with five response options for As in Study 2, two items made up a fifth factor again, but
Internet frequency: never, a few times a year, a few times a the remaining items were difficult to interpret in a five-factor
month, a few times a week, and daily. However, it was felt that solution. Common factor analysis with an oblique rotation was
there may be participants who used the Internet a few times a again performed and yielded a much more meaningful four-
week for instance (which could be interpreted by the researcher factor solution which explained 49.82% of the variance, and
as infrequently), but the participant may feel that they use the incorporated the two behavioral intention items into other fac-
Internet a lot. In other words, previous versions of the scale cap- tors with which they had a tolerable fit. Two items achieved
tured how often participants used the Internet in real terms but similar factor loadings across two factors in both the struc-
did not capture participants’ subjective feelings of how often ture and pattern matrix, so these items were deleted, reducing
they felt they use the Internet. As a result, the response choices the item pool to 25 items. Again, there were two further items
for the question “How often do you use the Internet?” were with low loadings, but they were kept because they conformed
changed to never, rarely, sometimes, often, and very often for semantically to the factors on which they loaded.
the current and subsequent administration of the scale. For the Having determined that the factor loadings remained rel-
reasons outlined in section 3, and because of the emerging diffi- atively consistent across three different samples, a reliability
culties with the behavioral intention factor, we also included the analysis was carried out. Cronbach’s alpha was calculated as
self-efficacy measurement instrument as part of the test battery a measure of internal consistency for the four subscales and
in this study and in Study 4 (see Appendix B). overall scale. The Cronbach’s alpha coefficients are reported in
The final sample of participants who completed Version 3 of Table 1.
the GIAS consisted of 657 individuals who were recruited via Each of the subscales and the total scale achieved satisfactory
e-mail using a snowballing sampling (a sampling method used reliability values.
to obtain participants from extended associations through par-
ticipants initially identified by the researcher; Morgan, 2008).
The final sample consisted of 370 male and 287 female partic-
ipants. Participants were from a number of different countries TABLE 1
but primarily from Ireland, United Kingdom, United States, Reliability Values for Subscales and Total Scale, General
and Australia. Participants varied in age; however, the major- Internet Attitude Scale Version 3
ity of participants (82.95%) were between 25 and 54 years of
age. Most participants used the Internet regularly, with 553 par- Name No. of Items Cronbach’s α
ticipants (84.17%) indicating that they used the Internet very Internet Affect 9 .87
often and 85 participants (12.94%) indicating that they used it Internet Exhilaration 5 .73
often. Social Benefit of the Internet 7 .79
The methods of analyses to identify factor retention were Internet Detriment 4 .66
again repeated as per Study 1. The eigenvalues that were gener- Total Scale 25 .86
ated from the data are displayed in the Scree plot in Figure 1.
 MEASURING ATTITUDES TOWARDS THE INTERNET 511

The factorial structure at this stage was similar to that of TABLE 2

Study 2, and with the deleting of two items that had been prob- Standardized Solutions by Confirmatory Factor Analysis for
lematical at Study 2 stage, the overall reliability score for the Model 2
scale was .86. GIAS was considered to have reached a stable
form at this stage. Factor

Internet Internet Social Benefit Internet

Item Affect Exhilaration of the Internet Detriment
8. STUDY 4
The fourth and final step in the development of the GIAS 15 .70
involved completing a confirmatory factor analysis (CFA) to 19 .79
test how well the measured variables (scale items) represent the 5 .71
small number of latent constructs that had been identified (Hair, 2 .69
Black, Babin, & Anderson, 2010). CFA predicts the goodness 10 .76
of fit of the hypothesized models in relation to acceptable cutoff 8 −.56
values, which are outlined next. 24 −.60
Downloaded by [Monash University Library] at 09:01 25 November 2015

The theory that informs the fit of the measured variables 22 .64
within their constructs is based on the results obtained from 12 .56
the third and final iteration of the GIAS. The final scale con- 20 .92
sisted of 25 variables from four underlying constructs. A four- 9 .50
factor model of Internet attitudes was thus specified in which 6 .80
Internet Affect, Internet Exhilaration, Social Benefit of the 14 .67
Internet, and Internet Detriment were the four latent factors. 11 .72
The four latent factors were permitted to be correlated based 21 .70
on prior analyses which indicated a relationship among the 1 .56
factors. 25 .58
The final sample for this study consisted of 841 participants 4 .52
who were recruited via e-mail using the snowballing sampling 16 .82
technique. The final sample consisted of 314 male partici- 13 .53
pants (37.3%) and 527 female participants. Participants were 3 .59
again from a number of different countries including Ireland,
United Kingdom, United States, and Australia. Participants var-
ied in age from 16 years and older; the majority of participants
(64.69%) were between 25 and 44 years of age. Most par- Internet Affect, Internet Exhilaration, Social Benefit of the
ticipants used the Internet regularly, with 97.50% participants Internet, and Internet Detriment.
indicating that they used the Internet very often or often. As in The factor loadings for Model 2 are presented in Table 2.
Study 3, the measurement of self-efficacy was also part of the
test battery.
The first confirmatory model (modeled on the factors emerg- 8.1. Proposed Model Hypotheses (H2) Results
ing from Study 3) did not achieve a satisfactory goodness of
The results of the specified model yielded a comparative fit
fit with χ 2 /df = 4.182, which is more than twice the rec-
index value of .918 and a Tucker–Lewis index of .906. The root
ommended cutoff value of 2.0 (Tabachnick & Fidell, 2007),
mean square error of approximation value was .058. The chi-
although the comparative fit index, Tucker–Lewis index, and
square was significant, at 695.033, with a degree of freedom
root mean square error of approximation results were excel-
of 183, resulting in the value for χ 2 /df of 3.798—still above
lent. Inspection of the path estimates identified two items that
the recommended cutoff point but tolerable bearing in mind the
achieved extremely low loadings of .24 and .29, respectively.
large sample size (n = 841). Model 2 achieved excellent fit in
Hair et al. (2010) suggested that loadings should be at least
the CFA after problematic item deletion.
.50 and ideally .70 or higher. These items were the same two
items that consistently loaded on a fifth factor in the exploratory
factor analyses in Studies 2 and 3. Two other items, which had
consistently received relatively low loadings in Studies 2 and 3, 8.2. Reliability
also achieved relatively weak factor loadings here (.45 and .47). Having identified an excellent model fit for H2, a reliability
All four of these items were deleted. analysis was subsequently carried out. The Cronbach’s alpha
The second model was then tested for goodness-of-fit. Model coefficients are reported in Table 3. Each of the subscales and
2 consisted of the same four factors as in the previous analysis: the total scale achieved satisfactory reliability values.
 512 M. JOYCE AND J. KIRAKOWSKI

TABLE 3 Internet might be more frequently used by men, so it was of

Reliability Values for Subscales and Total Scale, General interest to determine if there would be accompanying gender
Internet Attitude Scale Version 4 differences in Internet attitudes. An independent samples t test
was conducted on the mean attitude scores between male and
Name No. of Items Cronbach’s α female participants to investigate if there were any statistically
Internet Affect 9 .87 significant differences between the groups. It was found that
Internet Exhilaration 3 .76 there were no significant differences in attitude scores between
Social Benefit of the Internet 6 .79 male and female participants on total attitude scores. The effect
Internet Detriment 3 .67 size was small. These results suggest that male and female
Total Scale 21 .85 individuals hold similar attitudes towards the Internet. The find-
ings of Lewis (2002) with the Post Study System Usability
Questionnaire also supports the finding of no effect for gender.

8.3. Summary So Far 8.5. Age Differences

Following four rounds of data collection and analyses, Because the Internet is a relatively recent phenomenon, it
Downloaded by [Monash University Library] at 09:01 25 November 2015

efforts to apply the three-component model of attitudes uncov- is possible that a division exists between individuals who have
ered an underlying structure of four factors, namely Internet grown up immersed in an Internet environment, and those to
Affect, Internet Exhilaration, Social Benefit of the Internet, and whom the Internet is a new concept which was introduced at a
Internet Detriment. The final version of the GIAS consists of later stage in life. Researchers such as van Duersen, van Dijk,
four subscales with satisfactory reliability values and contains and Peters (2011) suggested that younger generations are par-
21 statements, the reliability of which is also satisfactory. The ticularly skilled users of the Internet, as they have had exposure
scale as it stands is accepted currently as the GIAS. to the Internet throughout their entire life. The authors referred
Although the three-component model of attitudes has not to research carried out by de Haan and Rijken (2002), who
been explicitly replicated in the underlying structure of Internet suggested that seniors did not have the opportunity to acquaint
attitudes, it was found that statements representing the origi- themselves with the Internet at school, thus lack the same level
nally proposed three components tended to “group” together in of use and ownership of Internet skills. Although it is thus
the four factors that emerged in the Internet attitude data. During logical to assume that there may be differences in Internet atti-
the CFA, four items were deleted from the GIAS. These four tudes between younger and older age groups, few studies have
items represented three of the originally proposed behavioral actually investigated such differences. In this study, partici-
intention items and one cognitive item. Thus, following appro- pants were asked to indicate which age group they belonged
priate modification during the CFA to achieve the best model fit, to. The age groups were < 18 years, 18–24 years, 25–34 years,
the final 21-item GIAS contains only one behavioral intention 35–44 years, 45–54 years, 55–64 years, and 65 years +. A one-
item and 20 items representing affect and cognitive elements way between-groups analysis of variance was conducted to
of Internet attitudes. Two of the four subscales (Internet Affect explore differences in average total attitude scores across the
and Internet Exhilaration) consisted of statements relating to seven age groups. There was a statistically significant differ-
the emotional response (Affect) and the other two subscales ence in average total attitude scores for the seven age groups:
(Social Benefit of the Internet and Internet Detriment) consisted F(6, 834) = 3.14, p < .01. The effect size was 0.22. Post hoc
of statements describing beliefs (Cognition). analyses indicated that the difference between the means of the
< 18 years group (M = 3.49, SD = 0.47) and 25–34 years group
(M = 3.95, SD = 0.49) approached significance (p = .06).
8.4. Gender Differences
A steady decline in attitude scores was observed over the
Numerous studies (e.g., Tsai et al., 2001; Weiser, 2000) have remainder of the age groups as age increased beyond the
investigated gender differences with technology in an effort 25–34 years age group.
to better understand how male and female participants inter-
act with it. Conflicting evidence for the existence of gender
differences has been reported in such studies. It is possible 9. DISCUSSION
that this conflicting evidence results from the use of question- After three iterations of data collection and appropriate
naires in which the construct under investigation is not clearly statistical analyses, followed by a fourth confirmatory factor
established or psychometrically robust. Research carried out by analysis, the final scale consists of 21 items, with four subscales
Joiner et al. (2005) and Helsper (2010) reported less exten- (see Appendix A). The scale has demonstrated satisfactory reli-
sive Internet use in women compared to men. More recently, ability. The four factors which emerged from the data and
research carried out by Ofcom (2013) still indicates that men were confirmed in the final scale iteration were Internet Affect,
have a higher estimated weekly volume of Internet use than Social Benefit of the Internet, Internet Exhilaration, and Internet
women. These studies thus lend evidence to suggest that the Detriment.
 MEASURING ATTITUDES TOWARDS THE INTERNET 513

The original three-component model of attitude was not 25–34 years age group, Internet attitude scores decreased, indi-
replicated in the Internet attitude data analyses, which con- cating that the older age groups have less favorable attitudes
sistently produced four factors. However, the original attitude towards the Internet. These results are also in line with the
components tended to group together into two groups in the aforementioned research, which identified that as age increases,
four-factor model. For example, although the original cognitive frequency of Internet use decreases. Results from this research
statements distributed across two different factors, the factors also found Internet attitude scores to be strongly correlated with
on which these items loaded tended to almost solely consist of Internet frequency, suggesting that individuals who report high
cognitive items. The same can be observed in the affective fac- Internet use hold more positive attitudes towards the Internet
tors. What is of particular significance in this research is the than those with less Internet interaction.
deletion of behavioral statements from the scale that occurred
as a gradual process during scale development. By the final ver-
sion of the GIAS there was only one remaining statement of a 9.1. Internet Attitude and Internet Self-Efficacy
behavioral nature on the scale that was placed under the Internet At the beginning of this article, we outlined that many previ-
Exhilaration factor. These findings are of significance in attitude ous studies in this research area included self-efficacy as part
theory in general, as little empirical evidence existed prior to of an Internet attitude. Although weaknesses were identified
this research to support the three-component model of attitude;
Downloaded by [Monash University Library] at 09:01 25 November 2015

in previous scale development, it is nevertheless possible that

the three-component model was theoretically important, and these two constructs are related. However, there is minimal lit-
although no research had been done that refuted it, no research erature that explores the empirical relationship between these
actively supported it, either. The current findings suggest that two constructs. To our knowledge, there is only one study that
evaluations of attitude ought to focus on affective and cognitive has done so (Wu & Tsai, 2006). Although positive correlations
elements in attitude measurement, which is in line with posi- of a medium strength were obtained between participants’ atti-
tions taken by early attitude theorists such as Thurstone (1931) tudes and the two subscales of the Internet self-efficacy scale
and Allport (1935), and more recently by Eagly and Chaiken developed in Wu and Tsai’s (2006) study, participants’ Internet
(2007). attitude scores were only weakly correlated with their over-
The results of this research identified no significant differ- all Internet self-efficacy scores. The authors did not offer any
ences between males and females in their Internet attitudes. reasons for such observations.
Although the results of this research are somewhat contrary to Within the studies reported in this article, we consistently
previous studies where it had been suggested that males hold attempted to include and possibly augment the eight initial
more positive Internet attitudes than females (e.g., Durndell & items relating to the behavioral intention component of attitude
Haag, 2002; Tsai et al., 2001), the lack of secure scales for to the Internet. In four studies, we found that items relating
the measurement of Internet attitudes puts the validity of these to this component fell out of the factor analytic matrix; in
previous findings into question. More recent studies by Jones, Study 4, despite our attempts to include the remaining two
Johnson-Yale, Millermaier, and Perez (2009); Helsper (2010); behavioral intention items in the Internet Exhilaration factor,
and most recently Ofcom (2013) found that male individuals we found that a better solution was obtained when these items
use the Internet more often than female, so it was reasonable were removed. We therefore conclude that behavioral inten-
to speculate that with such differences in Internet use between tion is not a component of attitude to the Internet. However,
the sexes, there may also be gender differences in their Internet it is possible that the theoretically unrelated construct of self-
attitudes. Nevertheless, the current research identified no such efficacy could well represent the behavioral manifestation of
differences, and male and female participants achieved similar attitude that many previous attitude theorists have attempted
scores on our Internet attitude measure. to include in their models, and that previous Internet attitude
Although there is minimal research that investigated age researchers also considered was of value to include in their
differences in Internet attitudes, we speculated that there may scales.
be age differences in attitudes towards the Internet as a result These conclusions are in keeping with the standard social
of differences across age groups in exposure to the Internet. psychological definition of attitude as “a psychological ten-
Zickuhr and Madden (2012) outlined that older age groups dency that is expressed by evaluating a particular entity with
continuously report lower levels of Internet use than younger some degree of favour or disfavour” (Eagly & Chaiken, 2007,
age groups. Similarly, research carried out by OxIS (Dutton & p. 598) and in contrast with self-efficacy, which is understood as
Blank, 2011) and Ofcom (2013) identified similar trends with “beliefs in one’s capabilities to organise and execute the courses
younger age groups (e.g., 16–24-year-olds) reporting higher of action required to produce given attainments” (Bandura,
levels of Internet use than older age groups (e.g., 65 years +). 1997, p. 3). It should also be noted that the appropriate recom-
In line with these findings, the results from the analyses in the mended methodologies for measuring these two constructs are
current study did identify age differences in Internet attitudes. markedly different (Joyce & Kirakowski, 2014).
It was found that the age group with the highest attitude scores Nonetheless, there is a medium, positive correlation between
were the 25–34 years age group. As age increased beyond the the 10-item Internet Self-Efficacy Scale and 21-item GIAS
 514 M. JOYCE AND J. KIRAKOWSKI

(r = .43, n = 841, p < .001), with positive Internet atti- Durndell, A., & Haag, Z. (2002). Computer self efficacy, computer anxiety,
tudes associated with high Internet self-efficacy. We take this as attitudes towards the Internet and reported experience with the Internet, by
gender, in an East European sample. Computers in Human Behaviour, 18,
confirming the theoretical intention of previous researchers in 521–535.
human–computer interaction in an unambiguous and defensible Dutton, W. H., & Blank, G. (2011). Next generation users: The Internet in
manner. Britain (Oxford Internet Survey 2011). Oxford Internet Institute, University
of Oxford, UK.
Eagly, A. H., & Chaiken, S. (2007). The advantages of an inclusive definition
of attitude. Social Cognition, 25, 582–602.
9.2. Future Research Eastin, M. S., & LaRose, R. (2000). Internet self-efficacy and the psychology
of the digital divide. Journal of Computer Mediated Communication, 6,
It is taken as a given in social psychology theory that the 1–8.
attitude a person has to an object or activity will influence the Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2010). Multivariate
way they use the object or engage in the activity (e.g., Hogg & data analysis: A global perspective (7th ed.). Upper Saddle River, NJ:
Vaughan, 2011). Bandura (1997) demonstrated that within edu- Pearson Education.
Heinssen, R. K., Glass, C. R., & Knight, L. A. (1987). Assessing computer
cational psychology, self-efficacy is a good individual predictor anxiety, development and validation of the Computer Anxiety Rating Scale.
of attainment. When we look at studies in human–computer Computers in Human Behavior, 3, 49–59.
interaction which evaluate devices and tools in a technological Helsper, E. J. (2010). Gendered Internet use across generations and life stages.
Communication Research, 37, 352–374.
Downloaded by [Monash University Library] at 09:01 25 November 2015

world, we generally find that research in this area has failed to

Henson, R. K., & Roberts, J. K. (2006). Use of exploratory factor anal-
evaluate individual differences in the context of such evaluations ysis in published research: Common errors and some comment on
and that, at the same time, there is much variance present in both improved practice. Educational and Psychological Measurement, 66,
performance and subjective usability measures (as highlighted 393–416.
by Molich et al., 2010). Hogg, M. A., & Vaughan, G. M. (2011). Social psychology (6th ed.). Harlow,
UK: Pearson Education.
The establishment of reliable methods for quantifying both Horn, J. L. (1965). A rationale and test for the number of factors in factor
attitude and self-efficacy that we have presented in this article, analysis. Psychometrika, 30, 179–185.
and in Joyce and Kirakowski (2013, 2014), now puts us in a Jackson, L. A., von Eye, A., Barbatsis, G., Biocca, F., Zhao, Y., & Fitzgerald,
H. E. (2003). Internet attitudes and Internet use: Some surprising findings
position to investigate to what extent it is individual differences
from the HomeNetToo project. International Journal of Human–Computer
between participants that drives such variation, and not, for Studies, 59, 355–382.
instance, lack of precision in measurement practice or the use of Joiner, R., Gavin, J., Duffield, J., Brosnan, M., Crook, C., Durndell,
poor or unstandardized measurement tools. However, this is at A., . . . Lovatt, P. (2005). Gender, Internet identification, and Internet
anxiety: Correlates of Internet use. Cyberpsychology & Behaviour, 8,
present a hypothesis that requires testing in well-differentiated 371–378.
application areas. Jones, S., Johnson-Yale, C., Millermaier, S., & Perez, F. S. (2009). U.S. col-
lege students; Internet use: Race, gender and digital divides. Journal of
Computer-Mediated Communication, 14, 244–264.
Joyce, M., & Kirakowski, J. (2013). Development of a General Internet Attitude
REFERENCES Scale. In A. Marcus (Ed.), Design, user experience, and usability. Design
philosophy, methods, and tools (Vol. 8012, pp. 303–311). Las Vegas, NV:
Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), A handbook of social
Springer-Verlag.
psychology (pp. 798–844). Worcester, MA: Clark University Press.
Joyce, M., & Kirakowski, J. (2014). Measuring confidence in Internet use: The
Aronson, E., Wilson, T. D., & Akert, R. (2002). Social psychology (4th ed.).
development of an Internet self-efficacy scale. In A. Marcus (Ed.), Design,
New York, NY: Prentice Hall.
user experience, and usability. Theories, methods and tools for designing
Bandalos, D. L., & Finney, S. J. (2010). Factor analysis: Exploratory and
the user experience (Vol. 8517, pp. 250–260). Heraklion, Greece: Springer-
confirmatory. In G. Hancock & R. O. Mueller (Eds.), The reviewer’s guide
Verlag.
to quantitative methods in the social sciences (pp. 93–114). New York, NY:
Kaiser, H. F. (1960). The application of electronic computers to factor analysis.
Routledge.
Educational and Psychological Measurement, 20, 141–151.
Bandura, A. (1997). Self–efficacy: The exercise of control. New York, NY: W.
Katz, D., & Stotland, E. (1959). A preliminary statement to a theory of attitude
H. Freeman.
structure and change. In S. Koch (Ed.), Psychology: A study of a science
Bartlett, M. S. (1950). Tests of significance in factor analysis. British Journal of
(Vol. 3, pp. 423–475). New York, NY: McGraw-Hill.
Psychology, 3, 77–85.
Kirakowski, J., & Corbett, M. (1993). SUMI—The Software Usability
Bartlett, M. S. (1951). A further note on tests of significance in factor analysis.
Measurement Inventory. British Journal of Educational Technology, 24,
British Journal of Psychology, 4, 1–2.
210–212.
Breckler, S. J. (1984). Empirical validation of affect, behaviour, and cogni-
Krech, D., Crutchfield, R. S., & Ballachey, E. L. (1962). Individual in society:
tion as distinct components of attitude. Journal of Personality and Social
A textbook of social psychology. New York, NY: McGraw-Hill.
Psychology, 47, 1191–1205.
Lewis, J. R. (2002). Psychometric evaluation of the PSSSQ using data from
Cattell, R. B. (1966). The Scree test for number of factors. Multivariate
five years of usability studies. International Journal of Human Computer
Behavioural Research, 1, 245–276.
Interaction 14, 463–488.
Cliff, N. (1987). Analyzing multivariate data. San Diego, CA: Harcourt, Brace,
Molich, R., Chattratichart, J., Hinkle, V., Jensen, J., Kirakowski, J., Sauro, J.,
Jovanovich.
. . . Traynor, B. (2010). Rent a car in just 0, 60, 240 or 1,217 seconds?—
Davis, F. D., Bagozzi, R. P., & Warshaw, P. R. (1989). User acceptance of com-
Comparative usability measurement, CUE 8. Journal of Usability Studies,
puter technology: A comparison of two theoretical models. Management
6, 8–24.
Science, 35, 982–1003.
Morgan, D. L. (2008). Snowballing sampling. In L. Given (Ed.), The Sage ency-
de Haan, J., & Rijken, S. (2002). The digital divide in the Netherlands: The
clopedia of qualitative research methods (pp. 816–817). Thousand Oaks,
influence of material, cognitive and social resources on the possession and
CA: Sage.
use of ICTs. Electronic Journal of Communication, 12(1&2).
 MEASURING ATTITUDES TOWARDS THE INTERNET 515

Morse, B. J., Gullekson, N. L., Morris, S. A., & Popovich, P. M. (2011). The Weiser, E. B. (2000). Gender differences in Internet use patterns and Internet
development of a general Internet attitudes scale. Computers in Human application preferences: A two-sample comparison. Cyberpsychology &
Behaviour, 27, 480–489. Behaviour, 3, 167–178.
Nickell, G. S., & Pinto, J. N. (1986). The computer attitude scale. Computers in Woodmansee, J. J., & Cook, S. W. (1967). Dimensions of verbal racial attitudes:
Human Behavior, 2, 301–306. Their identification and measurement. Journal of Personality and Social
Ofcom. (2013). Adults’ media use and attitudes report 2013. Retrieved from Psychology, 7, 240–250.
http://stakeholders.ofcom.org.uk/binaries/research/media-literacy/adult- Wu, Y-Y., & Tsai, C-C. (2006). University students’ Internet attitudes
media-lit-13/2013_Adult_ML_Tracker.pdf and Internet self-efficacy: a study at three universities in Taiwan.
Rosenberg, M. J., & Hovland, C. I. (1960). Cognitive, affective, and behavioural Cyberpsychology & Behaviour, 9, 441–450.
components of attitude. In M. J. Rosenberg, C. I. Hovland, W. J. McGuire, Zhang, Y. (2007). Development and validation of an Internet use attitude scale.
R. P. Abelson, & J. W. Brehm (Eds.), Attitude organisation and change: Computers and Education, 49, 243–253.
An analyses of consistency among attitude components (pp. 1–14). Oxford, Zickuhr, K., & Madden, M. (2012). Older adults and Internet use. Retrieved
UK: Yale University Press. http://www.pewinternet.org/∼/media//Files/Report/2012/PIP_Older_
Schumacher, P., & Morahan-Martin, J. (2001). Gender, Internet and com- adults_and_internet_use.pdf
puter attitudes and experiences. Computers in Human Behaviour, 17,
95–110.
Spector, P. E. (1992). Summated rating scale construction: An introduction
(Sage University Paper Series on Quantitative Applications in the Social ABOUT THE AUTHORS
Sciences, No. 82). Newbury Park, CA: Sage.
Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics (5th ed.). Mary Joyce’s research, which arose from her Ph.D studies,
Downloaded by [Monash University Library] at 09:01 25 November 2015

Boston, MA: Pearson Education. is concerned with the measurement of Internet attitudes and
Thurstone, L. L. (1931). The measurement of social attitudes. Journal of Internet self-efficacy and the use of such scales in usability eval-
Abnormal and Social Psychology, 27, 249–269.
Torkzadeh, G., & Van Dyke, T. P. (2001). Development and validation of
uation practice. She is also interested in using these scales to
an Internet self-efficacy scale. Behaviour & Information Technology, 20, examine demographic differences (e.g. age, gender) in online
275–280. environments. Dr. Joyce intends to promote GIAS and the ISES
Tsai, C.-C., & Lin, C.-C. (2004). Taiwanese adolescents’ perceptions and atti- through uxp.ie, the User Experience Solutions website.
tudes regarding the Internet: Exploring gender differences. Adolescence, 39,
725–734. Jurek Kirakowski comes from a practical computer science
Tsai, C.-C., Lin, S. S. J., & Tsai, M-J. (2001). Developing an Internet attitude and psychology background. His specialization is quantita-
scale for high school students. Computers & Education, 37, 41–51. tive measurement in human-computer interaction (HCI). Dr.
van Deursen, A. J. A. M., van Dijk, J. A. J. M., & Peters, O. (2011). Rethinking
Kirakowski’s research interests continue to be user experi-
Internet skills: The contribution of gender, age, education, Internet expe-
rience and hours online to medium- and content-related Internet skills. ence in HCI, as well as the philosophy and methodology of
Poetics, 39, 125–144. human measurement. Dr. Kirakowski and his research group
Velicer, W. F. (1976). Determining the number of components from the matrix (PaT) are well known for the development of the Software
of partial correlations. Psychometrika, 41, 321–327.
Watkins, M. W. (2000). Monte Carlo PCA for parallel analysis [Computer
Usability Measurement Inventory and Website Analysis and
software]. State College, PA: Ed & Psych Associates. Measurement Inventory.
 516 M. JOYCE AND J. KIRAKOWSKI

APPENDIX A
TABLE A1
General Internet Attitude Scale: 21 Items and Corresponding Subscales
Item Internet Affect Internet Exhilaration Social Benefit of the Internet Internet Detriment
15 I feel bewildered by the
Internet
19 I feel intimidated by the
Internet
5 I feel overwhelmed by the
Internet
2 The Internet makes me
feel anxious
10 The Internet makes me
feel uncomfortable
Downloaded by [Monash University Library] at 09:01 25 November 2015

8 The Internet does not

threaten me
24 I feel at ease using the
Internet
22 I feel disheartened at the
thought of using the
Internet
12 The Internet makes me
feel annoyed
20 The thought of going on the
Internet is exciting to me
9 I would like to stay on the
Internet for as long as I can
6 The idea of going on the
Internet gives me a thrill
14 The Internet makes a great
contribution to human life
11 The use of the Internet is
enhancing our standard of
living
21 The Internet is bringing us into
a bright new era
1 The Internet makes a positive
contribution towards society
25 The Internet is responsible for
many of the good things we
enjoy
4 The Internet makes life more
efficient
16 Using the Internet is harmful to
people
13 The Internet is dehumanising
to society
2 Using the Internet can cause
health problems
 MEASURING ATTITUDES TOWARDS THE INTERNET 517

APPENDIX B
Internet Self-Efficacy Scale
How often do you do the following activities on the Internet?
Never Rarely Sometimes Often Very often
Communication
Social networking/Personal
involvement
Obtaining information
Entertainment/Media
consumption
Shopping/Buying items
E-commerce
Booking events /trips
Financial services
Blogging/Contributing to
Downloaded by [Monash University Library] at 09:01 25 November 2015

websites/Discussion boards
Education and training

How confident do you feel that you can do each of the following Internet activities as of now?
Not at all Slightly Somewhat Moderately Considerably Very Completely
confident confident confident confident confident confident confident
Communication
Social networking/personal
Involvement
Obtaining information
Entertainment/media
consumption
Shopping/buying items
E-Commerce
Booking Events/trips
Financial services
Blogging/contributing to
websites/ discussion boards
Education and training