ArticlePDF Available

Predicting the Intention to Use Condoms and Actual Condom Use Behaviour: A Three-Wave Longitudinal Study in Ghana

December 2016
Applied Psychology Health and Well-Being 9(1)

DOI:10.1111/aphw.12082

Authors:

Enoch Teye-Kwadjo

University of Ghana

Ashraf Kagee

Stellenbosch University

Hermann Swart

Stellenbosch University

Background: Growing cross-sectional research shows that the theory of planned behaviour (TPB) is robust in predicting intentions to use condoms and condom use behaviour. Yet, little is known about the TPB's utility in explaining intentions to use condoms and condom use behaviour over time. Methods: This study used a longitudinal design and latent variable structural equation modelling to test the longitudinal relationships postulated by the TPB. School-going youths in Ghana provided data on attitudes, subjective norms, perceived control, intentions, and behaviour regarding condom use at three time points, spaced approximately three months apart. Results: As predicted by the TPB, the results showed that attitudes were significantly positively associated with intentions to use condoms over time. Contrary to the TPB, subjective norms were not significantly associated with intentions to use condoms over time. Perceived control did not predict intentions to use condoms over time. Moreover, intentions to use condoms were not significantly associated with self-reported condom use over time. Conclusion: These results suggest that school-going youths in Ghana may benefit from sex education programmes that focus on within-subject attitude formation and activation. The theoretical and methodological implications of these results are discussed.

Hypothesised longitudinal model of the theory of planned behaviour's (TPB) standard constructs. The three cross-lagged paths coming out of Time 1 attitude, subjective norm, perceived behavioural control respectively depicted by the three boldfaced downwardly sloped lines to Time 2 intention and the single cross-lagged path emanating out of Time 2 intention depicted by the single boldfaced downwardly sloped line to Time 3 actual condom use behaviour represent the hypotheses of this study. For ease of reading, indicators, error terms, and disturbance terms are not shown.

…

Comparison between Autoregressive and Cross-lagged Longitudinal Models

…

Cross-lagged longitudinal model showing structural paths with main longitudinal hypotheses depicted by downwardly sloped paths between theory of planned behaviour components across time (bold). All structural paths are indicated. Unstandardised parameter estimates are reported. Model fit: v 2 (1191) = 1985.45, p < .001, v 2 /df = 1.67, comparative fit index = .946, root mean square error of approximation = .031; 90% CI [.029, .034], standardised root mean square residual = .051. * p < .05; *** p < .001; N = 684.

…

Figures - uploaded by Hermann Swart

Content may be subject to copyright.

Content uploaded by Hermann Swart

Content may be subject to copyright.

Predicting the Intention to Use Condoms and

Actual Condom Use Behaviour: A Three-Wave

Longitudinal Study in Ghana

Enoch Teye-Kwadjo*

University of Ghana, Ghana

Ashraf Kagee and Hermann Swart

Stellenbosch University, South Africa

Background: Growing cross-sectional research shows that the theory of planned

behaviour (TPB) is robust in predicting intentions to use condoms and condom use

behaviour. Yet, little is known about the TPB’s utility in explaining intentions to

use condoms and condom use behaviour over time. Methods: This study used a

longitudinal design and latent variable structural equation modelling to test the lon-

gitudinal relationships postulated by the TPB. School-going youths in Ghana pro-

vided data on attitudes, subjective norms, perceived control, intentions, and

behaviour regarding condom use at three time points, spaced approximately three

months apart. Results: As predicted by the TPB, the results showed that attitudes

were signiﬁcantly positively associated with intentions to use condoms over time.

Contrary to the TPB, subjective norms were not signiﬁcantly associated with inten-

tions to use condoms over time. Perceived control did not predict intentions to use

condoms over time. Moreover, intentions to use condoms were not signiﬁcantly

associated with self-reported condom use over time. Conclusion: These results

suggest that school-going youths in Ghana may beneﬁt from sex education pro-

grammes that focus on within-subject attitude formation and activation. The theo-

retical and methodological implications of these results are discussed.

Keywords: attitudes, condom use, subjective norms, theory of planned behaviour

INTRODUCTION

There is ample evidence that cross-sectional data support the utility of the theory

of planned behaviour (TPB) in understanding young people’s condom use. For

example, see Bryan, Kagee, and Broaddus (2006), Jemmott et al. (2007),

* Address for correspondence: Enoch Teye-Kwadjo, Department of Psychology, University of

Ghana, P.O. Box LG 84, Legon, Accra-Ghana. Email: eteye-kwadjo@ug.edu.gh

APPLIED PSYCHOLOGY: HEALTH AND WELL-BEING, 2016

doi:10.1111/aphw.12082

bs_bs_banner

Montanaro and Bryan (2014), and Schaalma et al. (2009). However, there is lit-

tle evidence regarding the TPB’s utility in understanding young people’s inten-

tions to use condoms and actual condom use behaviour over time. The TPB, in

its present form, postulates a longitudinal relationship between its standard con-

structs over time. Speciﬁcally, that attitudes, subjective norms, and perceived

behavioural control will inﬂuence an individual’s intentions towards a given

behaviour (e.g. condom use) over time, and that such intentions will predict the

likelihood of that individual actually engaging in that given behaviour at some

point in the future (Ajzen, 1991). The TPB, therefore, implies a temporal relation

between its standard constructs over time.

Cross-Sectional vs. Longitudinal Designs and the TPB

Whereas cross-sectional data are useful in investigating how outcome variable

measures differ among participants who possess varying levels of a predictor

variable, such cross-sectional data provide little information about the temporal

relations between the variables over time, in theoretical models that describe lon-

gitudinal relationships (Cole & Maxwell, 2003). Related to this, Maxwell, Cole,

and Mitchell (2011) noted that “a variable that is found to be a strong mediator

in a cross-sectional analysis may not be a mediator at all in a longitudinal analy-

sis”(p. 816). Further, other research has shown that the use of cross-sectional

designs to estimate longitudinal mediation effects can lead to bias in parameter

estimates (Maxwell & Cole, 2007; Shrout, 2011). As noted earlier, the TPB is a

theoretical model that describes longitudinal mediation relationships between its

standard constructs. Arguably, the TPB’s utility in understanding young people’s

condom use may be extended by studies that use longitudinal designs.

Although longitudinal designs cannot test true “causal”assumptions (which

are better tested with experimental designs), they offer many advantages over

cross-sectional designs. Compared with cross-sectional designs, longitudinal

designs allow researchers to better capture the inﬂuence of conceptual “third

variables”, including putative mediators, and to control for autoregressive rela-

tionships (Maxwell et al., 2011; Shrout, 2011; West, 2011). In the case of the

TPB, a putative mediator would be “behavioural intentions”. In addition, longi-

tudinal designs aid in testing construct stationarity, a prerequisite in theoretical

models that describe longitudinal relationships (Cole & Maxwell, 2003; Swart,

Hewstone, Christ, & Voci, 2011). Kenny (1979) explained that there is construct

stationarity when the observed sizes of correlations among measures of the same

construct obtained across time (i.e. within-wave correlations) remain unchanged.

Correspondingly, scholars in the ﬁeld of health behaviour and theory testing

have recently begun to express concerns about the widespread use of cross-

sectional designs in health behaviour research (see, for example, Weinstein,

2006; Weinstein & Rothman, 2005; West, 2011). For example, Weinstein

(2006) argued that cross-sectional designs used to test the TPB may have

2TEYE-KWADJO ET AL.

produced estimates of current and past behaviour, instead of producing estimates

of future behaviour. Moreover, various authors have recommended that psychol-

ogists interested in the processes through which antecedent factors inﬂuence out-

come measures (mediation) should measure each construct at a minimum of

three time points (Cole & Maxwell, 2003; Maxwell & Cole, 2007; Reichardt,

2011). For example, Cole and Maxwell (2003) explained that a minimum of

three waves of data is required to test the stationarity assumption inherent in lon-

gitudinal mediation designs. They argued that a minimum of three waves of data

would allow researchers (a) to conﬁrm that sufﬁcient autoregressive stability

exists for each measure over time, and (b) to control for prior levels of the

dependent variable on itself (autoregressive effects) when testing for cross-

lagged effects (see also Swart et al., 2011). In addition, Cole and Maxwell

(2003) and Selig and Preacher (2009) recommended that researchers measure all

variables at all measurement occasions in order to have the opportunity to partial

out any effects that a variable might have on itself over time (see also Reichardt,

2011). These considerations motivated the choice of the three measurement

occasions used in the present study, and also served to explain why we assessed

all variables at all measurement occasions.

As can be expected, a complete test of the full longitudinal mediation effects

described by the TPB would require ﬁrst and second waves of data (Time 1 and

Time 2) to examine the paths from attitudes, subjective norms, and perceived

behavioural control to intentions, and then second and third waves of data (Time

2 and Time 3) to examine the path from intentions to overt behaviour.

The Current Study

Recent national sentinel survey reports have shown that young Ghanaians aged

15–24 years accounted for 28 per cent of all new infections of HIV in 2013

(Ghana Aids Commission, 2014). HIV prevalence among young Ghanaians aged

15–19 years and those aged 15–24 years in 2012 was 0.7 per cent and 1.3 per

cent, respectively (Ghana Aids Commission, 2014). Sexually transmitted infec-

tions (STIs) and unintended pregnancy have also been recorded among young

people in Ghana (Morhe, Tagbor, Ankobea, & Danso, 2012; Ohene & Akoto,

2008). Previous research has shown that heterosexual intercourse is driving the

spread of the HIV virus in Ghana (Ghana Health Service, 2009). Yet, by all

accounts, condom use to reduce one’s risk of acquiring sexually transmitted HIV

as well as other STDs remains low among Ghanaian school-going youths

(Abdul-Rahman, Marrone, & Johansson, 2011; Adu-Mireku, 2003; Odonkor,

Nonvignon, Adu, Okyere, & Mahami, 2012; Sallar, 2008).

Beliefs and misconceptions about condom use and HIV that make condom

use unattractive to young people (e.g. HIV is caused by witchcraft, that condom

use results in pleasure loss, that condom use makes sex unnatural and indicates

lack of trust in one’s sexual partner) are also reported to be widespread among

INTENTIONS AND BEHAVIOUR IN CONDOM USE 3

young Ghanaians (Tenkorang, 2013; Tenkorang, Gyimah, Maticka-Tyndale, &

Adjei, 2011). Meta-analyses and systematic reviews have shown that consistent

condom use prevents sexually transmitted HIV by between 87 per cent and 96

per cent (Weller, 1993; Weller & Davis-Beaty, 2002).

Despite the risks and negative health outcomes associated with young peo-

ple’s sexual behaviour in Ghana, there is little research effort to examine this

health problem. There is the need to identify and understand psychosocial factors

that determine why some young Ghanaians use condoms but others do not. This

understanding is crucial because the Ghana Aids Commission (2013) has made

the halting and reversing of the spread of sexually transmitted HIV in Ghana a

key national, public health goal. Unfortunately, in Ghana interventions to reduce

young people’s sexual risk behaviour have achieved only limited success

because they have been largely based on non-governmental organisation initia-

tives, which have not been theory-guided. Theory-guided studies are central to

evidence-based research and interventions (Michie et al., 2005). Nevertheless, to

date, there are no known Ghanaian theory-based models advanced to adequately

explain HIV, STD, and pregnancy risk reduction among young people in the

country.

Relatedly, health behaviour researchers have noted that it is important in evi-

dence-based sexual behaviour research to utilise a theory that reﬂects the charac-

teristics of the population and dimensions of the particular behaviour under

investigation (Eaton, Flisher, & Aarø, 2003; Michie, Johnston, Francis,

Hardeman, & Eccles, 2008). Further, they argued that such a theory ought to

have a parsimonious framework and well-deﬁned components; and should not

only be robust but also should be applicable to both men and women. In addi-

tion, they stressed that such a theoretical framework should suggest research

questions or indicate hypotheses to help focus research on key variables of inter-

est to the researcher. One health behaviour theory that appears to satisfy these

requirements is Ajzen’s (1991) theory of planned behaviour (TPB). These con-

siderations informed the choice of the TPB to guide the present research.

The TPB aims to explain human social behaviour in speciﬁc situations by

making two important assumptions. First, it assumes that an individual’s inten-

tions to engage in a speciﬁed behaviour (e.g. using condoms consistently) is

determined by his or her attitudes, his or her perceived control beliefs, and by

the prevailing normative beliefs within his or her social context regarding that

behaviour. Second, it assumes that the most important determinant of overt beha-

viour (e.g. condom-protected sexual behaviour) is an individual’s behavioural

intentions towards that target behaviour. Taken together, Ajzen (1991) argued

that individuals who have favourable attitudes towards a speciﬁed behaviour,

positive subjective norms regarding that behaviour, and who perceive themselves

to have the necessary control over performing that speciﬁed behaviour, are more

likely to execute their behavioural intentions when the need arises. The aim of

this study was to undertake a three-wave longitudinal study to test the

4TEYE-KWADJO ET AL.

longitudinal relationships postulated by the TPB in understanding heterosexual

high school youths’intentions to use condoms and actual condom use behaviour

over time.

Study Hypotheses

Using the theory of planned behaviour’s framework, we speciﬁed the following

longitudinal structural models (see Figure 1, for the hypothesised model):

Hypothesis 1: (a) Attitude towards condom use, (b) subjective norms regarding

condom use, and (c) perceived behavioural control over condom use at Time 1

would each be signiﬁcantly positively associated with increased intentions to use

condoms at Time 2, even after controlling for the autoregressive effects of the

intentions to use condoms at Time 1.

Hypothesis 2: Intention to use condoms at Time 2 would be signiﬁcantly positively

associated with increased condom use behaviour at Time 3, even after controlling

for the autoregressive effects of condom use behaviour at Time 1 and Time 2.

Hypothesis 3: Intention to use condoms at Time 2 would mediate the relationship

between attitudes, subjective norms, and perceived behavioural control at Time 1,

and condom use behaviour at Time 3.

METHOD

Participants and Procedure

Participants (N=1,023) in this study were recruited from a large municipal,

public senior high school in the Eastern Region of Ghana. The corresponding

author obtained a letter of permission from the Director-General, Ghana Educa-

tion Service and presented it, including a synopsis of the study as well as ethical

approval letters, to the school authorities. Students were recruited at an assembly

forum in the participating school with the help of the school authorities. To be

eligible for participation in the present study, students must never have been

married and must have attained the age of 14 years prior to participation in the

survey at the ﬁrst assessment. Also, they must have indicated a willingness to

participate in the study. Of the number recruited, 13 students were below the age

of 14 years and were thus excluded from the study. At the Time 1 assessment,

983 students completed the survey. Twenty-seven potential participants were

absent for various reasons, including illness, other important co-curricular

engagements, and not being in school because of being a day student. Of the

INTENTIONS AND BEHAVIOUR IN CONDOM USE 5

983 original participants, 956 and 835 provided the data at Time 2 and Time 3

assessments, respectively.

Surveys were completed in classrooms on the school compound. Participants

completed assent or consent forms depending on their self-declared age. Because

Attitude towards

condom use

Subjective norm

regarding condom

use

Perceived

behavioural control

regarding condom

use

Intention regarding

future condom use

Actual condom use

behaviour

Attitude towa rds

condom use

Subjective norm

regarding condom

use

Perceived

behavioural contro l

regarding condom

use

Intention regarding

future condom use

Actual condom use

behaviour

Attitude towa rds

condom use

Subjective norm

regarding condom

use

Perceived

behavioural contr ol

regarding condom

use

Intention regarding

future condom use

Actual condom use

behaviour

Time 1 Time 2 Time 3

FIGURE 1. Hypothesised longitudinal model of the theory of planned beha-

viour’s (TPB) standard constructs. The three cross-lagged paths coming out of

Time 1 attitude,subjective norm,perceived behavioural control respectively

depicted by the three boldfaced downwardly sloped lines to Time 2 intention

and the single cross-lagged path emanating out of Time 2 intention depicted by

the single boldfaced downwardly sloped line to Time 3 actual condom use

behaviour represent the hypotheses of this study. For ease of reading,

indicators, error terms, and disturbance terms are not shown.

6TEYE-KWADJO ET AL.

the majority of the students lived on the school compound, far away from their

parents or guardians, and were fed by school authorities, and also because they

were not allowed to use cell phones on the school compound, school authorities

liaised with the school’s parent-teacher association (PTA) to waive parental con-

sent to enable interested students aged 14 to 17 years, who were not legally eli-

gible to sign consent themselves, to participate in the study. Permission,

informed consent and assent were obtained only once at the ﬁrst assessment and

had a prospective effect on subsequent assessments.

All participants were provided with instructions to help them generate their

own alpha-numeric string identiﬁers on top of each survey they completed.

These alpha-numeric string identiﬁers were aimed at enabling the research team

to match the responses of the participants across all three waves of data collec-

tion. However, because some participants failed to generate their own alpha-

numeric string identiﬁers, and because other participants provided incomplete

alpha-numeric string identiﬁers or alpha-numeric string identiﬁers that differed

from wave to wave, only a total of 684 surveys with correct alpha-numeric string

identiﬁers were matched across the three waves of data collection. And because

we used only cases that we were able to match across time, the sample size was

reduced from 983 to 684, representing an overall retention rate of 69.6 per cent

for the main analyses. Using Jackson’s (2003) ratio of observations (N) to model

parameters to be estimated (q) procedure of (N:q=20:1; i.e. 20 observations per

one estimated parameter), our retained sample size (N=684) was deemed ade-

quate for performing the conﬁrmatory and structural analyses with the maximum

likelihood estimation. In addition, a post-hoc power analysis in G*Power (Faul,

Erdfelder, Buchner, & Lang, 2009) indicated that with an estimated population

effect size of (0.30; medium) and signiﬁcance level of (.05), the retained sample

size of 684 was enough to provide us with sufﬁcient statistical power (.99) to

detect correlational effects.

The current analyses relate to the 684 matched sample (n=335 males;

n=349 females). Participants were between the age range of 14 and 20 years.

About 418 of the matched sample were boarding students (i.e. students who lived

on the school compound and were fed by the school) and 266 were day students

(i.e. students who lived at home while attending school). This study was approved

by the Institutional Review Board of the Noguchi Memorial Institute for Medical

Research (#034/12-13), University of Ghana and by the Health Ethics Research

Committee of Stellenbosch University (#S12/06/179), South Africa.

Measures

Attitudes towards Condom Use. Five items adapted from previous research

(Basen-Engquist et al., 1999; Carmack & Lewis-Moss, 2009) assessed partici-

pants’attitudinal beliefs about condom use. Sample items included “I believe

condoms should always be used if a person my age has sex”, and “I believe

INTENTIONS AND BEHAVIOUR IN CONDOM USE 7

condoms should always be used if a person my age has sex, even if the two peo-

ple know each other very well.”Response scale for the items ranged from 1

(strongly disagree)to7(strongly agree). Scale responses were scored such that

higher scores indicated more favourable attitudinal beliefs about condom use.

Composite mean score, standard deviation, and coefﬁcient alpha for the attitudes

towards condom use were, Time 1 (M=5.64, SD =0.06, a=.64); Time 2

(M=5.72, SD =0.13, a=.66); and Time 3 (M=5.97, SD =0.15, a=.69).

Subjective Norms Regarding Condom Use. Five items adapted from previ-

ous research (Basen-Engquist et al., 1999; Carmack & Lewis-Moss, 2009)

assessed participants’perception of descriptive normative inﬂuences on them

regarding condom use. Sample items included “Most of my friends believe con-

doms should always be used if a person my age has sex, even if the two people

trust each other very well”, and “Most of my friends will say ‘no’to sex if a boy-

friend or girlfriend won’t use a condom.”Response scale for the items ranged

from 1 (strongly disagree)to7(strongly agree). Scale responses were scored

such that higher scores indicated greater perceived normative inﬂuence and moti-

vation to comply with peer norms regarding condom use. Composite mean

score, standard deviation, and coefﬁcient alpha for the subjective norms regard-

ing condom use were, Time 1 (M=5.55, SD =0.04, a=.62); Time 2

(M=5.66, SD =0.15, a=.66); and Time 3 (M=5.82, SD =0.03, a=.71).

Perceived Behavioural Control over Condom Use. Four items adapted

from previous research (Carmack & Lewis-Moss, 2009; Jemmott et al., 2007)

assessed participants’perceived behavioural control over condom use. Sample

items included “I can use a condom correctly”, and “I can get my boyfriend or

girlfriend to use a condom, even if he or she doesn’t want to do so.”Response

scale for the items ranged from 1 (strongly disagree)to7(strongly agree). Scale

responses were scored such that higher scores reﬂected a greater perceived efﬁ-

cacy or controllability over condom use. Composite mean score, standard devia-

tion, and coefﬁcient alpha for the perceived behavioural control over condom

use were, Time 1 (M=5.25, SD =0.90, a=.67); Time 2 (M=5.43,

SD =0.23, a=.66); and Time 3 (M=5.66, SD =0.14, a=.75).

Behavioural Intentions towards Future Condom Use. Six items adapted

from previous research (DeHart & Birkimer, 1997) assessed participants’inten-

tions to use condoms in the future (i.e. over the coming three months). Sample

items included “I am determined to use condoms in the next 3 months, if I have

sex”, and “I planned to use condoms, if I were going to have sex in the next

3 months.”Scale responses were scored such that higher scores indicated more

favourable intentions towards condom use over the next three months. Response

scale for the items ranged from 1 (strongly disagree)to7(strongly agree).

Composite mean score, standard deviation, and coefﬁcient alpha for the

8TEYE-KWADJO ET AL.

intentions to use condoms were, Time 1 (M=5.59, SD =0.19, a=.78); Time

2(M=5.74, SD =0.18, a=.78); and Time 3 (M=5.85, SD =0.13, a=.84).

Self-Reported Condom Use Behaviour. Seven items adapted from previous

research (Holland & French, 2012; Walsh, Senn, Scott-Sheldon, Vanable, &

Carey, 2011) assessed participants’self-reported condom use over the past three

months. Multiple items that sought to tap relevant aspects of condom use such as

condom use frequency (e.g. “How often have you had sex with your regular boy-

friend/girlfriend with a condom in the past 3 months?”), condom non-use fre-

quency (e.g. “How often did you refuse to have sex with a partner because they

would not use a condom in the past 3 months?”), and type of partners (e.g.

“How often have you had sex with someone who is not your boyfriend/girlfriend

(casual sex) with a condom in the past 3 months?”) were used in this study. The

response scale for this construct ranged from 1 (never)to7(all the time). Scale

responses were coded such that higher scores indicated greater self-reported con-

dom use behaviour over the past three months. Composite mean score, standard

deviation, and coefﬁcient alpha for the self-reported condom use behaviour were,

Time 1 (M=2.91, SD =0.14, a=.88); Time 2 (M=3.06; SD =0.19,

a=.90); and Time 3 (M=3.33, SD =0.15, a=.89).

RESULTS

Preliminary Analyses

To test for attrition effects, we compared participants who completed all three

waves of data (completers) with those who dropped out after Time 1 and those

who dropped out after Time 2 (attriters). Taking a pair of waves, we deﬁned

attriters as those participants who provided data at Time 1 but not at Time 2, and

then those who provided data at Time 1 but not at Time 3, by creating a missing

data indicator. We ran two multivariate analyses of variance on the continuous

variables, using the missing data indicator as a grouping variable. We then ran

two binary logistic regression analyses on the categorical variables, using the

missing data indicator as an outcome variable.

Continuous Variables. Attrition analyses revealed signiﬁcant differences

between the longitudinal sample (completers) and those who dropped out after

Time 1 (attriters), Wilks’k=.961, F(5, 975) =7.97, p<.001, partial g

=.04.

However, there were no signiﬁcant differences between completers and Time 2

attriters, Wilks’k=.996, F(5, 976) =815, p=.539, partial g

=.004. An

inspection of the univariate differences between completers and Time 1 attriters,

using a Bonferroni adjusted alpha level of (p<.01), showed that only subjective

norm, F(1, 979) =16.80, p<.001, partial g

=.02; perceived behavioural

INTENTIONS AND BEHAVIOUR IN CONDOM USE 9

control, F(1, 979) =27.57, p<.001, partial g

=.03; intention, F(1,

979) =23.29, p<.001, partial g

=.02; and condom use behaviour, F(1,

979) =9.55, p<.01, partial g

=.01, reached statistical signiﬁcance. As per

the Bonferroni adjusted alpha level, there were no signiﬁcant univariate differ-

ences between completers and Time 1 attriters in attitudes towards condom use,

F(1, 979) =4.81, p=.028, partial g

=.01.

Moreover, inspection of the mean scores indicated that completers scored

slightly higher on subjective norms (M=46.38, SD =.62) than did Time 1

attriters (M=43.63, SD =.62). Completers perceived slightly greater control

over condom use (M=51.88, SD =.34) than did Time 1 attriters (M=47.26,

SD =.81). In addition, completers possessed slightly greater intentions towards

condom use (M=46.38, SD =.30) than did Time 1 attriters (M=42.62,

SD =.72), and completers reported slightly more condom-protected sexual

behaviour (M=30.26, SD =.50) than did Time 1 attriters (M=26.29,

SD =1.18). We note that despite the signiﬁcant univariate differences, Cohen’s

(1988) effect size calculation criteria showed that these differences were small,

suggesting that completers did not differ signiﬁcantly from Time 1 attriters. As

can be expected, our attrition analyses seemed to indicate that data were missing

at random.

Categorical Variables. Logistic regression analyses, using the missing data

indicator as outcome variable and gender, age, and student status (day or board-

ing) as predictor variables, distinguished between completers and Time 1 attriters

(3) =14.89, p=.002. Only age made a signiﬁcant contribution to the model

(p=.001). Participant gender (p=.377) and student status (day or boarding;

p=.099) did not contribute signiﬁcantly to the model. Compared with Time 1

attriters, completers were 2.14 times more likely to be older (age; odds ratio

(OR) =2.14, 95% CI [1.39, 3.28], p=.001). However, there were no signiﬁ-

cant differences between completers and Time 2 attriters, v

(3) =5.27,

p=.153.

Further, the matched sample (N=684) was compared with the full Time 1

sample (N=983) along the continuous and categorical variables of interest in

this research. Multivariate analysis of variance showed that the matched sample

did not differ signiﬁcantly from the full Time 1 sample, Wilks’k=.990, F(5,

977) =1.93, p=.087, partial g

=.01. In addition, logistic regression revealed

a non-signiﬁcant difference between the matched sample and the full Time 1

sample along our categorical variables of gender, age, and student status (day or

boarding), v

(3) =4.61, p=.203.

Following the attrition analyses, we ﬁrst screened the study variables for nor-

mality at each time point, using West, Finch, and Curran’s (1995) cut-off criteria

of skewness (2.00 and +2.00) and of kurtosis (7.00 and +7.00) in IBM SPSS

Statistics (v20). All variables under consideration were normally distributed.

Second, we checked construct dimensionality at each time point by conducting

10 TEYE-KWADJO ET AL.

exploratory factor analyses, using maximum likelihood estimation. Each con-

struct demonstrated adequate unidimentionality at each time point. We then used

item-parcelling (Little, Rhemtulla, Gibson, & Schoemann, 2013) to reduce the

number of items on scales with more than four items. Bivariate correlations were

examined across time. Generally, items demonstrated signiﬁcant intra-construct

correlations. The correlation matrices are available from the corresponding

author on written request.

Main Analyses

Latent Variable Structural Equation Model. All structural equation analy-

ses were conducted in Mplus (v6.0) with maximum likelihood robust (MLR)

estimation. As noted earlier, the monotone pattern of drop-out in this study cou-

pled with the results from our attrition analyses suggested that data were missing

at random. Research has shown that when data are missing at random, the full

information maximum likelihood (FIML) estimation provides unbiased parame-

ter estimates (Enders, 2001). Missing data were handled using the FIML estima-

tion in Mplus. Parameter estimates were calculated using robust maximum

likelihood (MLR) estimation.

We conducted all model comparisons with the Satorra-Bentler scaled chi-

square difference test (Bryant & Satorra, 2012). In accordance with Anderson

and Gerbing’s (1988) two-step approach, we ﬁrst ﬁtted the longitudinal measure-

ment model and then ﬁtted the longitudinal structural model. We used multiple

goodness-of-ﬁt indices to determine model ﬁt: chi-square test statistic (v

) with

degrees of freedom, chi-square/degrees of freedom (v

/df <3.0) ratio, compara-

tive ﬁt index (CFI ≥.95), root mean square error of approximation (RMSEA

<.07) with conﬁdence interval, and the standardised root mean square residual

(SRMR <.05). We did not carry out any suggested model modiﬁcations in these

structural analyses.

Longitudinal Measurement Model. We ﬁtted the longitudinal measurement

model including all observed and latent variables across time, using longitudinal

conﬁrmatory factor analysis (CFA; Little, Preacher, Selig, & Card, 2007). The

longitudinal measurement model demonstrated adequate measurement across

time, v

(1102, N=684) =1850.95, p<.001, v

/df =1.68, CFI =.949,

RMSEA =.032; 90% CI [.029, .034]; SRMR =.040, suggesting that the hypothe-

sised structural relationships among the latent constructs of the theory of planned

behaviour could be tested. All factor loadings were signiﬁcant at (p<.001).

Factorial Invariance Model. Invariance constraints (Widaman, Ferrer, &

Conger, 2010) were imposed by constraining the factor loadings for the respec-

tive indicators of a given construct to equality over time. The factorial invariance

model showed a good ﬁt to the data, v

(1126) =1850.95, p<.001, v

INTENTIONS AND BEHAVIOUR IN CONDOM USE 11

df =1.67, CFI =.949, RMSEA =.031; 90% CI [.029, .034], SRMR =.042. A

comparison between the factorial invariance model and the longitudinal CFA

model, using the Satorra-Bentler scaled chi-square difference test, showed that

the factorial invariance model did not differ signiﬁcantly from the longitudinal

CFA model, Dv

(24) =21.75, p>.05.

Longitudinal Structural Model of the Theory of Planned Behaviour. Using

the better-ﬁtting longitudinal measurement model (i.e. the model with partial

measurement invariance) as the starting point, a series of longitudinal structural

models was ﬁtted to the data. The ﬁtted longitudinal models addressed questions

of direct effects—autoregressive effects and cross-lagged effects. The cross-

lagged effects constitute the longitudinal hypotheses of the current analyses.

Autoregressive longitudinal structural model of the TPB: To assess autore-

gressive stationarity among the ﬁve TPB constructs under consideration over

time, a series of increasingly restrictive autoregressive models was ﬁtted to the

data (see Table 1). These models set various equality constraints on the within-

construct relationships over time. The best-ﬁtting longitudinal autoregressive

structural model exhibited partial stationarity and showed acceptable model ﬁt,

(1195) =1996.59, p<.001, v

/df =1.67, CFI =.945, RMSEA =.031;

90% CI [.029, .034], SRMR =.053 (see Figure 2).

In the best-ﬁtting autoregressive structural model, the within-construct rela-

tionships between attitudes and subjective norms were constrained to equality

with one another over time, whereas the within-construct autoregressive relation-

ship of behaviour was constrained to equality over time. The within-construct

autoregressive relationships of perceived behavioural control and intentions were

freely estimated over time. This is because it was not possible to achieve accept-

able model ﬁt by imposing within-construct equality constraints on perceived

behavioural control and intentions over time.

Cross-lagged longitudinal structural model of the TPB: The cross-lagged

structural paths that constituted the longitudinal hypotheses of the current

analyses were added to the best autoregressive longitudinal model illustrated in

Figure 2. These cross-lagged structural paths test whether, beyond the within-

construct autoregressive effect that each TPB construct has on itself over time,

there also exist between-construct (cross-lagged) relationships among the latent

constructs of the TPB over time. To test the cross-lagged longitudinal structural

effects (i.e. longitudinal hypotheses), a freely estimated cross-lagged longitudinal

model was ﬁrst ﬁtted to the data. This model demonstrated adequate ﬁt, v

(1187) =1977.00, p<.001, v

/df =1.67, CFI =.946, RMSEA =.031; 90%

CI [.029, .034], SRMR =.050.

Next, to establish parsimony all cross-lagged structural paths in the longitudi-

nal model were constrained to between-construct equality across time. This con-

strained cross-lagged longitudinal structural model described the data well, v

(1191) =1985.45, p<.001, v

/df =1.67, CFI =.946, RMSEA =.031; 90%

12 TEYE-KWADJO ET AL.

TABLE 1

Comparison between Autoregressive and Cross-lagged Longitudinal Models

Model Model ﬁt

Model

comparisons

Corrected chi-square

difference test

1a v

(1191) =1989.53***; CFI =.95; RMSEA =.031; SRMR =.055

1b v

(1196) =2018.34***; CFI =.94; RMSEA =.032; SRMR =.053 1b vs. 1a Dv

(5) =18.20, p<.05

1c v

(1195) =2007.52***; CFI =.95; RMSEA =.032; SRMR =.053 1c vs. 1a Dv

(4) =10.57, p<.05

1d v

(1194) =1997.44***; CFI =.95; RMSEA =.031; SRMR =.053 1d vs. 1a Dv

(3) =7.16, p>.05

1e v

(1195) =1996.59***; CFI =.95; RMSEA =.031; SRMR =.053 1e vs.1d Dv

(1) =.37, p>.05

2a v

(1187) =1977.00***; CFI =.95; RMSEA =.031; SRMR =.050

2b v

(1191) =1985.45***; CFI =.95; RMSEA =.031; SRMR =.051 2b vs. 2a

2b vs. 1e

(4) =6.04, p>.05

(4) =10.04, p<.05

Note:CFI =comparative ﬁt index; RMSEA =root mean square error of approximation; SRMR =standardised root mean square residual. 1a =autoregressive model (freely esti-

mated theory of planned behaviour (TPB) model relationships); 1b =autoregressive model (within-construct equality constraints for all ﬁve TPB components); 1c =autoregressive

model (within-construct equality constraints for attitude, subjective norm, intention, and behaviour with perceived behavioural control freely estimated); 1d =autoregressive model

(within-construct equality constraints for attitude, subjective norm, and behaviour with intention and perceived control freely estimated); 1e =autoregressive model (within-construct

equality constraints for behaviour, with between-constraints for attitude and subjective norm, and intention and perceived control freely estimated). 2a =cross-lagged structural model

(all ﬁve TPB components freely estimated); 2b =cross-lagged structural model (within-construct equality constraints for all ﬁve TPB components).

How to determine which comparison model describes the data well: (a) when you compare two versions of a model with one being less restrictive (has no equality constraints,

freely estimated) and the other being more restrictive (has equality constraints) from the same category (e.g. autoregressive models), the more restrictive model should yield a non-

signiﬁcant p-value (p>.05) for the model ﬁt to be considered acceptable, and (b) when you compare two models from two different categories (e.g. autoregressive model versus

cross-lagged structural model) the model that improves model ﬁt(p<.05) is the one to be accepted.

***p<.001; all relative v

statistic <3.1; N=684.

INTENTIONS AND BEHAVIOUR IN CONDOM USE 13

Time 1 Time 2 Time 3

R2=.23 R2=.42

R2=.23 R2=.36

R2=.13 R2=.37

R2=.14 R2=.30

R2=.40 R2=.65

.52*** .52***

.33*** .58***

.35*** .53***

.69*** .69***

Attitude

Norm

Control

Intention

Attitude Attitude

Behaviour

Norm

Control

Norm

Control

Behaviour

Intention

Behaviour

x1x2

x4x5x6

x13

x12

x11

x16

x15

x14

x7x8x9x1

y17

y16

y15

y14 y34

y33

y31

y13

y12

y11 y30

y29

y28

y7y8y9y10 y24 y25 y2y27

y21 y22 y23

y4y5y6

y18 y19 y20

y1y2y3

FIGURE 2. Autoregressive longitudinal structural model showing unstandard-

ised parameter estimates and explained variances. For ease of reading, within-

wave correlations among the error terms, disturbance terms, and within-wave

between-construct correlations are not shown. Model ﬁt: v

(1195, N=684) =

1996.59, p<.001, relative v

=1.67, comparative ﬁt index =0.945, root mean

square error of approximation [90% CI] =0.031 [0.029, 0.034], standardised root

mean square residual =0.053. *p<.05; *** p<.001.

14 TEYE-KWADJO ET AL.

CI [.029, .034], SRMR =0.051. Compared with the freely estimated cross-

lagged longitudinal structural model, the Satorra-Bentler scaled v

-difference test

revealed that the constrained cross-lagged longitudinal model did not differ sig-

niﬁcantly from the freely estimated model, Dv

(4) =6.04, p>.05 (see Table 1

[2b vs. 2a]). Further, a comparison between the best cross-lagged longitudinal

structural model and the best autoregressive longitudinal structural model

showed that the best cross-lagged longitudinal structural model ﬁtted the data

signiﬁcantly better than did the best autoregressive longitudinal structural model,

(4) =10.04, p<.05 (see Table 1 [2b vs. 1e]).

The results of the best cross-lagged longitudinal structural model suggest that

the hypothesised longitudinal structural relationships between the theory of

planned behaviour’s constituent components achieved limited support (see Fig-

ure 3). Speciﬁcally, the path from attitudes towards condom use at Time 1 to

intentions to use condoms at Time 2 was statistically signiﬁcant (b=.11,

p<.05), even after controlling for prior effects of intentions to use condoms at

Time 1. On the contrary, the speciﬁc path from subjective norms regarding con-

dom use at Time 1 to intentions to use condoms at Time 2 was not statistically

signiﬁcant (b=.06, p=.261).

Similarly, the speciﬁc path from perceived behavioural control over condom

use at Time 1 to intentions to use condoms at Time 2 did not reach statistical

signiﬁcance (b=.03, p=.537), at least after controlling for prior levels of

intentions to use condoms at Time 1. The path from intentions to use condoms

at Time 2 to self-reported condom use behaviour at Time 3 was nonsigniﬁcant

(b=.04, p=.318), at least after controlling for prior levels of self-reported

condom use behaviour at Time 2. On the basis of these results, the longitudinal

mediation hypothesis (i.e. Hypothesis 3) of the current study was not supported.

Despite the limited support obtained for the longitudinal structural relationships

postulated by the theory of planned behaviour, the cross-lagged longitudinal

structural model accounted for a substantial portion of the variance in attitudes

towards condom use (Time 2: R

=24%, Time 3: R

=44%), subjective norms

regarding condom use (Time 2: R

=24%, Time 3: R

=38%), perceived beha-

vioural control over condom use (Time 2: R

=13%, Time 3: R

=38%),

intentions to use condoms (Time 2: R

=15%, Time 3: R

=31%), and self-

reported condom use behaviour (Time 2: R

=40%, Time 3: R

=65%) over

time.

DISCUSSION

Pathway between Attitude and Intention

This study simultaneously tested the theory of planned behaviour’s (TPB) utility

in explaining adolescent condom use over time. Overall, the results demonstrated

INTENTIONS AND BEHAVIOUR IN CONDOM USE 15

only limited support for the postulated longitudinal relationships of the TPB’s

constituent components. Consistent with the TPB, attitudes towards condom use

at Time 1 were longitudinally associated with intentions to use condoms three

.47***

Time 1 Time 2 Time 3

.53*** .53***

.34*** .60***

.29***

.68***

.11* .11*

Attitude

Norm

Control

Intention

Attitude Attitude

Behaviour

Norm

Control

Norm

Control

Behaviour

Intention

Behaviour

x1x2

x4x5x6

x13

x12

x11

x16

x15

x14

x7x8x9x1

y17

y16

y15

y14 y34

y33

y32

y31

y13

y12

y11 y30

y29

y28

y7y8y9y10 y24 y25 y26 y27

y21 y22 y23

y4y5y6

y18 y19 y20

y1y2y3

-.03

.06

-.03

.06

.04 .04

FIGURE 3. Cross-lagged longitudinal model showing structural paths with

main longitudinal hypotheses depicted by downwardly sloped paths between

theory of planned behaviour components across time (bold). All structural paths

are indicated. Unstandardised parameter estimates are reported. Model ﬁt: v

(1191) =1985.45, p<.001, v

/df =1.67, comparative ﬁt index =.946, root mean

square error of approximation =.031; 90% CI [.029, .034], standardised root

mean square residual =.051. *p<.05; *** p<.001; N=684.

16 TEYE-KWADJO ET AL.

months later at Time 2, and attitudes towards condom use at Time 2 were

longitudinally associated with intentions to use condoms another three months

later at Time 3. This result is consistent with a recent two-wave panel study in

sub-Saharan Africa that found attitude to be the strongest predictor of intentions

to use condoms among young people (Molla,



Astrøm, & Brehane, 2007).

The current ﬁnding is consistent with a systematic review among African adoles-

cents that found attitudes to be a more reliable predictor of condom use inten-

tions, and behavioural intentions to be a poor predictor of actual behaviour

(Paul-Ebhohimhen, Poobalan, & van Teijlingen, 2008).

In addition, a previous meta-analysis revealed that attitude is an important pre-

dictor of intentions to use condoms (Albarracin, Johnson, Fishbein, & Mueller-

leile, 2001). Similarly, our results are consistent with ﬁndings from other

previous longitudinal studies using structural equation modelling with high

school adolescents (Beadnell et al., 2007). Moreover, the current ﬁnding regard-

ing attitude–intention correspondence compared favourably with those reported

in a longitudinal investigation by Morrison, Baker, and Gillmore (1998) and by

Reinecke, Schmidt, and Ajzen (1996), using the theory of planned behaviour’s

(TPB) framework. To our knowledge, as far as the TPB sexual behaviour litera-

ture is concerned, this is the ﬁrst research to examine the longitudinal association

between the TPB’s components with three waves of data, using latent variable

structural equation modelling. Consequently, the ﬁnding relative to attitude

strength would contribute signiﬁcantly to our current knowledge of attitude–

intention correspondence.

Pathway between Subjective Norms and Intention

In the present analyses, the longitudinal pathways between subjective norms and

behavioural intentions were not supported. And neither were the longitudinal

pathways between perceived behavioural control and behavioural intentions. The

current ﬁndings regarding subjective norms and perceived behavioural control

contrast with those of Beadnell et al. (2007). The differences in these results

may reﬂect a methodological difference between their research and the present

study. This is because Beadnell et al.’s (2007) research was a two-wave panel

study with data collection spaced one year apart, predicting the intentions to

have sex, compared to the present study that used three waves of data collection

spaced three months apart, predicting the intentions to use condoms. Also,

Beadnell et al. (2007) assessed intentions to have sex only at Time 1 but not at

Time 2, as compared to the present study that assessed all TPB model constructs

at all three time points.

In this study, items measuring the subjective norm construct focused on

descriptive norms, speciﬁcally peer norms (what friends say and do) and not on

injunctive norms (what signiﬁcant others expect the adolescent to do). It is possi-

ble that descriptive peer norms are not salient for condom use intentions and

INTENTIONS AND BEHAVIOUR IN CONDOM USE 17

behaviour for the current sample. Correspondingly, there is evidence that the cul-

tural and societal restrictions on adolescent sexual behaviour in Ghana are mak-

ing sexual relations among young people strictly private and conﬁdential affairs

(Darteh, Doku, & Esia-Donkoh, 2014).

Pathway between Perceived Behavioural Control and

Intention

The current study did not conﬁrm the longitudinal relationship between per-

ceived behavioural control over condom use and intentions to use condoms. This

result is comparable to the ﬁndings of a previous panel study (Reinecke et al.,

1996). A possible explanation for the nonsigniﬁcant longitudinal associations

between perceived behavioural control and intentions may be traced to the dya-

dic nature of sexual behaviour. That is, since sexual intercourse involves two

people, it is possible for one partner to perceive greater control to use condoms,

but the non-availability of a sex partner or even failure to convince a sex partner

regarding the advantages of condom use may prevent this perception of control

from being enacted. Therefore, one’s sex partner’s cooperation is a central aspect

of condom negotiation and may serve to reinforce the notion held by sexual

behaviour researchers that condom use requires more than an individual sex part-

ner’s volitional control.

Pathway between Intention and Behaviour

Moreover, the hypothesised longitudinal relationship between intentions to use

condoms and condom use behaviour was not supported in the present study.

Although this ﬁnding is contrary to the postulate of the TPB, it is consistent with

other structural equation modelling investigations of condom use that reported

nonsigniﬁcant longitudinal associations between intentions and behaviour

(Carvajal, Estrada, & Estrada, 2005). The Ghanaian high school system pro-

scribes adolescent sexual behaviour. For example, high school students are not

allowed to use cell phones while in school, and pregnant students are often

excluded from school (Ghana News Agency, 2014). It is most likely that

students with intentions to engage in protected sexual behaviour may not have

adequate time to undertake preparatory behaviours (such as purchasing condoms

for an intended sexual activity), for fear this preparatory behaviour may expose

them to school authorities. This situation may serve to attenuate the effect of

behavioural intentions to use condoms on actual condom use behaviour. Relat-

edly, in an attempt to explain why some people act on their intentions and others

don’t, Fishbein, Hennessy, Yzer, and Douglas (2003) noted that “although inten-

tion is viewed as the primary determinant of behaviour, the model recognizes

that a lack of skills (or abilities) and/or environmental constraints may prevent

one from acting on his or her intentions. Thus, intentions alone are not the sole

18 TEYE-KWADJO ET AL.

determinant of behaviour, and different factors may attenuate or enhance the

intention/behaviour relationship”(p. 3).

Implications of Results

The current results suggest that school-aged adolescents in eastern Ghana possess

attitudinally controlled intentions. Consequently, it seems clear that sexual risk

reduction programmes in Ghana that target the antecedents of attitude formation

and activation may help strengthen positive attitudes towards condom use. Hav-

ing positive attitudes towards condom use may be an important ﬁrst step for sex-

ual behaviour modiﬁcation. This study raised theoretical questions relating to the

construct validity of some of the TPB components. In the present study, attitudes

explained variance in intentions to use condoms over time, as postulated by the

TPB. Conversely, subjective norms and perceived behavioural control each failed

to predict intentions to use condoms over time, contrary to the postulate of the

TPB. Also, there was no longitudinal association between intentions to use con-

doms and actual condom use behaviour. These nonsigniﬁcant results are incon-

sistent with the postulated model relationships of the TPB. In accordance with

the current results, other ﬁndings from an experimental test of the TPB (using a

different health behaviour) questioned the postulated relations between intentions

and behaviour (Sniehotta, 2009; see also Sniehotta, Presseau, & Ara

ujo-Soares,

2014). Thus, our results seem to call into question the sufﬁciency assumption of

the TPB, and may provide opportunities for theory reﬁnement.

Empirical questions raised by our results reﬂect the methodological quality of

previous tests of the TPB in the extant sexual behaviour literature. The limited

support obtained for the postulated relationships of the TPB in the current analy-

ses cannot be attributed to measurement artefacts. This is because our longitudi-

nal measurement model and the factorial invariance model both ﬁtted the data

well. Again, the latent variables of the TPB exhibited adequate autoregressive

stationarity over time. From an empirical perspective, it seems plausible that the

current results contrast with most previous ﬁndings reported in the TPB test liter-

ature because of variations in study design, measurement, and in analytic strat-

egy. Strikingly, Reinecke et al. (1996) evaluated their longitudinal research

ﬁndings by concluding that “had we relied solely on the cross-sectional data

available at the end of the 12-month period, we would have been led to the con-

clusion that the theory of planned behaviour accounts for a substantial amount of

variance in intention to use condoms with new sex partners, as well as in actual

condom use”(p. 765).

Limitations and Future Research

This study used self-report measures and a convenience sampling technique. It

also used global constructs of the TPB to investigate adolescent condom use.

INTENTIONS AND BEHAVIOUR IN CONDOM USE 19

The TPB test literature may beneﬁt from future research in Ghana that employs

differentiated/decomposed constructs of the TPB. Moreover, we did not carry

out an elicitation study to identify condom use relevant beliefs as recommended

by Ajzen (1991). Instead, general measures of adolescent condom use, obtained

from previous research, were used to assess participants’intentions to use con-

doms and condom use behaviour. It is possible that the use of the general mea-

sures, global constructs, and the non-use of an elicitation study accounted for the

fairly low internal consistency reliabilities, for example, attitudes (a=.64, at

Time 1) and subjective norms (a=.62, at Time 1), reported in this study. How-

ever, we note that the latent variable structural modelling procedures used in this

study took into account the measurement error associated with items on each

scale as well as the scale as a whole by way of the disturbance terms and mod-

elled them. Thus, the fairly low reliability coefﬁcients reported were not

expected to bias the parameter estimates.

In addition, methodologists have recently suggested that longitudinal data

may provide opportunities for disaggregating between-person and within-person

effects (see, for example, Curran & Bauer, 2011; Hoffman & Stawski, 2009).

For theoretical and methodological reasons, the current research did not use

advanced modelling statistical techniques such as multilevel models or latent

growth statistical models to elucidate within-person associations (stability) from

between-person effects (change). We note that within-person effects may not

occur in a vacuum. Thus, there appears to be the need for future research to over-

come this limitation by considering the capabilities offered by multilevel and

growth statistical models to clarify the dependencies associated with between-

person and within-person effects in longitudinal autoregressive models. Further,

although we believe that the participant attrition rate in this study emanated from

a design issue, we suspect that the participant-generated alpha-numeric string

identiﬁer used to match the data across time in this study may have contributed

to the reduction in the ﬁnal sample size used in the analysis. We note that the

errors associated with some of the identiﬁers on completed surveys may have

emanated from the rather long string identiﬁers required of participants (i.e.

between 10 and 15 strings). Future research should consider the possibility of

using shorter string identiﬁers to help reduce participant errors.

Another possible limitation to note is the three months’time lag between mea-

surement occasions used in this study. Reviews of the TPB literature indicated

that time lags of three months were preferable and optimal (Noar, Cole, &

Carlyle, 2006; see also Dormann & Grifﬁn, 2015). However, given that the pre-

sent participants were in-school youths, it would seem that a longer time lag

may have been more appropriate to provide enough time for the effect of one

variable on another to manifest (e.g. effect of intention on behaviour). Given the

limited support obtained in the present study for the TPB components, future

research may focus on clarifying the intention–behaviour correspondence longi-

tudinally, using appropriate study designs and robust statistical techniques. This

20 TEYE-KWADJO ET AL.

research may further our current understanding of the psychosocial constraints

preventing adolescents from acting on their intentions. Arguably, an immediate

next step may be to replicate and validate the current ﬁndings in a sample of

public high school students drawn from a cluster of schools in Ghana.

ACKNOWLEDGEMENT

This work was supported by the Graduate School of Arts and Social Sciences at

Stellenbosch University in the form of a doctoral scholarship to the correspond-

ing author. The research reported on here emerged from the doctoral dissertation.

Thus, opinions expressed and conclusions arrived at are those of the authors and

are not to be attributed to the Graduate School or Stellenbosch University.

REFERENCES

Abdul-Rahman, L., Marrone, G., & Johansson, A. (2011). Trends in contraceptive use among

female adolescents in Ghana. African Journal of Reproductive Health,15(2), 45–55.

Adu-Mireku, S. (2003). Family communication about HIV/AIDS and sexual behaviour

among senior secondary school students in Accra, Ghana. African Health Sciences,3

(1), 7–14.

Ajzen, I. (1991). The theory of planned behavior. Organizational Behavior and Human

Decision Processes,50, 179–211.

Albarracin, D., Johnson, B.T., Fishbein, M., & Muellerleile, P.A. (2001). Theories of rea-

soned action and planned behavior as models of condom use: A meta-analysis.

Psychological Bulletin,127, 142–161. doi:10.1037//0033-2909.127.1.142.

Anderson, J.C., & Gerbing, D.W. (1988). Structural equation modeling in practice: A

review and recommended two-step approach. Psychological Bulletin,103(3), 411–423.

Basen-Engquist, K., Masse, L.C., Coyle, K., Kirby, D., Parcel, G.S., Banspach, S., et al.

(1999). Validity of scales measuring the psychosocial determinants of HIV/STD-

related risk behavior in adolescents. Health Education Research,14,25–38.

Beadnell, B., Wilsdon, A., Wells, E.A., Morison, D.M., Gillmore, M.R., & Hoppe, M.

(2007). Intrapersonal and interpersonal factors inﬂuencing adolescents’decisions about

having sex: A test of sufﬁciency of the theory of planned behavior. Journal of Applied

Social Psychology,37(12), 2840–2876.

Bryan, A., Kagee, A., & Broaddus, M.R. (2006). Condom use among South African ado-

lescents: Developing and testing theoretical models of intentions and behavior. AIDS

Behaviour,10, 387–397. doi:10.1007/s10461-006-9087-5.

Bryant, F.B., & Satorra, A. (2012). Principles and practice of scaled difference chi-square

testing. Structural Equation Modeling,19, 372–398. doi:10.1080/10705511.2012.

687671.

Carmack, C.C., & Lewis-Moss, R.K. (2009). Examining the theory of planned behavior

applied to condom use: The effect-indicator vs. causal-indicator models. Journal of

Primary Prevention,30, 659–676. doi:10.1007/s10935-009-0199-3.

Carvajal, S.C., Estrada, A.L., & Estrada, B. (2005). Longitudinal prediction of unpro-

tected sex in predominantly Latino male IDUs. Journal of Applied Behavioural

Research,10(3), 133–148. doi:10.1111/j.1751-9861.2005.tb00008.x.

INTENTIONS AND BEHAVIOUR IN CONDOM USE 21

Cohen, J.W. (1988). Statistical power analysis for the behavioral sciences (2nd edn.).

Hillsdale, NJ: Lawrence Erlbaum Associates.

Cole, D.A., & Maxwell, S.E. (2003). Testing mediational models with longitudinal data:

Questions and tips in the use of structural equation modelling. Journal of Abnormal

Psychology,112, 558–577. doi:10.1037/0021-843X.112.4.558.

Curran, P.J., & Bauer, D.J. (2011). The disaggregation of within-person and between-per-

son effects in longitudinal models of change. Annual Review Psychology,62, 583–

619. doi:10.1146/annurev.psych.093008.100356.

Darteh, E.K.M., Doku, D.T., & Esia-Donkoh, K. (2014). Reproductive health decision

making among Ghanaian women. Reproductive Health,11, 23. Retrieved from http://

www.reproductive-health-journal.com/content/11/1/23

DeHart, D.D., & Birkimer, J.C. (1997). Trying to practice safer sex: Development of the

sexual risks scale. Journal of Sex Research,34,11–25. doi:10.1080/

00224499709551860.

Dormann, C., & Grifﬁn, M.A. (2015). Optimal time lags in panel studies. Psychological

Methods,20(4), 489–505. Advance online publication. http://dx.doi.org/10.1037/me

t0000041

Eaton, L., Flisher, A.J., & Aarø, L.E. (2003). Unsafe sexual behaviour in South African

youth. Social Science and Medicine,56, 149–165.

Enders, C.K. (2001). The impact of nonnormality on full information maximum-likeli-

hood estimation for structural equation models with missing data. Psychological

Methods,6, 352–370. doi:10.1037//I082-989X.6.4.352.

Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009). Statistical power analyses

using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research

Methods,41, 1149–1160. doi:10.3758/BRM.41.4.1149.

Fishbein, M., Hennessy, M., Yzer, M., & Douglas, J. (2003). Can we explain why some

people do and some do not act on their intentions? Psychology, Health & Medicine,8,

3–18. doi:10.1080/1354850021000059223.

Ghana Aids Commission (2013). National HIV and AIDS, STI policy. Retrieved from

http://www.healthpolicyproject.com/pubs/153_Policyfinal.pdf

Ghana Aids Commission (2014). Country AIDS response progress report –Ghana.

Retrieved from http://www.unaids.org/sites/default/files/country/documents//GHA_na

rrative_report_2014.pdf

Ghana Health Service (2009). Ghana demographic and health survey (2008): Key

ﬁndings. Calverton, MD: GSS, GHS, and ICF Macro. Retrieved from http://www.

measuredhs.com/pubs/pdf/FR221/FR221.pdf

Ghana News Agency (2014, 16 June). Director calls on educational heads to

maintain pregnant girls in school. Retrieved from http://www.ghananewsagency.

org/social/director-calls-on-educational-heads-to-maintain-pregnant-girls-in-school-

76124

Hoffman, L., & Stawski, R.S. (2009). Persons as contexts: Evaluating between-person

and within-person effects in longitudinal analysis. Research in Human Development,

6,97–120.

Holland, K.J., & French, S.E. (2012). Condom negotiation strategy use and effectiveness

among college students. Journal of Sex Research,49, 443–453. doi:10.1080/

00224499.2011.568128.

22 TEYE-KWADJO ET AL.

Jackson, D.L. (2003). Revisiting sample size and number of parameter estimates: Some

support for the N:q hypothesis. Structural Equation Modeling: A Multidisciplinary

Journal,10, 128–141. doi:1.1207/s15328007ESM1001_6.

Jemmott, J.B. III, Heeren, G.A., Ngwane, Z., Hewitt, N., Jemmott, L.S., Shell, R., et al.

(2007). Theory of planned behaviour predictors of intention to use condoms among

Xhosa adolescents in South Africa. AIDS Care,19, 677–684. doi:10.1080/

09540120601084308.

Kenny, D.A. (1979). Correlation and causality. New York: Wiley.

Little, T.D., Preacher, K.J., Selig, J.P., & Card, N.A. (2007). New developments in latent

variable panel analyses of longitudinal data. International Journal of Behavioral

Development,31, 357–365. doi:10.1177/01650254070777757.

Little, T.D., Rhemtulla, M., Gibson, K., & Schoemann, A.M. (2013). Why the items ver-

sus parcels controversy needn’t be one. Psychological Methods,18, 285–300. doi:10.

1037/a0033266.

Maxwell, S.E., & Cole, D.A. (2007). Bias in cross-sectional analyses of longitudinal

mediation. Psychological Methods,12,23–44. doi:10.1037/1082-989X.12.1.23.

Maxwell, S.E., Cole, D.A., & Mitchell, M.A. (2011). Bias in cross-sectional analyses of

longitudinal mediation: Partial and complete mediation under an autoregressive model.

Multivariate Behavioral Research,46,816–841. doi:10.1080/00273171.2011.606716.

Michie, S., Johnston, M., Abraham, C., Lawton, R., Parker, D., & Walker, A. (2005).

Making psychological theory useful for implementing evidence based practice: A con-

sensus approach. Quality and Safety in Health Care,14,26–33. doi:10.1136/qshc.

2004.011155.

Michie, S., Johnston, M., Francis, J., Hardeman, W., & Eccles, M. (2008). From theory to

intervention: Mapping theoretically derived behavioural determinants to behaviour

change techniques. Applied Psychology: An International Review,57, 660–680.

doi:10.1111/j.1464-0597.2008.00341.x.

Molla, M.,



Astrøm, A.N., & Brehane, Y. (2007). Applicability of the theory of planned

behavior to intended and self-reported condom use in a rural Ethiopian population.

AIDS Care,19, 425–431. doi:10.1080/09540120600722692.

Montanaro, E.A., & Bryan, A.D. (2014). Comparing theory-based condom interventions:

Health belief model versus theory of planned behavior. Health Psychology,33, 1251–

1260. doi:10.1037/a0033969.

Morhe, E.S.K., Tagbor, H.K., Ankobea, F.K., & Danso, K.A. (2012). Reproductive expe-

riences of teenagers in the Ejisu-Juabeng district of Ghana. International Journal of

Gynecology and Obstetrics,118, 137–140. doi:10.1016/j.ijgo.2012.03.035.

Morrison, D.M., Baker, S.A., & Gillmore, M.R. (1998). Condom use among high-risk

heterosexual teens: A longitudinal analysis using the theory of reasoned action.

Psychology & Health,13, 207–222. doi:10.1080/08870449808406747.

Noar, S.M., Cole, C., & Carlyle, K. (2006). Condom use measurement in 56 studies of

sexual risk behavior: Review and recommendations. Archives of Sexual Behavior,35,

327–345.

Odonkor, S.T., Nonvignon, J., Adu, J., Okyere, M., & Mahami, T. (2012). Sexually trans-

mitted diseases (STDs) among adolescents in second cycle institutions in Accra,

Ghana: Trends in sexual behaviours. International Journal of Recent Trends in

Science and Technology,3(1), 9–17.

INTENTIONS AND BEHAVIOUR IN CONDOM USE 23

Ohene, S.-A., & Akoto, I.O. (2008). Factors associated with sexually transmitted infec-

tions among young Ghanaian women. Ghana Medical Journal,42(3), 96–100.

Paul-Ebhohimhen, V.A., Poobalan, A., & van Teijlingen, E.R. (2008). A systematic

review of school-based health interventions to prevent STI/HIV in sub-Saharan Africa.

BMC Public Health,8, 4. doi:10.1186/1471-2458-8-4.

Reichardt, C.S. (2011). Commentary: Are three waves of data sufﬁcient for assessing

mediation? Multivariate Behavioral Research,46, 842–851. doi:10.1080/00273171.

2011.606740.

Reinecke, J., Schmidt, P., & Ajzen, I. (1996). Application of the theory of planned beha-

vior to adolescents’condom use: A panel study. Journal of Applied Social

Psychology,26, 749–772. doi:10.1111/j.1559-1816.1996.tb01128.x.

Sallar, A.M. (2008). Sexual behavior and attitude toward condoms among unmarried in-

school and out-of-school adolescents in a high-HIV prevalence region in Ghana.

International Quarterly of Community Health Education,29, 161–181. doi:10.2190/

IQ.29.2.e.

Schaalma, H., Aarø, L.E., Flisher, A.J., Mathews, C., Kaaya, S., Onya, H., et al. (2009).

Correlates of intention to use condoms among sub-Saharan African youth: The appli-

cability of the theory of planned behaviour. Scandinavian Journal of Public Health,

37,87–91. doi:10.1177/1403494808090632.

Selig, J.P., & Preacher, K.J. (2009). Mediation models for longitudinal data in develop-

ment research. Research in Human Development,6, 144–164. doi:10.1080/

15427600902911247.

Shrout, P.E. (2011). Mediation analysis, causal process, and cross-sectional data [Peer

commentary on the paper “Bias in cross-sectional analyses of longitudinal mediation:

Partial and complete mediation under an autoregressive model”, S.E. Maxwell, D.A.

Cole, & M.A. Mitchell]. Multivariate Behavioral Research,46, 852–860. doi:10.1080/

00273171.2011.606718.

Sniehotta, F. (2009). An experimental test of the theory of planned behavior. Applied

Psychology: Health and Well-Being,1, 257–270. doi:10.1111/j.1758-0854.2009.

01013.x.

Sniehotta, F.F., Presseau, J., & Ara

ujo-Soares, V. (2014). Time to retire the theory of

planned behaviour. Health Psychology Review,8,1–7. doi:10.1080/17437199.2013.

869710.

Swart, H., Hewstone, M., Christ, O., & Voci, A. (2011). Affective mediators of intergroup

contact: A three-wave longitudinal study in South Africa. Journal of Personality and

Social Psychology,101, 1221–1238. doi:10.1037/a0024450.

Tenkorang, E.Y. (2013). Myths and misconceptions about HIV transmission in Ghana:

What are the drivers? Culture, Health & Sexuality,15, 296–310. doi:10.1080/

13691058.2012.752107.

Tenkorang, E.Y., Gyimah, S.O., Maticka-Tyndale, E., & Adjei, J. (2011). Superstition,

witchcraft and HIV prevention in sub-Saharan Africa: The case of Ghana. Culture,

Health & Sexuality,13, 1001–1014. doi:10.1080/13691058.2011.592218.

Walsh, J.L., Senn, T.E., Scott-Sheldon, L.A.J., Vanable, P.A., & Carey, M.P. (2011). Pre-

dicting condom use using the information-motivation-behavioral skills (IMB) model:

A multivariate latent growth curve analysis. Annals of Behavioral Medicine,42, 235–

244.

24 TEYE-KWADJO ET AL.

Weller, S.C. (1993). A meta-analysis of condom effectiveness in reducing sexually trans-

mitted HIV. Social Science & Medicine,36, 1635–1644. doi:10.1016/0277-9536(93)

90352-5.

Weller, S.C., & Davis-Beaty, K. (2002). Condom effectiveness in reducing heterosexual

HIV transmission [Review Art. No: CD003255]. Cochrane Database of Systematic

Reviews. doi:10.1002/14651858.CD003255.

Weinstein, N.D. (2006). Misleading tests of health behavior theories. Annals of

Behavioral Medicine,33,1–10.

Weinstein, N.D., & Rothman, A.J. (2005). Commentary: Revitalising research on health

behavior theories. Health Education Research,20, 294–297. doi:10.1093/her/cyg125.

West, S.G. (2011). Editorial: Introduction to the special section on causal inference in

cross-sectional and longitudinal mediational models. Multivariate Behavioral

Research,46, 812–815. doi:10.1080/00273171.2011.606710.

West, S.G., Finch, J.F., & Curran, P.J. (1995). Structural equation models with non-nor-

mal variables: Problems and remedies. In R. Hoyle (Ed.), Structural

equation modeling: Concepts, issues, and applications (pp. 56–75). Thousand Oaks,

CA: Sage.

Widaman, K.F., Ferrer, E., & Conger, R.D. (2010). Factorial invariance within longitudi-

nal structural equation models: Measuring the same construct across time. Child

Development Perspectives,4,10–18. doi:10.1111/j.1750-8606.2009.00110.x.

INTENTIONS AND BEHAVIOUR IN CONDOM USE 25

Development and Evidence of the Validity of the Condom Use Attitudes Scale for Youth and Adults in a Chilean Context

Article

Full-text available

Dec 2021

Condom use is the most effective preventive behavior against HIV transmission, and its inadequate use is a public health problem that occurs mostly among youth and young adults. Although there are scales that measure condom use, those that exist correspond to English-speaking developments or do not have psychometric evidence to support them, so it is possible that the available adaptations of instruments do not adequately reflect the phenomenon in the Chilean population. Thus, the study aims to develop a scale to assess attitudes toward condom use in Chilean youth and young adults. Initially, a sample of students between 18 and 39 years (n = 520) was used for debugging the instrument. Then, a second sample was taken from the general population aged 18 to 40 (n = 992) to confirm the factor structure of the proposed model. The final scale has 10 items and 3 attitudinal dimensions (affective, cognitive, and behavioral). The results show that the identified structure provides adequate levels (ω > 0.7) or at least sufficient of reliability (ω > 0.6) and presents evidence of validity, based on the internal structure of the test, through ESEM (CFI = 0.993; TLI = 0.984; RMSEA = 0.056). In addition, evidence of validity was obtained based on the relationship with other variables and strong invariance between the scores of men and women. It is concluded that the scale developed has adequate psychometric properties to assess, in brief form, condom use attitudes in equal samples for research and screening purposes.

Perceived norms, personal agency, and postpartum family planning intentions among first-time mothers age 15-24 years in Kinshasa: A cross-sectional analysis

Article

Full-text available

Jul 2021
PLOS ONE

Unintended pregnancy is an important global health problem and frequently occurs during the immediate postpartum period. However, few studies have examined postpartum family planning (PPFP) intentions among adolescent girls and young women. This study assessed whether perceived norms and personal agency predicted PPFP intentions among first-time mothers age 15-24 in Kinshasa, the Democratic Republic of Congo. Data were derived from the 2018 Momentum Project baseline survey. Analysis was based on 2,418 nulliparous pregnant women age 15-24 who were approximately six months pregnant with their first child in six health zones of Kinshasa. Overall PPFP intentions were low and ten to thirteen percent of women stated they were very likely to discuss PPFP next month with (a) their husband/male partner and (b) a health worker, and to (c) obtain and (d) use a contraceptive method during the first six weeks following childbirth. The results of multivariable linear regression models indicated that the PPFP intention index was predicted by description norms, perceptions of the larger community's approval of PPFP, normative expectations, perceived behavioral control, self-efficacy, and autonomy. Rejection of family planning myths and misconceptions was also a significant predictor. Interaction terms suggested that the association of normative expectations with PPFP intentions varied across ethnic groups and that the positive association of injunctive norms with PPFP intentions was significantly increased when the larger community was perceived to disapprove of PPFP use. Normative expectations and PPFP-related self-efficacy accounted for two-thirds of the variance in PPFP intentions. The results suggested that understanding different normative influences may be important to motivate women to use contraception in the immediate postpartum period. In addition to addressing institutional, individual, and social determinants of PPFP, programs should consider integrating norm-based and empowerment strategies.

Problem Gambling and Suicidal Behaviours in Young Adult men in Ghana: A Moderated Mediation Model of Psychological Distress and Social Support

Article

Full-text available

Apr 2024
J GAMBL STUD

Problem gambling has been recognised as an important public health issue because of its harmful consequences at the personal, interpersonal, and societal levels. There is a proliferation of gambling activities in Ghana owing to increased access to the internet, soaring smartphone penetration, and an influx of sports betting companies. Yet, very little research has addressed the harm associated with problem gambling in Ghana. This study assessed if the effect of problem gambling severity on suicidal behaviours (ideation, planning, and attempts) through psychological distress is contingent on the level of perceived social support. A total of 350 young adult men in the Greater Accra Region of Ghana provided the data for the current analysis. The data were analysed using a moderated mediation approach. The results showed that psychological distress is an important intervening mechanism through which problem gambling severity influences suicidality. In addition, the positive association between psychological distress and suicidal behaviours was found to be conditional on the levels of perceived social support. In other words, our results showed that problem gambling may first trigger episodes of psychological distress, which, in turn, can lead to suicidality among problem gamblers who report low to moderate levels of perceived social support. Our findings suggest that young adult problem gamblers require higher levels of social support from family, friends, and significant others to prevent those experiencing psychological distress from engaging in suicidal behaviours.

Salient beliefs and intention to use pre-exposure prophylaxis among pregnant and breastfeeding women in Zambia: Application of the Theory of Planned Behaviour

Article

Full-text available

Jan 2023

Primary HIV prevention is a priority for pregnant and breastfeeding women in sub-Saharan Africa; however, such services should be designed to optimise uptake and continuation. Between September and December 2021, we enrolled 389 women who were not living with HIV into a cross-sectional study from antenatal/postnatal settings at Chipata Level 1 Hospital. We used the Theory of Planned Behaviour to study the relationship between salient beliefs and intention to use pre-exposure prophylaxis (PrEP) among eligible pregnant and breastfeeding women. On a seven-point scale, participants had positive attitudes towards PrEP (mean = 6.65, SD = 0.71), anticipated approval of PrEP use from significant others (mean = 6.09, SD = 1.51), felt confident that they could take PrEP if they desired (mean = 6.52, SD = 1.09) and had favourable intentions to use PrEP (mean = 6.01, SD = 1.36). Attitude, subjective norms, and perceived behavioural control significantly predicted intention to use PrEP respectively (β = 0.24; β = 0.55; β = 0.22, all p < 0.01). Social cognitive interventions are needed to promote social norms supportive of PrEP use during pregnancy and breastfeeding.

Theory of Planned Behavior in predicting condom use in Latin-American Women

Article

Full-text available

Apr 2022

Introducción: La Teoría de la Conducta Planeada (TCP) se utiliza para predecir el uso del condón. El presente estudio evaluó la TCP en un contexto cultural diferente al cual fue propuesto. Objetivo: Evaluar las relaciones causales entre las variables que propone la TCP para determinar si éstas predicen el uso consistente del condón en dos grupos de mujeres latinas e identificar las semejanzas inter grupales en ambos colectivos y las discrepancias con lo que propone el modelo para la población norteamericana. Método: Se evaluaron en dos muestras de mujeres latinas (mexicanas y argentinas) un conjunto de variables como: la consistencia del uso del condón, la intención de usarlo en el próximo encuentro sexual, la norma subjetiva de amigos y padres, las actitudes hacia el preservativo y el control conductual percibido para usarlo. Se utilizó un análisis de regresión lineal. Resultados: Se encontró que la intención del uso del condón es el mejor predictor del uso del condón. La norma subjetiva fue más importante que las actitudes; aunque esta última variable, no se configuró como un predictor en las mujeres argentinas. Conclusiones: Las asociaciones entre las variables de la TCP y los valores predictivos de las variables, pueden cambiar de acuerdo con el grupo cultural evaluado; el control conductual percibido, en ambas muestras, no resultó ser un predictor directo de la conducta para usar condón, pero si para la intención de utilizarlo. Además, se discute la relevancia de las variables que propone la TCP en relación al contexto cultural de los latinos y el uso crítico del modelo en poblaciones diferentes a la norteamericana.

Explicating factors that explain condom use intention among in-school adolescents in Botswana: a structural equation modelling approach

Article

Full-text available

Dec 2021

Knowledge with respect to adolescents’ intentions to engage in protective sexual behaviours is still deficient in numerous countries around the world, particularly in Sub-Saharan Africa (SSA) where HIV prevalence is the highest. Increasing cross-sectional research suggests that the theory of planned behaviour (TPB) is robust in predicting condom use intentions. This study used secondary, baseline data from a study involving 794 in-school adolescents. A structural equation model was applied to the data. Latent variables were used to validate the applicability of the TPB in HIV/AIDS prevention among adolescents in the Botswana context. The predictors of interest were all derived from the TPB. The results revealed that the TPB predictors, apart from affective and instrumental attitude, were predictive of condom use intention among Batswana adolescents. The independent variables explained 57% of the variance in the model. These results suggest that the TPB is recommended as a framework to establish the predictors of condom use intention among Batswana in-school adolescents. Policy makers working on developing HIV education programmes or interventions targeted at adolescents should improve the intention to use condoms via promotion of positive instrumental attitudes, subjective norms and perceived behavioural control beliefs of condom use.

"PrEP protects us": Behavioural, normative, and control beliefs influencing pre-exposure prophylaxis uptake among pregnant and breastfeeding women in Zambia

Article

Full-text available

Apr 2023

Background: Although pre-exposure prophylaxis (PrEP) is recommended for pregnant and breastfeeding women at elevated HIV risk, uptake has been low in Zambia. Methods: In in-depth interviews, we explored beliefs about PrEP among 24 HIV-negative pregnant and breastfeeding Zambian women. Thematic analysis was used to identify behavioural, normative and control beliefs likely to influence PrEP uptake. Results: Most women viewed PrEP as a good method of protecting themselves and their babies from HIV infection. Partners were cited as key referents in decision making about PrEP use. Many women felt that PrEP use was not entirely in their control. Most reported that they would not use PrEP if their partners did not approve. Health care providers with negative attitudes, long distance to clinics, and extended waiting times were cited as barriers to PrEP uptake. Conclusion: HIV-negative pregnant and breastfeeding women had a positive attitude towards PrEP but barriers to uptake are multifaceted.

Understanding User Perception and Intention to Use Smart Homes for Energy Efficiency: A Survey

Preprint

Full-text available

Dec 2022

The positive impact of Smart Homes on energy efficiency is heavily dependent on how consumers use the system after adoption. While the technical aspects of Smart Home systems and their potential to reduce energy usage is a focus of various studies, there is a limited consideration of behavioral psychology while designing systems for energy management. To investigate users' perception and intention to use Smart Homes to support energy efficiency, we design a research model by combining a theory of planned behavior and the norm activation model. We design a questionnaire and conduct a survey targeting current smart home users (over 350 responses). To analyze the survey results, we extend the partial least squares structural equation modeling (PLS-SEM) by a random forest algorithm. The findings suggest that personal norms have the strongest influence on behavioral intention to use Smart Homes for energy efficiency, followed by the ascription of responsibility. Furthermore, the results support the effects of attitudes, subjective norms, awareness of consequences, as well as the moderating effect of past behavior on the relationship between personal norms and behavioral intentions.

Predictors of intention to use modern contraceptives among female senior secondary school students in the Kpando Municipality, Ghana

Article

Full-text available

Sep 2021
Afr Health Sci

Background: Improving the reproductive health of young women in developing countries requires access to safe and effective methods of fertility control. Volta Region records one of the highest prevalence of teenage pregnancy and adolescents aged 15-19 years are the least acceptors of contraceptives in the Region. Guided by the Theory of Planned Behaviour, this study determined predictors of intention to use modern contraceptives among female Senior Secondary School students in the Kpando Municipality, Ghana. Method: A cross-sectional design was adopted, collecting data among a multistage sample of 270 participants, using a pretested self-administered questionnaire and analysing them using Stata software Version 16 at the 0.05 level of significance. Results: The mean age of the respondents was 16.78 ± 1.31. About 40.2% of the sexually experienced participants used a modern contraceptive during ther first sexual encounter. However, the majority (69.3%) had the intention to use modern contraceptives. Regarding subjective norms, the majority of the significant others (63.0%) were not supportive of modern contraceptive use and more than half of the respondents (59.3%) had a positive attitude towards modern contraceptive use, while 54.1% perceived that they had control over modern contraceptive use. The majority (69.3%) had the intention to use modern contraceptives. Perceived behavioural control was the only significant predictor of the intention to use modern contraceptives [AOR= 9.80 (C.I: 5.11, 18.77); p< 0.001]. Conclusion: Interventions to increase the perception of control over contraceptive use is of the essence. This will help increase their intention to use modern contraceptives.

Pornografija, stav o uporabi kondoma i uporaba kondoma kod riječkih adolescenata

Article

Jan 2021

Jednostavnost pristupa internetskoj pornografiji izazvala je zabrinutost odgojnih stručnjaka i opće javnosti zbog negativnih učinaka pornografije na psihosocijalno i seksualno sazrijevanje adolescenata. Pretpostavljena povezanost između uporabe pornografije i rizičnog seksualnog ponašanja, posebice nezaštićenog spolnog odnosa, često se nalazi u središtu pozornosti. Prema dosadašnjim istraživanjima, stav o uporabi kondoma važan je prediktor stvarne uporabe kondoma. Međutim, razmjerno je malo studija analiziralo povezanost između uporabe pornografije, stava o uporabi kondoma i uporabe kondoma među adolescentima. Konceptualni okvir ove studije čini nedavno razvijeni model akvizicije, aktivacije i aplikacije (3AM). Cilj je ove studije analiza longitudinalne povezanosti između učestalosti uporabe pornografije te stava o uporabi kondoma i uporabe kondoma pri posljednjem spolnom odnosu. Podaci su prikupljeni metodom ankete u dva vala među učenicima/ama 14 riječkih srednjih škola od 2015. do 2018. godine (N=227; Mdob=17,3; SD=0,5). Kontrolirajući traženje uzbuđenja i permisivne stavove o seksualnosti, provedena je višegrupna ukrižena analiza primjenom strukturalnog modeliranja. Utvrđena je longitudinalna povezanost između učestalosti uporabe pornografije i negativnijeg stava o uporabi kondoma, ali samo kod mladića. Nije utvrđena longitudinalna povezanost između uporabe pornografije i uporabe kondoma. Povrh rasprave o rezultatima, zaključno se razmatraju preporuke za daljnja istraživanja i praktične implikacije. Dobiveni nalazi mogu poslužiti stručnjacima iz područja obrazovanja, medijske pismenost i javnog zdravstva, ali i široj javnosti.

New developments in latent variable panel analyses of longitudinal data

Article

Full-text available

Jan 2007

New developments in latent variable panel analyses of longitudinal data

Article

Full-text available

Jul 2007

We review fundamental issues in one traditional structural equation modeling (SEM) approach to analyzing longitudinal data — cross-lagged panel designs. We then discuss a number of new developments in SEM that are applicable to analyzing panel designs. These issues include setting appropriate scales for latent variables, specifying an appropriate null model, evaluating factorial invariance in an appropriate manner, and examining both direct and indirect (mediated), effects in ways better suited for panel designs. We supplement each topic with discussion intended to enhance conceptual and statistical understanding.

Validity of scales measuring the psychosocial determinants of HIV/STD-related risk behavior in adolescents

Article

Full-text available

Feb 1999

Karen Basen-Engquist

We examined the content, construct and concurrent validity of scales to assess beliefs and self-efficacy related to adolescents' sexual risk behavior. We addressed content validity in the scale development process by drawing on literature and theory, and by pre-testing items with focus groups. We used confirmatory factor analysis of two models, an intercourse involvement model and a condom use model, to assess construct validity. The final intercourse involvement model included three scales: norms about sexual intercourse, attitudes about sexual intercourse and self-efficacy in refusing sex. The final condom use model included five scales: norms about condoms, attitudes about condom use, self-efficacy in communicating about condoms, self-efficacy in buying/using condoms and barriers to condom use. After two alterations to the models, the χ 2 and other indices indicated that the data fit the models well. Supporting the concurrent validity of the scales, high school students who had never had sexual intercourse had more negative attitudes toward sexual intercourse among teenagers, perceived norms toward sexual intercourse among teenagers to be more negative and expressed greater self-efficacy in refusing sex than did those who had experienced sexual intercourse. Consistent condom users had more positive attitudes and 25 norms about condoms, had higher self-efficacy in communicating about and buying/using con-doms, and perceived fewer barriers to condom purchase and use than did inconsistent con-dom users.

Optimal Time Lags in Panel Studies

Article

Full-text available

Aug 2015

Cross-lagged regression coefficients are frequently used to test hypotheses in panel designs. However, these coefficients have particular properties making them difficult to interpret. In particular, cross-lagged regression coefficients may vary, depending on the respective time lags between different sets of measurement occasions. This article introduces the concept of an optimal time lag. Further, it is demonstrated that optimal time lags in panel studies are related to the stabilities of the variables investigated, and that in unidirectional systems, they may be unrelated to the size of possible true effects. The results presented also suggest that optimal time lags for panel designs are usually quite short. Implications are (a) that interpreting cross-lagged regression coefficients requires taking the time lag between measurement occasions into account, and (b) that in much research, far shorter time lags than those frequently found in the literature are justifiable, and we call for more "shortitudinal" studies in the future. (PsycINFO Database Record (c) 2015 APA, all rights reserved).

The Theory of Planned Behavior

Article

Full-text available

Dec 1991

Icek Ajzen

Research dealing with various aspects of* the theory of planned behavior (Ajzen, 1985, 1987) is reviewed, and some unresolved issues are discussed. In broad terms, the theory is found to be well supported by empirical evidence. Intentions to perform behaviors of different kinds can be predicted with high accuracy from attitudes toward the behavior, subjective norms, and perceived behavioral control; and these intentions, together with perceptions of behavioral control, account for considerable variance in actual behavior. Attitudes, subjective norms, and perceived behavioral control are shown to be related to appropriate sets of salient behavioral, normative, and control beliefs about the behavior, but the exact nature of these relations is still uncertain. Expectancy value formulations are found to be only partly successful in dealing with these relations. Optimal rescaling of expectancy and value measures is offered as a means of dealing with measurement limitations. Finally, inclusion of past behavior in the prediction equation is shown to provide a means of testing the theory*s sufficiency, another issue that remains unresolved. The limited available evidence concerning this question shows that the theory is predicting behavior quite well in comparison to the ceiling imposed by behavioral reliability.

Reproductive health decision making among Ghanaian women

Article

Full-text available

Mar 2014
Reprod Health

Women's reproductive health decision-making and choices, including engaging in sexual intercourse and condom use, are essential for good reproductive health. However, issues concerning sexual intercourse and condom use are shrouded in secrecy in many sub-Saharan African countries. This study investigates factors that affect decision making on engaging in sexual intercourse and use of condom among women aged 15-49. A nationally representative sample (N = 3124) data collected in the 2008 Ghana Demographic and Health Survey was used. Multivariate logistic regression was used to study the association between women's economic and socio-demographic characteristics and their decision making on engaging in sexual intercourse and use of condom. One out of five women reported that they could not refuse their partners' request for sexual intercourse while one out of four indicated that they could not demand the use of condoms by their partners. Women aged 35-49 were more likely to make decision on engaging in sexual intercourse (OR = 1.35) compared to those aged 15-24. Furthermore, the higher a woman's education, the more likely that she would make decision regarding condom use. Also, if a woman had primary (OR = 1.37) or secondary (OR = 1.55) education, she is more likely to make decision regarding engaging in sexual intercourse compared to a woman who had no formal education. Compared to women in the Greater Accra region (the capital city region), women in the Western region (OR = 2.10), Central region (OR = 2.35), Brong Ahafo (OR = 1.70), Upper East (OR = 7.71) and Upper West (OR = 3.56) were more likely to make decision regarding the use of condom. Women who were in the richest, rich and middle wealth index categories were more likely to make decision regarding engaging in sexual intercourse as well as condom use compared to the poorest. Interventions and policies geared at empowering women to take charge of their reproductive health should focus particularly on women from less wealthy backgrounds and those with low educational attainments.

Time to Retire the Theory of Planned Behaviour

Article

Full-text available

Nov 2013
Health Psychol Rev

http://www.tandfonline.com/doi/full/10.1080/17437199.2013.869710

Factors Associated with Sexually Transmitted Infections Among Young Ghanaian Women

Article

Full-text available

Oct 2008

To identify factors associated with a history of sexually transmitted infections in Ghanaian women 15 - 24 years. The study was a cross-sectional data analysis of 1280 sexually experienced females from the 2003 Ghana Demographic Health Survey. Using chi square and t-test, those with a history or symptoms of STI were compared with those denying such a history on demographic, individual and partner level variables. Significant variables were entered into logistic regression to identify variables associated with STI. The STI group comprised 12% of the study population. Compared to those denying a history of STI, the STI group was less likely to know where to get condoms (37% vs 23%, p=.001) but more likely to use a condom at their last sexual encounter (27% vs 17%, p=.003). Women in the STI group were significantly less likely to discuss family planning with their partners but more likely to have 2 or more partners in the preceding 12 months. Logistic regression showed that factors associated with STI among sexually active Ghanaian female youth included not knowing where to get condoms and not discussing family planning with partner. Majority of sexually experienced Ghanaian female youth do not know where to get condoms. Lack of knowledge of source of condoms was identified as a risk factor for STI for these women.

Statistical power analysis for the behavioral sciences. Plenum, New York 43. Storch EA, Larson MJ, Muroff J, Caporino N, Geller D, Reid JM et al (2010) Predictors of functional impairment in pediatric obsessive-compulsive disorder.

Article

Jan 1988
J ANXIETY DISORD

J. Cohen

The relative performance of full -information maximum likelihood estimation for missing data in structural equation models

Article

Jan 1999

Craig Enders

A Monte Carlo study was designed to compare the performance of four missing data methods in structural equation models: full information maximum likelihood (FIML), listwise deletion, pairwise deletion, and similar response pattern imputation. The effects of three independent variables were examined (factor loading magnitude, sample size, and missing data rate) on four outcome measures: convergence problems, parameter estimate bias, parameter estimate efficiency, and model goodness-of-fit. The performance of the four estimation methods was assessed in two different simulations. In the first, missing values were missing completely at random (MCAR), while the second simulated missing at random (MAR) data. Results indicated that FIML estimation was superior across all conditions of the design. FIML estimates were unbiased and more efficient than the other methods across both the MCAR and MAR simulations. In addition, FIML yielded the lowest rate of convergence problems and provided near-optimal Type 1 error rates across both simulations. When data were MCAR, pairwise deletion generally yielded satisfactory performance. Parameter estimates were unbiased and convergence rates and efficiency were only slightly worse than FIML. However, inflated Type 1 error rates were observed, particularly when the magnitude of factor loadings was high. However, under MAR pairwise deletion parameter estimates were substantially biased. Both listwise deletion and similar response pattern imputation performed substantially worse than FIML across most conditions of the study. The methods yielded high rates of nonconvergence, and inefficient parameter estimates relative to FIML. These efficiency differences became more pronounced as the missing data rate increased. Under MAR, both methods yielded substantially biased parameter estimates. Although listwise deletion rejection rates were acceptable, those of similar response pattern imputation were substantially inflated. Based on the results of this study, it is recommended that applied researchers discontinue the use of popular ad hoc methods such as listwise and pairwise deletion in favor of FIML.

Predicting the Intention to Use Condoms and Actual Condom Use Behaviour: A Three-Wave Longitudinal Study in Ghana

Abstract and Figures

Recommended publications

Youth Exercise Intention and Past Exercise Behavior

Determinants of condom use among heterosexual young men and women in southeastern Ghana: a mediation...

Does Past Condom Use Moderate the Future Condom Use Intention-Behavior Relationship?Results from a G...

Condom use negotiation among high school adolescents in Ghana: the role of gender

Lag as Moderator Models for Longitudinal Data