You are on page 1of 21

The Dimensional Structure of the Perceived

Behavioral Control Construct1

PAUL SPARKS,* CAROL A. GUTHRIE, AND RICHARD SHEPHERD


Department of Consumer Sciences
Institute of Food Research, Reading, England

Following concerns in the literature about the conceptualization and operationalization


of the perceived behavioral control (PBC) construct within the Theory of planned
behavior (Ajzen, 1985, 1991). 2 studies were conducted to examine whether
questionnaire items explicitly assessing perceived difficulty and perceived control
issues would attract a different pattern of responses. In applications of the theory
of planned behavior to the reduction of red meat consumption and the reduction of
potato chip (French fries) consumption, Principal Components Analysis showed that
items reflecting perceived difficulty and items reflecting perceived control loaded
onto different components. Furthermore, measures of perceived difficulty and not
measures of perceived control contributed independent predictive effects in multiple
regressions of respondents’ behavioral intentions to make the dietary changes.

. . . when we define a word we are merely inviting others to use


it as we would like it to be used. . . the purpose of definition is
to focus argument upon fact. . . the proper result of good defini-
tion is to transform arguments over terms into disagreements
about fact, and thus open arguments to further inquiry. (Mills,
1959, p. 34)
The theory of reasoned action (Ajzen & Fishbein, 1980; Fishbein 8z Ajzen,
1975) is a theory of attitude-behavior relationships which links attitudes, subjec-
tive norms (akin to perceived social pressure), and behavioral intentions to
(volitional) behavior. Volitional behavior is posited as resulting from behavioral
intentions,which in turn are said to arise out of a combination of a person’s attitude
toward performing the behavior in question and their perceptions of the social
pressure (subjective norm) put upon them to perform that behavior. Attitudes

‘The preparation of this article and the research that it reports were supported by the
European Union through their funding of the research project “The Development of Models
for Understanding and Predicting Consumer Food Choice.” We would like to thank Sarah
Grugeon and Chris Armitage for their assistance with the production of this article.
*Correspondence concerning this article should be addressed to Paul Sparks, Institute of
Food Research, Earley Gate, Whiteknights Road, Reading RG6 6BZ, U.K.

418

Journal ofApplied Social Psychology, 1997, 27, 5, pp. 418-438.


Copyright 0 1997 by V. H.Winston & Son, Inc. All rights reserved
PERCEIVED BEHAVIORAL CONTROL 419

are seen as deriving from a combination of a person’s salient beliefs about


behavioral outcomes and their evaluations of those outcomes (e.g., Ajzen,
Nichols, & Driver, 1995). Subjective norms are proposed as having origins in
a combination of people’s perceptions that important others think they should
or should not perform the behavior in question and their motivation to comply
with others’ wishes. The theory is clearly outlined in Ajzen and Fishbein
(1980) and has been widely applied in a number of domains (e.g., Ajzen &
Fishbein, 1980; Manstead, Proffitt, & Smart, 1983; Pagel & Davidson, 1984;
Pomazal & Jaccard, 1976).
Ajzen’s theory ofplanned behavior (TPB; Ajzen, 1985,1988,1991), which
essentially adds a perceived behavioral control (PBC) construct to the earlier
theory of reasoned action, has also received wide empirical support (e.g.,
Ajzen, 1991; Conner & Sparks, 1996). The TPB indicates that for behaviors
characterized by incomplete volitional control, it is useful to include an assess-
ment of people’s perceptions of the existence of behavioral constraints and
facilitators, alongside measures of attitude and subjective norm, in studies
designed to predict people’s intentions and behavior. Ajzen has described PBC
as “the person’s belief as to how easy or difficult performance of the behavior
is likely to be” (Ajzen & Madden, 1986, p. 457) and has likened and compared
it to Bandura’s notion of self-efficacy (e.g., Ajzen, 1991; Ajzen & Madden,
1986). He suggests (Ajzen, 1985, 1988) that control factors may be either
internal to the person (e.g., skills, abilities, power of will, compulsions) or
external to the person (e.g., time, opportunity, dependence on others). It is
apparent from these early writings that Ajzen conceptualizes perceived control
and perceptions of easeldifficulty as largely synonymous.
In an early application, concerning students’ class attendance, Ajzen and
Madden (1986) used the following three items in order to construct a measure
of PBC (7-point response scale anchors are indicated in parentheses): “How
much control do you have over whether you do or do not attend this class every
session?” (complete control to very little control); “For me to attend every
session of this class is (easy to dijjjcult)”; and “If I wanted to, I could easily
attend this class every session” (extremely likely to extremely unlikely). These
items (a = .74) were summed to form a measure of PBC which was sub-
sequently shown to be an independent predictor of behavioral intentions in
multiple regressions of intentions on attitude, subjective norm, and PBC. In
subsequent applications of the TPB, PBC has been measured by different
questionnaire items, but most of these have been similar in structure to those
suggested by Ajzen in this early work.
However, in spite of a growing number of successful applications of the
theory, several authors have commented on problems of conceptualization and
operationalization with the PBC construct (e.g., Beale & Manstead, 1991; Chan
420 SPARKS, GUTHRIE. AND SHEPHERD

& Fishbein, 1993; Sparks, 1994; Terry, 1993; Terry & O’Leary, 1995). The
concerns expressed by these authors about the PBC construct reflect different,
but related issues.
First, although some authors report acceptable interitem reliabilities for
PBC items, other authors have experienced difficulties in this regard. As noted
above, Ajzen and Madden (1986) reported acceptable interitem reliabilities for
PBC items (Study 1, 3 items, a = .74; Study 2, 5 items, a = .69 and .79 [items
were measured on two separate occasions]). In fact, Ajzen and Timko (1 986)
felt able to employ a single PBC item (with an easy-difficult response scale) in
a study of several health-related behaviors, since this correlated well with other
PBC measures in the Ajzen and Madden (1986) research. However, Beale and
Manstead (1991), for example, reported low interitem reliability for three items
used to assess PBC, and subsequently had to drop one of them in order to
construct a measure of PBC. Similar problems have been reported by other
researchers (e.g., Chan & Fishbein, 1993; Sparks, 1994).
Second, Chan and Fishbein (1993) and Sparks (1994) noted how there are
probably conceptual differences in the way that questions about control and
diff;culty are likely to be interpreted by respondents (see also Fishbein &
Stasson, 1990). While these authors have not elaborated on the likely nature of
such differences in the way that people interpret control and difficulty items,
an example provided by Chan and Fishbein at least illustrates one kind of
situation where measures of control and difficulty may be expected to diverge.
In a study of women’s intentions to tell their sexual partners to use condoms,
Chan and Fishbein (1993) suggest:
With respect to the behavior investigated here, many women find
it embarrassing to tell their partners to use condoms and thus
view this behavior as difficult to perform. At the same time,
however, this behavior is also clearly under one’s volitional
control. (pp. 1457-1458)3
The clear implication here is that there are likely to be behaviors which are
perceived to be under personal control yet difficult to carry out.
Third, there have recently been some suggestionsthat PBC might be reconcep-
tualized to make reference only to external constraining influences on behavior
(Terry, 1993; Terry & O’Leary, 1995; White, Terry, & Hogg, 1994). In studies
utilizing Bandura’s (1977) notion of self-efficacy, Terry and colleagues suggest
that although Ajzen’s conceptualizationof PBC includes reference to both internal

3Elsewhere, Fishbein (1993) also suggests that measures of PBC that assess the perceived
ease or difficulty of carrying out a behavior are equivalent to affective measures of attitude.
We do not share this view.
PERCEIVED BEHAVIORAL CONTROL 421

and external constraining factors on behavior, behavioral control essentially


reflects “external control factors” (Terry, 1993, p. 138). They propose that PBC
and self-efficacy can be conceptually distinguished in terms of an internal-external
dimension, with PBC incorporating influences on behavior and self-efficacy
incorporating internal influences, since self-efficacy perceptions are suggested
to be “based primarily on consideration of control factors that emanate from the
person, rather than from external control factors” (Terry, 1993, p. 138).
In order to investigate this latter suggestion, Terry and O’Leary (1995)
constructed separate measures of PBC and self-efficacy in a study of attitudes
toward taking regular exercise. Self-efficacy measures but not PBC measures
were found to be independent predictors of intentions to exercise regularly
(although PBC was an independent predictor of actual behavior). Similarly, in
a study of condom use and discussion of condom use with new partners, White
et al. (1994) found independent predictive effects of self-efficacy but not of
PBC on behavioral intentions. PBC was a significant predictive influence on
discussion of condom use with new partners. The authors conclude that their
findings suggest that a distinction does need to be made in the TPB between
PBC and self-efficacy. Importantly for the issues raised in this article, in
operationalizing self-efficacy in these studies, Terry and colleagues made
extensive use of questionnaire items that made reference to the difficulty or
ease of behavioral performance.
While we would agree with Terry and O’Leary (1995) that the PBC con-
struct requires some conceptual clarification, our own view departs somewhat
from their position since we do not share their view that the fundamental issue
here is one of distinguishing self-efficacy from PBC. Rather, our concern is
focused on the proposition that laypeople may conceptualize the notions of
control and difficulty in different ways (Chan & Fishbein, 1993; Sparks, 1994)
and that many problems of interitem reliability with PBC items relate to such
conceptual distinctions. Furthermore, we would not wish to reconceptualize
PBC as referring simply to external barriers to action (as Ajzen has quite
clearly noted, his conceptualization of PBC refers to internal as well as external
constraining influences). Specifically, we would raise the following objections
to the position adopted by Terry and colleagues.
First, Terry and colleagues reconceptualize perceived control as referring
to external control factors. As we, and they (e.g., Terry & O’Leary, 1995;
White et al., 1994) have noted, this is at odds with Ajzen’s own position with
regard to PBC. They suggest that Ajzen’s interpretation departs from the
“traditional” sense of perceived control which they describe as “a reflection of
the person’s appraisal of the extent to which other people or events will
interfere with the performance of the behavior“ (Terry & O’Leary, 1995,
p. 202), but they do not really offer further evidence to support this latter
422 SPARKS, GUTHRIE, AND SHEPHERD

interpretation. Indeed, it appears to be at odds with various other positions in


the literature that discuss internal control problems (e.g., Hoch & Loewenstein,
1991; Rotter, 1966; Weiner, 1979): Conceptualizations of perceived control in
the psychology literature usually acknowledge both internal and external con-
straining factors on behavior (cf. Peterson, 1990, on personal control as a
transactional concept). Ajzen’s conceptualization is not essentially different
from that of other researchers in this regard, and we do not view his position as
idiosyncratic.
Second, Terry and colleagues claim some support for their distinction
between self-efficacy and PBC since Bandura (1977) is keen to distinguish
efficacy expectations (“the conviction that one can successfully execute the
behavior required to produce the outcomes,” p. 193) from outcome expectan-
cies (“a person’s estimate that a given behavior will lead to certain outcomes,”
p. 193) and because, they suggest, outcome expectancies are “similar to the
notion of perceived behavioral control (unconfounded with efficacy expectan-
cies)” (Terry & O’Leary, 1995, p. 202). We would strongly disagree with this
suggestion that outcome expectancies can be likened to PBC. Rather, outcome
expectancies are similar to behavioral beliefs within the model structure of the
theory of reasoned action (cf. “Each behavioral belief links the behavior to a
certain outcome, or to some other attribute such as the cost incurred by
performing the behavior,” Ajzen, 1988, p. 120).
Third, citing Triandis (1977), Terry and O’Leary (1995) suggest that the
perception of the existence of internal barriers may not match the perception of
the existence of external barriers (since, e.g., “there may not be a correspon-
dence between the person’s perception of the extent to which external barriers
may impede the performance of the behavior and his or her judgment that the
behavior will be easy to perform,” Terry & O’Leary, 1995, p. 202). They imply
that this is an argument for not including reference to both internal barriers
and external barriers within the same concept. This is a non sequitur: The fact
that different control factors may either facilitate or impede behavior enactment
is not a valid argument against incorporating these within a single construct.
Besides, it is also true, for example, that some external factors may facilitate
carrying out a behavior while other external factors may impede it; for exam-
ple, time constraints on shopping or food preparation may impede certain
dietary changes, but adequate financial resources may facilitate those changes
(other things being equal).
In the two studies reported in this article, we address the issues of
(a) whether people may respond to perceived control and perceived difficulty
questionnaire items in different ways, and (b) the extent to which measures of
perceived control and perceived difficulty are predictive of people’s behavioral
intentions. On the evidence of the research that we have already alluded to, we
PERCEIVED BEHAVIORAL CONTROL 423

expected respondents to interpret and respond to perceived control and perceived


difficulty questions in different ways. Moreover, we expected to obtain greater
predictive effects for perceived difficulty measures than for perceived control
measures, since the former measures are closer to the way in which PBC is
conceptualized and because we believed that there would be a greater corre-
spondence of the barriers that people experience (or envisage) with their ratings
on difficulty questions than with their ratings on control questions. Throughout
the article, we shall be referring to PBC in the way in which it is defined by
Ajzen-the person’s belief as to how easy or difficultperformance of the behavior
is likely to be; when we refer to perceived control and perceived difficulty from
now on, we shall be referring only to those measures described in this text.
The applied research issue addressed here is people’s attitudes and inten-
tions toward making dietary changes that may be considered beneficial to
health: reducing consumption of red meat and reducing consumption of
potato chips (French fries). Both of these foods make a significant contribu-
tion to fat intake in the U.K., which is considerably higher than expert commit-
tees concerned with diet and health have deemed advisable (Cannon, 1992;
Committee on Medical Aspects of Food Policy, 1984; National Advisory
Committee on Nutrition Education, 1983; World Health Organization, 1990).
While there is some evidence in the U.S. that fat intake levels fell between the
mid-1960s and mid-1980s (Stephen & Wald, 1990), fat consumption levels (in
terms of percentage energy intake) in the U.K. remained more or less constant
during the 1970s and 1980s (Secretary of State for Health, 1991). It is
therefore important to understand the barriers to the adoption of dietary
change. While some of these can be investigated at a more macroscopic
societal level and while people’s attitudes themselves are clearly potential
targets for persuasive communications, there is some evidence that PBC is
likely to be a significant factor in people’s intentions to participate in health-
related behaviors (e.g., Ajzen & Timko, 1986) and behavior change (e.g.,
Sparks, Shepherd, Wieringa, & Zimmermanns, 1995).

Study 1

Method

Respondents

Ninety-one people (66 female, 25 male) approached in the foyer of a


supermarket in Reading, U.K., agreed to participate in some research examin-
ing people’s food choices. All respondents were identified as consumers of red
meat; they were not paid for their participation.
424 SPARKS, GUTHRIE, AND SHEPHERD

Materials
A questionnaire addressing red meat consumption was constructed.
Twenty-six of the questions were concerned with measuring PBC, while the
remaining questions assessed, among other things, attitudes, subjective norm
and intentions. These questions were as follows (unless otherwise indicated,
response scales, indicated in parentheses, were 7-point and only the endpoints
were anchored; items were reverse scored where necessary):
Perceived behavioral control. The questionnaire contained 26 PBC ques-
tion~~ 24: of these were ostensibly TPB items and 2 were self-efficacy items.
Seven of the items made explicit reference to difficulty in either the questions
or the accompanying response scales (e.g., “For me to reduce the amount of red
meat that I eat from now on would be . . .,” easy-dificult); likewise, seven of
the items made reference to control (e.g., “How much control do you have over
whether you do or do not reduce the amount of red meat that you eat from now
on?” complete control-very little contror); the remaining items made no ex-
plicit reference to these concepts and will be referred to hereafter as miscella-
neous PBC items (e.g., “It is mostly up to me whether or not I reduce the
amount of red meat that I eat from now on,” strongly agree-strongly disagree).
The two self-efficacy items were included in this latter category. All PBC items
were collected from published applications of the TPB and modified for use in
the present study.5 Cronbach’s alpha for the 25 items was .93, and the items
were summed to form an initial, general measure of PBC.
Attitude. Five !My anchored questions assessed attitude via the use of seman-
tic differential scales: “My reducing the amount of red meat that I eat from
now on would be . . .” (extremely unenjoyable-extremely enjoyable; extremely
foolish-extremely wise; extremely unpleasant-extremely pleasant; extremely
harmful-extremely beneficial; extremely good-extremely bad). Cronbach’s
alpha for these items was .87, and they were summed to form a measure of
attitude. A second, more general attitude, question was also included: “My
attitude toward reducing the amount of red meat that I eat from now on is . . .”
(extremely unfavorable-extremely favorable; cf. Ajzen & Fishbein, 1980).
Subjective norm. Three fully anchored subjective norm items were in-
cluded: “I feel under social pressure to reduce the amount of red meat that I
eat from now on” (disagree strongly-agree strongly); “Most people who are
important to me think that I should reduce the amount of red meat that I eat
from now on” (extremely likely-extremely unlikely); and “Are there people who
are likely to influence your decision to reduce the amount of red meat that you

4Due to a presentation flaw, many respondents failed to respond to the final PBC question.
This question was therefore excluded from subsequent analyses.
5 A list of these questions and their sources is available from the authors.
PERCEIVED BEHAVIORAL CONTROL 425

eat from now on? (yes, no). If you answered yes, please complete the following
question: The views or attitudes of those people toward my reducing the
amount of red meat that I eat from now on are . . .” (extremely unfavorable-
extremely favorable). Cronbach’s alpha for these items was only .38. When
responses to the last of these three items was removed, the coefficient increased
to .63: Therefore, the subjective norm measure was constructed only from the
first two items.
Intention. Three intention questions were included: “I will make an effort
to reduce the amount of red meat that I eat from now on” (dejnitely true-definitely
false); “I will try to reduce the amount of red meat that I eat from now on”
(definitely will-definitely will not); and “I intend to reduce the amount of red
meat that I eat from now on” (definitely do-definitely do not). Cronbach’s alpha
for these items was .92; the items were summed to form an intention measure.
Respondents were also asked their gender, age, marital status, education,
and whether or not they had ever tried to reduce the amount of red meat that
they ate. Other measures assessing desire, perceived moral obligation, unreal-
istic optimism, and perceived relative consumption were included as filler
items (between the PBC items) but are not discussed further here.
Design and Procedure
The PBC questions (presented in a fixed random order) were interspersed
with the other questionnaire items listed above (also presented in a fixed random
order). Respondents completed the questionnaire in the supermarket foyer.
Results
A principal components analysis of the 25 PBC items revealed five compo-
nents explaining 70.3% of the variance. Rotation of the first two factors
(accounting for 56.2% of the variance) showed that there was a distinct split
between difficult/easy items and control items, with difficult/easy items load-
ing on Component 1 (accounting for 38.0% of the variance) and control items
loading predominantly on Component 2 (accounting for 18.3% of the vari-
ance). Table 1 shows the component loadings for these various items.
A measure that we shall termperceived difficulty was constructed from five
of the items6 loading most heavily on Factor 1 (a= .90), and a measure that we

6The items chosen were the highest loading items that were deemed most suitable for
adaptation for Study 2. Some exclusion of high loading items occurred because question wording
or the response scales were viewed as problematic; in one case, an item was excluded because
it was judged to be too similar to an already included item. Full details are available from the
authors. In a separate analysis, when perceived difficulty and perceived control measures were
constructed from the five highest loading items on each component, the subsequent results of
multiple regressions were similar to those reported here.
426 SPARKS, GUTHRIE, AND SHEPHERD

shall term perceived control was constructed from five of the items loading
most heavily on Factor 2 (a= .83).The correlation between these two measures
was .24 (p < .05). The mean score for perceived control was 3 1.5 (range = 5 to
35, SD = 4.72; higher scores indicate more control), and the mean score for
perceived difficulty was 25.80 (range = 5 to 35, SD = 8.68; higher scores
indicate less difficulty). However, the diversity of the rating scales accompa-
nying the items used to construct these measures precludes any clear conclu-
sions from a comparison of these scores.
Multiple regressions of people’s intentions to reduce meat consumption on
their attitudes, subjective norm, perceived difficulty, and perceived control,
with attitudes and subjective norm entered at Step 1, and perceived difficulty
and perceived control entered at Step 2, showed that the variables entered at
Step 2 did not result in a significant increase to the amount of variance
explained (Fchange = 1.63, ns). Final beta values indicated a significant effect
for attitudes (p = .69,p < .OOl), a marginally significant effect for the perceived
difficulty measure (p = .14, p < .08) but no effect for either the perceived
control measure (p = -.01, ns) or the subjective norm measure (p = .05,ns)
(Table 2). When the initial, general measure of PBC (constructed from 25
items) was included in the regression, in place of the perceived difficulty and
perceived control measures, it contributed a marginally significant predictive
effect on behavioral intentions (final CJ = .13, p < .09).

Discussion
The findings from this study show that when multiple items were used to
measure PBC, items which related explicitly to perceived difficulty and items
which related explicitly to perceived control showed distinct patterns of inter-
correlations. Moreover, a measure constructed to reflect perceived difficulty
contributed a marginally significant effect to the prediction of behavioral
intentions, whereas a measure reflecting perceived control did not. In a second
study, we sought to assess whether the findings of the first study could be
reproduced in a shorter questionnaire assessing intentions to perform a differ-
ent behavior, with a different subject sample.

Study 2

Method

Respondents
Ninety-seven visitors (43 female, 5 1 male [3 did not record their gender])
approached at a careers fair in Reading (U.K.) city center agreed to participate
PERCEIVED BEHAVIORAL CONTROL 427

Table 1

Component Loadings From PCA (Red Meat)

Component 1 (38.0%) Component 2 (1 8.3%)

Miscellaneous 1 358 .025


Di@dtyl .848 .009
Miscellaneous2 (s.e.) 330 .101
Diff?culty2 .819 .207
Difficulty3 .804 .215
Difficulty4 303 .088
Miscellaneous3 .798 .076
Miscellaneous4 .779 -.161
Difficulty5 .775 .329
Difficulty6 .765 ,126
Miscellaneous5 .757 .3 13
Miscellaneous6 .730 ,300
Difficulty7 .668 .322
Control6 .573 .254
Control7 .545 .508
Miscellaneous7 (s.e.) .545 -.092
Miscellaneous8 .544 -.011
Miscellaneous9 .243 .799
Controll .131 .784
Control2 .112 .775
Miscellaneous 10 .130 .688
Control3 .24 1 ,687
Control4 -.lo7 .591
Control5 -.085 .559
Miscellaneous 11 .095 .465

Note. Difficulty items make explicit reference to difficultylease in either the question or
the response scale; control items make explicit reference to control; miscellaneous
items make direct reference to neither difficulty nor control (those marked “s.e.” are
self-efficacy items). Italicized items are those that were used to construct the perceived
difficulty and perceived control measures reported in Studies 1 and 2.
428 SPARKS, GUTHRIE. AND SHEPHERD

Table 2

Hierarchical Regressions of Intentions to Reduce Consumption of Red Meat

Increment Final
Step Predictor R R2 to R2 Fchange P
1 .80 .63 .63 74.95+**
Attitudes .69***
Subjective norm .05

2 .81 .65 .o 1 1.63


Perceived control -.01
Perceived difficulty .14t

Note. n = 89.
t p < .lo. ***p < .001.

in some research examining people’s food choices. Although their ages ranged from
15 to 44 years, the respondents were predominantly 15-16year olds (86 respondents
belonged to this age group). Respondents were not paid for their participation.
Materials
A questionnaire comprised of 20 questions addressing consumption of potato
chips (French fries) was constructed. Ten of these questions were concerned
with measuring PBC. The remaining questions assessed the central components of
the theory of reasoned action, along with one question about behavioral expecta-
tions and one concerning frequency of consumption. These questions were as
follows (unless otherwise indicated, response scales were 7-point and only the
endpoints were anchored; items were reversed scored where necessary):
Perceived behavioral control. The questionnaire contained 10 PBC ques-
tions: nine of these were TPB questions and one was ostensibly a self-efficacy
question. These questions were adapted from the perceived difficulty and
perceived control questions used in Study 1. These PBC questions were inter-
spersed with those questions referred to below. Cronbach’s alpha for these
items was .80; the items were summed to form an initial, general PBC measure
(the items are listed in the Appendix).
Am-tz.de.There were two fully anchored attitude questions which were designed
to assess cognitive and affective components of attitude: “My reducing the
amount of chips that I eat from now on would be . . .” (extremely unpleasant-
extremely pleasant; extremely harmful-extremely beneJicial). The correlation
PERCEIVED BEHAVIORAL CONTROL 429

for these items was only .33, so they were treated in the analyses as separate
affective and cognitive components of attitude.
Subjective norm. Two fully anchored subjective norm questions were in-
cluded: “I feel under social pressure to reduce the amount of chips that I eat
from now on” (disagree strongly-agree strongly); and “Most people who are
important to me think that I should reduce the amount of chips that I eat from
now on” (extremely likely-extremely unlikely). As the correlation between
these items was .06, only the second was used in subsequent analyses as a
measure of subjective norm.
Intention. Intentions were assessed by the items, “I will try to reduce the
amount of chips that I eat from now on” (definitely will-definitely will not); and
“I will make an effort to reduce the amount of chips that I eat from now on”
(definitely true-definitelyfalse). The correlation between these items was .72;
the items were summed to form an intention measure.
Expectation. Expectation was assessed by a single item, “All things consid-
ered, how likely is it that you will actually reduce the amount of chips that you
eat from now on?” (extremely likely-extremely unlikely).
Consumption. Consumption was measured by a single item, “On average,
how often do you eat chips?” (less than once a week-every day).
Respondents were also asked to indicate their age and gender.
Design and Procedure
As with Study 1, the PBC questions (presented in a fixed random order)
were interspersed with the other questionnaire items listed above (also presented
in a fixed random order). The order of questions was reversed for half of the
respondents. Respondents completed the questionnaire in the location where
they were handed out.

Results

Principal components analysis (with varimax rotation) of the PBC items


revealed three components explaining 64.2% of the variance. Again, it was
apparent that there was a distinct split between different kinds of items, with
difficulty/ease items loading on Component 1 (27.5%), and control items
loading on Component 2 (24.4%; Table 3). A measure of perceived difficulty
from the four items (a= .79) loading most heavily on Factor 1 and a measure
of perceived control from the four items (a = .76) loading most heavily on
Factor 2 were constructed. The correlation between these two measures was .55
(p < .001).The mean score for perceived control was 23.70 (range = 4 to 28,
SD = 3.97) and the mean score for perceived difficulty was 22.30 (range = 4 to
28, SD = 5.03).
430 SPARKS, GUTHRIE. AND SHEPHERD

Table 3

Component Loadings From PCA (Chips)

Component 1 Component 2 Component 3


(27.5%) (24.4%) (12.3%)

Difficulty 1 .791 .224 .080


Miscellaneous4 .780 .049 .027
Difficulty2 .768 .273 .034
Difficulty4 .548 .465 -.05 1
Control3 .282 .805 .132
Miscellaneous9 -.103 .749 .128
Control2 .426 .709 .115
Control 1 .323 .592 .04 1
Control4 -.130 .164 .874
Miscellaneous2 .489 .082 .638

Table 4

Hierarchical Regressions of Intentions to Reduce Consumption of Chips

Increment Final
Step Predictor R R2 to R2 Fchange P
1 .53 .28 .28 10.86***
Affective attitude .24*
Cognitive attitude -.03
Subjective norm .37***

2 .60 .37 .08 5.13**


Perceived control -.09
Perceived difficulty .34**

Note. n = 86.
*p < .05. **p < .01. ***p < .001.
PERCEIVED BEHAVIORAL CONTROL 431

Hierarchical regressions of intentions to reduce chip consumption on atti-


tudes and subjective norm (Step l), and perceived difficulty and perceived
control (Step 2) revealed a significant increase in the variance explained at
Step 2 (Fchange = 5 . 1 3 , <
~ .01). Final beta values showed significant effects for
affective attitude (p = .24, p < .05), subjective norm (p = .37, p < .001), and
perceived difficulty (p =.34, p < .01; see Table 4). However, there was no
independent effect for either perceived control (p = -.09, ns) or cognitive
attitude (p = -.03, ns). When the initial, general measure of PBC (constructed
from 10 items) was included in the regression, in place of the perceived
difficulty and perceived control measures, it did not contribute a significant
predictive effect on behavioral intentions (final p = .14, ns)
Discussion

The results of the second study largely mirror the results of the first:
Respondents from a different general age group and in the context of a different
behavior gave distinct patterns of ratings to items that made explicit reference
to control and to difficulty issues. As in Study 1, the predictive value of
perceived difficulty, and not that of perceived control, was confirmed.
General Discussion
The results of the two studies match our predictions very closely, presenting
clear evidence of a different pattern of responses to what we have termed perceived
difficulty and perceived control items. Given that we have also demonstrated
independent predictive effects (on intentions) for perceived difficulty but not
for perceived control, this would suggest that considerable care needs to be
taken when constructing measures of PBC in applications of the TPB.
Our empirical findings are essentially compatible with those of Terry and
colleagues (e.g., Terry & O’Leary, 1995; White et al., 1994), perhaps largely
because our measures, if not our arguments, have been similar. We would refrain
from contrasting PBC with self-efficacy for the reasons that we have outlined
in the introduction. That is, while Terry and colleagues suggest that self-effi-
cacy and PBC should be contrasted in terms of their respective reference to
internal and external constraints on behavior and because of the judged simi-
larity between PBC and outcome expectancies, we prefer to retain an interpre-
tation of PBC that includes reference to internal and external constraints, and
we do not equate PBC with outcome expectancies. Rather, our explanation of
the empirical data reported in this article is focused on the conceptual distinc-
tions between the control and difficulty issues that are embodied in PBC items.
We agree wholeheartedly with Terry and colleagues that more attention
needs to be paid to the PBC construct. Future research would be well advised
432 SPARKS, GUTHRIE, AND SHEPHERD

to follow those authors in including behavioral measures, allowing the predic-


tive effects of behavioral intentions and PBC on behavior to be further assessed
(cf. Terry & O’Leary, 1995). As we noted in the introduction, PBC measures
have demonstrated predictive success in a number of applications of the TPB:
It is of considerable concern that measures of PBC demonstrate adequate
construct validity in order for research concerned with understanding the
determinants or predictors of PBC to proceed on a sound footing.
We would also suggest that more attention needs to be paid to how people
interpret questionnaire items, not only about difficulty and control, but also
about self-efficacy, especially as the antecedents of self-efficacy (like those of
PBC) are of considerable research interest (Gist & Mitchell, 1992). Of course,
self-efficacy itself has been measured in a number of different ways (Lust,
Celuch, & Showers, 1993). Some of these measures make explicit reference to
difficulty/ease, similar to the items employed by Terry and colleagues (e.g.,
De Vries, Backbier, Kok, & Dijkstra, 1995; De Vries, Dijkstra, & Kuhlman,
1988). However, many applications of self-efficacy do not make explicit
reference to difficulty/ease in this way; rather, most applications assess
strength and magnitude components separately (Bandura, 1977; Gist &
Mitchell, 1992), or measure the construct with the use of scales that assess how
sure (e.g., Basen-Engquist & Parcel, 1992) or how confident (e.g., Yordy &
Lent, 1993) respondents are that they could carry out a behavior, or how likely
(e.g., McCaul, Sandgren, O’Neill, & Hinsz, 1993) they thought it was that they
could carry out a behavior. Although the ambiguities surrounding PBC have
been the focus of this article, the lack of clarity in the self-efficacy construct
has been commented on elsewhere (Gist & Mitchell, 1992). Both constructs
merit more critical scrutiny: Comparisons between them in future research
would be well advised to take into account the various ways in which each
construct is operationalized in the research literature.
We would like to draw attention to two further methodological issues that
should be borne in mind in an assessment of the research reported here. First,
we should point out that we included a large number of disparate PBC items in
our questionnaires. The methodology may have predisposed people to assimi-
late the interpretation of the control and difficulty items in terms of each other
(e.g., Strack, 1992). To the extent that this might be true, one would expect this
to increase the similarity of ratings on the two kinds of questions. However,
other related questionnaire context effects (e.g., contrast effects) may have
played a role, such that we might expect a different pattern of findings if
matched groups of respondents received either perceived control or perceived
difficulty questions. Second, in this article we have been concerned with
reductions in the consumption of certain foods. As such, it may be argued that
this does not reflect a behavior as such, but rather the difference between
PERCEIVED BEHAVIORAL CONTROL 433

consumption at two unspecified time points (i.e., a change in behavior). While


it could be argued that the TPB is consequently not strictly applicable here, the
issue is somewhat analogous to the Schifter and Ajzen (1985) TPB research in
which attitudes toward weight reduction were assessed.
Conceptually,there is much work to be carried out with the notion of perceived
control. We have only been able to hint at the importance of this issue here. For
example, it may be that people sometimes interpret control in terms of physical
possibilities but with little regard to the social and physical difficulties that are
associated with a course of action, as Chan and Fishbein (1993) hint in the
quote that we provided earlier. Within such an interpretation, people may view
control as a dichotomous issue-an action is either controllable or uncon-
trollable. This would have a number of implications for research: from the
need for distributions of PBC item ratings to be carefully examined and borne
in mind in correlation and regression analyses, to the important connotations
of responsibility and blame that may accompany such an interpretation.
As Peterson ( 1 990) has noted, “I think it fair to say that the measurement of
personal control has lagged behind other aspects of its study, giving us a line
of research with poor reliability but spectacular validity” (p. 243). Although our
expectations about the predictive success of our measure of perceived difficulty
were confirmed in the studies reported in this article, we would not propose that
measures of PBC should necessarily reflect the kind of perceived difficulty items
used in the research reported here. Rather, unless a more universally applicable
formulation of PBC becomes available, measures may have to be tailored to the
kind of behavioral issue under consideration. For example, in the case of behaviors
associated with control problems that may relate to internal compulsions or urges,
the control problems relate to restraining oneself from, or resisting, carrying out
the behavior in question rather than to the difficulties faced in carrying it out. It
would be of little value to discover that everyone (including those who do not
experience “chocolate problems” as well as those who do) finds it easy to consume
chocolate, for example! What is important here is to ascertain who finds it difficult
not to carry out the behavior in question; for such kinds of control problems (e.g.,
those that involve problems of internal restraint or external coercion), appro-
priate measures will need to be constructed. Moreover, rather than discounting
the role of items that explicitly assess perceived control, it is necessary to bear
in mind the research which has shown measures of PBC (constituted from items
explicitly addressing control issues) to have independent predictive effects on
behavior (Terry & O’Leary, 1995; White et al., 1994).
At an applied level, it is hoped that further research into the problems that
people envisage and experience in adopting and maintaining health-related
behaviors (as well as research into the factors that help them to carry out those
behaviors) will assist in health promotion efforts. This clearly involves the
434 SPARKS, GUTHRIE, AND SHEPHERD

necessity, in the first place, of accurately assessing or measuring the problems


that people face. In applications of the TPB, this involves having adequate
measures of PBC. Research is needed to address the immediate problems of
interitem reliability when diverse PBC measures are used, the conceptual
differences between control and difficulty, and the predictive effects of differ-
ent PBC measures for different samples and in diverse behavioral domains. In
advance of this research, however, those involved in applied empirical work
need to make decisions about construct measurement in applications of the
TPB. For many applications of the TPB to behaviors whose performance is
associated with difficulties, we would therefore tentatively conclude that PBC
measures constructed out of items that explicitly make reference to the diffi-
culty or ease of carrying out the behavior are likely to demonstrate the best
predictive effects on behavioral intentions.

References

Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In


J. Kuhl& J. Beckmann (Eds.), Action control: From cognition to behavior
(pp. 11-39). Berlin, Germany: Springer.
Ajzen, I. (1988). Attitudes, personality and behaviour. Milton Keynes, England:
Open University Press.
Ajzen, I. (1 99 1). The theory of planned behavior. Organizational Behavior and
Human Decision Processes, 50, 179-21 1.
Ajzen, I., & Driver, B. L. (1991). Prediction of leisure participation from
behavioral, normative, and control beliefs. An application of the theory of
planned behavior. Leisure Sciences, 13, 185-204.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting
social behavior. Englewood Cliffs, NJ: Prentice-Hall.
Ajzen, I., & Madden, T. J. (1986). Prediction of goal-directed behavior: Atti-
tudes, intentions and perceived behavioral control. Journal of Experimen-
tal Social Psychology, 22,453-474.
Ajzen, I., Nichols, A. J., & Driver, B. L. (1995). Identifying salient beliefs
about leisure activities: Frequency of elicitation versus response latency.
Journal ofApplied Social Psychology, 25, 1391-1410.
Ajzen, I., & Timko, C. (1986). Correspondence between health attitudes and
behavior. Basic and Applied Social Psychology, 7,259-276.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral
change. Psychological Review, 84, 191-2 15.
Basen-Engquist, K., & Parcel, G. S. (1992). Attitudes, norms and self-efficacy:
A model of adolescents’ HIV-related sexual risk behavior. Health Educa-
tion Quarterly, 19, 263-277.
PERCEIVED BEHAVIORAL CONTROL 435

Beale, D. A., & Manstead, A. S. R. (1991). Predicting mothers’ intentions to


limit frequency of infants’ sugar intake: Testing the theory of planned
behavior. Journal of Applied Social Psychology, 21,409-43 1.
Cannon, G. (1992). Food and health: The experts agree. London, England:
The Consumers’ Association.
Chan, D. K. S., & Fishbein, M. (1993). Determinants of college women’s
intentions to tell their partners to use condoms. Journal of Applied Social
Psychology, 23, 1455-1470.
Committee on Medical Aspects of Food Policy (COMA), Department of Health
and Social Security. (1984). Diet and cardiovascular disease. London,
England: HMSO.
Conner, M. T., & Sparks, P. (1996). Applying the theory of planned behaviour.
In M. T. Conner & P. Norman (Eds.), Social cognition in health: Predict-
ing behaviour (pp. 12 1 - 162). Milton Keynes, England: Open University
Press.
De Vries, H., Backbier, E., Kok, G., & Dijkstra, M. (1995). The impact of
social influences in the context of attitude, self-efficacy, intention and
previous behavior as predictors of smoking onset. Journal of Applied
Social Psychology, 25,237-257.
De Vries, H., Dijkstra, M., & Kuhlman, P. (1988). Self efficacy: The third
factor besides attitude and subjective norm as a predictor of behavioral
intentions. Health Education Research, 3,273-282.
Fishbein, M. (1993). Introduction. In D. J. Terry, C. Gallois, & M. McCamish
(Eds.), The theory of reasoned action: Its application to AIDS-preventative
behavior (pp. xv-xxv). Oxford, England: Pergamon.
Fishbein, M., & Ajzen, I. (1975). Belie$ attitude, intention and behavior.
Reading, MA: Addison-Wesley.
Fishbein, M., & Stasson, M. (1990). The role of desires, self-predictions, and
perceived control in the prediction of training session attendance. Journal
ofApplied Social Psychology, 20, 173-198.
Gist, M. E., & Mitchell, T. R. (1992). Self-efficacy: A theoretical analysis of
its determinants and malleability. Academy of Management Review, 17,
183-211.
Hoch, S. J., & Loewenstein, G. F. (1991). Time inconsistent preferences and
consumer self control. Journal of Consumer Research, 17,492-507.
Kimiecik, J. (1 992). Predicting vigorous physical activity of corporate employ-
ees: Comparing the theories of reasoned action and planned behavior.
Journal of Sport and Exercise Psychology, 14, 192-206.
Lust, J. A., Celuch, K. G.,& Showers, L. S. (1993). A note on issues concerning
the measurement of self-efficacy. Journal of Applied Social Psychology,
23, 1426-1434.
436 SPARKS, GUTHRIE, AND SHEPHERD

Madden, T. J., Ellen, P. S., & Ajzen, I. (1992). A comparison of the theory of
planned behavior and the theory of reasoned action. Personality and Social
Psychology Bulletin, 18, 1, 3-9.
Manstead, A. S. R., Proffitt, C., & Smart, J. L. (1983). Predicting and under-
standing mothers’ infant-feeding intentions and behavior: Testing the
theory of reasoned action. Journal of Personality and Social Psychology,
44, 657-671.
McCaul, K. D., Sandgren, A. K., O’Neill, H. K., & Hinsz, V. B. (1993). The
value of the theory of planned behavior, perceived control, and self-efficacy
expectations for predicting health-protective behaviors. Basic and Applied
Social Psychology, 14,23 1-252.
Mills, C. W. (1959). The sociological imagination. Oxford, England: Oxford
University Press.
National Advisory Committee on Nutrition Education. (1983). A discussion
paper on proposals for nutritional guidelines for health education in
Britain. Prepared for the NACNE by an ad-hoc working party under the
chairmanship of Professor W. P. T. James. London, England: HMSO.
Netemeyer, R. G., & Burton, S. (1990). Examining the relationship between
voting behavior, intention, perceived behavioral control and expectation.
Journal of Applied Social Psychology, 20, 661-680.
Pagel, M. D., & Davidson, A. R. (1984). A comparison of three social psycho-
logical models of attitude and behavioral plan: Prediction of contraceptive
behavior. Journal of Personality and Social Psychology, 47, 5 17-533.
Peterson, C. (1990). Personal control and health promotion: A psychological per-
spective. In J. Rodin, C. Schooler, & K. W. Schaie (Eds.), Self-directedness:
Cause and effects throughout the life course (pp. 24 1-246). Hillsdale, NJ:
Lawrence Erlbaum.
Pomazal, R. J., & Jaccard, J. J. (1976). An informational approach to altruistic
behavior. Journal of Personality and Social Psychology, 33, 3 17-326.
Randall, D. M., & Gibson, A. M. (1991). Ethical decision making in the
medical profession: An application of the theory of planned behavior.
Journal of Business Ethics, 10, 11 1- 122.
Rotter, J. B. (1966). Generalized expectancies for internal versus external control
of reinforcement. Psychological Monographs, 80, 1(Whole No. 609).
Schifter, D. E., & Ajzen, I. (1985). Intention, perceived control, and weight
loss: An application of the theory of planned behavior. Journal of Person-
ality and Social Psychology, 49, 843-85 1.
Schwarzer, R., & Fuchs, S. (1996). Self-efficacy and health behaviors. In M.
Conner & P. Norman (Eds.), Predicting health behavior: Research and
practice with social cognition models. Milton Keynes, England: Open
University Press.
PERCEIVED BEHAVIORAL CONTROL 437

Secretary of State for Health. (1991). The health of the nation: A consultative
document for health in England. London, England: HMSO.
Sparks, P. (1994). Attitudes towards food: Applying, assessing and extending
the theory of planned behaviour. In D. R. Rutter & L. Quine (Eds.), The
social psychology of health and safety: European perspectives (pp. 25-
46). Aldershot, England: Avebury.
Sparks, P., Hedderley, D., & Shepherd, R. (1992). An investigation into the
relationship between perceived control, attitude variability, and the con-
sumption of two common foods. European Journal of Social Psychology,
22,55-71.
Sparks, P., Shepherd, R., Wieringa, N., & Zimmermanns, N. (1995). Perceived
behavioural control, unrealistic optimism and dietary change: An explora-
tory study. Appetite, 24,243-255.
Stephen, A. M., & Wald, N. J. (1990). Trends in individual consumption of
dietary fat in the United States, 1920-1984. American Journal of Clinical
Nutrition, 52, 457-469.
Strack, F. (1992). “Order effects” in survey research: Activation and informa-
tion functions of preceding questions. In N. Schwarz & S. Sudman (Eds.),
Context effects in social and psychological research (pp. 23-34). New
York, NY: Springer.
Terry, D. J. (1993). Self efficacy expectancies and the theory of reasoned
action. In D. J. Terry, C. Gallois, & M. McCamish (Eds.), The theory of
reasoned action: Its application to AIDS-preventative behavior (pp. 135-
15 1). Oxford, England: Pergamon.
Terry, D. J., & O’Leary, J. E. (1995). The theory of planned behaviour: The
effects of perceived behavioural control and self-efficacy. British Journal
of Social Psychology, 34, 199-220.
Triandis, H. (1 977). Interpersonal behavior. Monterey, CA: Brooks/Cole.
Weiner, B. (1979). A theory of motivation for some classroom experiences.
Journal of Educational Psychology, 71,3-25.
White, K . M., Terry, D. J., & Hogg, M. A. (1994). Safe sex behavior: The role
of attitudes, norms, and control factors. Journal ofApplied Social Psychol-
ogy, 24,2164-2192.
World Health Organization. (1990). Diet, nutrition and the prevention of
chronic diseases. Geneva, Switzerland: WHO.
Yordy, G. A., & Lent, R. W. (1993). Predicting aerobic exercise participation-
social cognitive, reasoned action and planned behavior models. Journal of
Sport and Exercise Psychology, 15,363-374.
438 SPARKS, GUTHRIE. AND SHEPHERD

Appendix

These PBC questions are listed in the sequence that they appeared in the
questionnaire used in Study 2 (the order was reversed for half of the respon-
dents). Unless otherwise indicated, response scales are 7-point and only the
endpoints are anchored (endpoints are indicated in parentheses):
Dfficultyl: “For me to reduce the amount of chips that I eat from now on
would be . . .” (easy and difficult; adapted from Ajzen & Driver, 1991).
Miscellaneous9: “It is mostly up to me whether or not I reduce the amount
of chips that I eat from now on” (strongly agree and strongly disagree; adapted
from Netemeyer & Burton, 1990).
Dificulty2: “If I wanted to, I could easily reduce the amount of chips that
I eat from now on” (extremely likely and extremely unlikely; adapted from
Ajzen & Madden, 1986).
Control2: “How much personal control do you feel you would have over
whether or not you reduce the amount of chips that you eat from now on?”
(complete control and no control; adapted from Randall & Gibson, 1991).
Control3: “How much control do you have over reducing the amount of
chips that you eat from now on?” (complete control and absolutely no control;
adapted from Madden, Ellen, & Ajzen, 1992).
Miscellaneousl: “What is the likelihood that if you tried you would be able
to reduce the amount of chips that you eat from now on?” (likely and unlikely;
adapted from Kimiecik, 1992).
Miscellaneous2: “How certain are you that you could reduce the amount of
chips that you eat from now on?” (not at all certain and very certain; adapted
from Schwarzer & Fuchs, 1996).
ControN: “The number of events outside my control which could prevent
me from reducing the amount of chips that I eat from now on are . . .”
(numerous and very few; adapted from Madden et al., 1992).
Controll: “How much control do you have over whether you do or do not
reduce the amount of chips that you eat from now on?” (complete control and
very little control; adapted from Ajzen & Madden, 1986).
Dfficulty4: “For me to reduce the amount of chips that I eat (to the extent
that I would like to) from now on would be . . .” (extremely easy to extremely
dfficult; adapted from Sparks, Hedderley, & Shepherd, 1992).

You might also like