Do intentions predict condom use? Metaanalysis and examination of six moderator variables

British Journal of Social Psychology (1998), 37, 231-250 0 1998 The British Psychological Society

Printed in Great Britain 231

Do intentions predict condom use? Meta- analysis and examination of six moderator

variables

Paschal Sheeran* and Sheina Orbell Department of Pyhology, University of Shefield, Shefield S10 ZTN, UK

This study used meta-analysis to quantify the relationship between intentions and behaviour in prospective studies of condom use. The effects of six moderator variables were also examined : sexual orientation, gender, sample age, time interval, intention versus expectation and condom use with ‘steady’ versus ‘casual’ partners. Literature searches revealed 28 hypotheses based on a total sample of 2532 which could be included in the review. Overall, there was a medium to strong sample-weighted average correlation between intentions and condom use (r+ = .44), and this correlation was similar to the effect sizes obtained in previous reviews. There were too few studies of gay men to permit meaningful comparison of effect sizes between homosexual versus heterosexual samples. Gender and measurement of intention did not moderate the intention-behaviour relationship. However, shorter time intervals, older samples and condom use with ‘steady’ rather than ‘casual ’ partners were each associated with stronger correlations between intentions and condom use. Factors which might explain the significant effects of moderator variables are discussed and implications of the study for future research on intention-behaviour consistency are outlined.

Unprotected penetrative sex is the primary transmission route for human immunodeficiency virus (HIV), the agent which causes AIDS (acquired immunodeficiency syndrome). Condom use can prevent sexual transmission of HIV and is more effective than reducing numbers of sexual partners (Reiss & Leik, 1989). Although time trend analyses show that condom use has increased among both heterosexuals (e.g. Catania, Stone, Binson & Dolcini, 1995 ; DeVroome, Paalman, Dinglestad, Kolker & Sandfort, 1994; Robertson, 1995) and gay men (e.g. Flowers, Sheeran, Smith & Beail, 1997; Hospers & Kok, 1995; Stall, Coates & Hoff, 1988), the absolute level of condom use remains low. For example, a nationally representative survey of people in the UK and France found that 40-60% of the sexually active sample had never used a condom in the previous 12 months (Bajos e t al., 1995).

Social psychology can contribute to reducing the spread of HIV/AIDS by identifying psychological prerequisites of HIV-preventive behaviours such as condom use (Abraham & Sheeran, 1993, 1994). There have been several applications

* Requests for reprints.

232 Paschal Sbeeran and Sbeina Orbell

of social psychological models of behaviour to condom use. These accounts propose that a person’s intention to use a condom is the most immediate, and important, predictor of that behaviour. Given the complexity of sexual behaviour, however, it remains an open question whether intentions do indeed predict future condom use. The present study uses meta-analysis (e.g. Rosenthal, 1984) to address this question by quantifying the extent to which behavioural intentions have been associated with condom use in prospective studies to date.

Social pybological models of condom use

Perhaps the most important social psychological models of behaviour that have been applied to condom use are the theory of reasoned action (Ajzen & Fishbein, 1980) and the theory of planned behaviour (Ajzen, 1985, 1991). These models specify different predictors of intention. According to the theory of reasoned action, people’s intentions to perform a behaviour are predictable from their attitude towards the behaviour, their positive or negative evaluation of their performing the behaviour (e.g. ‘For me, using a condom would be good/bad’), and from their subjective norm, their beliefs about what significant others think that they should do (e.g. ‘Most people who are important to me think that I should/should not use a condom’). The theory of planned behaviour posits an additional variable which influences intention and which may also directly affect behaviour : perceived behavioural control (Ajzen & Madden, 1986). Perceived behavioural control refers to the person’s perceptions of the ease or difficulty of performing the behaviour (e.g. ‘Whether or not I use a condom is entirely under/outside my control ’) and is closely related to the notion of self-eflcacy (Bandura, 1992; see, however, Terry & O’Leary, 1995 for a study of the distinctiveness of the two constructs).

While the theories of reasoned action and planned behaviour differ on the proposed determinants of intention, both models regard forming an intention (e.g. ‘I intend using a condom the next time I have sex with someone new’) as the prerequisite of behavioural performance. Behavioural intentions are presumed to mediate the effects of variables extraneous to the models such as demographic characteristics as well as attitudes and subjective norms (though not necessarily perceived behavioural control). The intention construct therefore provides a summary of the person’s motivational orientation towards performing a behaviour. Ajzen (1991), for example, states that:

Intentions are assumed to capture the motivational factors that influence a behavior; they are indicators of how hard people are willing to try, of how much effort they are planning to exert, in order to perform the behavior (p. 181).

Previous meta-analytic reviews have shown that intentions are good predictors of behaviours in a variety of domains. Across 88 studies, Sheppard, Hartwick & Warshaw (1988) obtained an average correlation of .53 between intention and behaviour, while Randall & Wolff’s (1994) review of 98 studies obtained an average correlation of .45. While these findings are encouraging, there are reasons to suspect that intentions may not predict condom use as well as intentions predict other behaviours. Condom use is less an individual than a joint behaviour which requires

Intentions and condom use 233

the cooperation of a sexual partner (Kashima, Gallois & McCamish, 1993). Because sexual partners may have different intentions regarding condom use, intentions obtained from one partner may not be predictive of their joint behaviour. Condom use also requires resources (e.g. having a condom available) and opportunity (e.g. a prospective sexual partner). Both of these factors are thought to attenuate the relationship between intentions and behaviour (Liska, 1984).

The first aim of the present study, therefore, is to systematically examine the extent to which behavioural intentions are associated with condom use among heterosexual and gay men using meta-analytical procedures. The second aim is to examine a number of potential moderators of the intentionxondom use relationship. Specifically, sample factors such as sexual orientation, gender and age, and methodological issues such as the time interval between measures of intention and behaviour and the measurement of both intention (intention versus expectation) and condom use (‘casual ’ versus ‘steady ’ partner) could each influence the relationship between intentions and condom use.

Sexual orientation

Since there are no theoretical grounds for supposing that the correlations between intentions and condom use should differ for gay versus heterosexual samples, no predictions were made regarding the influence of this variable.

Gender

Several researchers have suggested that men possess greater power in heterosexual relationships than women (Holland, Ramazaglou & Scott, 1990a; Holland, Ramazaglou, Scott, Sharpe & Thomson, 19906; Wight, 1992). This may mean that women are less able to translate their intentions to use a condom into action than men. Consistent with this view, Abraham, Sheeran, Abrams & Spears (1996) found that while intentions were significantly associated with condom use among men in their sample, the correlation between intention and condom use was not significant among women.

Researchers have argued that sexual scripts (Gagnon & Simon, 1974) which provide implicit understandings of gender-appropriate roles and behaviours in sexual contexts accord men a more ‘agentic’ role in terms of initiating and coordinating sexual relations (Rose & Frieze, 1989) and that men’s sexual pleasure is privileged in such scripts (Nicolson, 1993). Women may also face emotional and/or physical coercion from men when they try to use a condom. Experiences of frequently being ‘pressured’ to engage in intercourse by men have been reported by women from a wide variety of backgrounds (Biglan, Noel, Ochs, Smolkowski & Metzler, 1995; Holland e t al., 19906) and these reports of pressured intercourse are associated with condom non-use (Biglan et al., 1995).

A related consideration is women’s self-efficacy to use condoms. Morrison, Rogers Gillmore & Baker (1995) point out that a woman depends more on a male partner’s cooperation than vice versa, since using a condom is a behaviour that he, rather than she, performs. This lack of direct control over the behaviour may have


a negative impact upon women’s self-efficacy to use condoms (Kasen, Vaughan & Walter, 1992). Since self-efficacy or perceived behavioural control can have a direct impact upon behaviour which is not mediated by intention (Ajzen & Madden, 1986), this might mean that women are less able to act upon their intentions to use a condom than men.

Evidence also suggests that adolescents may be less able to translate their intentions to use condoms into action compared to undergraduate and adult samples (Fisher, Fisher & Rye, 1995). Two factors might be responsible. First, intentions to use condoms may be very unstable among this group. For example, Stanton e t al. (1996) found that 58 per cent of their sample of 9-15-year-olds changed their intentions over a six-month interval. Similarly, Reinecke, Schmidt & Ajzen (1 996) found relatively small correlations (.39 < rs < .46) between measures of intention taken one year apart among a representative sample of German youth. This temporal inconsistency is important because unstable intentions have been found to attenuate the relationship between intentions and behavioural performance (Bagozzi & Yi, 1989; Doll 8c Ajzen, 1992).

Second, age and sexual experience are positively correlated (Dunne, Donald, Lucke, Nilsson, Ballard & Raphael, 1994). This means that older samples are likely to have greater knowledge of sexual scripts and greater condom use self-efficacy and may be better able to implement their intentions to use a condom. Kashima e t a/ . (1 993) have shown that previous experience of condom use increases the consistency between intentions and subsequent use. In their study, intenders with direct experience of condom use were more likely to use a condom than intenders who had no prior experience. Thus, intention stability and lack of direct experience may mean that adolescents’ intentions are less predictive of condom use than the intentions of older samples.

Time interval between measurements of intention and behaviour

Ajzen (1985) and Ajzen & Fishbein (1980) have argued that stronger relationships between intention and behaviour will be obtained when the time interval between the two measures is shorter than when it is longer. The time interval may influence the intention-behaviour correlation because intentions may become unstable over time or because unforeseen obstacles prevent action (Ajzen, 1985 ; Cote, McCullough & Reilly, 1985). In a meta-analytic review of this issue, however, Randall & Wolff (1 994) found no significant relationship between the length of delay between assessment of intention and behaviour and the strength of the intention-behaviour correlation (r = - .06, n.s.).

Examination of the data employed in Randall & Wolffs (1994) study suggests caution in accepting their conclusion that time interval does not affect intention- behaviour relations. Randall & Wolff examined 98 hypotheses which were distributed across five time intervals (less than one day, less than one week, less than one month, less than one year, greater than one year) and seven ‘types of behaviour’


(food/beverage, sexual/reproductive, drug/alcohol, political/voting, leisure/ exercise, school/work/job/career, and ‘other behaviours’). This yields a 5 x 7 matrix of time interval by behaviour type. Inspection of the numbers of behaviours in each cell reveals that there are no data available for 10 of the cells while a further 8 cells contain just one datum. Thus, time interval and behaviour type would seem to be confounded.

Randall & Wolff also analysed the impact of time interval within each behaviour type (excluding ‘ other behaviours’) and found a significant association between time interval and the intention-behaviour correlation for just one of the six behaviour types-drugs/alcohol behaviours. Even within each behaviour type, however, there is at least one empty time interval cell for each of the six types of behaviour. Since the missing time interval cell varies for different types of behaviour, time interval and behaviour type would again seem to be confounded.

We would argue that a stronger test of the effects of temporal contiguity on intention-behaviour relations would be provided by examining the effects of time interval in the context of a single behaviour than in the context of several different behaviours of the same ‘type’. Consistent with Azjen’s (1985) analysis, we hypothesize that longer time intervals will attenuate the strength of the intention-condom use relationship here.

Behavioural intention versus behavioural expectation

Sheppard e t a/ . (1988) and Warshaw & Davis (1985) have drawn attention to a distinction between behavioural intentions and behavioural expectations. Whereas intention refers to what one intends or plans to do (e.g. ‘ I intend using a condom the next time I have sexual intercourse 7, behavioural expectation refers to self- predictions about what one is likely to do (e.g. ‘How likely is it that you will use a condom the next time you have sexual intercourse?’). Measures of behavioural expectation are thought to encompass people’s perceptions of factors which may impede performance of a behaviour, such as situational constraints or lack of ability, and may therefore provide better predictors of behaviour than traditional measures of intention (Warshaw & Davis, 1985). Support for this view comes from Sheppard e t al.’s (1988) meta-analysis of the theory of reasoned action. They found that behavioural expectations were more strongly correlated with behaviour than behavioural intentions. Randall & Wolff (1994), on the other hand, found that intention versus expectation did not moderate the relationship between time interval and the intention-behaviour correlation.

‘ Casual’ versus ‘ steah ’ partner

Sheeran & Abraham (1994) showed that measures of condom use employed in most studies of HIV-preventive behaviour do not specify the type of partner (e.g. ‘new ’, ‘casual’ or ‘steady’ partner) with whom a condom was used. Research suggests, however, that intentions may be better predictors of condom use with ‘steady’ or ‘regular’ partners than condom use with ‘casual’ or ‘new’ partners. Morrison e t a/ . (1995) argued that the theory of reasoned action should better predict condom use

236 Paschal Sheeran and Sheina Orbell

among steady partners than casual partners because the beliefs and attitudes of casual partners are less well known to the actor, leading to ‘greater ambiguity in the formation of, and follow-through on, intentions to use condoms’ (p. 654). Findings appear to support this view. Morrison e t af. (1995) and Galligan & Terry (1993) both found that condom use was more predictable for steady partners than for casual partners.

The present stub In summary, the present study uses meta-analysis to determine the strength of the relationship between intentions and condom use among heterosexual and gay respondents. The effects of six potential moderators of this relationship are also examined : (i) sexual orientation, (ii) gender, (iii) age, (iv) time interval, (v) intention versus expectation and (vi) type of partner.

Method

Sample of studies Several methods were used to generate the sample of studies: (u) computerized searches of social scientific and medical databases (PsychLit, PsychINFO, Social Science Citation Index (BIDS), Medline, Index Medicus, AIDSline, Dissertation Abstracts Online and the Conference Papers Index) from the first report of HIV/AIDS (January 1981) to the time of writing (May 1997), (b) reference lists in each article identified above were evaluated for inclusion, and (c) the authors of published articles were contacted and requests were made for unpublished studies and studies in press.

1. Studies had to include a measure of intention and a measure of self-reported condom use. Studies which did not disaggregate condom use from other measures of HIV-preventive behaviour, such as abstinence or non-penetrative sex, were excluded. While these studies are informative, as DiClemente (1992) points out, composite measures of HIV-preventive behaviour mean that the effects of predictors as they specifically relate to condom use cannot be isolated. Studies which did not disaggregate condom use from other measures of contraception were also excluded for this reason.

2. A bivariate statistical relationship between intention and condom use had to be retrievable from studies. Where studies did not include relevant statistics, the authors of the study were contacted and requests were made for bivariate associations. Almost all authors provided these data (Boldero, Moore & Rosenthal, 1992; Morrison, 1993; Morrison et ul., 1995; Reinecke etul., 1995; Rye, 1995; White, Terry & Hogg, 1994).

3. Studies had to measure intention at time To and measure condom use behaviour at some later time T,. Studies which reported contemporaneous measures of intentions and behaviour were excluded because cross-sectional designs do not permit causal inferences (Basen-Engquist, 1992; Basen- Engquist & Parcel, 1992; Brown, DiClemente & Park, 1992; Cochran, Mays, Ciarletta, Caruso & Mallon, 1992; Hernandez & DiClemente, 1992; Jemmott & Jemmott, 1991 ; Macey & Boldero, 1992; Schaalma, Kok & Peters, 1993; Trefie, Juggemann & Ross, 1992).

Using these inclusion criteria, a total of 28 tests of the association between intention and condom use could be used in the review. Of these, just two hypotheses came from samples of gay men. The remainder involved exclusively or predominantly heterosexual samples. The 18 studies which yielded the 28 effect sizes are preceded by an asterisk in the reference list. These 18 studies include 2 unpublished papers (yielding three hypotheses: Morrison, 1993; Rye, 1995).

Study characteristics were coded independently by the authors. Reliabilities were uniformly high, ranging from 94 to 100 per cent. Disagreements were jointly resolved. Table 1 presents the characteristics and effect sizes obtained from each study.

There were several inclusion criteria for the review:

Tab

le 1

. Stu

dies

of

the

rela

tion

ship

bet

wee

n be

havi

oura

l int

enti

ons

and

hete

rose

xual

con

dom

use

Tim

e in

terv

al

Res

pond

ent

Aut

hor(

s)

Sam

ple

Inte

ntio

n ve

rsus

exp

ecta

tion

(wee

ks)

Typ

e of

par

tner

se

x N

r

Abr

aham

, Sh

eera

n,

Abr

ams

& S

pear

s (1

996)

Bol

dero

, Moo

re &

R

osen

thal

(19

92)

Bre

akw

ell,

Mill

war

d &

Fife

-Sha

w (

1994

)

Bry

an, A

iken

&

Wes

t (19

96)

de W

it, v

an

Grie

nsve

n, Kok

Sa

ndfo

rt (1

993)

Fish

er (

1984

)

Ran

dom

sam

ple

of

hete

rose

xual

ad

oles

cent

s (1

6-19

ye

ars)

Het

eros

exua

l un

derg

radu

ates

Ran

dom

sam

ple

of

hete

rose

xual

ad

oles

cent

s (1

6-20

ye

ars)

Het

eros

exua

l col

lege

st

uden

ts

Gay

men

atte

ndin

g M

unic

ipal

Hea

lth

Clin

ic in

Am

ster

dam

(m

ean

age =

41.

2 ye

ars)

Het

eros

exua

l un

derg

radu

ates

Inte

ntio

n, 1

ite

m (‘In

futu

re, I

inte

nd

to u

se a

con

dom

if I

hav

e se

x w

ith

som

eone

new

’; 5

-poi

nt s

cale

, ‘ s

tron

gly

agre

e ’ to

‘ st

rong

ly

disa

gree

’)

Inte

ntio

n“,

1 ite

m (

‘Stre

ngth

of

inte

ntio

n’,

5-po

int

scal

e, ‘v

ery

dete

rmin

ed n

ot t

o us

e a

cond

om’

to

‘ver

y de

term

ined

to

use

a co

ndom

’)

Expe

ctat

ion,

2 it

ems,

[‘7

-poi

nt s

cale

(‘I

do

not

expe

ct t

o ha

ve s

ex’

to ‘

I de

finite

ly w

ill d

o th

is’

for

two

inst

ance

s of

con

dom

use

) (a

lway

s us

e co

ndom

s/us

e co

ndom

s w

hen

not

certa

in a

bout

the

oth

er p

erso

n’s

sexu

al h

isto

ry) ’1

Ex

pect

atio

n, 4

item

s (e

.g. ‘

How

lik

ely

is it

tha

t yo

u w

ill u

se a

co

ndom

the

nex

t tim

e yo

u ha

ve

inte

rcou

rse?

’, re

spon

se o

ptio

ns n

ot

repo

rted)

, al

pha =

.77

Inte

ntio

n, 1

item

(‘D

o yo

u in

tend

to

use

a c

ondo

m w

hen

you

have

an

al in

terc

ours

e w

ith a

cas

ual

part

ner

in t

he f

utur

e?’;

5-p

oint

sc

ale;

‘ce

rtain

ty n

ot’

to ‘

yes,

ce

rtain

ly ’)

Expe

ctat

ion,

1 i

tem

(‘ u

nlik

ely-

likel

y I

will

alw

ays

use

cond

oms’

; num

ber

of p

oint

s on

the

scal

e no

t sp

ecifi

ed)

52

‘ New

’ pa

rtner

W

omen

81

.1

5 M

en

41

.33

6 N

ot s

peci

fied

Wom

en

95

.31

Men

49

.4

0

52

Not

spe

cifie

d

I

P a.

6 N

ot s

peci

fied

2:

Wom

en

81

.69

2

26 4

I. C

asua

l ’ p

artn

er

Not

spe

cifie

d

Men

24

4 .2

1

Men

44

.55

P;,

CJ

4

N

w

00

Tab

le 1

. (co

nt.)

Tim

e in

terv

al

Res

pond

ent

Aut

hor(

s)

Sam

ple

Inte

ntio

n ve

rsus

exp

ecta

tion

(wee

ks)

Type

of

partn

er

sex

N

1

Fish

er, F

ishe

r &

C

onve

nien

ce s

ampl

e

recr

uite

d fr

om g

ay

orga

niza

tions

Rye

(19

95)”

of

gay

men

Het

eros

exua

l un

derg

radu

ates

Het

eros

exua

l 9th

gr

ade

high

sch

ool

pupi

ls (

adol

esce

nts)

Gal

ligan

& T

erry

H

eter

osex

ual

(1 99

3)

unde

rgra

duat

es

Gal

lois

, Kas

him

a,

Pred

omin

antly

Te

rry,

McC

amis

h,

hete

rose

xual

Ti

mm

ins

&

conv

enie

nce

sam

ple

Cha

uvin

(1 99

2)

obta

ined

thr

ough

st

uden

t gr

oups

and

so

cial

net

wor

ks

Inte

ntio

n, 1

item

(‘If

I h

ave

inse

rtive

8

Not

spe

cifie

d M

en

29d

.59

anal

inte

rcou

rse

in t

he n

ext t

wo

mon

ths,

I in

tend

to

alw

ays

use

late

x co

ndom

s’;

5-po

int s

cale

, ‘ve

ry

likel

y’ to

‘ve

ry u

nlik

ely’

)

anal

inte

rcou

rse

in th

e ne

xt t

wo

Inte

ntio

n, 1

item

(‘I

f I

have

rec

eptiv

e 8

Not

spe

cifie

d

mon

ths,

I in

tend

to

alw

ays

use

late

x co

ndom

s ’ ; 5

-poi

nt s

cale

, ‘ ve

ry

likel

y’ to

‘ve

ry u

nlik

ely’

) In

tent

ion,

1 it

em (

‘If I

hav

e se

x du

ring

the

next

tw

o m

onth

s, I

in

tend

to

alw

ays

use

late

x co

ndom

s’; 5

-poi

nt s

cale

, ‘ve

ry

likel

y’ to

‘ve

ry u

nlik

ely’

) In

tent

ion,

1 it

em (

‘If I

hav

e se

x du

ring

the

next

tw

o m

onth

s, I

in

tend

to

alw

ays

use

late

x co

ndom

s ’ ; 5

-poi

nt s

cale

, ‘ ve

ry

likel

y’ to

‘ve

ry u

nlik

ely’

)

8 N

ot s

peci

fied

4 N

otsp

ecif

ied

0

P

29

.ll

Men

P

Expe

ctat

ion,

1 it

em (

‘Ove

r th

e ne

xt

12

‘Reg

ular

’ pa

rtner

sc

Mix

ed

50

.63

thre

e m

onth

s I

will

def

inite

ly u

se

12

‘Cas

ual/n

ew’

Mix

ed

27

.38

cond

oms

with

reg

ular

(ca

sual

/new

) pa

rtner

s pa

rtner

s’;

7-po

int

scal

e, ‘

very

un

likel

y’ to

‘ve

ry li

kely

’)

Inte

ntio

n, 1

item

(‘W

heth

er t

hey

8 N

otsp

ecif

ied

inte

nded

to

use

a co

ndom

dur

ing

thei

r ne

xt s

exua

l enc

ount

er’;

7-po

int

scal

e, ‘

defin

itely

not

inte

nd’ t

o ‘ d

efin

itely

inte

nd ’)

Mix

ed

144

.49

Gal

lois

, Ter

ry,

Het

eros

exua

l In

tent

ion,

1 it

em (

‘Whe

ther

the

y Ti

mm

ins,

Kas

him

a un

derg

radu

ates

in

tend

ed t

o pe

rfor

m t

heir

sexu

al

& M

cCam

ish

(199

4)

activ

ities

with

the

mse

lves

or

thei

r

Mor

rison

(19

93)

Mor

rison

, R

oger

s G

illm

ore

& B

aker

(1

995)

Rei

neck

e, S

chm

idt

& A

jten

(199

6)

Rye

(19

95)

Sand

erso

n &

Je

mm

ott

(199

6)

Het

eros

exua

l tee

nage

rs

at s

exua

lly

trans

mitt

ed d

isea

ses

(STD

) cl

inic

s and

ju

veni

le d

eten

tion

cent

res

Het

eros

exua

l adu

lt ST

D c

linic

atte

nder

s (m

ean

age =

27.

7 ye

ars)

Ran

dom

hou

seho

ld

surv

ey o

f ad

oles

cent

s (p

redo

min

antly

he

tero

sexu

al)

Pred

omin

antly

he

tero

sexu

al

unde

rgra

duat

es

Pred

omin

antly

he

tero

sexu

al

unde

rgra

duat

es

partn

er u

sing

a c

ondo

m o

n th

eir

next

sex

ual e

ncou

nter

’; 7-

poin

t sc

ale,

‘def

inite

ly d

o no

t int

end’

to

‘def

inite

ly in

tend

’) Ex

pect

atio

n, 1

item

(‘H

ow l

ikel

y ar

e yo

u to

use

con

dom

s w

ith y

our

stea

dy/c

asua

l pa

rtner

(s)

over

the

next

3 m

onth

s?’,

‘ver

y un

likel

y’ to

‘v

ery

likel

y’)

Expe

ctat

ion,

1 it

em (

‘How

lik

ely

are

you

to u

se c

ondo

ms

with

you

r st

eady

/cas

ual

partn

er(s

) ov

er t

he

next

3 m

onth

s?’;

‘ver

y un

likel

y’ to

‘v

ery

likel

y’)

Inte

ntio

n, 3

item

s (‘

I in

sist

on

usin

g a

cond

om w

ith n

ew s

exua

l par

tner

s ev

en if

my

partn

er d

oes

not

wan

t to

’; 3

-poi

nt s

cale

, ‘ye

s, tr

ue’,

‘do

n’t

know

’, ‘n

o, f

alse

’)

Inte

ntio

n, n

o de

tails

Expe

ctat

ion,

2 it

ems

(e.g

. ‘H

ow

likel

y is

it th

at y

ou w

ill u

se

cond

oms

if yo

u de

cide

to h

ave

sex

in th

e ne

xt 3

mon

ths?

’; 5

-poi

nt

scal

e, ‘ v

ery

unlik

ely ’

to ‘ v

ery

likel

y’),

alph

a =

.72

8 12

12

52 8 12

Not

spe

cifie

d

Cas

ual p

artn

er

Stea

dy p

artn

er

Cas

ual p

artn

er

Stea

dy p

artn

er

Cas

ual p

artn

er

Stea

dy p

artn

er

Cas

ual p

artn

er

Stea

dy p

artn

er

‘New

’ par

tner

s

Not

spe

cifie

d

Not

spe

cifie

d

Wom

en

91

.60

Men

70

.5

3

Wom

en”

Wom

en

Men

M

en

Wom

en

Wom

en

Men

M

en

Mix

ed

Wom

en

Mix

ed

43

140 32

77

38

163 52

105

172 56

85

.18

.49

.32

.45

.28

.31

.26

.49

.22

1.24

, .2

0, .2

2]’

.55

[.50,

.60

]’

.66

Tab

le 1

. (c

ont.)

Tim

e in

terv

al

Inte

ntio

n ve

rsus

exp

ecta

tion

(wee

ks)

Aut

hor(

s)

Stan

ton,

Li,

Bla

ck,

Ric

ardo

, Gal

brai

th,

Feig

elm

an &

Kal

jee

(1 99

6)

van

der

Vel

de,

Hoo

ykaa

s &

va

n de

r Pl

igt

(1 99

6)

Sam

ple

Het

eros

exua

l you

ng

peop

le a

ged

9-15

ye

ars

in p

ublic

ho

usin

g de

velo

pmen

ts

clin

ic a

ttend

ers

(age

>

17

year

s)

Het

eros

exua

l STD

~ ~~

~

~~

~

Expe

ctat

ion,

1 it

em (

‘How

lik

ely

is it

th

at y

ou w

ill u

se a

con

dom

the

next

tim

e yo

u ha

ve s

ex?’

; %

poin

t sc

ale,

‘li

kely

’, ‘u

ncer

tain

’, ‘u

nlik

ely’

)

24

Inte

ntio

n, 1

item

(In

tend

to

use

16

cond

oms

with

priv

ate/

pros

titut

ion

16

cont

acts

; 5-

poin

t sc

ale,

‘def

inite

ly

no’

to ‘

defin

itely

yes

’)

Res

pond

ent

Type

of

partn

er

sex

N

f

Not

spe

cifie

d M

ixed

24

.0

5

‘Pri

vate

’ par

tner

s*

Mix

ed

100

.35

\

partn

ers

2 a\

‘ Pro

stitu

tion ’

Mix

ed

147

.42

3 a W

hite

, H

ogg

Terr

y &

H

eter

osex

ual

Inte

ntio

n, 3

item

s (e

.g. ‘

I int

end

to

4 N

ot s

peci

fied

Mix

ed

164

.80

P s

(199

4)

univ

ersi

ty s

tude

nts

use

a co

ndom

eve

ry ti

me

I ha

ve s

ex

5 du

ring

the

next

mon

th’;

7-p

oint

sc

ale,

’ extre

mel

y un

likel

y’ to

‘e

xtre

mel

y lik

ely’

), al

pha =

.96

2.

3 0

Bol

der0

et u

l. (1

992)

em

ploy

ed two

mea

sure

s of i

nten

tion

to u

se a

con

dom

: a ‘

prio

r int

entio

n’ m

easu

re a

nd a

mea

sure

of ‘

inte

ntio

n in

act

ion’

. T

he la

tter

mea

sure

ref

ers

to re

spon

dent

s’ p

erce

ptio

ns o

f the

ir in

tent

ions

imm

edia

tely

pri

or to

inte

rcou

rse.

Bcc

ause

inte

ntio

n in

act

ion

was

mea

sure

d at

the

sam

e tim

e as

con

dom

use

, onl

y th

e pr

ior

inte

ntio

n m

easu

re is

incl

uded

her

e.

Indi

cate

s a

mix

ed sex s

ampl

e. D

ata

wer

e no

t dis

aggr

egat

ed f

or m

en a

nd w

omen

. Th

ree

inde

pend

ent

sam

ples

wer

e st

udie

d.

Res

pond

ents

hav

ing

inse

rtive

or r

ecep

tive

anal

inte

rcou

rse

are

not i

ndep

ende

nt. I

n or

der t

o co

mpu

te th

e ov

eral

l int

entio

n-co

ndom

use

effe

ct s

ize,

the

aver

age-

wei

ghte

d co

rrel

atio

n fo

r the

two m

easu

res

was

em

ploy

ed a

nd th

e la

rges

t N in

the

anal

ysis

(cf.

Ger

rard

, Gib

bons

& B

ushm

an, 1

996)

. ‘R

espo

nden

ts w

ith ‘casual’ a

nd ‘s

tead

y’ p

artn

ers

are

not i

ndep

ende

nt. I

n or

der t

o co

mpu

te th

e ov

eral

l int

entio

n-co

ndom

use

effe

ct size, th

e av

erag

e w

eigh

ted

corr

elat

ion

for

the two

mea

sure

s w

as e

mpl

oyed

and

the

larg

est

N in

the

anal

ysis

(cf.

Ger

rard

et uJ

., 19

96).

’Thr

ee m

easu

res o

f co

ndom

use

wer

e em

ploy

ed.

’“Pr

ivat

e’ p

artn

ers

refe

r to

resp

onde

nts

who

onl

y ha

d pr

ivat

e pa

rtner

s. Sa

mpl

es fo

r priv

ate

and

pros

titut

ion

part

ners

are

ther

efor

e in

depe

nden

t. T

wo

mea

sure

s of i

nten

tion

wer

e em

ploy

ed.


Meta-anahtic strateg))

The effect size estimate employed here was a weighted average of the sample correlations, r+. r+ describes the direction and strength of the relationship between two variables with a range of - 1.0 to + 1 .O. Computing the weighted average effect size requires a transformation of the correlation from each relevant hypothesis into Fisher’s Z. The following formula is then employed :

J(N, x r z o

JN, Average Z value =

where rzt = the Fisher’s Z transformation of the correlation from each study i, N , = number of persons in study i.

In this way correlations based on larger samples receive greater weight than those from smaller samples. The average Z value is then backtransformed to give r+ (see Hedges & Olkin, 1985; Hunter, Schmidt & Jackson, 1982).

Homogeneity analyses were conducted using the chi-square statistic (Hunter e t a/., 1982) to determine whether variation among the correlations was greater than chance. The degrees of freedom for the chi- square test is A- 1, where k is the number of independent correlations. If chi-square is non-significant, then the correlations are homogeneous and the average weighted effect size, r+s can be said to represent the population effect size.

Transformations of other statistics (e.g. t , contingency tables) to statistic r and computation of weighted average effect size and homogeneity statistics were conducted using Schwarzer’s (1988) Meta computer program.

Multiple samples and multiple measures. Where studies included more than one sample and reported separate statistical tests for each sample, then the correlation from each sample was used as the unit of analysis. Where studies included more than one measure of condom use (e.g. condom use with ‘casual’ versus ‘steady’ partners) (Morrison, 1993; Morrison et a/ . , 1995), then the weighted average correlation was computed within each independent sample of that study. The largest N in that sample was then employed in computing the overall effect size (cf. Gerrard, Gibbons & Bushman, 1996). These procedures retain the richness of the data without violating the independence assumption which underlies the validity of meta-analytic procedures.

Results

The overall intention-condom use relationship

The sample size-weighted average correlation between intention and condom use was r+ = .44 (95 per cent confidence interval = .41-.47, A = 28, N = 2532). In order to ensure that this statistic was not biased by the preponderance of published studies, the effect sizes for published versus unpublished studies were compared. The average correlations for published studies (r+ = .44, k = 25, N = 2259) and unpublished studies (r+ = .44, A = 3, N = 273) were identical. To determine the robustness of the average correlation obtained here, we estimated the number of unpublished studies containing null results which would be required to invalidate this study’s conclusion that intention and condom use are significantly related ( p < .05). The ‘Fail-safe N’ (Rosenthal, 1984) was 217. Since there are unlikely to be so many unpublished studies with null results which we were unable to locate, the r+ obtained can confidently be viewed as significantly different from zero. While the average correlation is robust, the homogeneity statistic shows considerable variation in the correlations reported in previous studies (x2 (21) = 161.65, p < .OOl). This heterogeneity encourages a search for moderators.


Effects of moderator variables on the intention-condom use relationship

The first moderator variable we had hoped to examine was sexual orientation. Unfortunately, since there were only two hypotheses involving gay men with a combined sample size of N = 273, meaningful comparison of the effect sizes from gay versus heterosexual samples was not possible. This view is supported by findings showing that the Fail-safe N for the average correlation between intentions and behaviour for gay men was 8. This value is considerably less than Rosenthal’s (1984) guidelines for regarding a correlation as ‘robust’ (5k + 10, or 20 studies in the present case). More studies of gay men are required in order to determine whether sexual orientation moderates the intention-condom use relationship.

We adopted two strategies to examine the effects of other moderators (Hunter & Schmidt, 1990). First, correlations between r+ and each of the moderator variables, gender, sample age, time interval and intention versus expectation were computed (see Table 2). Second, we treated each moderator as a categorical variable. We computed the effect size for each level of the moderator and used Fisher’s Z test for the comparison of independent correlations to test the significance of the difference between effect sizes. Table 3 presents the separate effect sizes obtained for men and women, adolescents and older samples, shorter and longer time intervals, and behavioural intention and behavioural expectation (analyses for condom use with a ‘steady’ versus a ‘casual’ partner were more complex and are described later).

Gender. We hypothesized that there would be a stronger correlation between intention and condom use for (heterosexual) men than for (heterosexual) women. However, the average effect sizes for men (r+ = .45) and women (r+ = .44) did not differ significantly (Z = 0.22, n.s.) and gender and effect size were not associated ( r = .02, n.s.). Thus, men and women do not appear to differ in their capacity to implement their intentions to use condoms.

Sample age. There was a significant correlation between sample age and the strength of the intention-condom use relationship ( r = - .69, p < .OOl). Consistent with our hypothesis, the effect size for adolescents (r+ = .25) was significantly smaller than the effect size for older samples (r+ = .50, Z = 6.48, p < .OOl). Adolescents were less able to implement their intentions to use condoms than undergraduate and adult samples.

Time interval. The correlation between the logarithmic transformation of time interval and strength of the intention-condom use relationship was also significant (r = - .59, p < .OOl). Longer delays between the assessment of intention and the assessment of condom use were associated with attenuation of the intention- behaviour correlation. Dividing time interval at the sample median (Mdn = 10 weeks), the average correlation for ‘ short’ intervals (r+ = .59) was significantly bigger than the correlation for ‘long’ intervals (r+ = .33; Z = 8.28, p < .OOl).

It should be noted that sample age and time interval were negatively correlated ( r = - .58, p < .OOl), indicating that studies of adolescent samples have generally employed longer time intervals while studies of undergraduates and adults have


Table 2. Correlations between intention-condom use effect size and moderator variables

1 2 3 4 5

1. Gender' 1 .oo - .08 - .06 - .08 .02 2. Sample ageb 1 .oo - .58* .13 .69* 3. Time interval' 1 .oo .02 - .59* 4. BI vs. BEd 1 .oo .05 5. Intention-condom use r+ 1 .oo * p < .001.

Gender was coded men = 0, women = 1. Analyses for gender do not include data for gay men. For correlations involving gender, N = 1310. N = 2532 for all other correlations. * Sample age was coded adolescents = 0, other = 1. Time interval was computed as a logarithmic transformation (base 10 log) of the delay in assessment

between intention and condom use (in weeks). As Cohen & Cohen (1983) and Randall & Wolff (1994) point out, a logarithmic transformation of time is more Likely than an untransformed variable to be linearly related to the dependent variable.

Behavioural expectation (BE) was coded = 0, behavioural intention (BI) was coded = 1.

Table 3. Intention-condom use effect sizes obtained for each moderator variable

Moderator k' Nb r+c 95% CId Chi-square'

Women Men Adolescents Older samples ' Short' time interval 'Long' time interval Intention Expectation

9 8 9

19 14 14 18 10

825 485 661

1871 1040 1492 1700 832

.44

.45

.25

.50

.59

.33

.44

.43

.38-.50

.37-.52

.17-.32

.46-.53

.54-.62

.2&.37

.40-.48

.37-.48

41.76* 11.55 15.41

119.23* 91.77* 42.92*

11 1.12* 50.68*

* p < .001. Number of correlations. Sample size upon which sample-weighted average correlation is based. Sample-weighted average correlation between intentions and condom use. 95 % confidence interval. Chi-square test for homogeneity of sample correlations.

employed shorter intervals. To ensure that the effects of time interval were independent of the effects of sample age, we compared the correlations between time interval and r+ within the adolescent and older samples. There remained a significant difference between longer and shorter time intervals for adolescents (r+ = .16 and .36, respectively, Z = 2.72, p < .O l ) and for older samples (r+ = .37 and .60 respectively, Z = 6.57, p < .OOl).

In order to ensure that the effects of sample age were independent of the effects of time interval, the effect sizes for sample age were compared within each level of time interval. There was a significant difference between the effect sizes for adolescents


(r+ = .25) and older samples (r+ = .37) for time intervals greater than 10 weeks (Z = 2.51, p < .Ol). There was also a significant difference between adolescents and older samples for shorter time intervals (r+ = .18 and .60, respectively, Z = 3 . 8 4 , ~ < .Ol). Longer time intervals and younger age, therefore, both attenuate the strength of the correlation between behavioural intention and condom use.

Bebavioural intention versus behavioural expectation. The correlation between intention versus expectation and r+ was not significant ( r = .05, n.s.) and when the effect sizes for intention (r+ = .44) and expectation (r+ = .43) were compared, the difference was not significant (Z = 0.29, n.s.). Measures of behavioural intention versus behavioural expectation appeared to have similar average correlations with condom use.

‘ Casual’ versus ‘steah’ partner. Galligan & Terry (1993), Morrison (1993) and Morrison e t al. (1995) present intention-condom use correlations separately for ‘casual’ and ‘ steady’ partners (see Table 1). We therefore computed separate effect sizes for the two types of partner. Consistent with our hypothesis, the correlation between intention and condom use with ‘ steady’ partners seemed to be stronger (r+ = .45, k = 5, N = 535, 95 % CI = .38-.51, x2 = 9.42, n.s.) than the correlation with ‘casual’ partners (r+ = .27, k = 5, N = 192,95% CI = .13-.40, xa = 0.91, n.s.). We cannot statistically compare these effect sizes because these ‘casual ’ and ‘ steady ’ samples are not independent and because the Ns differ for the two correlations.

In a second analysis we combined the effect sizes for ‘steady’ partners from Morrison (1993), Morrison e t al. (1995) and Galligan & Terry (1993) and compared the result with the combined effect sizes for ‘casual’/‘new’ partners from Abraham e t al. (1996), de Wit et al. (1993) and Reinecke e t al. (1996). The average correlation between intention and condom use with ‘steady’ partners (r+ = .45) was significantly bigger than the average correlation for ‘casual’ partners (r+ = .21, k = 4, N = 538, 95% CI = .13-.29, x2 = 0.99, n.s., Z = 4.44,p < .OOl).

Discussion

A sample size-weighted average correlation coefficient of .44 was obtained between intentions and condom use. Because condom use requires the cooperation of a sexual partner, we had anticipated that the intention-behaviour effect size here would be lower than that obtained in previous reviews. Contrary to expectations, r+ = .44 is very similar to the average correlations obtained by Randall & Wolff (1994) (r+ = .45) and Sheppard e t al. (1988) (r+ = .53) in their meta-analyses of the theory of reasoned action. Condom use is not, it seems, less predictable from intentions than are other behaviours.

Literature searches revealed just two longitudinal studies of intentions and condom use among gay men. This small sample precluded meaningful comparison of effect sizes for gay versus heterosexual samples. It is, perhaps, worrying that despite the large number of longitudinal studies of gay men (see Flowers e t al., 1997, for review), the most proximate predictor of condom use-intentions to use one-has rarely been measured. Clearly, this is a serious omission in studies to date, which needs to be rectified in future research.

Our expectation was that gender would influence the intention-condom use


relationship. Previous research showed that men have greater power in heterosexual relationships (Holland e t al., 1990a, b), which suggested that men might be better able to implement their intentions to use condoms than women. We found that the strength of the intention-condom use correlation did not differ for men and women, however. Future research will need to examine whether variables characterized by substantive gender differences in previous research such as sex roles, experiences of sexual coercion, or condom use self-efficacy might directly influence the enactment of intentions to use condoms.

We hypothesized that the intention-condom use relationship would be weaker among adolescent samples compared to older samples based on research suggesting that intentions to use condoms are very unstable among adolescents (Reinecke e t al., 1996; Stanton e t al., 1996). This hypothesis was supported both by the overall analyses and by analyses which controlled for the confounding effect for time interval. The effect of sample age upon the intention-behaviour correlation does not appear to have been examined in previous research. However, despite the significant effect obtained here, it remains unclear why age moderated the relationship between intentions and condom use. We have suggested that intention stability (Bagozzi & Yi, 1989; Doll & Ajzen, 1992) or lack of direct experience with the behaviour (Kashima e t al., 1993) might be responsible. Future research should directly address these hypotheses taking account of the need to increase adolescents’ capacity to enact their intentions to practise safer sex.

A previous meta-analysis of the effects of time interval on the intention-behaviour correlation (Randall & Wolf€, 1994) concluded that the strength of the intention- behaviour relationship does not diminish as the delay between the assessment of intention and behaviour increases. We argued that there were insufficient data in Randall & Wolff s (1994) meta-analysis to appropriately address this hypothesis and that a stronger test would be afforded by determining the effects of time interval in the context of a single behaviour. Time interval had a significant negative relationship with the intention-condom use correlation in the present study, and this effect remained significant even when sample age was controlled.

This finding supports the position repeatedly stressed by Fishbein and Ajzen (e.g. Ajzen, 1985; Ajzen & Fishbein, 1980; Ajzen & Madden, 1986) that the measure of intention should be as close as possible to the performance of the behaviour. This is not to suggest that intentions are necessarily very poor predictors of behaviour over longer time periods. Cohen (1992) suggests that a weighted average correlation of .10 should be characterized as ‘small’, a value of .30 as ‘medium’, and a value of .50 as ‘ large ’. Our findings therefore indicate that the average correlation between intention and behaviour over longer time intervals is ‘medium’ rather than small (r+ = .32), while the average correlation over shorter time intervals is ‘large’ (r+ = .56).

We also examined whether a ‘measure of behavioural intention versus behavioural expectation influenced the intention-condom use correlation. Contrary to Sheppard e t al.’s (1988) meta-analysis, but consistent with Randall & Wolff s (1994) findings, we found no difference between the average correlations between expectations and condom use versus intentions and condom use. Our data indicate that the type of measure of behavioural intention does not influence the predictive validity of that measure.


The final moderator of intention-condom use consistency examined here was condom use with ‘casual’l‘new ’ partners versus condom use with ‘steady’ partners. We hypothesized that the intention-condom use correlation would be stronger for ‘steady’ sexual partners than for ‘casual’ partners because the beliefs and attitudes of these partners are better known to actors and communication about contraceptive behaviour is more likely (Morrison e t al., 1995). Although relatively few studies specified the type of partner with whom a condom was used, our predictions were supported. Condom use with a ‘steady’ partner was better predicted by intention than was condom use with a ‘casual’ partner. This finding underlines the need to specify type of sexual partner in future psychosocial studies of condom use (Sheeran & Abraham, 1994). Specifying the type of partner with whom a condom is used would contribute to research in this area by enabling researchers to identify the unique determinants of condom use in different types of relationship. This would enable more careful targeting of psychological variables in AIDS education campaigns.

Possible criticisms of our meta-analysis should be addressed. The present research is based upon a relatively small number of hypotheses (k = 28) compared to previous reviews (Randall & WoH, 1994; Sheppard e t aL, 1988). Mullen (1984) points out that the validity of meta-analysis does not depend upon the number of studies included in a review, but depends upon the extent to which the studies which have been included are representative of the population of studies on that topic. Since the present study involved an exhaustive literature search (including unpublished research), we believe that the findings obtained here are valid. Moreover, since the present study focused upon a single behaviour, inferences about the effects of moderator variables can be made with confidence.

Our meta-analysis also has the difficulty that the measurement of condom use relies upon self-reports. Randall & Wolff (1994) have shown that there are stronger intention-behaviour correlations when self-report measures of behaviour are employed compared to more objective behaviour measures. This is a difficulty for research on sexual behaviour and for other behaviours which are sensitive, private or illegal. Catania, Gibson, Chitwood & Coates (1990) point out that there is no ‘gold standard’ for the measurement of condom use and that the employment of self- reports of behaviour is unavoidable. This does not represent a serious problem for our research, however, because test-retest reliability analyses and validation of self- reports against reports of sexual partners indicate that self-report measures of condom use do have satisfactory reliability and validity (Blake, Sharp & Temoshok, 1992; Catania e t al., 1990; Sheeran & Abraham, 1994).

In conclusion, this review finds that there is a medium to strong correlation between intentions to use condoms and condom use. The weighted average correlation obtained here does not differ substantively from the correlations found in previous meta-analyses of intention-behaviour relationships. While gender and measures of intention versus expectation did not moderate the intention-condom use relationship, shorter time intervals, older samples and condom use with a ‘steady’ rather than a ‘casual’ sexual partner were each associated with stronger correlations between intention and behaviour. Future research will need to examine variables such as intention stability and condom communication which may mediate the effects


of time interval, age and types of sexual partner on the intention-condom use relationship.

Acknowledgements The authors would like to thank Jennifer Boldero, Diane Morrison, Jost Reinecke, Barbara J. Rye, Catherine A. Sanderson and Katy White for their cooperation in providing additional data. We are particularly grateful to Jennifer Boldero, Diane Morrison and Barbara J. Rye for providing unpublished data. We would also like to thank Christine Galloway, Olivia Rickerby and Janette Watson of the Department of Information Studies, University of Sheffield for conducting the computerized literature search. We thank Penny Ditchburn for production of the tables.

References * Studies included in meta-analysis are preceded by an asterisk. Abraham, S. C. S. & Sheeran, P. (1993). In search of a psychology of safer sex promotion: Beyond

beliefs and texts. Health Education Research, 8, 245-254. Abraham, S. C. S. & Sheeran, P. (1994). Modelling and modifying young heterosexuals’ HIV-

preventive behaviour : Theories, findings and implications. Patient Education and Counseling, 23, 173-186.

* Abraham, S. C. S., Sheeran, P., Abrams, D. & Spears, R. (1996). Psychosocial determinants of condom use consistency among sexually active Scottish teenagers. Pychology and Health, 11, 641455.

Ajzen, I. (1985). A theory of planned behavior. In J. Kuhl & J. Beckmann (Eds), Action Control. New

Ajzen, I. (1 991). The theory of planned behavior. Organizational Behavior and Human Decision Processes,

Ajzen, I. & Fishbein, M. (1980). Understanding Attitudes and Predicting Behavior. Englewood Cliffs, NJ: Prentice-Hall.

Ajzen, I. & Madden, T. J. (1986). Prediction of goal-directed behavior: Attitudes, intentions and perceived behavioral control. Journal of Experimental Social Psychology, 22, 453474.

Bagozzi, R. P. & Yi, Y. (1989). Degree of intention formation as a moderator of the attitudebehavior relationship. Social Psychology Quarterly, 52, 266-279.

Bajos, N., Wadsworth, J., Ducot, B., Johnson, A. M., Le Pont, F., Wellings, K., Spira, A., Field, J. & the ACSF Group (1995). Sexual behaviour and HIV epidemiology: Comparative analysis in France and Britain. A I D S , 9, 735-743.

Bandura, A. (1992). A social cognitive approach to the exercise of control over AIDS infection. In R. J . DiClemente (Ed.), Adolescents and A I D S : A generation in Jeopar4. Newbury Park, CA: Sage.

Basen-Engquist, K. (1992). Psychosocial predictors of ‘safer sex’ behaviors in young adults. A I D S Education and Prevention, 4, 12G134.

Basen-Engquist, K. & Parcel, G. S. (1992). Attitudes, norms and self-efficacy: A model of adolescents’ HIV-related sexual risk behaviour. Health Education Quarterly, 19, 263-277.

Biglan, A., Noell, J., Ochs, L., Smolkowski, K. & Metzler, C. (1995). Does sexual coercion play a role in the high-risk sexual behavior of adolescent and young adult women? Journul of Behavioral Medicine,

Blake, S. M., Sharp, E. S. & Temoshok, L. (1992). Methodological considerations in developing measures of HIV risk-relevant behaviors and attitudes : An empirical illustration. Psychology and Health, 6, 265-280.

* Boldero, J.. Moore, S. & Rosenthal, D. (1992). Intention, context and safe sex: Australian

* Breakwell, G. M., Millward, L. J. & Fife-Shaw, C. (1994). Commitment to ‘safer’ sex as a predictor

Brown, L. K., DiClemente, R. J. & Park, T. (1992). Predictors of condom use in sexually active

* Bryan, A. D., Aiken, L. S. & West, S. G. (1996). Increasing condom use: Evaluation of a theory-

York: Springer-Verlag.

50, 1-33.

18, 549-568.

adolescents’ responses to AIDS. Journal of Applied Social Psychology, 22, 1374-1 398.

of condom use amongst 16-20 year olds. Journal of Applied Social Psychology, 24, 189-217.

adolescents. Journal of Adolescent Health, 13, 651-657.

248 Paschal Sheeran and Sbeina Orbell

based intervention to prevent sexually transmitted diseases in young women. Health Psychology, 15,

Catania, J. A., Gibson, D. R., Chitwood, D. D. & Coates, T. J. (1990). Methodological problems in AIDS behavioral research : Influences on measurement error and participation bias in studies of sexual behavior. Psychological Bulletin, 108, 339-362.

Catania, J. A., Stone, V., Binson, D. & Dolcini, D. D. (1995). Changes in condom use among heterosexuals in Wave 3 of the AMEN survey. Journal of Sex Research, 32, 193-200.

Cochran, S. D., Mays, V. M., Ciarletta, J., Caruso, C. & Mallon, D. (1992). Efficacy of the theory of reasoned action in predicting AIDS-related risk reduction among gay men. Journal of Applied Social

371-382.

Psychology, 22, 1481-1501. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159. Cohen, J. & Cohen, P. (1 983). Applied Multa$le RegressionlCorrelation Anabsis for the Behavioral Sciences.

Hillsdale, N J : Erlbaum. Cote, J. A., McCullough, J. & Reilly, M. (1985). Effects of unexpected situations on behavior-intention

differences : A garbology analysis. Journal of Consumer Research, 12, 18tF194. DeVroome, E. M. M., Paalman, M. E. M., Dinglestad, A. A. M., Kolker, L. & Sandfort, T. G. M.

(1994). Increase in safe sex among the young and non-monogamous: Knowledge, attitudes and behavior regarding safe sex and condom use in The Netherlands from 1987 to 1993. Patient Edwation and Counseling, 24, 279-288.

* de Wit, J. B. F., van Griensven, G. J. P., Kok, G. & Sandfort, T. G. M. (1993). Why do homosexual men relapse to safe sex? Predictors of resumption of unprotected anogenital intercourse with casual partners. AIDS, 7 , 11 13-1 11 8.

DiClemente, R. J. (1992). Psychosocial predictors of condom use among adolescents. In R. J. DiClemente (Ed.), Adolescents and A I D S : A Generation in Jeoparh. Newbury Park, CA: Sage.

Doll, J. & Ajzen, I. (1992). Accessibility and stability of predictors in the theory of planned behavior. Journal of Personality and Social Psychology, 63, 754-765.

Dunne, M. P., Donald, M., Lucke, J., Nilsson, R., Ballard, R. & Raphael, B. (1994). Age-related increase in sexual behaviours and decrease in condom use among adolescents in Australia. International Journal of S T D and AIDS, 5, 41-47.

* Fisher, W. A. (1984). Predicting contraceptive behavior among university men: The role of emotions

* Fisher, W. A., Fisher, J. D. & Rye, B. (1995). Understanding and promoting AIDS preventive

Flowers, P., Sheeran, P., Smith, J. A. & Beail, N. (1997). The role of psychosocial factors in HIV risk- reduction among gay and bisexual men: A quantitative review. Psychology and Health, 12, 197-230.

Gagnon, J. & Simon, W. (1974). Sexual Conduct: The Social Sources of Human Sexualig. London: Hutchinson.

* Galligan, R. F. & Terry, D. (1993). Romantic ideals, fear of negative implications and the practice of safe sex. Journal of Applied Social Psychology, 23, 1685-1711.

* Gallois, C., Kashima, Y., Terry, D., McCamish, M., Timmins, P. & Chauvin, A. (1992). Safe and unsafe intentions and behavior: The effects of norms and attitudes. Journal of Applied Social

* Gallois, C., Terry, D., Timmins, P., Kashima, Y. & McCamish, M. (1994). Safe sexual intentions and behavior among heterosexuals and homosexual men : Testing the theory of reasoned action. Psychology and Health, 10, 1-16.

Gerrard, M., Gibbons, F. X. & Bushman, B. J. (1996). The relation between perceived vulnerability to

Hedges, L. V. & Olkin, I. (1985). Statistical Methods for Meta-analysis. New York: Academic Press. Hernandez, J. T. & DiClemente, R. J. (1992). Self-control and ego identity development as predictors

of unprotected sex among late adolescent males. Journal of Adolescence, 15, 437-477. Holland, J., Ramazaglou, G. & Scott, S. (1990~). Managing risk and experiencing danger: Tensions

between government AIDS health education policy and young women’s sexuality. Gender and Education, 2, 125-146.

Holland, J., Ramazaglou, C., Scott, S., Sharp, S. & Thomson, R. (1990b). Sex, gender and power: Young women’s sexuality in the shadow of AIDS. Sociology of Health and Illness, 12, 336350.

and behavioral intentions. Journal of Applied Social Psychology, 14, 104-123.

behavior: Insights form the theory of reasoned action. Health Psychology, 14, 255-264.

psycho lo^, 22, 1521-1545.

HIV and precautionary behavior. Psychological Bulletin, 119, 390-409.


Hospers, H. J . & Kok, G. (1995). Determinants of safe and risk-taking sexual behavior among gay

Hunter, J. E. & Schmidt, F. L. (1990). Methods of Meta-analysis: Correcting Error and Bias in Research

Hunter, J. E., Schmidt, F. L. & Jackson, G. B. (1982). Meta-analysis: Cumulating Research Findings Across

Jemmott, L. S . & Jemmott, J. B. (1991). Applying the theory of reasoned action to AIDS risk

Kasen, S., Vaughan, R. D. & Walter, H. J. (1992). Self-efficacy for AIDS-preventive behaviors among

Kashima, Y., Gallois, C. & McCamish, M. (1993). The theory of reasoned action and cooperative

Liska, A. E. (1984). A critical examination of the causal structure of the Fishbein-Ajzen model. Social

Macey, L. P. & Boldero, J . M. (1992). The prediction of condom use by adult males and females.

* Morrison, D. M. (1993). Condom use with steady and casual partners: Results from a longitudinal study of high-risk adolescents and adults. NICHD Conference on Behavioral Research on the Role of Condoms in Reproductive Health, Rochille, M D .

* Morrison, D. M., Rogers Gillmore, M. & Baker, S. A. (1995). Determinants of condom use among high-risk adults: A test of the theory of reasoned action. Journal of Applied Social Psychology, 25, 651-676.

men: A review. A I D S Education and Prevention, 7 , 74-94.

Findings. Newbury Park, CA: Sage.

Studies. Beverly Hills, CA: Sage.

behaviour : Condom use among black women. Nursing Research, 40, 228-234.

tenth grade students. Health Education Quarterly, 19, 187-202.

behaviour: It takes two to use a condom. British Journal of Social Psychology, 32, 227-239.

Psychology Quarterb, 47, 61-74.

Annual Meeting of Australian Social Psychologists, Auckland, New Zealand, April.

Mullen, B. (1989). Advanced BASIC Meta-analysis. Hillsdale, NJ : Erlbaum. Nicolson, P. (1993). Deconstructing sexology : Understanding the pathologisation of female sexuality.

Randall, D. M. & Wolff, J. A. (1994). The time interval in the intention-behaviour relationship: Meta-

* Reinecke, J., Schmidt, P. & Ajzen, I. (1996). Application of the theory of planned behavior to

Reiss, I. L. & Leik, R. K. (1989). Evaluating strategies to avoid AIDS: Numbers of partners vs. use

Robertson, B. (1985). Sexual behaviour and risk of exposure to HIV among 18-25-year-olds in

Rose, S. & Frieze, I. H. (1989). Young singles’ script for a first date. Gender and Socieg, 3, 258-268. Rosenthal, R. R. (1984). Meta-analytic Procedures for Social Research. Beverly Hills, CA: Sage. * Rye, B. J . (1995). Unpublished data. Department of Psychology, University of Western Ontario,

London, Ontario, Canada. * Sanderson, C. A. & Jemmott, J. B. (1996). Moderation and mediation of HIV-prevention

interventions : Relationship status, intentions and condom use among college students. Journal of Applied Social Psychology, 26, 20762099.

Schaalma, H., Kok, G. & Peters, L. (1993). Determinants of consistent condom use by adolescents: The

Schwarzer, R. (1988). Meta: Programs for Seconday Data Anabsis. Berlin: Free University of Berlin. Sheeran, P. & Abraham, S. C. S. (1994). Measurement of condom use in 72 studies of HIV-preventive

behaviour : A critical review. Patient Education and Counseling, 24, 199-216. Stall, R. D., Coates, T. J. & Hoff, C. (1988). Behavioral risk reduction for HIV infection among gay

and bisexual men. American Psychologist, 43, 878-885. Sheppard, B. H., Hartwick, J. & Warshaw, P. R. (1988). The theory of reasoned action: A meta-

analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15, 325-343.

* Stanton, B. F., Li, X., Black, M. M., Ricardo, I., Galbraith, J., Feigelman, S. & Kaljee, L. (1996). Longitudinal stability and predictability of sexual perceptions, intentions, and behaviors among early adolescent African-Americans. Journal of Adolescent Health, 18, 1 G19.

Terry, D. J. & O’Leary, J. E. (1995). The theory of planned behaviour: The effects of perceived

]ournal of Reproductive and Infant Psychology, 11, 191-201.

analysis. British Journal of Social Psychology, 33, 405-418.

adolescents’ condom use: A panel study. Journal of Applied Social Psychology, 26, 749-772.

of condoms. Journal of Sex Research, 26, 181-196.

Scotland: Assessing change 1988-1993. AIDS, 9, 285-292.

impact of experience of sexual intercourse. Health Education Research, 8, 255-269.

behavioural control and self-efficacy. British Journal of Social Psychology, 34, 199-220.


Treffke, H., Juggemann, M. & Ross, M. W. (1992). The relationship between attitude, assertiveness and

Warshaw, P. R. & Davis, F. D. (1985). Disentangling behavioral intention and behavioral expectation.

* van der Velde, F. W., Hooykaas, C. & van der Pligt, J. (1996). Conditional versus unconditional risk

* White, K. M., Terry, D. J. & Hogg, M. A. (1994). Safer sex behavior: The role of attitudes, norms

Wight, D. (1992). Impediments to safer heterosexual sex: A review of research with young people.

condom use. Psychology and Health, 6, 45-52.

]onrnal of Experimental Social Psychology, 21, 21 3-228.

estimates in models of AIDS-related risk behaviour. Psychology and Health, 12, 87-100.

and control factors. Journal of Applied Social Psychology, 24, 2164-2192.

A I D S Care, 4, 11-21.

Received 9 October 1996; revised version received 21 Mg 1997

Do intentions predict condom use? Metaanalysis and examination of six moderator variables

Documents